Subunit 8 o f Chloroplast F0Fr ATPase and O SC P of
Mitochondrial F0Fr ATPase: a Comparison by CD-Spectroscopy
Siegfried Engelbrechta, Jennifer Reedb, Francois Peninc, Daniele C. G autheronc, and
W olfgang Jungea
a Biophysik, Fachbereich Biologie/Chemie, Universität Osnabrück, D-4500 Osnabrück
b Institut für experimentelle Pathologie, Deutsches Krebsforschungszentrum,
D -6900 Heidelberg 1
c Laboratoire de Biologie et Technologie des Membranes du C. N . R. S.,
Universite Claude Bernard de Lyon, F-69622 Villeurbanne Cedex
Z. N aturforsch. 46c, 7 5 9 -7 6 4 (1991); received May 31, 1991
C F,, M F ,, OSCP, C D-Spectroscopy, Subunit 8
C D spectra have been recorded with subunit 5 from chloroplast C F0CF, and with OSCP
from mitochondrial M F0M F ,. These subunits are supposed to act similarly at the interface
between proton transport through the F0-portion and ATP-synthesis in the F,-portion o f their
respective F0F,-A TPase. Evaluation o f the data for both proteins revealed a very high a-helix
content o f —85% and practically no ß-sheets. Despite their low hom ology on the primary
structure level (23% identity) and their different electrostatic properties (pl-values differ by
3 units), spinach 8 and porcine OSCP are indistinguishable with respect to their secondary
structure as measured by CD. Prediction and analysis o f consensual a-helices even in poorly
conserved regions indicate a high degree o f structural similarity between chloroplast 8 and
OSCP. In view o f the topology and function o f 8 and OSCP in intact F0F, these findings are
interpreted to indicate the dominance o f secondary and tertiary structure over the primary
structure in their supposed function between proton flow and ATP-synthesis.
Introduction
ATP synthesis at the expense o f a protonm otive
force is catalyzed by F 0F,-ATPases in thylakoids,
m itochondria and many microorganisms. F 0F,ATPases consist o f a membrane-embedded proton
channel F 0 and a water-soluble F,-portion, which
carries the active sites. Upon removal from F0, F,
catalyses ATP hydrolysis. F, consists of five poly­
peptides named a, ß, y, 5, and e in order of de­
creasing molecular mass. Subunit 5 of chloroplast
and E. coli F, and their mitochondrial counterpart
OSCP are small proteins of ~ 21 kD a mass. They
are located between the F, and F 0 parts and indis­
pensable for functional F 0F,. The position of the
‘small’ subunits (y, 8 /OSCP, e) at the interface be­
tween F 0 and F, points to a special role of these
Abbreviations: CFqCF,, chloroplast F0F,-ATPase; CF0,
chloroplast proton channel (membrane-embedded);
C F,, chloroplast ATPase (soluble part); M F0, M F ,, EF0
and EF, are the respective terms for the mitochondrial
and E. coli proteins; OSCP, mitochondrial oligomycin
sensitivity-conferring protein; C D, circular dichroism.
Reprint requests to Dr. S. Engelbrecht, Universität Os­
nabrück, Biophysik, Barbarastraße 11, D-4500 Osna­
brück, F.R .G .
Verlag der Zeitschrift für Naturforschung, D-7400 Tübingen
0939-5075/91/0900-0759 $01.30/0
subunits with respect to the link - whatever its na­
ture - between H M ranslocation and ATP syn­
thesis/liberation. For general reviews on F 0F,ATPases see ref. [1] to [3], the role of subunits 8
and OSCP has been reviewed in ref. [4],
In view of their location and supposedly similar
function in the FqFj complex it is surprising that
subunit 5 is an acidic protein (pi = 5.7, S. Engelbrecht, unpublished data) whereas OSCP is basic
(pi = 8.5, F. Penin, unpublished data). Based on
similar quaternary structure of F-type ATPases
from different sources, a common mechanism is
expected and also similar structures of the protein
subunits. On the basis of sequence homologies
alone this expectation is not met. O f the five CF,subunits only the sequences o f a and ß are highly
homologous to other species, while y, 5 and s show
only weak similarities. This trend is even more
pronounced for the subunits of the F 0-portion.
