|
|
|
|||||||
|
|||||||||
(Received for publication, November
20, 1995; and in revised form, February 21, 1996) From the
The architecturally complex secretory granules of Paramecium, known as trichocysts, have two unusual and
seemingly contradictory features: their protein contents have
crystalline organization (Sperling, L., Tardieu, A., and
Gulik-Krzywicki, T.(1987) J. Cell Biol. 105, 1649-1662),
yet these proteins are a heterogeneous set of molecules encoded by a
large multigene family (Madeddu, L., Gautier, M.-C.,
Vayssié, L., Houari, A., and Sperling, L.(1995) Mol. Biol. Cell 6, 649-659). We present here the first
complete sequences of three genes coding for three different precursors
of the trichocyst crystalline matrix proteins. The deduced protein
sequences indicate that each precursor gives rise to two of the mature
polypeptides found in the crystalline trichocyst matrix. Analysis of
putative processing sites suggests that a series of reactions, some of
which may involve a novel endopeptidase, are involved in their
proteolytic maturation. Each of the 6 mature polypeptides contains
heptad segments. Characterization of the heptad segments leads us to
propose that the mature polypeptides that compose the crystalline
trichocyst matrix, despite their different primary structures, all
share a unique protein fold, probably a 4
Regulated secretion provides a means of communication between
cells by allowing storage of biologically active molecules in
specialized organelles for subsequent release in response to
extracellular stimuli. Regulated secretion is restricted to a few
differentiated cell types in multicellular organisms (e.g. neurons, endocrine, and exocrine cells) and is absent from most
microorganisms such as yeast. However, Paramecium, like other
ciliates(1) , does possess regulated secretion, which seems to
be involved in defense against certain predators(2) . Each Paramecium bears around 1000 secretory granules, known as
trichocysts, docked at specialized exocytotic sites in the plasma
membrane, ready for rapid secretion in response to an appropriate
stimulus (see (3) , for review). Genetic studies of secretory
mutants have identified more than 20 genes involved in different steps
of this secretory pathway: organelle biogenesis, transport and docking
at the plasma membrane, and
exocytosis(3, 4, 5, 6) . Our
interest lies in using the Paramecium model to study the
biogenesis of secretory granules, which involves the coordination of
membrane traffic events with cargo protein processing, sorting, and
condensation (see (7) , for review). Given their architectural
complexity, Paramecium trichocysts also present an intriguing
problem of morphogenesis and molecular design. Trichocysts are
3-4 µm in size and consist of a carrot-shaped body surmounted
by a tip by which they attach to the cortical docking sites (Fig. 1a). This shape, which is genetically
determined(6) , is a property of the protein contents of the
granules, which has crystalline organization (Fig. 1b).
Indeed, structural studies have shown that the trichocyst contents is a
true protein crystal with periodicities in all three dimensions at low
resolution (
Figure 1:
Trichocyst shape and crystalline
organization. The panels on the left show phase-contrast light
microscope images of the carrot-shaped intracellular (a) and
needle-shaped extracellular (c) trichocysts. Note that despite
variation in size, all the trichocysts have the same shape. The bar represents 5 µm. The corresponding panels on the right show freeze-fracture electron microscope images (see (8) )
of unfixed trichocysts of either the compact intracellular (b)
or the extended extracellular (d) form and illustrate their
crystalline organization. The same bar represents 500 Å. The
electron micrographs are courtesy of T.
Gulik-Krzywicki.
The
proteins that assemble into the crystalline trichocyst matrix (trichocyst matrix proteins, TMPs) ( Perhaps the most
unusual feature of the system is that the mature TMPs which compose the
crystalline matrix are a heterogeneous set of immunologically related,
small acidic polypeptides: at least 30 major and as many as 100
different spots are revealed by high resolution two-dimensional gel
electrophoresis of purified trichocysts(12, 13) .
Moreover, this heterogeneity appears to be situated entirely at the
level of primary structure. Analysis of PCR-generated gene fragments,
corresponding to N-terminal microsequences obtained for several mature
TMP polypeptides, showed that TMP heterogeneity is the result of
expression of a large multigene family(14) . The greatest
challenge in understanding trichocyst design is therefore to relate the
disorder at the molecular level ( We set out to clone complete TMP
genes in order to gain insight into trichocyst design as well as to
obtain information necessary for further study of the
post-translational processing that controls matrix assembly. We present
the first complete TMP sequences, corresponding to three different
precursor proteins. Analysis of the deduced amino acid sequences shows
that each precursor gives rise to two mature matrix polypeptides, and
that several different enzymatic reactions are likely to be involved in
their processing. The parts of the precursors corresponding to the
mature polypeptides contain heptad repeats. Characterization of the
heptad segments allows us to identify a probable unique protein fold
for the mature polypeptides and to propose a model of their arrangement
in the precursor molecules.
