Proc. Natl. Acad. Sci. USA
Vol. 94, pp. 6025–6029, June 1997
The NG domain of the prokaryotic signal recognition particle
receptor, FtsY, is fully functional when fused to an unrelated
integral membrane polypeptide
ADRIAN ZELAZNY, ANDREI SELUANOV, ARIE COOPER, AND EITAN BIBI*
Department of Biochemistry, Weizmann Institute of Science, Rehovot 76100, Israel
Communicated by H. Ronald Kaback, University of California, Los Angeles, CA, April 10, 1997 (received for review February 26, 1997)
coli possesses an essential targeting system for integral mem-
brane proteins, similar to the mammalian signal recognition
particle (SRP) machinery. One essential protein in this
system is FtsY, a homologue of the ?-subunit of the mamma-
lian SRP-receptor (SR-?). However, E. coli does not possess a
close homologue of the integral membrane protein SR-?,
which anchors SR-? to the membrane. Moreover, although
FtsY can be found as a peripheral membrane protein, the
majority is found soluble in the cytoplasm. In this study, we
obtained genetic and biochemical evidence that FtsY must be
targeted to the membrane for proper function. We demon-
strate that the essential membrane targeting activity of FtsY
is mediated by a 198-residue-long acidic N-terminal domain.
This domain can be functionally replaced by unrelated inte-
gral membrane polypeptides, thus avoiding the need for
specific FtsY membrane targeting factors. Therefore, the N
terminus of FtsY constitutes an independent domain, which is
required only for the targeting of the C-terminal NG domain
of FtsY to the membrane.
Recent studies have revealed that Escherichia
Targeting of newly synthesized membrane and secretory pro-
teins to the endoplasmic reticulum in mammalian cells is
mediated cotranslationally by the signal recognition particle
(SRP)-machinery (1, 2). In Escherichia coli, the targeting of
posttranslationally (3). Despite this difference, E. coli contains
essential genes encoding Ffh and FtsY with a significant
similarity to proteins of the eukaryotic SRP machinery (4, 5)
and 4.5S RNA which resembles part of the mammalian 7SL
RNA (6). Recent studies have elucidated the need for such a
system in prokaryotes, demonstrating that two components of
the E. coli SRP pathway, FtsY (7) and Ffh (8–10), are utilized
preferentially for biogenesis of polytopic membrane proteins.
One essential protein in this machinery is FtsY, the E. coli
SRP-receptor (SR) which contains a large domain (the NG
domain, see Ref. 11 for details) that is homologous to the NG
domain of the mammalian SR-? protein (see Fig. 1B). By
analogy to SR-?, the ?-subunit of the mammalian SRP-
receptor, it has been suggested that FtsY acts as a peripheral
membrane protein (12, 13), although a large fraction of FtsY
is found soluble in the cytoplasm (13). In mammalian cells,
SR-? is targeted cotranslationally to the membrane (14) and
interacts via a 140-residue-long N-terminal domain with its
integral membrane ?-subunit (refs. 15 and 16; Fig. 1B). In E.
coli, no SR-? homologue has been found and the mechanisms
by which FtsY reaches its destination and binds to the mem-
brane are not known. The N-terminal 198-amino acid se-
quence of FtsY shares no homology with the mammalian
receptor and it is highly charged and enriched with acidic
residues (Fig. 1A). It has been hypothesized that FtsY might
interact directly with membrane phospholipids. However, the
observation that a large fraction of FtsY is soluble (13)
supports the idea that there may be a limited number of
membrane targeting factors or specific binding sites for FtsY.
Here we show that FtsY must be targeted to the membrane
for proper function. The membrane localization function is
mediated by a 198-residue-long N-terminal domain of the
protein. However, the essential targeting and attachment of
FtsY to the membrane can be mediated also by unrelated
integral membrane polypeptides fused to the N terminus of the
NG domain of FtsY.
Materials. Spectinomycin, ampicillin, arabinose, 3-?-
indoleacrylic acid, and phenylmethylsulfonyl fluoride were
purchased from Sigma. Restriction enzymes were obtained
from New England Biolabs, and modifying enzymes were from
Boehringer Mannheim. Antibodies to FtsY were a gift from J.