Polarity inversions within hom ologous protein
complexes isolated from different sources are not
uncommon, e.g. cytochrome c-oxidase/cytochrome c [5]. It is also not uncom m on that differ­
ent prim ary structures result in very similar folding
patterns and dom ain structure, e.g. the two do­
mains o f the bifunctional enzyme N-S'-phosphoribosylanthranilate
isomerase/indole-3-glycerol-
Unauthenticated
Download Date | 6/16/17 5:22 PM
760
S. Engelbrecht et al. • Subunit 8 o f Chloroplast F0F,-A TPase
phosphate synthase [6 ]. This prom pted us to study
the secondary structure of chloroplast 5 and m ito­
chondrial OSCP.
CD spectroscopy offers a sensitive m ethod for
determining the proportion of certain secondary
structure elements. Dupuis, Zaccai, and Satre have
previously measured the CD spectrum of beef
heart OSCP [7] and calculated 43% a-helical struc­
ture. With improved instrum entation and data
evaluation we studied porcine OSCP for com pari­
son with spinach chloroplast 5.
meter coupled to a Jasco J-DPY data processor.
Curves were recorded digitally and fed through the
data processor for signal averaging and baseline
subtraction. Samples at a concentration of
50-100 ng protein/ml in 10 m M Tris/HCl pH 7.5
were scanned from 190-240 nm in a dichroically
neutral quartz cuvette with a path length of
1.0 mm. The sensitivity was 2.0 m°/cm, time con­
stant 2 s, scanning speed 5 nm/min. Spectra were
averaged over four scans. A signal-averaged base­
line was subtracted.
F or estimation of secondary structure content,
points taken at 0.5 nm intervals were processed us­
ing the CD application package o f C O N TIN [12]
run on a VAX 11/780 computer. This program
analyses a given CD spectrum as a linear com bina­
tion of the CD spectra o f 16 proteins whose sec­
ondary structure content is known, and gives the
result as percent a-helix, ß-sheet and ‘rem ainder’
(a mixture of extended coils and reverse turns). A
total of four different preparations was measured
both with spinach chloroplast 5 and pig heart mi­
tochondrial OSCP. Secondary structure predic­
tions and predicted pFs were calculated with the
University of Geneva PcGene program.
Materials and Methods
Spinach 5 was prepared from CFj by anion ex­
change chrom atography in the presence of deter­
gent, followed either by rechrom atography or by
hydrophobic interaction chrom atography as de­
scribed [8 , 9], Porcine OSCP was prepared by se­
lective extraction o f preextracted m itochondrial
membranes, followed by cation exchange chrom a­
tography as described [10]. All preparations were
SDS-electrophoretically homogeneous (data not
shown).
Protein determ inations were perform ed accord­
ing to Sedmak and Grossberg [11]. Alternatively,
the Pierce bicinchoninic acid version o f the Lowry
procedure was used (Pierce Europe B.V., POB
1512, NL-BA Oud Beijerland, The Netherlands).
Lysozyme, bovine serum albumin and ovalbumin
were used as standards. W hereas spinach 5 gave
the same response with both assays, porcine OSCP
was grossly underestim ated by the Sedmak and
Grossberg procedure. D ata were therefore cross­
checked by amino acid analysis.
Circular dichroism spectra were measured with
a Jasco J-500 autom atic recording spectropolari-
Results and Discussion
Fig. 1 shows representative CD spectra for spin­
ach 8 (sample 2) and porcine OSCP (sample 1).
M easurements are indicated by points, the line
represents a fit as calculated according to ref. [ 1 2 ].
Fit param eters are given in Table I. The relative
proportions of secondary structure elements are
summarized in Table II. Both proteins are highly
a-helical to about 85%. Whereas OSCP seems to
be virtually devoid of any ß-sheet structure, the
Table I. Summary o f possible solutions for fits o f measured CD spectra.
All data are in %.
S2
5,
5
a-Helix:
ß-Sheet:
Remainder:
85
3
12
82
4
15
86
6
8
OSCP,
OSCP
a-Helix:
ß-Sheet:
Remainder:
85
0
15
79
0
21
89
9
3
OSCP-,
93
0
4
81
0
19
91
0
9
53
84
71
23
6
81
16
3
OSCP3
OSCP4
89
0
11
89
0
11
85
2
13
85
0
15
Unauthenticated
Download Date | 6/16/17 5:22 PM
S. Engelbrecht et al. ■Subunit 5 o f Chloroplast F0F,-A TPase
Fig. 1. CD spectra o f spinach chloroplast 5
(sample 2, left part o f the figure) and porcine
heart mitochondrial OSCP (sample 1, right
part o f the figure). Points are experimental,
curves are theoretical fits.