The library was screened
using T1, T2, and T4 subfamily specific
Figure 2:
TMP gene cloning. Restriction maps of the
Two-dimensional gel electrophoresis was carried
out as described previously(14) . Isoelectric focusing was
performed in 0.8% pH 3-10 and 1.2% pH 4-6.5 ampholines
(Pharmacia) in the presence of a chemical spacer (50 mM MOPS)
as described by Tindall(12) . The SDS-polyacrylamide gel
electrophoresis second dimension analysis was performed according to
Laemmli(19) , on 13% acrylamide gels.
Figure 3:
TMP gene sequences. Complete nucleotide
sequences of the genes (a) T1-b, (b) T2-c, and (c) T4-a are shown along with the deduced amino acid
sequences. Coding nucleotides are in upper case and noncoding
nucleotides, including the introns that interrupt each sequence, are in lower case. Note that in Paramecium, TAA and TAG code
for glutamine. GenBank accession numbers for the T1-b, T2-c, and T4-a
nucleotide sequences are U47115, U47116, and U47117,
respectively.
Initial evaluation of heptad segments
relied on the COILS program(27) ; further characterization was
by visual inspection. The alignment of the three proteins was used to
identify positions with apolar residues (Leu, Ile, Phe, Val, Met, Tyr,
and Ala) in all sequences, which defined likely heptad segments
(repetitions of heptads of the form abcdefg with apolar
residues at a and d)(28) . Homology
searches of the protein sequence data bases (combined non-redundant
GENBANK/EMBL/SWISSPROT) and of the SBASE protein domain data base (29) relied on the FASTA (30) and BLAST (31) algorithms; statistical significance of the results was
evaluated using the RDF2 program(32) .
In order to clone one member of each
of the three subfamilies under investigation (named T1, T2, and T4
after the original microsequences), we used subfamily specific probes
to screen a The sequences of the
three genes (which correspond to the T1-b, T2-c, and T4-a gene
fragments reported in (14) ) are shown in Fig. 3. The
DNA sequences of T1-b, T2-c, and T4-a reveal a single reading frame of
1224, 1192, and 1143 nucleotides, respectively, interrupted by 4, 1,
and 2 introns of 23 to 29 base pairs (the standard size in Paramecium) whose existence has been confirmed by reverse
transcriptase-PCR experiments ((14) , and results not shown).
The encoded proteins have lengths of 369, 387, and 363 amino acids,
compatible with the 40-45 kDa size of the trichocyst precursor
proteins(9, 10) . For one of the genes, T4-a, the
transcription start site was determined by primer extension (not
shown). As for other characterized Paramecium genes(36, 37, 38, 39) , the
untranslated leader sequence is very short: transcription begins only
13 nucleotides upstream of the ATG codon that initiates translation
(ATTAATAAAAAAAATG). The T1-b and T4-a genes contain a consensus
polyadenylation signal (AATAAA), found in some but not all ciliate
genes(40) , located some 85 nucleotides downstream of the TGA
stop codon (Fig. 2). The T2-c gene does not contain this
sequence. The predicted sizes of the mRNA molecules are consistent with
the 1.4 kb measured by Northern blot experiments(14) .
The minor discrepancies between the
composition determined experimentally and that calculated for T1-b,
T2-c, and T4-a may arise either from the fact that the complete protein
sequences, not just the regions corresponding to the mature
polypeptides (which account for approximately 75% of the amino acids;
see below), were used for each calculation or from the fact that the
three genes we have cloned constitute a random sampling of the 30 or
more mature polypeptides present in the trichocyst matrix. Nonetheless
the similarities outweigh the differences and speak for a close family
resemblance among the proteins that constitute the trichocyst matrix.
Figure 4:
Alignment of deduced TMP protein
sequences. The T1-b, T2-c, and T4-a deduced protein sequences were
aligned using the Clustal program. Identical amino acids are shown on a
black background and similar amino acids on a gray
background.
Figure 5:
Organization of TMP precursor molecules.
The same sequence alignment as in Fig. 4is shown. Above the
sequences, asterisks (*) denote positions with identical amino
acids in all sequences and dots (
To gain support for
the organization of the precursors suggested by the sequence data, we
performed pulse-chase experiments. We had previously investigated TMP
proteolytic maturation by pulse-chase experiments, using polyclonal
antibodies raised against the entire set of mature polypeptides, which
allowed us to obtain a global picture of the processing. The
immunoprecipitated polypeptides were analyzed on one-dimensional
SDS-polyacrylamide gel electrophoresis gels which revealed the
conversion of a family of 40-45-kDa precursors to a family of
15-20-kDa products(9) . Since antibodies that recognize
individual TMPs are not yet available, we performed pulse-chase
experiments using high resolution two-dimensional gels to analyze the
proteins immunoprecipitated by the polyclonal antibodies in order to
detect changes in isoelectric point as well as more subtle changes in
molecular mass. A typical experiment is shown in Fig. 6.