Luirink (Biocentrum Amsterdam). Goat anti-rabbit antibodies
conjugated to horseradish peroxidase were obtained from
Jackson ImmunoResearch. Antibodies to the C-terminal tail
of LacY were kindly provided by H. R. Kaback (University of
California, Los Angeles). Prestained protein molecular weight
molecular weight markers were from GIBCO?BRL. Gene-
Clean and Mermaid glassmilk DNA purification kits were
obtained from Bio 101, and Wizard Mini Prep kits were from
Promega. All other materials were reagent grade and obtained
from commercial sources.
Bacterial Strains and Plasmids. E. coli UT5600[ompT?]
obtained from the E. coli Genetic Stock Center at Yale
University (strain 7092) was used for mutagenesis and expres-
sion studies. E. coli N4156::pAra14-FtsY? (13) was obtained
from J. Luirink. Plasmid pT7-5(lacY) encoding lactose per-
mease under the lac promoter was described elsewhere (17)
and served for cloning of ftsY and DNA manipulations.
Plasmid pCL1921 (18) (alone or containing various ftsY mu-
tants) was used to transform E. coli N4156::pAra14-FtsY?
(polA?) for growth and expression experiments. Plasmid
pATH2 (19) was used as a source for the gene encoding the
N-terminal 326-amino acid residues of TrpE which was fused
to the 5? end of the gene encoding a deletion mutant of FtsY
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
© 1997 by The National Academy of Sciences 0027-8424?97?946025-5$2.00?0
Abbreviations: SRP, signal recognition particle; SR, SRP-receptor;
LB, Luria–Bertani medium; IPTG, isopropyl ?-D-thiogalactoside.
*To whom reprint requests should be addressed at: Incumbent of Dr.
Department of Biochemistry, Weizmann Institute of Science, Reho-
vot 76100 Israel. e-mail: BCBIBI@WEIZMANN.WEIZMANN.
Construction of ftsY Mutants and lacY–ftsY Gene Fusions.
The ftsY gene was amplified by PCR from the E. coli chro-
mosome using the following sense and antisense deoxyoligo-
nucleotides containing sites for enzymes BamHI and HindIII
(underlined), respectively: 5?-TATATGGATCCATGGC-
GAAAGAAAAAAAACGTGGC and 5?-TATATAAGCT-
The amplified DNA was cloned instead of the lacY gene into
pT7-5(lacY) under the lac promoter?operator producing plas-
mid pTftsY. The ftsY mutants and the lacY–ftsY hybrids (see
Fig. 1 C and D) were constructed as follows: ?198 plasmid
pTftsY was digested by NcoI and BssHII, treated by Klenow
fragment, and the large DNA fragment was ligated on itself.
The resulting plasmid was linearized by BssHII, treated by
digested by NcoI and AgeI, treated by Klenow fragment, and
the large DNA fragment was ligated on itself. ?(30–92)
plasmid pTftsY was digested by AgeI and the large fragment
was ligated on itself. N6-?198 plasmid pTftsY was digested by
BssHII, treated with Klenow fragment, and then digested by
PvuI to create a 1.5-kb DNA fragment. This fragment was
ligated to the 2.5-kb DNA fragment released from pT7-5lacY
by XhoI, Klenow treatment, and then PvuI. N4-?92 plasmid
pTftsY was digested by AgeI, treated by Klenow fragment, and
then digested by PvuI to create a 1.8-kb DNA fragment. This
fragment was ligated to the 2.3-kb DNA fragment released
from pT7-5lacY by XhoI, Klenow treatment, and then PvuI.
Plasmid pCLtrpE-?198 was made by ligating the 5.5-kb DNA
fragment of pCLftsY (treated with EcoRI, BssHII, and Klenow
fragment) with the 1.4 kb of pATH2 (treated with BamHI,
PvuII, and Klenow fragment). All the gene constructs were
analyzed by restriction enzymes and DNA sequencing.
Growth of Cells, Expression, and Immunoblotting. E. coli
constructs was grown overnight at 37°C in Luria–Bertani
medium (LB) supplemented with ampicillin (100 ?g?ml).
Cultures were then diluted 1:100, grown to 0.6 OD600unit, and
induced by 0.5 mM isopropyl ?-D-thiogalactoside (IPTG) for
2 hr. Cells were harvested, washed in buffer A (7% sucrose?
