Published online before print
July 17, 2003, 10.1101/gr.1555203
Genome Res. 13:1787-1799, 2003
©2003 by Cold Spring Harbor Laboratory Press; ISSN 1088-9051/03 $5.00
Integrated Mapping, Chromosomal Sequencing and Sequence Analysis of Cryptosporidium parvum
Alan T. Bankier1,
Helen F. Spriggs1,
Berthold Fartmann2,
Bernard A. Konfortov1,
Martin Madera1,
Christine Vogel1,
Sarah A. Teichmann1,
Al Ivens3 and
Paul H. Dear1,4
1 Medical Research Council (MRC) Laboratory of Molecular Biology, Cambridge
CB 2 2QH, UK
2 MWG Biotech, D-85560 Ebersberg, Germany
3 The Wellcome Trust Sanger Institute, The Wellcome Trust Genome Campus,
Hinxton, Cambridge CB10 1SA, UK
 |
ABSTRACT
|
|---|
The apicomplexan Cryptosporidium parvum is one of the most
prevalent protozoan parasites of humans. We report the physical mapping of the
genome of the Iowa isolate, sequencing and analysis of chromosome 6, and
 0.9 Mbp of sequence sampled from the remainder of the genome. To
construct a robust physical map, we devised a novel and general strategy,
enabling accurate placement of clones regardless of clone artefacts. Analysis
reveals a compact genome, unusually rich in membrane proteins. As in
Plasmodium falciparum, the mean size of the predicted proteins is
larger than that in other sequenced eukaryotes. We find several predicted
proteins of interest as potential therapeutic targets, including one
exhibiting similarity to the chloroquine resistance protein of
Plasmodium. Coding sequence analysis argues against the conventional
phylogenetic position of Cryptosporidium and supports an earlier
suggestion that this genus arose from an early branching within the
Apicomplexa. In agreement with this, we find no significant synteny and
surprisingly little protein similarity with Plasmodium. Finally, we
find two unusual and abundant repeats throughout the genome. Among sequenced
genomes, one motif is abundant only in C. parvum, whereas the other
is shared with (but has previously gone unnoticed in) all known genomes of the
Coccidia and Haemosporida. These motifs appear to be unique in their
structure, distribution and sequences.
Cryptosporidium parvum is an apicomplexan parasite, conventionally
placed in the class Coccidia, order Eimeriida, family Cryptosporidiidae. This
is in the same order as the parasites Toxoplasma, Sarcocystis, and
Eimeria, and in the same phylum (Apicomplexa) as Plasmodium,
Babesia, and Theileria
(http://www.ncbi.nlm.nih.gov/Taxonomy/taxonomyhome.html/ ).
Speciation within the genus is still controversial but ten species, infecting
a range of vertebrate hosts, have been named
(Chappell and Okhuysen 2002).
On the basis of a large number of polymorphic loci, at least eight genotypes
of C. parvum have been identified
(Sulaiman et al. 2002;
Xiao et al. 2002), of which
genotypes 1 and 2 are responsible for most human infections
(Guyot et al. 2001;
McLauchlin et al. 2000).
Genotype 1 is confined almost exclusively to humans, whereas genotype 2
infects a range of wild and domesticated mammals in addition to humans. The
taxonomic status of the genotypes is unclear, but C. parvum genotype
1 was recently renamed as Cryptosporidium hominis
(Morgan-Ryan et al. 2002). The
Iowa isolate used in this study was previously identified as genotype 2, based
on RFLP analysis (Carraway et al.
1997).
C. parvum is an obligate intracellular parasite, transmitted as
highly durable oocysts in feces. Ingested oocysts excyst in the ileum,
releasing sporozoites which infect the intestinal epithelium. Subsequent
development includes both a cyclic asexual reproduction and the production of
gametes giving rise to further oocysts, which are either excreted or reinfect
the host. Symptoms are mainly gastrointestinal (diarrhea, abdominal cramps,
vomiting), coupled with fever and headache. Human cryptosporidiosis was first
reported as recently as 1976 (Meisel et
al. 1976; Nime et al.
1976), but is now recognized as highly prevalent
(Casemore et al. 1997;
Fayer et al. 1997). Although
the parasite is normally eradicated from the gut after 1–2 weeks,
infection persists in immunosuppressed patients, and can be life-threatening.
There is currently no effective therapy. In addition to sporadic cases,
massive outbreaks occur periodically, typically through contamination of water
supplies or of food (e.g., Glaberman et
al. 2002; Howe et al.
2002; Rose et al.
2002).
Conventional approaches to genome sequencing—whether based on
physical mapping or on shotgun sequencing—are critically dependent upon
the fidelity and representation of the clone libraries on which they are
based. Dependence on a cloned resource both as a sequencing substrate and for
determining the long-range order of the sequence data makes these approaches
vulnerable to gaps, cloning artefacts, and large, well conserved repeat
elements. We therefore set out to develop an alternative strategy which
integrated complementary mapping techniques to overcome these drawbacks.
Electrophoretic analysis suggests that the C. parvum genome
contains eight chromosomes of between 0.9 and 1.5 Mbp, giving a total genome
size of 10.4 Mbp (Blunt et al.
1997; Caccio et al.
1998). We had already made
(Piper et al. 1998a) a
complete medium-resolution map of the genome of the Moredun isolate by HAPPY
mapping (Dear and Cook 1993;
Dear 1997). Also, a library of
PAC clones made from this isolate (Piper
et al. 1998b) showed that the majority of such clones were stable
in culture, suggesting that an integrated mapping/sequencing strategy which
combined the HAPPY and PAC-based approaches would be feasible. The strategy
which we devised for the present study uses HAPPY mapping to define the
positions of cloned fragments in the genome, before confirming their overlaps
to create a robust, `HAPPily anchored' physical map
(Fig. 1). This approach greatly
accelerates physical mapping and, by combining clone-independent (HAPPY) with
clone-dependent (overlap) data, is robust against artefacts and repeats. Where
gaps remain in the physical map, adjacent contigs are oriented by virtue of
the HAPPY data, aiding directed gap-closure.
View larger version (10K):
[in this window]
[in a new window]
|
Figure 1 Construction of HAPPily-anchored physical map. (A) Members of a
genomic PAC library are end-sequenced (filled rectangles). (B) One
end-sequence of each clone is HAPPY-mapped to establish its position in the
genome; additional sequence-tagged sites (triangles) are also mapped.
(C) Overlaps between nearby clones are determined, by using PCR to
test each clone for its content of nearby mapped markers, creating contigs.