Figure 6:
Two-dimensional gel analysis of a
pulse-chase experiment. Wild type cells were labeled then collected and
lysed at the chase times indicated. TMP proteins were
immunoprecipitated from the lysates with polyclonal antibodies raised
against the complete set of mature TMP polypeptides and analyzed on
two-dimensional gels. The basic end of the first dimension isoelectric
focusing is on the left(-) and the acidic end (+)
is on the right. The previously determined pH range for the
isoelectric points of the mature (15-20 kDa) polypeptides is 5.5
to 4.7(12) . The different sets of spots discussed in the text
are annotated on the 30 min gel as follows: 1, precursors; 2, weak intermediates; 3, basic mature polypeptides; 4, acidic mature polypeptides.
First of all, the precursor molecules are on average more basic than
the mature polypeptides, and many of the mature polypeptides are more
acidic than any of the precursors, consistent with removal of the basic
region as postulated above. Second, at 20 min of chase the more basic
mature polypeptides are present, and their pattern changes little
between 20 and 40 min. In contrast, the more acidic spots are barely
visible at 20 min, and the acidic half of the pattern of mature
polypeptides evolves between 20 and 40 min of chase (but not
thereafter; data not shown). Finally, the gels present a few weak spots
of intermediate size, which progressively disappear as the more acidic
low molecular mass polypeptides appear. These data support the idea
that a temporally ordered series of reactions are involved in the
conversion of the precursors to the mature polypeptides, and clearly
demonstrate net loss of basic residues.
For each of the three proteins,
the COILS program(27) , an implementation of the algorithm of
Parry(44) , gave scores (>1.3 with a window of 28)
consistent with ability to form coiled-coils over most or all of the
regions that correspond to the mature polypeptides, i.e. the
unshaded portions of Fig. 5; neither the pro nor the basic
regions were predicted to form coiled-coils. The heptad, however, is a
motif typical not only of long rod-shaped Within globular proteins,
helices characterized by heptad repeats tend to pack in anti-parallel
bundles rather than forming extended rod-like domains(46) .
These heptad segments are shorter than those found in 2- and 3-stranded
coiled-coils, and are generally not well defined by the COILS
program(47) . ( As
shown in Fig. 5, each of the mature polypeptides contains 4
short segments of roughly 3-4 heptads, which have been labeled A,
B, C, and D. In some of the sequences, the heptad segments are
separated or flanked by proline residues, consistent with
Figure 7:
Proposed folding motif for TMPs. a, schematic drawing of a 4
We have presented the results of cloning and sequence
analysis of three genes coding for three different Paramecium secretory granule precursor proteins. Alignment of the three
protein sequences, which share only about 25% amino acid identity,
indicates common organization of the precursor molecules, each of which
gives rise to 2 mature polypeptides of the crystalline trichocyst
matrix. The organization deduced from the sequence data is supported by
two-dimensional pulse-chase experiments. Analysis of the aligned
sequences, characterized by heptad repeats, provides a picture of TMP
structure: the basic fold of all the mature polypeptides is very likely
a 4
Other
cleavage sites in these molecules are clearly not the targets of
subtilisin-like processing enzymes. The junction pro/first mature
polypeptide has a consensus TG-G/D. The sequence N-terminal to the
second mature polypeptide (VEAN-F for T1-b) is different but still not
a target for a (di)basic processing enzyme. Given the dearth of
knowledge of protozoan processing enzymes, we cannot exclude
involvement of a novel endopeptidase in TMP maturation. The high
The
folding pattern we propose for TMPs is based on similar identification
of the periodic disposition of apolar residues in the sequences,
consistent with coiled-coil packing of An example of a
large multigene family coding for proteins that share little sequence
identity but all fold into the same structure is provided by the
variant surface antigens of the African trypanosome, the agent of
sleeping sickness. A repertoire of some 1000 genes codes for the
surface proteins of the parasite. Although only one gene is expressed
at a time, expression can switch to a different antigen to escape the
host's immune response. X-ray crystallographic structure
determination has revealed that several antigens, with quite different
primary structures, have nearly identical tertiary structures,
suggesting that all of the variant surface antigens, representing some
1000 different sequences, correspond to a unique protein fold (54, 55, 56) . In the trypanosome example,
sequence variation of the surface antigen genes would have evolved to
fool the host's immune system. In our example, in which as many
as 30 different sequences may share a common protein fold, the
selective pressure for sequence diversification may be related to the
constrained trichocyst shape which, as genetic analysis has shown, is
necessary for successful
exocytosis(5, 6, 9, 11) .