100 mM Tris?HCl, pH 8?50 mM NaCl?1 mM EDTA?1 mM
phenylmethylsulfonyl fluoride) and resuspended in the same
buffer. Cell suspensions (500 ?l of 10 OD420 units) were
sonicated, and cell debris were removed by centrifugation (2
min at 13,000 rpm). E. coli (N4156::pAra14-FtsY?) cells har-
boring plasmid pCL1921 or its derivatives as indicated were
grown in LB broth with ampicillin (100 ?g?ml), spectinomycin
(100 ?g?ml) and arabinose (0.2%). The overnight cultures
were washed once in LB broth and diluted (1:2000) in LB
containing the same antibiotics with or without arabinose as
indicated. Cell density (OD600) was measured every hour for
the construction of growth curves. For expression studies, cells
LacY, and LacY–FtsY hybrids. (A) The distribution of charged
residues (line a for Glu or Asp, and line b for Arg or Lys) along the
primary sequence of SR-? and FtsY. The figure was adopted from the
output of the program DNA STRIDER. The shaded region highlights the
acidic N-terminal domain of FtsY. (B) Schematic alignment of SR-?
and FtsY. Stippled boxes represent the homologous C-terminal NG
domains. The shaded (box) N-terminal 140-residue-long domain of
SR-? is implicated in membrane attachment (16). The shaded box in
FtsY represent the highly acidic N-terminal domain. Restriction
enzymes used for mutagenesis are shown under the FtsY box. (C)
Schematic presentation of the N-terminal FtsY truncated mutants
(?198 and ?92) and the deletion mutant ?(30–92). The deleted
regions are not shown (?198 and ?92) or shown as a straight line
[?(30–92)]. (D) Schematic picture of lactose permease (LacY) con-
taining 12 transmembrane helices (shown as shaded boxes) and
hydrophilic loops (shown as straight lines between the boxes). The
enzymes used for the construction of the LacY–FtsY hybrids are
shown under the LacY diagram.
Schematic representation of FtsY and SR-?, FtsY mutants,
arabinose. Competent E. coli N4156::pAra14-FtsY? (13) was trans-
formed with pCL1291 derivatives harboring the various ftsY gene
constructs as indicated. The transformants were incubated in LB
supplemented with 0.2% arabinose at 37°C. One hour later, the
recovered transformants were centrifuged, washed once in LB without
arabinose, and spread on LB plates containing ampicillin (100 ?g?ml),
spectinomycin (100 ?g?ml), IPTG (0.1 mM), and with or without
arabinose. Photographs were taken after incubation for 36 hr at 37°C.
Formation of single colonies on agar plates with or without
6026Biochemistry: Zelazny et al. Proc. Natl. Acad. Sci. USA 94 (1997)
were grown without arabinose to 0.15 OD600unit and induced
by 0.5 mM IPTG for 2 hr. Cells were then harvested and
treated as described above. Membranes were collected by
ultracentrifugation (30 min at 150,000 ? g). Protein concen-
trations in the extracts and in the membrane fractions were
measured according to a modified Lowry procedure in the
presence of 2.5% SDS using bovine serum albumin as a
standard. An aliquot (10 ?g of proteins or as specified) from
each sample was then mixed with 2? sample buffer, incubated
at 37°C for 15 min, and subjected to SDS?PAGE (7.5–10% as
indicated). After electroblotting, the nitrocellulose was incu-
bated with anti-FtsY or anti-LacY serum as indicated, washed
twice, and incubated with goat anti-rabbit antibodies conju-
gated to horseradish peroxidase. The nitrocellulose paper was
briefly soaked in the fluorescent substrate solution and ex-
posed to film for ?10 sec.
RESULTS AND DISCUSSION
We have speculated that like in SR-? (16), the membrane
targeting and attachment domain of FtsY is located within the
N terminus. To identify this domain, full ftsY DNA was
amplified from the chromosome and inserted into pT7-5(lacY)
(17), instead of the lacY coding region, under the control of the
lac promoter?operator (plasmid pTftsY). This plasmid was
used to construct N-terminally manipulated FtsY mutants,
leaving the conservative C-terminal NG domain intact (Fig.