Table II. Summary o f secondary structure com position
as evaluated from CD spectra (n = 4).
a-H elix [%]
ß-Sheet [%]
Remainder [%]
5
OSCP
83.6 ± 6 .1
8.7 ± 7.5
8.0 ± 4.1
85.8 ± 4 .2
0.3 ± 0 .7
13.6 ± 4.9
figures for 5 lie between 0 and 23% . 5 3 seems to be
somewhat anom alous. On the average, OSCP
shows about 14% o f ‘rem ainder’ and 5 8 %.
O f the 187 amino acids o f spinach 5, 156 thus
are in helical conform ation and about 16 in
ß-sheets. For OSCP, 163 am ino acids out of 190
are in helical conform ation. A lthough CD d ata do
not yield inform ation concerning the alignm ent of
secondary structure elements along the prim ary
structure, the sheer am ount o f a-helix and the near
identity o f this am ount for both 8 and OSCP sug­
gest that these two proteins share a similar threedimensional structure.
Dupuis et al. have previously calculated an
am ount of 43% a-helical structure for bovine
OSCP [7]. Although the prim ary structure o f p o r­
cine OSCP is not known, porcine and bovine
OSCP are highly similar: the am ino acid analyses
are very close, the 18 N-terminal residues are the
same and an epitope for monoclonal antibodies is
shared by both proteins [13]. Therefore the dis­
crepancy between Dupuis et al. evaluation and the
one presented here is likely due to the improved
evaluation method and also to the extended range
of measurement (190-240 nm vs. 200-240 nm).
Commonly used prediction program s for second­
ary structure [14, 15] grossly underestim ate the
a-helix content both o f spinach 8 (3 9 -5 0 % ) and
bovine OSCP (49-62% ). Only the use of the D ou­
ble-Prediction method [16], which takes into ac­
count the predicted structural class of the protein,
improves the prediction of secondary structure ele­
ments. Indeed, with this method OSCP is predict­
ed as an all a-protein with 79% a-helix and almost
no ß-sheet (3%) and chloroplast 8 is predicted as
an a + ß protein in very good agreement with the
CD data.
Since CD data do not yield tertiary structure in­
formation, prediction methods have been used
here for this purpose. It is possible to constrain the
prediction program to fit the secondary structure
content as measured by CD, using optimized deci­
sion constants [14], Fig. 2 shows the alignment and
predicted structures after application o f the D ou­
ble Prediction program [17]. E. coli 8 is included in
Unauthenticated
Download Date | 6/16/17 5:22 PM
762
S. Engelbrecht et al. ■ Subunit 5 o f Chloroplast F0F,-A TPase
10
1
40
E. coli 8
MSEFITVARPYAKAAFDFAVEHQSVERWQDMLAFA AEVTKNEQ
spinach 8
VDSTASRYASALADVADVTGTLEATNSDVEKLIRIFSEEP
OSCP
FAKLVRPPVQIYGIEGRYATALYSAASKQNKLEQV EKELLRVGQILKEPK
• •••
••
•••••••••••••••••••••••••••••••••••••
consensus
kk
k
k
k
90
E. coli 8
.A E L L S G A L A P E T L A E S F I A V C G .EQLDENGQNLIRVMAENGRLNALPD
spinach 8
VYYFFANPVISIDNKRSVLDEIITT SGLQPHTANF INILIDSERINLVKE
OSCP
MAASLLNPYVKRSVKVKSLSDMTAKEKFSPLTSNL INLLAENGRLTNTPA
................... .............................
•••••••
==•••••••••••===•
.....
••
+ . .............................
•••••=•=••••••••••••
••••••••
consensus
* •*
•
••
•••••*•
•
••••
••
E. coli 8
V LEQFIHLRAVSEATA E V D V I S A A A L S E Q Q L A K I S A A M E K R L S R . .KVKL
spinach 8
I L N E F E D V F N K I T G T E V A W T S W K L E N D H L A Q I A K G V Q K .ITGAKNVRI
OSCP
VISAFSTMMSVHRGEVPCTVTTASALDETTLTELK TVLKSFLSKGQVLKL
• • • • • • • • • • • • • • • • • • • • • • • • • • • » • • • • • • • • • • • • • • • A
••••
consensus
E. coli 8
N C K I D K S V M A G V I I R ...AGDMVIDGSVRGRLERLADVLQS
spinach 8
KTVIDPSLVAGFTIRYGNEGSKLVDMSVKKQLEEI AAQLEMDDVTLAV
OSCP
E V K I D P S I M G G M I V R . ..IGEKYVDMSAKTKIQKLSRAMREIL
==
.......................