Chimeric clones or deletions (zigzag line) become apparent. The orientations
of contigs which are separated by uncloned portions (heavy parallel diagonal
lines) are known from the HAPPY map, and additional linking clones (hatched
rectangle) can be sought by screening the same or other libraries with the
markers adjacent to the gap (dashed vertical lines).
|
|
Based on our experiences with Cryptosporidium, we feel that this
integrated approach overcomes some of the limitations of established
strategies, and may be relevant to the sequencing of other genomes.
|
RESULTS
|
|---|
Construction of a `HAPPily Anchored' Physical Map of the Genome
We first constructed a genomic PAC library of the Iowa isolate of C.
parvum (mean insert size 35 kbp). The end-sequences of 1172 clones,
chosen at random from this library, were determined, and the positions in the
genome of one end-sequence from each of 827 clones (as well as a further 269
STS markers, taken from previously published sequences) were determined using
HAPPY mapping. This is an in vitro technique which maps sequence-based markers
directly on native genomic DNA (see Methods).
To convert the HAPPY map into a robust physical map, each anchored clone
was screened by PCR for the presence of the seven markers which, according to
the HAPPY map, lay on either side of its mapped end-sequence
(Fig.1). This enabled the
direction and extent of overlaps between clones to be established, and the
local order of the mapped markers to be refined (e.g., where the marker
spacing was below the 5–10-kbp resolution limit of the HAPPY map).
Construction of the HAPPY map of the 10.4 Mbp genome
(Fig. 2) and its conversion to
a `HAPPily anchored' physical map took a total of about four person-weeks.
Finally, gaps in the genome-wide physical map were closed by screening the
remainder of the PAC library (and, in some cases, a BAC library; see Methods)
for the terminal markers of adjacent contigs. Only two regions on chromosome 6
remained intractable (i.e., physical gaps unspanned by any large clones;
Fig. 3). Although an
ApaI restriction fragment spanning both these gaps and the flanking
sequence was isolated, subclones of this fragment showed an extreme bias in
their distribution, most originating from the already known parts of the
fragment (data not shown). However, the minority of the subclones which did
fall within the gaps were HAPPY mapped at high resolution to produce dense
local maps of the unsequenced regions (see Methods).
View larger version (20K):
[in this window]
[in a new window]
|
Figure 3 C. parvum chromosome 6. The chromosome is shown in two halves
(heavy vertical lines). Start points of predicted coding sequences on the +
and - strands of the chromosome are indicated by short green horizontal lines
to the left and right, respectively (red: tRNA genes). Octamer palindrome
motifs are indicated by short black (TGCATGCA) and red (TGGCGCCA) bars,
respectively, polymorphisms by blue bars. Two gaps in the sequence are
indicated, and the STS markers mapped within them are shown in expanded form
(upper left). The A+T content (sliding window of 100 bp) of a
representative 50-kbp segment of the chromosome is shown in expanded form
(graph, lower left), aligned with protein-coding regions on the +
(green bars) and - (red bars) strands.
|
|
Comparison of this map with the previous, medium-resolution HAPPY map of
the Moredun isolate of C. parvum
(Piper et al. 1998a) did not
reveal any differences in marker order, beyond those expected given the
resolution of the earlier map.
Sequencing of Chromosome 6
A tiling path of 54 PAC clones (plus four BAC clones found during
gap-closure; see Methods) was selected, and each of these clones was sequenced
independently prior to alignment with consensus sequences from the overlapping
clones (see Methods). The mean coverage of the tiling path with large insert
clones was 1.6-fold. One gap was spanned by a previously sequenced PAC clone
from an earlier library of the Moredun isolate of C. parvum
(Piper et al. 1998b);
overlapping PCR primer pairs were designed at intervals to amplify the
corresponding region of the Iowa isolate, and these PCR products were
sequenced directly.
Gross Sequence Characteristics
The largely complete sequence of the chromosome comprises 1,162,753
basepairs, and includes two gaps which have been indicated in the EMBL
sequence submission (Accession no. BX526834
[GenBank]
) as blocks of 300 Ns. The total
size of these gaps is estimated at 159 kbp, based on restriction fragment
(ApaI) sizes; the local high-density HAPPY maps of the gap regions
(above) indicate that the second gap is <10 kb in size. One of the
telomeres was obtained (below), whereas restriction analysis indicates that
approximately 10 kbp of sequence is missing from the other telomere (data not
shown). Hence, the total length of the chromosome is predicted to be 1.33 Mbp.
This compares well with the previously published estimate of 1.44 Mbp
(Blunt et al. 1997) and with
our own PFG-based estimate (mean of several determinations) of 1.31 Mbp (data
not shown).
The overall G+C content of chromosome 6 is 31.1%, being approximately
uniform along the chromosome on a gross scale. Analysis of the 867 kbp of PAC
end-sequences scattered throughout the rest of the genome reveals a similar
figure (31.9%). Although this is comparable to that of another apicomplexan
parasite, Theileria annulata (32.9%), other apicomplexans display a
wide range of G+C contents (e.g., Plasmodium falciparum, 18.2%
[Gardner et al. 2002];
Eimeria tenella, 53.2%
[http://www.sanger.ac.uk/Projects/E_tenella/],
and Toxoplasma gondii, 53.6%;
[http://www.tigr.org/tdb/e2k1/tga1/]).
Centromere and Telomeres
The presumed centromeres in the most complete apicomplexan genome, that of
P. falciparum, are tracts of 2–3 kbp which are extremely
A+T-rich (>97% A+T) and contain imperfect tandem repeats
(Gardner et al. 2002). No
comparable feature is seen on C. parvum chromosome 6
(Fig. 3); it is possible that a
centromere lies within one of the gaps in our sequence, though the additional
STSs mapped within these gaps are not unusual in base composition.
A total of 17 nonredundant sequences containing more than two tandem copies
of the telomere repeat hexamer (CCTAAA) were obtained (Methods). The largest
telomere repeat motif contained 51 copies of the hexamer, interspersed with a
small number of imperfect or incomplete copies but, because the repeat-primer
used to obtain these fragments can prime at any point in a long tandem array
of telomere repeats, an upper limit on the number of repeats cannot be
set.
Only one sequence, cptel29, was assigned unambiguously to
chromosome 6 by digital blotting (Methods), and PCR was used to generate a
sequencing template linking this to the main PAC/BAC contig. None of the other
putative telomeric sequences was assigned experimentally to chromosome
termini, and clearly at least one must be an internal sequence or an
artefact.
Simple Tandem Repeats
Chromosome 6 was analyzed (P.H. Dear, unpubl. software) for the presence of
tandem repeats of significant lengths
(Table 1). Homopolymer tracts
of  10 bp are abundant but, strikingly, are almost exclusively poly-A or
poly-T—only four of the 236 tracts are poly-G or poly-C. Dinucleotide
repeats are also abundant (again biased strongly towards A/T sequences), with
progressively smaller numbers of repeats as the number of dinucleotide units
increases. The units of most of the longest repeats appear themselves to
contain imperfect trinucleotide repeats, implying reduplication of an earlier
repeat region. A broadly similar picture is seen in the 867 kbp of PAC
end-sequences from elsewhere in the genome.