1B, stippled boxes). One mutant contains an internal deletion
of codons 30–92 of ftsY [Fig. 1C, ?(30–92)] and the other two
are N-terminally truncated versions of FtsY (Fig. 1C, ?92 and
?198). To test the function of the mutants we used the E. coli
strain N4156::pAra14-FtsY? (13) which harbors a chromo-
somal copy of the essential ftsY gene under the control of the
tight araB promoter (instead of the native ftsY) and therefore
requires arabinose for growth. The deleterious effects of
arabinose depletion on cell growth and on the expression of
FtsY have been analyzed previously (7, 13). One technical
point worth mentioning is that the ftsY gene constructs were
transferred with their lac promoters to the low copy-number
plasmid, pCL1921 (18), which is compatible with the polA?
phenotype of strain N4156::pAra14-FtsY? [thereby generating
plasmids pCLftsY, pCL?(30–92), pCL?92 and pCL?198].
When competent N4156::pAra14-FtsY? cells are transformed
with pCLftsY, they form wild-type-like single colonies on LB
plates without arabinose and with 0.1 mM IPTG (Fig. 2).
However, when transformed with plasmids encoding the de-
letion mutants, single colonies are formed only on arabinose
supplemented LB plates (Fig. 2).
Clearly, one explanation for the observation that the mu-
tants do not support growth of FtsY-depleted cells is that the
deleted sequences are essential for enzymatic activity, or
alternatively are required for targeting and attachment of FtsY
to the membrane. To exclude a more trivial possibility that the
mutants are simply not expressed from the lac promoter,
immunoblots were carried out with extracts prepared from E.
coli cells (ftsY?) transformed with pT7-5 (control), pT?92, or
pT?198, using anti-FtsY antibodies. The results demonstrate
that mutants ?198 and ?92 are expressed from the lac
were grown overnight, diluted, and induced by IPTG at the exponential growth phase. Cell extracts were subjected to immunoblot analysis with
anti FtsY antibodies. (B) E. coli N4156::pAra14-FtsY? transformed with the indicated plasmids were grown overnight, diluted into LB without
arabinose, and induced by IPTG at OD600of ?0.15. Membrane fractions were separated by SDS?PAGE and subjected to immunoblot analysis
with anti-FtsY antibodies. (C) E. coli N4156::pAra14-FtsY? transformed with the indicated plasmids were grown overnight, diluted into LB without
arabinose, and 3 hr later induced for 2 hr by 0.2% arabinose (Ara, for expression of FtsY, lane 2) or by 100 ?g per ml of indoleacrylic acid (IAA,
for expression of hybrid TrpE-?198, lane 4) (19).
Expression of FtsY mutants, LacY–FtsY hybrids, and TrpE-?198 hybrid. (A) E. coli UT5600 cells transformed with the indicated plasmid
Biochemistry: Zelazny et al.Proc. Natl. Acad. Sci. USA 94 (1997)6027
promoter (Fig. 3A, lanes 1 and 2, respectively). Interestingly,
unlike wild-type FtsY which migrates extremely slowly (ap-
parent molecular weight 97 kDa compared with a calculated
molecular weight of 54 kDa; Fig. 3C, lane 2; see also refs. 13
and 20), the migration of mutant ?92 is only moderately slower
(?58 kDa) compared with a calculated molecular weight of 44
kDa and ?198 migrates according to its calculated molecular
weight (33 kDa). Therefore, the aberrant migration of FtsY in
SDS?PAGE is determined by the N-terminal sequence.
The observations that we have described lead to the con-
clusion that the N-terminal domain of FtsY carries an essential
function. One possibility referred to above is that this domain
mediates membrane targeting and attachment of FtsY. To test
this hypothesis we constructed hybrids between the N-
terminally truncated FtsY mutants and unrelated integral
membrane proteins which have the ability to guide and anchor
the mutant proteins to the membrane. We used integral
membrane polypeptides from the LacY of E. coli because it is
well characterized structurally (21) and contains 12 hydropho-
bic membrane spanning segments (Fig. 1D). The fusion joints,
connecting N-terminal fragments of LacY with the FtsY
mutants, were placed at known cytoplasmic locations accord-
ing to the secondary structure of the permease which has been
fused to the cytoplasmic loops between transmembrane seg-
ments 4 and 5 (to form N4-?92), and 6 and 7 (to form
N6-?198), respectively (Fig. 1D). Immunoblot analysis with
anti-FtsY antibodies demonstrates, despite of the occasional,
nonspecific cross reactivity of the polyclonal anti FtsY anti-
bodies with some unrelated proteins (Fig. 3B, lane 2), that the
fraction (Fig. 3B, lanes 1 and 3). As expected, hybrid N4-?92
contains part of the acidic N-terminal domain of FtsY and
therefore migrates slower (as a 74-kDa band, Fig. 3B, lane 1)
than its theoretical molecular weight (60 kDa). Hybrid N6-
?198 which does not contain the N-terminal acidic residues of
FtsY, migrates as expected (55 kDa, Fig. 3B, lane 3), with a
characteristic broad band, as do many other integral mem-
brane proteins. A putative proteolytic product which migrates
similarly to mutant ?198 is also apparent (Fig. 3B, lane 3).