...........................................................
consensus
the figure because the high m utational frequency
of this organism serves as a convenient means to
sort out ‘essential’ amino acids. Although E. coli Ö
has not been m easured here by CD spectroscopy,
the functional equivalence o f E. coli 5 and spinach
5 as observed in hybrid-reconstitution experiments
[17] leaves little doubt about the structural similar­
ity of these two proteins. OSCP appears to be
more closely related to E. coli 5 (25% homology)
than to chloroplast 5 (22%). The degree of similar­
ity of the three sequences is especially impressive in
the range of Pro 70 to Leu 87 and Pro 146 to Met
181. The presence o f six consensual a-helices clear­
ly appears in Fig. 2. Helical wheel projection of
these consensual a-helices revealed a comparative
distribution o f hydrophobic, hydrophilic, and
Fig. 2. Comparison be­
tween predicted secondary
structure o f E. coli 5, spin­
ach chloroplast 5 and bo­
vine OSCP. For spinach 5
and OSCP the prediction
method D O U BL E PRE­
D ICTIO N [16] was con­
strained to fit the content o f
secondary structure ob­
tained by C D (Table II),
using optimized decision
constants [14], The decision
constants chosen for spin­
ach 6 were -0 .1 7 for
a-helix, +0.07 for ß-sheet,
and 0 for turn and coil. For
OSCP, the respective values
were -0 .0 3 , +0.05, 0,
-0 .0 3 . For E. coli 8, all con­
stants were set to 0. A lign­
ment o f the three sequences
was done with the program
P A R SIM O N Y [26], Identi­
cal alignments were o b ­
tained with CLU STAL [27]
and
M U L T A L IN
[28].
Stars
indicate
perfect
matches between all three
sequences, small points in­
dicate identical residues in
two sequences. Symbols o f
predicted secondary struc­
ture: a-helix: # , ß-sheet: = ,
turn: + , coil: blank. C on­
sensual a-helical segments
are underlined. All calcula­
tions were performed by
using the software package
A N T H E PR O T [29],
charged residues between the three proteins (not
shown). However, the well conserved regions do
not necessarily belong to the consensual a-helices
but might be very im portant in maintaining a simi­
lar three-dimensional structure. The results leave
little doubt that chloroplast 8 , E. coli 5, and m ito­
chondrial OSCP share a similar three-dimensional
structure.
Based on small-angle neutron scattering Dupuis
et al. calculated m olecular dimensions of about
9 x 3 x 3 nm for bovine OSCP [7], Based on ro ta­
tional diffusion in solution Wagner et al. calculat­
ed dimensions o f about 1 0 x 2 . 8 nm for spinach 5
[18]. How do these findings relate to the high
a-helix content of 5 and OSCP? With a rise per re­
sidue of 0.15 nm 160 am ino acids would give a sin-
Unauthenticated
Download Date | 6/16/17 5:22 PM
S. Engelbrecht et al. ■Subunit 5 o f Chloroplast F0F|-A TPase
gle stretched-out a-helix o f 24 nm length. The
above dimensions then allow for only two parallel
helices spanning the entire length of the molecule.
Such an arrangement also fits into the model of
F 0F,-ATPases as proposed by Gogol et al. [19],
These workers observed a narrow stalk between
EF 0 and EF, (using electron microscopy). The di­
ameter of the stalk would allow for four to tive
closely packed a-helices. Subunit b o f E F 0 (I in
C F 0) is predicted to consist o f only two a-helices
outside the membrane [2 0 , 2 1 ],
E. coli 5 contains proline residues in positions 7,
52, and 89, spinach 5 contains prolines in positions
40, 48, 70, and 146, and OSCP contains prolines in
positions - 4 , - 3 , 39, 48, 70, 89, 107, and 146. It
remains to be seen in w hat m anner these proline
residues, especially those located in the center of
the molecules, are accom m odated in the tertiary
structure. Proline, if not starting/breaking helices,
is known to introduce kinks which loosen up the
packing density.
The present study was triggered by the observa­
tion that OSCP does not ‘fit the picture’ because it
has a basic pi as com pared to the acidic spinach
and E. coli 8 subunits. In view of the similarity on
the secondary structure level reported here, the
failure of OSCP to substitute for spinach 5 in hy­
brid reconstitution experiments with C F,(- 8 )
(S. Engelbrecht, unpublished data) is in contrast to
the behaviour of E. coli 8 [17] and may be ex­
plained by the reversal in charge.