View this table:
[in this window]
[in a new window]
|
Table 1. Tandem Repeat Motifs (Including Single-Base Tracts) in C.
parvum Chromosome 6 (1.163Mbp of sequence) and in the PAC End-Sequences
Lying in the Remainder of the Genome (867kbp of Sequence)
|
|
Hyperabundant Palindromic Octamer Motifs
We find two striking and highly repeated octamer motifs in the genome of
C. parvum. The first, TGGCGCCA, is palindromic; the second, TGCATGCA,
is doubly palindromic (i.e., a palindromic octamer composed of two repeats of
a palindromic tetramer). Both occur at very high frequencies compared to that
predicted for a random sequence of bases with the same overall nucleotide
composition. TGGCGCCA occurs 71 times on chromosome 6 (as opposed to an
expected frequency of around two, an overabundance of 35-fold), whereas
TGCATGCA occurs 166 times (expected frequency around nine, overabundance of
18-fold). Both motifs are present almost exclusively in the 20% of the
chromosome which is noncoding, and are scattered fairly uniformly along the
chromosome apart from a region of 300 kbp in which TGCATGCA is scarce and
TGGCGCCA is completely absent (Fig.
3). The bases flanking the motifs are not strongly conserved,
apart from a preference for A or T in the two bases on either side of both
types of octamer. Analysis of the 867 kbp of PAC end-sequences from the
remainder of the genome reveals a similar overabundance of these two octamer
sequences (data not shown).
These octamer motifs may also reflect the phylogeny of
Cryptosporidium and the other apicomplexans, and are considered in
this context below.
Gene Density, Number, and Organization
In the absence of a large data set of annotated C. parvum genes, a
variety of complementary methods was used for gene identification,
supplemented by extensive manual analysis (see Methods). All analyses were
integrated using the Artemis software package
(Rutherford et al. 2000).
A total of 474 protein-coding genes are predicted, giving a mean density of
one per 2.46 kbp. Only 122 of the genes (25.7%) have predicted
introns—delimited by the usual eukaryotic GT...AG motifs at either
end—and these have an average of 2.7 exons per gene. The mean size of
exons is 1277 bases, whereas that of introns is 154 bases; 78% of the sequence
appears to be coding (including introns). Introns and intergenic regions
generally have a higher AT content than protein-coding sequence, though the
difference is not clear or consistent enough to serve as a reliable predictor
of gene organization (Fig. 3).
Six loci contain potentially overlapping pairs of coding sequences. In
addition, eight tRNA genes are scattered across the chromosome.
One unusual feature of the predicted genes is that their mean coding length
(616 amino acids) is considerably greater than in other eukaryotic genomes
(e.g., about 470 amino acids for Saccharomyces cerevisiae or 475 for
Caenorhabditis elegans). Comparison reveals that in general, each
C. parvum sequence is of similar length to its orthologs in C.
elegans and S. cerevisiae, but that there is an apparent
underrepresentation of smaller genes and a relative overabundance of proteins
longer than 700 amino acids (Fig.
4). These larger proteins are not preferentially those with one or
multiple predicted transmembrane helices, nor do they appear to be
significantly biased in their amino acid composition or repetitive nature.
View larger version (16K):
[in this window]
[in a new window]
|
Figure 4 Distribution of protein lengths. Predicted lengths of proteins on
chromosome 6 of C. parvum and in the complete genomes of P.
falciparum and C. elegans were sorted into size-bins of 100
amino acids, and the proportion of proteins in each bin were plotted for each
species. Proteins longer than 2100 amino acids are not shown; the arrows on
the x-axis indicate the arithmetic mean length of all proteins in
each of the three species.
|
|
Comparisons with Previously Sequenced C. parvum Genes
Only 13 of the predicted genes on chromosome 6 have previously been
sequenced in any Cryptosporidium species, the majority being from
C. parvum. In most (10) cases, we find that our predicted protein
sequence matches perfectly either the only published counterpart, or one of
several sequences of a protein known to vary among isolates. In the remaining
three cases (histone deacetylase 56k.08, the sporozoite cysteine-rich protein
1MB.07, and the ATP-binding cassette 1MB.703), we find only one or two amino
acid differences (most of which are conservative) compared to the previously
published C. parvum sequences.
Structural and Functional Gene Classification
A range of tools was used to identify functional and structural motifs in
the predicted genes (see Methods). The domains in the predicted coding
sequences were characterized by matches against the hidden Markov models of
the Pfam (Bateman et al. 2002)
and SUPERFAMILY (release 1.59; Gough and
Chothia 2002) databases. The SUPERFAMILY database contains a
library of hidden Markov models for all domains of known three-dimensional
structure, as defined in the Structural Classification of Proteins (SCOP 1.59)
database (Lo Conte et al.
2000).
A total of 371 Pfam domains of 174 different types were identified in 206
(43%) of the 474 predicted proteins. SUPERFAMILY analysis found a total of 265
SCOP structural domains in 196 (41%) of the predicted proteins. In both the
Pfam and SCOP analyses, the majority of domain families are represented by
only one or two examples, whereas a small number of domain families are more
abundant. The predominant domains include P-loop-containing nucleotide
triphosphate hydrolases (41 instances), RNA-binding domains (22), and protein
kinase domains (11).
A small number of proteins contain uncommon or unique combinations of
domains. 1MB.635, a putative aminoacyl tRNA synthetase, contains a four-domain
combination to date seen only in Plasmodium spp. and S.
pombe, whereas 1MB.33 contains a combination of two domains seen only in
C. parvum.
The results of these analyses were integrated using Interpro
(Apweiler et al. 2000), and
this output was in turn used to assign GO (Gene Ontology) classification terms
(Ashburner et al. 2000) to the
predicted proteins. The distribution of functions and biological processes is
broadly similar to that seen in Plasmodium
(Fig. 5).
Of the 474 predicted proteins, only 215 have significant similarity to any
protein of known or inferred function. Therefore, no function can be inferred
for the remaining 259 (55%) of the predicted chromosome 6 proteins.
Genes of Potential Relevance to Infectivity, Diagnosis, or
Therapy
A variety of protein types are of interest for the identification and
characterization of Cryptosporidium, or as potential therapeutic
targets. Such genes include those predicted to encode membrane, extracellular,
or cell-surface proteins, transport proteins, and mitochondrial or plastid
proteins.