To test the activity of these hybrids, the genes were trans-
ferred into pCL1921 (as described above for the deletion
mutants) and the resulting plasmids were used to transform
N4156::pAra14-FtsY? cells. The transformants were plated on
LB plates with or without arabinose and IPTG. Both LacY–
FtsY expressing transformants were able to form colonies on
plates without arabinose as well as on plates with arabinose
(Fig. 2). Cells expressing N6-?198 produce smaller colonies,
indicating that although this hybrid complements FtsY-
depleted cells, it might not be fully functional or alternatively
its expression might be toxic to some extent. The ability of the
truncated FtsY mutants and of the LacY–FtsY hybrids to
support growth of FtsY-depleted cells was assayed also quan-
titatively, in liquid media (Fig. 4A). For this purpose, overnight
cultures grown in LB containing arabinose were harvested,
washed twice, and diluted 1:2,000 into fresh LB without
arabinose. The growth of cells harboring only the vector
(pCL1921) or mutants ?92, ?198, and ?(30–92) ceases after
4–5 hr at 0.3 OD600 unit. Mutants N4-?92 and N6-?198
continue to grow up to cell densities of 1.5, 1.4, and 0.8 OD600
unit, respectively, in the absence of arabinose. In fact, the
growth of N4-?92 in the absence of arabinose is similar to the
growth in the presence of arabinose of control cells harboring
‘‘activate’’ the NG domain of FtsY if fused to its N terminus,
an important control experiment was conducted. We con-
structed a hybrid composed of the N-terminal 326-amino acid
residues of the cytoplasmic protein, TrpE, fused to the N
terminus of ?198. Western blot analysis indicates that the
hybrid (TrpE-?198, Fig. 3C, lane 4) and wild-type FtsY (Fig.
3C, lane 2) are expressed to a comparable level. However,
TrpE-?198 is unable to support growth of FtsY depleted cells
(data not shown). Therefore, the NG domain of FtsY is
functional only when attached to a membrane protein but not
to a soluble protein.
Since it has been shown that FtsY is essential for expression
of integral membrane proteins, including LacY (7), we tested
the ability of the FtsY mutants and the LacY–FtsY hybrids to
support LacY expression from the chromosome, in FtsY-
depleted cells. Cultures were treated as in the growth exper-
iments and 4 hr after FtsY depletion the cultures were induced
by IPTG, and membrane preparations were analyzed by
immunoblotting using anti-LacY antibodies (7). As shown in
harboring pCLN4-?92 or pCLN6-?198 is significantly higher
than in cells harboring pCL?(30–92), pCL?198, or pCL?92
(see also ref. 7). Under the same conditions, the expression of
the cytoplasmic protein, ?-galactosidase, is not affected by
FtsY depletion (data not shown; see also ref. 7). Therefore, the
function of the inactive FtsY mutants, ?92 and ?198, is
restored when they are fused to cytoplasmic loops of an
integral membrane protein. We conclude from these experi-
ments that for proper function, FtsY must be targeted to the
membrane and that its N-terminal domain is indeed involved
in this process.