Crystallization of the trichocyst matrix from a mixture of proteins
with the same structure but slightly different chemical properties
might allow formation of a gradient of crystallization within the
maturing granules (which would give the carrot shape), much as pH
gradients can be formed from mixtures of carrier ampholytes with
slightly different pI values(57) . We should be able to test
this idea, using specific antibodies and/or epitope tagged transgenes.
We know that T1 and T4 belong to the group of heat soluble proteins
since N-terminal sequences of mature T1 and T4 polypeptides were
determined after purification from the pool of heat soluble proteins (34) . This is consistent with the absence of cysteine residues
in the deduced T1 and T4 protein sequences. T2 belongs to the class
of disulfide-bonded dimers(13, 59) , and the precursor
contains exactly 2 cysteine residues. One of the residues is situated
between the helices A1 and B1 of the first mature polypeptide while the
other is situated between the helices C2 and D2 of the second mature
polypeptide. The unique disulfide bond that can be formed in the
precursor molecule would join the mature polypeptides that emanate from
T2, yielding, as expected(59) , a disulfide bonded heterodimer. The presence of a disulfide bond in T2 imposes a constraint on the
way in which the two Although an experimental structure
determination is of course necessary to test the model, we consider it
likely that the same arrangement of the helical bundles found in T2
will also hold for the T1 and T4 precursors, despite the absence of the
disulfide bond. It is tempting to suggest that this arrangement,
determined by the initial folding of the precursor polypeptide chain,
remains after protein processing and is a feature of TMP packing in the
crystalline trichocyst matrix. However, the arrangement might be
metastable, once the pro and basic regions of the precursor were
removed. Upon exocytosis, the H
The nucleotide
sequence(s) reported in this paper has been submitted to the
GenBank(TM)/EMBL Data Bank with accession number(s)
U47115[GenBank],
U47116[GenBank], and
U47117[GenBank].
Volume 271,
Number 17,
Issue of April 26, 1996 pp. 10247-10255
©1996 by The American Society for Biochemistry and Molecular Biology, Inc.
A UNIQUE PROTEIN FOLD FOR A FAMILY OF POLYPEPTIDES WITH DIFFERENT
PRIMARY STRUCTURES (*)
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
-helical antiparallel
bundle.
30 Å)(8) . Upon exocytosis, as soon as
the crystalline trichocyst matrix comes into contact with the
Ca
and H
O of the external medium (i.e. within a few milliseconds), it elongates by a factor of
8, much like a spring suddenly released from a confined space (Fig. 1c). This dramatic and irreversible structural
transition propels the trichocyst matrix out of the cell. The
extracellular needle-shaped form, which is also an ordered array (Fig. 1d), remains insoluble(8) .
)are synthesized as 40-45 kDa precursor molecules
that are converted to 15-20 kDa polypeptides by proteolytic
maturation(9, 10) . Only the mature polypeptides are
able to crystallize within the maturing vesicles, and studies of
secretory mutants unable to produce functional trichocysts strongly
suggest that the protein processing controls the crystallization
process (9, 10, 11) .Å), owing to the complex
mixture of polypeptides, to the emergence of periodic order at the
electron microscope level (nm) and of shape at the light
microscope level (µm).
Cells and Culture Conditions
The wild type Paramecium cells used in all experiments were Paramecium
tetraurelia strain d4-2(15) . Cells were grown at 27
°C in wheat grass powder (Pines International Co., Lawrence, KA),
inoculated with Enterobacter aerogenes and supplemented with
0.4 µg/ml 
-sitosterol(16) .Isolation and Characterization of Genomic Clones
A
library constructed in the BamHI site of 
EMBL3 with Sau3A partially digested Paramecium tetraurelia (strain d4-2) genomic DNA was kindly provided by Eric Meyer
(Laboratoire de Génétique
Moléculaire, Ecole Normale
Supérieure, Paris).
P-labeled probes
(290, 400, and 287 base pairs, respectively; Fig. 2), generated
by inverse or direct PCR as described previously(14) . Selected
clones were isolated according to standard techniques(17) .
Inserts containing the entire coding regions of the T1-b, T2-c, and
T4-a genes were identified by restriction digestion and Southern blot
analysis using as P-labeled probes the subfamily-specific,
PCR-generated DNA fragments used for library screening as well as
gene-specific oligonucleotides (given in Fig. 2, legend). For
T1, a second larger (477 base pairs) subfamily-specific probe was also
used; the probe was generated with the same oligonucleotide primers by
inverse PCR amplification as described(14) , using as template RsaI digested and religated genomic DNA (RsaI
circles). HindIII (T1-b; 1.5 and 1.8 kb), EcoRI
(T2-c; 5 kb), and XbaI (T4-a; 2.3 kb) phage restriction
fragments were subcloned into the appropriate restriction sites of the
pUC18 plasmid. DNA sequences were determined on both strands according
to the dideoxy nucleotide chain termination method with
sequence-specific oligonucleotide primers using a T7 sequencing kit
(Pharmacia, Uppsala, Sweden).