Does this reversal allow for identification of
contact sites for 8 and OSCP on neighbouring sub­
units? Spinach 8 contains 13 aspartic acids and 12
lysines, whereas (bovine) OSCP contains 4 aspar­
tic acids and 20 lysines. Num bers for the other
charged residues are somewhat balanced. In order
to shift the pi of OSCP into the acidic range and
likewise the pi of 8 into the basic range, one would
have to exchange about 7 aspartic acids for lysines
and vice versa, or twice this am ount of each amino
acid alone. The sequence alignment of spinach and
E. coli 8 and bovine OSCP (ref. [4], Fig. 2) reveals
that the (spinach 8 ) aspartic acids, if not con­
served, are mostly substituted for by serines and
threonines in OSCP. Roughly the same is true for
the substitution of OSCP-lysines in spinach 8 . The
‘additional’ lysines in OSCP are scattered through­
out the sequence in a way which completely ob­
scures possible counterparts.
763
Subunits b of EF 0 and M F 0 and subunit I of
spinach C F 0 are considered to be the main binding
partners of 8 and OSCP [1-4]. Based on the amino
acid sequences [21-23], the predicted p i’s of these
proteins, 8 and OSCP are summarized in
Table III. It is evident that a charge reversal like
the one between 8 and OSCP has not occurred
with F0-subunits I and b. Furtherm ore, helical
wheel plots (not shown) reveal that any ‘pairing’ of
subunits is possible at least theoretically: R ota­
tional and translational shifts neither make salt
bridges nor regions of electrostatic repulsion ob­
vious. The complex F 0F r ATPase eludes further
structural conclusions at this level. Its variability
of primary structure remains enigmatic.
Table III. Nomenclature and theoretical isoelectric
points o f some F0F, subunits. Please note the difference
between experimentally determined p i’s and predicted
values (5.7 vs. 4.41 for SDinach 5 and 8.5 vs. 10.66 for
OSCP).
Subunit
(E . coli and
chloroplast)
Counterpart
in mitochondria
Predicted pi
EF0-6
C F0-I
M F0-6
EF,-5
CF,-8
OSCP
M F0-6
M F0-ft
5.9
8.6
9.7
4.71
4.41
10.66
OSCP
OSCP
The combined evidence presented here and
formerly [17] favors the view that spinach and
E. coli 8 and OSCP are very similar proteins not
on the primary, but on the secondary and most
likely also on the tertiary structure level. It has
been reported that the N-term inal half of OSCP
shows some sequence homology with E F 0- 6 [24].
In view of the high predicted content o f a-helices
in E F 0- and M F 0- 6 and C F 0-I [21-23] and the ob­
served highly a-helical structure of 8 and OSCP
this finding should not be interpreted to indicate
variations in small subunit arrangements between
F 0 and F, from various sources (OSCP being a
‘mixture’ of subunits 8 and b/l of bacteria or chlo­
roplasts). A m utant E. coli strain which carries
Unauthenticated
Download Date | 6/16/17 5:22 PM
764
S. Engelbrecht et al. • Subunit 5 o f Chloroplast F0F,-ATPase
C F0-I instead of EF 0-Z> grows similar to the wild
type [25] although C F 0-I and EF 0-Z> are even less
homologous than 5/OSCP. The molecular mecha­
nism of coupling proton movement to ATP libera­
tion most probably has been conserved, despite the
variability of prim ary structure. This points to the
dominance of secondary to quaternary structure in
the function of F 0F r ATPases.
Acknowledgements
[1] E. Schneider and K. Altendorf, M icrobiol. Rev. 51,
4 7 7 -4 9 7 (1 9 8 7 ).
[2] A. E. Senior, Physiol. Rev. 68, 177-231 (1988).
[3] R. H. Fillingame, The Bacteria XII, 345 -3 9 1 (1990).
[4] S. Engelbrecht and W. Junge, Biochim. Biophys.
Acta 1015, 3 7 9 -3 9 0 (1990).
[5] E. A. E. Garber, T. L. Luntz, and E. Margoliash,
Prog. Clin. Biol. Res. 274, 7 4 9 -7 6 9 (1988).
[6] J. P. Priestle, M. G. Grütter, J. L. White, M. G. Vin­
cent, M. Kania, E. W ilson, T. S. Jardetzky, K.