Of the 474 predicted genes, 118 were predicted (see Methods) to have at
least one transmembrane domain, with 26 containing more than six such domains.
Approximately half (52) of the putative membrane-spanning proteins also have
predicted signal peptides. In addition, the 11 genes which carry protein
kinase domains may warrant further investigation as potential regulators of
signal transduction or cell–cell interactions.
Thirteen of the predicted membrane-spanning proteins have sequence
similarities to known transport proteins from other species, including four
ABC transporters (genes 1MB.703, 1MB.800, 1MB.816, and 1MB.836). In addition,
we find a potential homolog (1MB.736) of the putative chloroquine resistance
transporter found in P. falciparum (Q9N623). Chloroquine has hitherto
been found to be only moderately effective as an anti-cryptosporidial agent in
in vitro tests (Armson et al.
1999), and the vulnerability of C. parvum to this drug
may be modulated by this gene, just as the Plasmodium homolog renders
chloroquine ineffective against some forms of malaria
(Su et al. 1997;
Nomura et al. 2001).
SUPERFAMILY analysis reveals 10 candidate extracellular or cell-surface
proteins carrying domains typically found in such proteins
(Table 2). These proteins could
be involved in parasite–host interactions, and hence may be of interest
for further experimental characterization.
We find several genes encoding oocyst wall proteins and other surface
antigens, known to vary in detail both between and within species of
Cryptosporidium (e.g., Spano et
al. 1997; McLauchlin et al.
2000; Akiyoshi et al.
2002). Predicted protein 1MB.63 encodes the previously known
sporozoite antigen glycoprotein GP15 (Q9U521). Gene 1MB.13 encodes a large
(3007 amino acids) protein, and is a perfect match for a known partial C.
parvum oocyst wall protein sequence (41 kD oocyst wall protein
[fragment], Q9U4U4). Gene 1MB.208 (1623 amino acids) differs by two amino
acids from a previously identified oocyst wall protein precursor (Q06550).
Another of the predicted genes (56k.19) has no matches to known genes, but
much of its sequence has significant identity (22%, over 1081 amino acids) to
the oocyst wall protein precursor Q06550 and, like the previous two predicted
wall proteins, also contains repetitive amino acid tracts. This gene therefore
may encode a new oocyst wall protein.
C. parvum had been reported to lack mitochondria
(Current 1989;
Tetley et al. 1998) though, in
contrast, there are reports of their presence in at least some developmental
stages (Riordan et al. 1999;
Beyer et al. 2000). As in an
earlier EST-based gene survey (Strong and
Nelson 2000), we find several genes which appear to encode
mitochondrial proteins. Predicted proteins 1MB.254 and 1MB.530 show convincing
sequence similarities to mitochondrial carrier proteins, but apparently lack a
signal for export to mitochondria
(http://mips.gsf.de/cgi-bin/proj/medgen/mitofilter;
Claros and Vincens 1996). It
is possible that the relevant signals in Cryptosporidium are unusual,
and hence are not detected by this software.
Also of interest are genes whose proteins may be associated with the
apicoplast, a relict plastid characteristic of the Apicomplexa, which has been
proposed as a vulnerable target for drugs against parasites of this phylum
(Fichera and Roos 1997). The
question of whether C. parvum possesses an apicoplast has been
debated (e.g., Blunt et al.
1997; Tetley et al.
1998; Zhu et al.
2000). In Plasmodium falciparum, 551 nuclear-encoded
proteins (about 10% of the total) were identified, with varying degrees of
confidence, as being potentially targeted to the apicoplast
(Gardner et al. 2002).
Forty-five of the 474 predicted proteins of C. parvum chromosome 6
were identified as potentially apicoplast-targeted, using an algorithm trained
on Plasmodium sequences (Zuegge
et al. 2001;
http://gecco.org.chemie.uni-frankfurt.de/pats/pats-index.ph).
However, in the absence of experimental evidence, these predictions must be
considered tentative.
Phylogeny and Comparison with Other Genomes
Among the apicomplexans, Eimeria and Toxoplasma both fall
within the Coccidia, whereas Plasmodium is within the Haemosporida
and Theileria is within the Piroplasmida. The gregarines (including
Monocystis and Gregarina spp.) are a fourth major group of
apicomplexans, believed to have diverged early from these other groups, and
for which little sequence data exist. Cryptosporidium has long been
placed in the Coccidia, but comparisons of small-subunit ribosomal RNA
(ssrRNA) and  -tubulin gene sequences suggested that it is more closely
related to the early-branching gregarines than to the other apicomplexans
(Carreno et al. 1999;
Leander et al. 2003).
On a gross level, the predicted proteins of C. parvum chromosome 6
reflect a greater similarity to other apicomplexans than to non-apicomplexan
species, though not by as great a margin as one might expect. For example,
based on a FASTA expectation value of e < 0.01, 237 (50%) of the predicted
proteins have matches to those in P. falciparum, whereas 184 (39%)
have matches to those of S. cerevisiae; both of these latter genomes
are fully sequenced, and all three genomes have comparable numbers of genes.
If the criteria for declaring matches are made more stringent, the number of
matches between species decreases and the number of matches to P.
falciparum becomes more similar to the number of matches to S.
cerevisiae.
To try to resolve relationships among the apicomplexa, we sought to compare
the protein sequences of those genes on C. parvum chromosome 6 with
their homologs in T. gondii, P. falciparum, and P. yoelii;
Schizosaccharomyces pombe was used as an outgroup. To identify
homologs of C. parvum proteins among the other available apicomplexan
protein sequences, the 474 predicted chromosome 6 proteins were searched
against the Non-Redundant Protein Database
(Jenuth 2000) with FASTA
(Pearson and Lipman 1988). In
addition to -tubulin (analyzed in earlier phylogenetic studies,
Leander et al. 2003), there
were four other proteins which had matches to both T. gondii and
Plasmodium sequences at expectation values lower than that for the
best S. pombe match, and hence were considered likely to be
orthologs. Putative orthologs were aligned using CLUSTAL W
(Thompson et al. 1994), and
the four alignments were concatenated. A phylogenetic tree was calculated from
the concatenated alignment using distances based on the PAM matrices
(Dayhoff et al. 1978) from
1833 positions in the alignments, and the neighbor-joining and bootstrapping
options in PHYLO_WIN (Galtier et al.
1996). As can be seen in Figure
6, Cryptosporidium is an outgroup to both the coccidian
Toxoplasma and the haemosporidan Plasmodium. Inclusion of
the -tubulin sequence in the calculation lead to an essentially
identical tree, as did calculations based on each of the five proteins
individually (data not shown). In all cases, the bootstrap support (a measure
of robustness of the tree) was 100%, providing strong additional evidence for
the deep-branching position of Cryptosporidium within the
Apicomplexa. Analysis of the same sequences using more sophisticated
algorithms (maximum likelihood analysis with heterogeneity corrections, and
probabilistic neighbor-joining algorithms;
Schmidt et al. 2002; see also
Teichmann and Mitchison 1999)
yielded an identical pattern of relationships between the species, again with
strong statistical support (data not shown).