clones selected for subcloning and of the pUC18 subclones used
for sequencing are shown. The thick dotted lines under the
maps represent the different probes used to screen the
bacteriophage library and/or to select regions for subcloning. The boxes on the pUC18 subclones pmgh4, pmgh9, pm2431, and pmgx5
represent the regions that were sequenced, with the coding sequences
shaded. T1, probes 1 and 2, and T4, probe 1, are described under
``Experimental Procedures.'' The sequences of
oligonucleotides used as probes are as follows: T2, probe 1:
5`-GTCRGTCAAGGTTTAGAG-3` (antisense); T2, probe 2:
5`-TGGAAGACAGATATGTTG-3` (sense); T4, probe 2:
5`-GTATTTTCTATTTACTATTAATAACTAG-3` (sense).
Pulse-Chase Experiments, Immunoprecipitation, and
Two-dimensional Gel Analysis
Pulse-chase experiments and
immunoprecipitations were carried out as described
previously(9) . Briefly, log phase P. tetraurelia cells were fed with E. aerogenes bacteria
metabolically labeled with 
SO
(Amersham Int.,
United Kingdom) for 10 min at 27° C (800 cells/ml; 5 
10
labeled bacteria/ml). To start the chase, cells were
centrifuged and resuspended in medium supplemented with unlabeled
bacteria. Aliquots of the cultures were removed at the times indicated
and centrifuged. The cell pellets were lysed by injection into equal
volumes of hot (95 °C) 0.4% SDS, followed by rapid reconstitution
of immunoprecipitation buffer (150 mM NaCl, 1% Nonidet P-40,
0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 8; final
concentrations). Cell lysates were cleared by centrifugation, and
incubated with saturating amounts of a polyclonal antiserum which
recognizes and efficiently immunoprecipitates most of the mature
trichocyst matrix polypeptides as well as their precursor molecules (9) . Immunocomplexes were collected using protein A-Sepharose
beads (Pharmacia). After extensive washing, final pellets were
resuspended directly in isoelectric focusing sample
buffer(18) .Characterization of Introns
Reverse
transcriptase-PCR was performed using total Paramecium RNA
according to standard procedures, as described previously(14) .
The absence of a putative intron sequence from the mRNA was evaluated
by comparison of the size of PCR products obtained using either cDNA or
genomic DNA as PCR template.Primer Extension
Primer extension was carried out
according to the method described by Di Rago and Colson(20) .
Briefly, 20 pmol of oligonucleotide primer complementary to bases
+166 to +186 of the T4-a gene sequence (Fig. 3c) was 5` end-labeled with T4 kinase according
to standard procedures (17) and hybridized to total Paramecium RNA. The primer was extended by avian
myeloblastosis virus-reverse transcriptase (Pharmacia; 0.5 unit/µl)
and analyzed on a 6% polyacrylamide-urea gel with products of a
sequencing reaction carried out with the same primer.
Sequence Analysis
Sequence assembly and initial
characterization of the DNA and protein sequences were performed using
the UWGCG sequence analysis package(21) . Signal sequences were
identified using the algorithm of von Heijne(22) . Protein
sequences were aligned using the Clustal program (23, 24) run with default parameters. Protein
secondary structure was evaluated with UWGCG software based on the
method of Garnier et al.(25) , the Heidelberg Profile
Network Prediction PHD(26) , and the Heidelberg Prediction of
Secondary Structural Content of Proteins from Their Amino Acid
Composition. ()
TMP Genes
The first two steps in cloning TMP
genes have been previously described. First, N-terminal microsequences
were obtained for several mature matrix
polypeptides(13, 33, 34) . Three of the
N-terminal sequences, chosen because they had been independently
determined in different laboratories, using different protein
purification procedures (reviewed in (35) ), were used to
design partially degenerate PCR primers. Genomic DNA fragments,
corresponding to the approximately 20 to 40 amino acids of the
N-terminal microsequences, were amplified. The PCR products were cloned
and sequenced and were also used as probes for genomic blot
experiments, revealing that a large multigene family codes for
TMPs(14) . The multigene family is organized in at least 10
subfamilies. Within each subfamily 4 to 8 genes, sharing 80-90%
identity at the nucleotide level, code for nearly identical proteins,
judging by analysis of gene fragments for 3 subfamilies. Genes
belonging to different subfamilies share only about 25% identity at the
amino acid level (see below). bacteriophage library of P. tetraurelia genomic DNA. 14 positive clones were isolated and characterized
for T1, 6 for T2, and 20 for T4. Maps of clones selected for
subcloning and pUC18 subclones, the probes used and the regions
sequenced are presented in Fig. 2. Restriction and Southern blot