Kirschner, and J. N . Jansonius, Proc. Natl. Acad.
Sei. U .S.A . 84, 5 6 9 0 -5 6 9 4 (1 9 8 7 ).
[7] A. D upuis, G. Zaccai, and M. Satre, Biochemistry
22, 59 5 1 -5 9 5 6 (1 9 8 3 ).
[8] S. Engelbrecht and W. Junge, FEBS Lett. 219, 321 325(1987).
[9] S. Engelbrecht, G. Althoff, and W. Junge, Eur. J.
Biochem. 189, 1 9 3 -1 9 7 (1 9 9 0 ).
[10] P. Archinard, F. Penin, C. G odinot, and D. C.
Gautheron, M ethods Enzymol. 126, 4 5 5 -4 6 0
(1986).
[11] J. J. Sedmak and S. E. Grossberg, Anal. Biochem.
79, 5 4 4 -5 5 2 (1 9 7 7 ).
[12] S. W. Provencher and J. Glöckner, Biochemistry 20,
3 3 -3 7 (1 9 8 1 ).
[13] C. G odinot, S. Colorio, F. Cretin, Infaurgarat,
G. D eleage, and B. R oux, Biochimie 71, 9 1 7 -9 2 9
(1989).
[14] J. Garnier, D. J. Osguthorpe, and B. Robson, J.
M ol. Biol. 120, 9 7 -1 2 0 (1978).
[15] O. Gascuel and J. L. Golmard, CABIOS 4, 3 5 7 -3 6 5
(1988).
[16] G. Deleage and B. Roux, Protein Engineering 1,
2 8 9 -2 9 4 (1 9 8 7 ).
[17] S. Engelbrecht, G. Deckers-Hebestreit, K. Altendorf, and W. Junge, Eur. J. Biochem. 181, 485—491
(1989).
[18] R. Wagner, E. C. Apley, S. Engelbrecht, and
W. Junge, FEBS Lett. 230, 1 0 9 - 115 (1988).
[19] E. P. G ogol, U . Lücken, and R. A. Capaldi, FEBS
Lett. 219, 2 7 4 -2 7 8 (1 9 8 7 ).
[20] J. E. Walker, M. Saraste, and N . J. Gay, Nature
298, 8 6 7 -8 6 9 (1 9 8 2 ).
[21] C. R. Bird, B. Koller, A. D . Auffret, A. K. Huttly,
C. J. Howe, T. A. Dyer, and J. C. Gray, EMBO J. 4,
13 8 1 -1 3 8 8 (1 9 8 5 ).
[22] J. E. Walker, M. Saraste, and N . J. Gay, Biochim.
Biophys. Acta 768, 1 6 4 -2 0 0 (1984).
[23] J. E. Walker, M. J. Runswick, and L. Poulter,
J. M ol. Biol. 197, 8 9 - 100 (1987).
[24] Y. A. O vchinnikov, N . N. M odyanov, V. A. Grin­
kevich, N . A. A ldanova, P. V. Kostetsky, O. E. Tru­
betskaya, T. Hundal, and L. Ernster, FEBS Lett.
175, 1 0 9 -1 1 2 (1 9 8 4 ).
[25] G. Schmidt, A. J. W. Rodgers, S. M. Howitt, A. L.
Munn, G. S. H udson, T. A. H olten, P. R. Whitfeld,
W. Bottom ley, F. G ibson, and G. B. Cox, Biochim.
Biophys. Acta 1015, 1 9 5 -1 9 9 (1990).
[26] D. F. Feng and R. F. D oolittle, J. M ol. Evol. 25,
3 5 1 -3 6 0 (1 9 8 7 ).
[27] D. G. Higgins and P. M. Sharp, Gene 73, 2 3 7 -2 4 4
(1988).
[28] F. Corpet, N ucl. Acids Res. 16, 1 0 8 8 1 -1 0 8 9 0
(1988).
[29] G. Deleage, F. F. Clerc, B. R oux, and D. C. G au­
theron, CABIO S 4, 351 - 3 5 6 (1988).
Expert technical assistance o f M artina Blomeier
and preparation o f Fig.l by Hella Kenneweg are
gratefully acknowledged. This work was support­
ed by a grant o f the Deutsche Forschungsgemein­
schaft (SFB 171/B3). Thanks are also due to Dr.
Gilbert Deleage for his expert contribution to the
prediction analysis.
Unauthenticated
Download Date | 6/16/17 5:22 PM