View larger version (7K):
[in this window]
[in a new window]
|
Figure 6 Protein sequence-based phylogenetic tree. The tree was calculated from the
aligned and concatenated sequences of four proteins from the five species
indicated. Branch lengths are distances calculated using PAM matrices;
bootstrap values are indicated at nodes. The proteins used and (in brackets)
the gene name on C. parvum chromosome 6 and the GenBank identifiers
for their sequences from P. falciparum, P. yoelii yoelii, S. pombe,
and T. gondii, respectively, are as follows: protein disulphide
isomerase (56K11, 23612738, 23481103, 19113783, 14494995);
glyceraldehyde-3-phosphate dehydrogenase (1MB519, 23509820, 23491258,
19112028, 13377044); heat shock protein 60 (1MB751, 23507957, 23479768,
19113806, 5052052); and protein phosphatase 2b (1MB598, 23612977, 23489838,
19112970, 22535354).
|
|
The two octamer DNA motifs noted above as being highly abundant in the
C. parvum genome may also shed some light on the phylogenetic
relationships of the apicomplexans. As far as we are aware, such motifs have
not hitherto been noticed in any of the previously sequenced apicomplexan
genomes. Our analysis finds that the first such motif, TGCATGCA, is common not
only in C. parvum but also in the coccidians E. tenella and
T. gondii (being about 18-fold overrepresented in the first two
genomes, and about 11-fold in the third), somewhat less common in the
haemosporidian P. falciparum (about sevenfold overrepresented), and
rare in the piroplasmid T. annulata. The second motif, TGGCGCCA, is
abundant only in C. parvum, and occurs no more than expected by
chance in the other genomes. There were insufficient sequence data to analyze
representatives of the other apicomplexan group, the Gregarina.
To compare the overall character of these genomes, we made chaos plots
(Jeffrey 1990), which allow
the frequency of every possible octamer in a sequence to be visualized.
Because the genomes have very different G+C contents, we used a modified
program (P.H. Dear, unpubl.) which plots the frequency of each octamer
relative to the frequency expected based only on the G+C content of
the sequence. As can be seen in Figure
7, the two coccidians Toxoplasma and Eimeria
have broadly similar characters (once normalized for base composition),
whereas chromosome 6 of C. parvum is significantly different from
these and more similar to that of the piroplasmid Theileria; the
haemosporidian Plasmodium is significantly different again. A similar
picture is seen when considering only the coding or noncoding regions of the
genomes, or when examining the 867kbp of nonchromosome 6 sequence from C.
parvum (data not shown). Numerical comparison of the chaos plots (the
root-mean-square difference in relative frequencies of each octamer between
two genomes to be compared) supports the visual interpretation of the plots
(data not shown).
View larger version (124K):
[in this window]
[in a new window]
|
Figure 7 Normalized chaos plots for apicomplexan genomes. Each pixel represents the
frequency of a given octamer sequence in the genome, relative to the frequency
expected in a randomly ordered sequence with the same base composition as the
genome in question (log scale; green <10-6; grayscale black
through white =10-6 through 5; red >5). In each plot, the
octamers [G]8, [C]8, [A]8, and
[T]8 are represented at the top left, top right, bottom
left, and bottom right corners, respectively.
|
|
Finally, we looked for syntenic relationships between C. parvum
and P. falciparum by comparing the positions, in the two genomes, of
the 237 C. parvum chromosome 6 genes for which homologs with FASTA
expectation values of <0.01 could be identified in P. falciparum.
No substantial region of the C. parvum chromosome appears to have
homologs preferentially from any one region of the P. falciparum
genome. There were four instances of microsynteny (which we define as two
genes separated by four or fewer intervening genes in one genome, with their
orthologs also separated by four or fewer genes in the other), but this is not
appreciably more than would be expected by chance alone, if the gene order
were randomly scrambled between the two genomes.
SNPs and Other Polymorphisms Within the Iowa Isolate
Because it is not possible to initiate infection from a single sporozoite
of C. parvum, isolates of this species cannot be guaranteed to be
clonal. We therefore looked for possible polymorphisms within the Iowa isolate
by comparing the overlapping regions of independently sequenced PACs and BACs;
such overlaps cover approximately 57% of the chromosome 6 sequence. We found
39 clear differences between overlapping clones; in all such cases, the
consensus sequence for each of the overlapping large-insert clones was
unambiguous, thereby excluding the possibility of sequencing errors. Such
inter-clone differences could in principle arise as artefacts within the
clones, and indeed one (a 54-bp insertion/deletion) was confirmed by PCR as
being such an artefact. The remaining 38 differences were single-base
substitutions or small insertions/deletions (often in repeat tracts). Cloning
artefacts should occur more or less independently of the coding structure of
the cloned fragments, but we find that most (25/38) of them are confined to
the small part of the genome (20%) which is noncoding. The few found in
predicted coding regions exhibit a bias (relative to random mutation) in favor
of silent substitutions or insertions/deletions which do not cause
frame-shifts (Table 3). Only
one apparent polymorphism disrupts a predicted gene, causing a frame-shift in
the first of three exons of 1MB.764; this may be a true instance of a mutation
arising within the PAC clones, or might reflect a mis-prediction in the
start-point of the gene (such that the polymorphism, in a homopolymer tract,
is not actually within the gene). With this possible exception, we believe
that most of the differences represent true polymorphisms.
View this table:
[in this window]
[in a new window]
|
Table 3. DNA Sequence Polymorphisms Found in Regions of Overlap Between PAC
Clones of the Iowa Isolate of C. parvum
|
|
The polymorphisms we observe are distributed nonrandomly along the
chromosome (Fig. 3), most
falling in the region between 500 and 800 kb
(Fig. 3). This may reflect a
true bias in the distribution of polymorphisms along the chromosome or may be
because, in this 300-kb region, more of the overlaps between the clones
are fortuitously represented by PACs from two different genotypes within the
Iowa isolate.
|
DISCUSSION
|
|---|
Mapping and Sequencing Strategy
Our strategy was intended to overcome the limitation inherent in both
shotgun sequencing and conventional physical mapping: the dependence on the
integrity and representation of clone libraries for both local sequence data
and long-range contiguity. The initial HAPPY mapping enabled the extremely
rapid production of an anchored physical map which was robust against clone
artefacts, and allowed directed gap closure to be initiated at an early stage.