analysis of the T1 clones identified 4 clones with overlapping
inserts corresponding to the same genomic region. The coding region of
the gene was contained within 2 adjacent HindIII fragments of
1.5 and 1.8 kb, which were subcloned in the pUC18 plasmid vector for
sequencing. Southern blot analysis of the T2 clones using
oligonucleotide probes specific for the T2-c gene (14) allowed
us to identify a single clone. A 5-kb EcoRI fragment
containing the coding region was subcloned in the pUC18 vector for
sequencing. Restriction and Southern blot analysis of the T4
clones revealed 7 overlapping clones. The T4-a coding region was
contained within an XbaI fragment of 2.3 kb, which was
subcloned in the pUC18 vector for sequencing.Analysis of the Protein Primary and Secondary
Structure
The amino acid composition of purified trichocyst
matrices was determined experimentally by Steers et
al.(41) , at a time when the heterogeneity of TMPs was not
suspected. Table 1compares the amino acid composition calculated
for the deduced T1-b, T2-c, and T4-a protein sequences with the earlier
experimental data. Interestingly, each of the three proteins has
essentially the same composition, which is very close to that
determined for the purified trichocyst matrices. The mature
polypeptides are acidic (pI 4.7-5.5) (12) and for each
precursor, 16% of the residues are acidic (Glu + Asp). The most
abundant amino acids in these proteins are, however, the apolar
residues alanine and leucine (Ala + Leu 
22%) while the least
abundant amino acid is cysteine. Glutamine is also quite abundant
(between 7 and 10%). We note that in Paramecium, TAA and TAG,
STOP codons in the universal genetic code, designate glutamine (42, 43) . In the TMP genes, over 50% of the glutamine
codons are TAA: use of the cloned genes to drive a bacterial expression
system will require changing 18, 25, and 23 TAA codons in T1-b, T2-c,
and T4-a, respectively.
TMP Organization
As mentioned in the Introduction,
TMPs are synthesized as 40-45 kDa precursor molecules which are
converted in the course of trichocyst biogenesis to 15-20 kDa
mature polypeptides; only the mature polypeptides participate in the
crystalline trichocyst matrix(9) . An alignment of the T1-b,
T2-c, and T4-a deduced protein sequences, which are of the size
expected to code for the precursors, is shown in Fig. 4. These
sequences share 25% amino acid identity and 45% amino acid
similarity. In Fig. 5, the experimentally determined N-terminal
amino acid sequences of the mature polypeptides are shown in white
letters on a black background. The positions of these N-terminal
sequences indicate that all three precursor molecules have the same
organization. Each precursor consists of a hydrophobic signal sequence
(dark shading) separated from the first mature polypeptide by a
pro-sequence (light shading). Note that the N-terminal sequence of the
first mature polypeptide has been determined experimentally for each of
the three molecules. We do not know where the first mature polypeptide
ends, but for one of the three precursors, T1-b, we do know where the
second mature polypeptide begins, thanks to the N-terminal
microsequence (FADQGAL . . . ). It is thus likely that each precursor
gives rise to 2 mature polypeptides; further evidence for this comes
from analysis of the secondary structure (see below). The shaded
region, the most basic part of each protein, separates the two mature
polypeptides. Given the size and isoelectric points of the mature
polypeptides(12) , this basic region is likely to be partially
or totally removed by protein processing.
), positions with
similar amino acids in all sequences. The processing of N-terminal
microsequences determined for mature polypeptides are shown in white
letters on a black background (cf.(35) for a review
of the microsequencing). Putative signal peptides are darkly shaded.
The pro-regions separating the signal peptides from the first mature
polypeptides are lightly shaded. The basic regions postulated to
separate the two mature polypeptides of each precursor are also shaded,
and basic residues within the region are in bold-face type.
The structural motifs of the heptad segments are underlined and the heptads are given in italics below the sequences.
Apolar residues in the a and d positions of the
heptads are in bold-face type. Each segment has been labeled
``helix A, B, C, or D'' in accordance with the proposition
that each group of heptad segments forms an -helical bundle. We
note that the D2 helices for T2 and T4 are less satisfactory than the
other helices. They can be improved if we remove the constraint of
perfect alignment of the heptads among the three sequences and situate
them closer to the C terminus. The 2 cysteine residues in the T2-c
sequence are in bold-face type.