Despite the functional similarity between the N-terminal
domains of SR-? (16) and FtsY (shown here) in mediating
peripheral interaction with the membrane, they differ mark-
edly in their primary structures and overall chemical proper-
ties: The N-terminal domain of FtsY contains a large number
of acidic amino acid residues (mainly glutamic acids) through-
out the entire sequence, whereas the membrane binding
domain of SR-? contains two hydrophobic stretches and a
basic region. These differences support a suggestion that the
LacY–FtsY hybrids. E. coli (N4156::pAra14-FtsY?) cells harboring the
once in LB, and diluted to OD600 ? 0.005 in LB with or without
arabinose as indicated. Growth curves (A) were constructed from the
average of two independent experiments. After 4 hr, 10 ml samples
were transferred from chosen cultures into separate Erlenmeyers and
induced by 0.5 mM IPTG for 2 hr. Membranes were prepared from the
induced cells and subjected to Western blot analysis using anti-LacY
Functional complementation of FtsY depleted cells by
6028Biochemistry: Zelazny et al.Proc. Natl. Acad. Sci. USA 94 (1997)
twoproteinsusedifferentmembranetargetingandattachment Download full-text
mechanisms or recognize and interact with different integral
membrane counterparts. In both cases, however, the SRP
receptor is probably responsible for an essential step late
during the targeting process, by placing the ribosome–nascent
chain–SRP complex near the translocation machinery on the
target membrane. An important challenge is to clarify the
mechanism and to identify the proteins that mediate the
following step in the targeting pathway where the translation
machinery is transferred from the SRP system to the translo-
This work was supported by the Leo and Julia Forchheimer Center
for Molecular Genetics, the Weizmann Institute of Science, and the
Dr. Josef Cohn Minerva Center for Biomembrane Research.
Nunnari, J. & Walter, P. (1992) Curr. Opin. Cell Biol. 4, 573–580.
Rapoport T. A. (1992) Science 258, 931–936.
Wickner, W. T. & Lodish, H. F. (1985) Science 230, 400–407.
Ro ¨misch, K., Webb, J., Herz, J., Prehn, S., Frank, R., Vingron, M.
& Dobberstein, B. (1989) Nature (London) 340, 478–482.
Bernstein, H. D., Poritz, M. A., Strub, K., Hoben, P. J., Brenner,
S. & Walter, P. (1989) Nature (London) 340, 482–486.
Ribes, V., Ro ¨misch, K., Giner, A., Dobberstein, B. & Tollervey,
D. (1990) Cell 63, 591–600.
Seluanov, A. & Bibi, E. (1997) J. Biol. Chem. 272, 2053–2055.
Ulbrandt, N. D., Newitt, J. A. & Bernstein, H. D. (1997) Cell 88,
9. de Gier, J.-W. L., Mansournia, P., Valent, Q. A., Phillips, G. J.,
Lurink, J. & von Heijne, G. (1996) FEBS Lett. 399, 307–309.
Macfarlane, J. & Muller, M. (1995) Eur. J. Biochem. 233,
Montoya, G., Svensson, C., Luirink, J. & Sinning, I. (1997) Nature
(London) 385, 365–368.
Gill, D. R. & Salmond, G. P. C. (1987) Mol. Gen. Genet. 210,
Luirink, J., Hagen-Jongman, C. M., der Weijden, C. C., Oudega,
B., High, S., Dobberstein, B. & Kusters, R. (1994) EMBO J. 13,
Young, J. C. & Andrews, D. W. (1996) EMBO J. 15, 172–181.
Miller, J. D., Tajima, S., Lauffer, L. & Walter, P. (1995) J. Cell
Biol. 128, 273–282.
Young, J. C., Ursini, J., Legate, K. R., Miller, J. D., Walter, P. &
Andrews, D. W. (1995) J. Biol. Chem. 270, 15650–15657.
Bibi, E. & Kaback, H. R. (1990) Proc. Natl. Acad. Sci. USA 87,
Lerner, C. G. & Inouye, M. (1990) Nucleic Acids Res. 18, 4631.
Koerner, T. J., Hill, J. E., Myers, A. M. & Tzagoloff, A. (1991)
Methods Enzymol. 194, 477–490.
Gill, D. R. & Salmond, G. P. C. (1990) Mol. Microbiol. 4,
Kaback, H. R., Frillingos, S., Jung, H., Jung, K., Prive, G. G.,
Ujwal, M. L., Weitzman, C., Wu, J. & Zen, K. (1994) J. Exp. Biol.
Ujwal, M. L., Jung, H., Bibi, E., Manoil, C., Altenbach, C.,
Hubbell, W. L. & Kaback, H. R. (1995) Biochemistry 34, 14909–
Biochemistry: Zelazny et al.Proc. Natl. Acad. Sci. USA 94 (1997)6029