Of our two remaining gaps, one is quite large, but our attempts to create
small-insert subclones of the restriction fragment spanning these gaps suggest
that a shotgun sequencing strategy would still have left this region
fragmented and unordered. Overall, the strategy was successful; we would
expect it to offer an advantage in sequencing the many genomes in which
extreme biases in base composition or other features make the production of
well ordered sequence by conventional means technically challenging.
Our experience in mapping other genomes (e.g.,
Dear et al. 1998;
Konfortov et al. 2000)
suggests that this approach is completely scaleable. The number and sizes of
clones to be mapped, and the resolution with which their end-sequences are
HAPPY-mapped in the first instance, can all be varied over a wide range.
Sequence Analysis
Given the absence of a comprehensive set of gene models for C.
parvum, it is reasonable to ask how effective and accurate our gene
predictions are. The high density of predicted genes means that there is
little room on the chromosome for short `hidden' genes. The number of genes
predicted on the chromosome 6 sequence (10% of the genome) implies a
total gene number of around 4500–5000, comparable to that of P.
falciparum (5300; Gardner et al.
2002). This again suggests that large numbers of genes have not
been missed by the predictions. Given the absence of biological data,
unambiguous assignment of the starting methionine of each gene cannot be made;
in some cases, therefore, the true gene lengths or exon numbers may differ
from those predicted. One must also ask whether chromosome 6 is representative
of the remainder of the C. parvum genome. None of our analyses
(including comparison with 867 kbp of clone-end sequences from the
remainder of the genome) reveals any obvious differences in character between
chromosome 6 and the remainder, nor are gross variations seen between the
different chromosomes in the genomes of Plasmodium spp. or of most
other fully sequenced genomes. We therefore think it likely that chromosome 6
of C. parvum is fairly typical of the genome.
With a mean density of one predicted gene per 2.46 kb, chromosome 6 of
C. parvum is particularly gene-dense. In comparison, the P.
falciparum genome (approximately twice the size of that of C.
parvum) contains one predicted gene per 4.3 kb
(Gardner et al. 2002) and,
among the fully sequenced eukaryotes, only S. cerevisiae and
Encephalitozoon cuniculi have higher densities of predicted genes
(one per 2 kb and one per 1.5 kb, respectively;
Goffeau et al. 1996;
Katinka et al. 2001).
Analysis of the predicted genes on chromosome 6 reveals three unusual
features, which collectively suggest that the gene set of C. parvum
is different in character both from that of P. falciparum and from
those of most other eukaryotes. First, the average size of the predicted
proteins (616 amino acids, about 30% longer than yeasts or C.
elegans) is surprisingly large (Fig.
4). P. falciparum also has an apparent bias toward larger
proteins (mean size 700 amino acids), though at least some of this bias is due
to the presence of extensive nonglobular domains (simple sequence repeats) in
otherwise conventional P. falciparum proteins; this does not appear
to be the case in C. parvum. The larger C. parvum proteins
taken as a whole do not appear to have any distinguishing feature other than
their size (e.g., they are not preferentially membrane proteins; nor are they
more likely to have homologs in the Plasmodium genome). The
significance of this size bias, therefore, remains unclear. Second, a large
number (118; 25%) of the predicted proteins contain predicted transmembrane
domains, and 26 (5%) contain more than six such domains, typical of channels
or receptors. The corresponding proportions for P. falciparum are 18%
and only 0.8%, respectively. Third, a significantly lower proportion (43%) of
the predicted proteins have recognizable structural motifs (Pfam domains) than
is the case for most genomes (mean 55%, standard deviation 6% for all fully
sequenced genomes). A similar situation is seen in P. falciparum,
where only 37% of the predicted proteins have matches to known Pfam
domains.
The two abundant octamer palindromes (TGCATGCA and TGGCGCCA) are unlike
other repeat classes in their structure and distribution. They differ in
character both from simple tandem repeats and from more complex (e.g.,
transposon-related) elements, each of which arises and persists in genomes
through partly understood mechanisms regardless of selective advantage. The
TGCATGCA motif is common in the binding sites for homeotic proteins in some
Drosophila enhancers (Manak et
al. 1995), but it is not clear whether there is any functional
equivalence to the C. parvum octamer. Neither octamer has a
consistent positional relationship with coding sequences or other obvious
sequence features. The distribution and palindromic nature of these motifs
suggest a possible role in chromatin packaging or dynamics and, given their
prevalence in C. parvum, E. tenella, and T. gondii (all of
whose life cycles involve a stable dormant phase), we speculate that they may
be involved in stabilizing the DNA during prolonged dormancy.
The phylogeny of the apicomplexans remains controversial. Our analysis of
protein sequences strongly supports earlier suggestions
(Carreno et al. 1999;
Leander et al. 2003) based on
the comparison of -tubulin and rRNA sequences, that
Cryptosporidium diverged early during the evolution of the
apicomplexans, before the divergence of Plasmodium (a haemosporidian)
from Toxoplasma (a coccidian). There were insufficient sequences from
other apicomplexan groups to provide new data on the relative ages of other
divergences (e.g., between the gregarines and other apicomplexans). Though a
comparison based on a larger range of sequences would be desirable, the fact
that each of our protein alignments alone led to the same relationship as the
concatenated alignment of all four proteins (and that the same relationship
was inferred in previous studies based on -tubulin sequences;
Carreno et al. 1999) argues
that the proteins chosen are reasonably representative of the remainder. The
lack of synteny between C. parvum and P. falciparum also
argues for a distant relationship between these two species, though this
cannot be placed in context until extensive genome sequences become available
for other apicomplexans. A further hint at the remoteness of these two species
comes from the relatively low degree of similarity between their predicted
proteomes. The distribution of the TGCATGCA motif, conversely, suggests that
Cryptosporidium is more closely related to the Eimeriida (coccidians)
than to the Haemosporida or Piroplasmida, although lack of data on other
apicomplexans (particularly the gregarines) again obscures the complete
picture. On a more optimistic note, it is clear that protein sequences are
sufficiently conserved between the apicomplexans to retain a strong
phylogenetic signal, and hence the phylogeny of this group is likely to be
resolved once more sequences from other species become available for
comparison.
The profiles of normalized octamer frequencies (which are more similar to
those of Theileria than to the other apicomplexans considered) appear
to reflect concerted changes in genome architecture which consistently
accompany changes in G+C content, rather than phylogenetic relationships. This
raises the more general question of why even closely related genomes can
differ markedly in base composition, and how the overall character of the
genome—the normalized frequencies of dimers and longer
sequences—varies in concert.
|
METHODS
|
|---|
Genomic DNA and Large-Insert Cloning
Purified oocysts of C. parvum (Iowa) were obtained from the
Sterling Parasitology Laboratory. Genomic DNA was prepared in agarose strings
(2x108 oocysts/mL) as described
(Piper et al. 1998a).