Several Processing Reactions Involved in TMP
Maturation
Examination of the putative cleavage sites on
substrate molecules can in theory help identify the enzymes responsible
for the cleavage. In the TMP sequences, the junction between the
pro-region and the first mature polypeptide, whose position is
absolutely certain for all three proteins, presents a short consensus
with TG on the N-terminal side of the cleavage (i.e. in the
P2-P1 positions) and either Gly or Asp on the C-terminal side (i.e. in the P1` position). As far as we are aware, this sequence does
not correspond to a known endopeptidase cleavage site, moreover
cleavage at a glycine residue is quite unusual. The sequence preceding
the second mature polypeptide is yet different. No consensus of even
two amino acids appears in this region, and there is no TG. The
endopeptidase that removes the pro-sequence from the first mature
polypeptide is thus not likely to be involved in liberating the second
mature polypeptide, and it moreover seems probable that several
distinct processing reactions are necessary to remove the pro and basic
regions and liberate mature TMP polypeptides.
Mature Polypeptides Contain Heptad Repeats
T1-b,
T2-c, and T4-a protein sequences were used to search the protein data
bases. No statistically significant similarity with known proteins was
found. The highest scores for each of the proteins were with the rod
portions of myosins, keratins, and intermediate filament proteins.
Different methods for the prediction of protein secondary structure all
indicate that the trichocyst proteins have very high -helical
content. We therefore looked for heptad repeats (repetitions of 7 amino
acids, with apolar residues in the first and fourth
positions(28) ), which are indicative of coiled-coil
interactions between -helices.-fibrous proteins, but
also of globular proteins containing bundles of -helices. In the
case of the TMPs, the sequence data are more consistent with globular
proteins than fibrous ones. The charged to apolar residue ratios for
the complete T1-b, T2-c, and T4-a sequences are 0.55, 0.73, and 0.74,
respectively; 2-stranded -fibrous proteins, for example, have
ratios greater than 1.0(45) .
)We therefore took advantage of the
homology of the three TMP sequences to look for heptad repeats whose
positions are conserved with respect to the sequence alignment.-turns:
before helix A1, between helices B1 and C1, and after helix D1 and
helix D2. Charged residues in positions e and g,
which stabilize parallel coiled-coils in -fibrous proteins, are
not notably present: this is consistent with a bundle of -helices
in a globular protein. Finally and most important, the percentage of
apolar residues in positions a and d of the heptad
segments is high. These features taken together argue strongly that the
major portion of each mature polypeptide is a 4 -helical bundle (Fig. 7a).
-helical antiparallel bundle
(modified from (46) ). The bundles have a left-handed tilt
necessary to optimize coiled-coil packing of the helices. The chain
connectivity has arbitrarily been drawn as right-handed. This is the
basic protein fold proposed for each mature TMP polypeptide. Labeling
of the helices corresponds to that in Fig. 5. b,
possible arrangement of two helical bundles in the disulfide-bonded T2
precursor molecule. c, end-on view of the same arrangement
showing that the connectivity is the same for each bundle and for the
precursor molecule as a whole. Thick lines, top; thin and dotted lines, bottom.
-helical bundle motif.TMP Processing: Novel Enzymes?
By far the most commonly
found processing enzymes in the regulated (and constitutive) secretory
pathways of metazoa and fungi are serine proteinases belonging to the
bacterial subtilisin superfamily known as kexins or prohormone
convertases(48, 49) . These enzymes cleave their
substrates at dibasic (or more rarely, monobasic, tribasic, or
tetrabasic) sites. In the basic region separating the two TMP mature
polypeptides, all three precursor sequences contain basic residues, and
T1 and T2 contain pairs of basic residues. This region could
potentially be cleaved by subtilisin-like processing enzymes, thus
accounting for the conversion of the precursors to 15-20 kDa
products characterized by pulse-chase experiments(9) .-helical content of TMPs may provide a clue. It has been shown
that the magainin peptides of Xenopus skin, which are toxic to
many microorganisms because of their pore-forming capacity, are
processed by a metalloendopeptidase that recognizes -helical
secondary structure(50) . Since it is probable that all TMP
precursor proteins have similar three-dimensional structures despite
different amino acid sequences, an endopeptidase designed for
structural rather than sequence specificity might be an efficient and
economical adaptation to the problem of trichocyst biogenesis.Many Sequences, One Structural Motif
Although much
progress has been made in the prediction of protein secondary structure
from primary sequence data, especially in cases where there is homology
with proteins of known structure(51) , it is not yet common to
be able to infer the folding pattern of a protein simply from its
sequence. -Fibrous proteins that form coiled-coils are a notable
exception (47) . The ``knobs into holes'' coiled-coil
packing (52) has long been recognized to impose regularities on
the primary sequence (apolar and charged residue periodicities) that
can be appreciated by Fourier transformation(53) .-helices(46) . Two
arguments add confidence to our assignment of a 4 -helical bundle
as the basic folding pattern for each of the mature TMP polypeptides.