Libraries of genomic DNA, size-selected after partial digestion with
Sau3A or EcoRI, were made in the vectors pCYPAC2
(Ioannou and de Jong 1996) or
pBACe3.6 (Frengen et al.1999),
respectively, in ElectroMAX DH10B Competent Cells (Life Technologies), using
essentially standard techniques (Piper et
al. 1998b).
Mapping
HAPPY mapping was performed essentially as described
(Piper et al. 1998a). Briefly,
10 µL of agarose-embedded DNA was melted in 1xPCR bufferII (Applied
Biosystems) at 65°C for 8 min, the DNA sheared by gentle inversion,
diluted 2300-fold in water, and 2-µL aliquots were dispensed into a 96-well
thermocycler plate (88 samples plus 8 x 2 µL water as negative
controls). Samples were preamplified by primer-extension pre-amplification
(PEP) as described (Zhang et al.
1992; Piper et al.
1998a) in a total volume of 5 µL. Each PEP product was diluted
to 150 µL with water, and 5-µL subfractions were dispensed into 30
replica mapping plates, overlaid with 30 µL of mineral oil, and stored at
-80°C until needed. PCR primers (forward, internal, and reverse) were
designed for one end-sequence of each PAC clone, plus additional sequences
taken from Piper et al.
(1998a) or from database
sequences, essentially as described (Piper
et al. 1998a; Konfortov et al.
2000). For mapping, the forward and reverse primers for between 96
and 200 markers were used in a Phase1 PCR reaction with one of the replica
mapping plates (total volume 10 µL per well, comprising 5 µL of diluted
PEP product, PCR Gold Buffer [Applied Biosystems], 2mM MgCl2, 0.2
µM each primer, 2U Taq Gold [Applied Biosystems], 200 µM each dNTP;
93°C x 9 min, then 25 cycles of 94°C x 20 sec, 55°C
x 30 sec, 72°C x 1 min). Phase 1 products were then diluted to
600 or 1200 µL, and 5 µL of each product was used in a Phase 2
heminested PCR reaction for each marker in turn (10 µL total volume
containing PCR Gold Buffer [Applied Biosystems], 1.5 mM MgCl2, 1
µM each of forward and reverse primer for the marker in question, 0.25 U
Taq Gold [Applied Biosystems], 200 µM each dNTP; 93°C x 9 min,
then 33 cycles of 94°C x 20 sec, 52°C x 30 sec, 72°C
x 1 min). PCR products were supplemented with 8 µL of loading dyes
(4x SyBr Green1, 15% w/v Ficoll 400, 0.1 mg/mL bromophenol blue),
resolved by brief electrophoresis, and imaged under UV illumination. Analysis
was as described (Dear 1997),
with markers being sorted into linkage groups at an LOD threshold of 6.
Linkage groups were assigned to chromosomes based on their content of
previously mapped markers (Piper et al.
1998a); assignment of some groups was verified by digital blotting
(below).
For conversion of the HAPPY map into a physical map, both PAC clones and
primers were rearrayed robotically according to their positional order defined
by the HAPPY map. Each PAC clone was screened using a standard single-phase
PCR for the presence of the seven markers which lay to either side of it
according to the HAPPY map. The marker content of each clone was then used to
deduce the clone overlaps (P.H. Dear, unpubl.).
Gap Closure
Gaps in the physical map were closed in the first instance by screening the
PAC and BAC libraries by PCR for further clones, using the HAPPY markers on
either side of the gap. In some cases, the end-sequences of clones extending
into the larger gaps were used to design new markers with which to rescreen
the libraries for clones, closing the gaps by a `walking' approach. Both of
the gaps which remained uncloseable by this approach on chromosome 6 were
found to lie in a 350-kbp ApaI restriction fragment; this fragment
was purified using standard procedures, digested with Sau3A, and
subfragments were cloned into pBluescriptIISK+ and M13mp19 and
sequenced using standard procedures. Sequences not matching the sequence of
the known part of the ApaI fragment were confirmed as originating
from the fragment by digital blotting (below), and were used to design further
HAPPY markers which were mapped as described above.
PAC and BAC Sequencing
Large-insert clone DNA (20 µg) was isolated using the double
acetate method
(http://www.genome.ou.edu/BAC_isoln_200ml_culture.html),
then sheared, concentrated, and desalted using standard protocols. DNA was
then end-repaired (30 min, 15°C, 100 µL reaction: 20 µg sheared DNA,
15 U T4 DNA polymerase, 10 U Klenow DNA polymerase [both MBI Fermentas,
Vilnius, Lithuania], 500 µM each dNTP, 10 µL Yellow Tango Buffer [MBI
Fermentas]), desalted, and tailed with an extra A residue (30 min, 50°C,
100 µL reaction: 20 µg sheared DNA, 50 µM each dCTP, dGTP, dTTP, 2mM
dATP, 20 U Taq polymerase [MBI Fermentas], 10 µL Yellow Tango buffer).
A-tailed DNA was then size-fractionated by electrophoresis, and the
1.0–1.5-kbp fraction was isolated and purified using essentially
standard methods before cloning into pGEM-T (Promega). Plasmid clones were
sequenced from both ends with standard primers using the Big Dye terminator
chemistry on ABI 3700 capillary sequencers (Applied Biosystems), to give a
mean coverage of approximately eightfold for each PAC or BAC. The Gap4 program
(Bonfield et al. 1995) was used
first to assemble complete BAC and PAC clone sequences, and then to assemble
contigs of these.
Telomere Cloning
Terminal EcoRI and BamHI restriction fragments carrying
telomere sequences from the whole genome were obtained using a combination of
linker-ligation and telomere-repeat-specific PCR, followed by cloning in
pPCR-Script (Stratagene), and identification of telomere-containing clones by
probing with a telomere-repeat-specific oligonucleotide
([AAACCT]5); positive clones were sequenced by standard methods.
For each putative telomere, PCR primers were designed against the unique
(nontelomere-repeat) part of the sequence and used to assign the sequence to
one of the five chromosomal bands resolvable by Pulsed Field Gel
Electrophoresis (PFGE), by digital blotting (below).
Digital Blotting
Digital blotting was devised as an alternative to Southern blotting for
assigning sequences to discrete DNA bands resolved by PFGE. Briefly,
chromosomes or restriction fragments are resolved by PFGE in low-melting-point
agarose; each band is excised and melted in an equal volume of water
(typically 10 ml for bands containing 100–500 ng of DNA), and 12 serial
threefold dilutions are made for each. Five microliters of each dilution of
each band are then screened using standard PCR conditions for the STS to be
assigned. Typically, all bands give a positive result at their highest
concentration (because PFGE does not perfectly resolve the DNA, and hence each
band is contaminated with material from each of the others) but, over
successive threefold dilutions, the amount of product decreases; the band in
which the STS lies continues to give a strong product for two or three more
dilutions than the other bands.