First, the apolar residue periodicity is manifest in the alignment of
the three protein sequences and indeed accounts for much of the
similarity shared by the three sequences; second, the six different
mature polypeptides all contain similar arrangements of heptad
segments. Although a particular example of such a bundle has been drawn
in Fig. 7for the sake of illustration, in the absence of
structural data we are unable to specify details of the fold, for
example, the handedness of the chain connectivity.Arrangement of the
Steers et al.(41) , who first
characterized TMPs, showed that these proteins are present in the
trichocyst matrix as disulfide bonded dimers. Using monoclonal
antibodies that recognize defined subsets of precursors and mature
polypeptides, Shih and Nelson (58, 59) were able to
demonstrate that most mature TMPs are disulfide-bonded heterodimers,
and that the corresponding precursor molecules contain intramolecular
disulfide bonds as judged by electrophoresis in the presence or absence
of reducing agents. Some of the mature TMPs are present as monomers,
and there are no disulfide bonds in the corresponding precursor
molecules. This latter group corresponds to the proteins that can be
solubilized from extended trichocyst matrices by heating(60) . -Helical Bundles in the Precursor
Molecules-helical bundle motifs can be arranged in the
precursor molecule. The loops connecting helices A1 and B1 of the first
mature polypeptide and helices C2 and D2 of the second mature
polypeptide are pinned together by the disulfide bond. Another
constraint comes from polypeptide chain continuity: the basic region
connects helix D1 of the first bundle with helix A2 of the second
bundle. An arrangement which accommodates both constraints is presented
in Fig. 7, b and c. We propose that in the
precursor molecules, the two helical bundles face each other, related
by a pseudo 2-fold symmetry axis.0 and Ca of the external medium would trigger the rearrangement of the
helical bundles into a thermodynamically more stable array, accounting
for the irreversible transition to the needle-shaped extracellular
form.
)))
We thank Carl Creutz, Roger Karess, and Roberto
Bruzzone for critical reading of the manuscript and Janine Beisson,
Jean Cohen, and Vittorio Luzzati for many useful discussions. We are
particularly indebted to David A. D. Parry for his kind advice on
analysis of heptad repeats and for pointing out to us the likelihood
that TMPs contain 4 -helical bundle motifs.
©1996 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E.-M. Ladenburger, I. M. Sehring, I. Korn, and H. Plattner Novel Types of Ca2+ Release Channels Participate in the Secretory Cycle of Paramecium Cells Mol. Cell. Biol., July 1, 2009; 29(13): 3605 - 3622. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. T. Cowan, G. R. Bowman, K. F. Edwards, J. J. Emerson, and A. P. Turkewitz Genetic, Genomic, and Functional Analysis of the Granule Lattice Proteins in Tetrahymena Secretory Granules Mol. Biol. Cell, September 1, 2005; 16(9): 4046 - 4060. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Wassmer, M. Froissard, H. Plattner, R. Kissmehl, and J. Cohen The vacuolar proton-ATPase plays a major role in several membrane-bounded organelles in Paramecium J. Cell Sci., July 1, 2005; 118(13): 2813 - 2825. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. R. Bradshaw, N. D. Chilcoat, J. W. Verbsky, and A. P. Turkewitz Proprotein Processing within Secretory Dense Core Granules of Tetrahymena thermophila J. Biol. Chem., January 31, 2003; 278(6): 4087 - 4095. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Haddad, G. R. Bowman, and A. P. Turkewitz New Class of Cargo Protein in Tetrahymena thermophila Dense Core Secretory Granules Eukaryot. Cell, August 1, 2002; 1(4): 583 - 593. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. D. Chilcoat, N. C. Elde, and A. P. Turkewitz An antisense approach to phenotype-based gene cloning in Tetrahymena PNAS, June 28, 2001; (2001) 151243498. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L Vayssie, N Garreau de Loubresse, and L Sperling Growth and form of secretory granules involves stepwise assembly but not differential sorting of a family of secretory proteins in Paramecium J. Cell Sci., January 3, 2001; 114(5): 875 - 886. [Abstract] [PDF]
|
|
|
|
|
|
F. Ruiz, L. Vayssié, C. Klotz, L. Sperling, and L. Madeddu Homology-dependent Gene Silencing in Paramecium Mol. Biol. Cell, April 1, 1998; 9(4): 931 - 943. [Abstract] [Full Text]
|
|
|
|
|
|
J. W. Verbsky and A. P. Turkewitz Proteolytic Processing and Ca2+-binding Activity of Dense-Core Vesicle Polypeptides in Tetrahymena Mol. Biol. Cell, February 1, 1998; 9(2): 497 - 511. [Abstract] [Full Text]
|
|
|
|
|
|
N. D. Chilcoat, N. C. Elde, and A. P. Turkewitz An antisense approach to phenotype-based gene cloning in Tetrahymena PNAS, July 17, 2001; 98(15): 8709 - 8713. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Molecular and Cellular Proteomics |
| Journal of Lipid Research | ASBMB Today |