Sequence Analysis
All analyses were integrated in the Artemis software package
(Rutherford et al. 2000).
Several complementary methods of gene prediction were used, including
BLASTX analysis against the SWALL (Swiss-Prot + TrEMBL) protein database,
mapping of the 567 EST sequences to the genome sequence, and application of
the GeneID prediction software, using a training set for Dictyostelium
discoideum (which has a similar G+C genome content) and a minimum gene
length of 120 coding bases. Other gene prediction tools (e.g., HMMGENE and
Genefinder) were also tested with the same training set, but were found to be
`correct' for a lower proportion of predicted genes.
Fasta analyses of the predicted genes were performed; first-pass assignment
of a function to the predicted CDS was primarily based on the extent and
degree of Fasta similarity (>40% identity and opt score >100 OR >20%
percent identity and opt score >200); those exhibiting weaker similarities
were initially classified as hypothetical predicted proteins. Gene models and
putative functions were refined manually for each CDS, based on a close
inspection of the similarity data and/or domain information (e.g., Pfam).
Predicted transmembrane domains and signal peptides were identified using
TMHMM (Sonnhammer et al. 1998)
and SignalP (Nielsen et al.
1997), respectively.
|
Acknowledgements
|
|---|
We thank J. Pachebat and J. McLauchlin for helpful comments in the
preparation of this manuscript, G. Nyakatura for advice on sequencing, and J.
Gough for providing SUPERFAMILY assignments to our proteins.
The publication costs of this article were defrayed in part by payment of
page charges. This article must therefore be hereby marked
"advertisement" in accordance with 18 USC section 1734 solely to
indicate this fact.
|
Footnotes
|
|---|
Article and publication are at
http://www.genome.org/cgi/doi/10.1101/gr.1555203.
4 Corresponding author. E-MAIL
phd{at}mrc-lmb.cam.ac.uk;
FAX 44 1223 412178. 
[Supplemental material is available online at www.genome.org. The sequence
data from this study have been submitted to EMBL. The sequence of Chromosome 6
appears under accession number BX526834
[GenBank]
; the end-sequences of the PAC clones
appear under accession numbers AJ561222
[GenBank]
–AJ563278.]
Article published online before print in July 2003.
|
REFERENCES
|
|---|
Akiyoshi, D.E., Feng, X., Buckholt, M.A., Widmer, G., and Tzipori,
S. 2002. Genetic analysis of a Cryptosporidium parvum
human genotype 1 isolate passaged through different host species.
Infect. Immun. 70:
5670-5675.[Abstract/Free Full Text]
Apweiler, R., Attwood, T.K., Bairoch, A., Bateman, A., Birney, E.,
Biswas, M., Bucher, P., Cerutti, L., Corpet, F., Croning, M.D., et al.
2000. InterPro—An integrated documentation resource for
protein families, domains and functional sites.
Bioinformatics 16:
1145-1150.[Abstract/Free Full Text]
Armson, A., Meloni, B.P., Reynoldson, J.A., and Thompson, R.C.A.
1999. Assessment of drugs against Cryptosporidium parvum
using a simple in vitro screening method. FEMS Microbiol.
Lett. 178:
227-233.[CrossRef][Medline]
Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H.,
Cherry, J.M., Davis, A.P., Dolinski, K., Dwight, S.S., Eppig, J.T., et al.
2000. Gene ontology: Tool for the unification of biology.
Nat. Genet. 25:
25-29.[CrossRef][Medline]
Bateman, A., Birney, E., Cerruti, L., Durbin, R., Etwiller, L.,
Eddy, S.R., Griffiths-Jones, S., Howe, K.L., Marshall, M., and Sonnhammer,
E.L. 2002. The Pfam protein families database. Nucleic
Acids Res. 30:
276-280.[Abstract/Free Full Text]
Beyer, T.V., Svezhova, N.V., Sidorenko, N.V., and Khokhlov, S.E.
2000. Cryptosporidium parvum (Coccidia, apicomplexa):
Some new ultrastructural observations on its endogenous development.
Eur. J. Protistol. 36:
151-159.
Blunt, D.S., Khramstov, N.V., Upston, S.J., and Montelone, B.A.
1997. Molecular karyotype analysis of Cryptosporidium
parvum: Evidence for eight chromosomes and a low-molecular-size molecule.
Clin. Diagn. Lab. Immunol.
4: 11-13.
Bonfield, J.K., Smith, K.F., and Staden, R. 1995. A
new DNA sequence assembly program. Nucleic Acids Res.
23:
4992-4999.[Abstract/Free Full Text]
Caccio, S., Camilli, R., La Rosa, G., and Pozio, E.
1998. Establishing the Cryptosporidium parvum karyotype
by NotI and SfiI restriction analysis and Southern
hybridizations. Gene
219: 73-79.[Medline]
Carraway, M., Tzipori, S., and Widmer, G. 1997. A new
restriction fragment length polymorphism from Cryptosporidium parvum
identifies genetically heterogeneous parasite populations and genotypic
changes following transmission from bovine to human hosts. Infect.
Immun. 65:
3958-3960.[Abstract]
Carreno, R.A., Martin, D.S., and Barta, J.R. 1999.
Cryptosporidium is more closely related to the gregarines than to
coccidia as shown by phylogenetic analysis of apicomplexan parasites inferred
using small-subunit ribosomal RNA gene sequences. Parasitol.
Res. 85:
899-904.[CrossRef][Medline]
Casemore, D., Wright, S., and Coop, R. 1997.
Cryptosporidiosis —Human and animal epidemiology. In
Cryptosporidium and cryptosporidiosis (ed. R. Fayer),
pp. 65-92. CRC Press, Boca Raton,
Florida.
Chappell, C.L. and Okhuysen, P.C. 2002.
Cryptosporidiosis. Curr. Opin. Infect. Dis.
15:
523-527.[Medline]
Claros, M.G. and Vincens, P. 1996. Computational
method to predict mitochondrially imported proteins and their targeting
sequences. Eur. J. Biochem.
241:
779-786.[Medline]
Current, W.L. 1989. Cryptosporidium spp. In
Parasitic infections of the immunocompromised host
(eds. P.W. Walzer and R.M. Genta), pp. 251-341.
Marcel Dekker, NY.
Dayhoff, M.O., Schwartz, R.M., and Orcutt, B.C. 1978.
A model of evolutionary change in proteins. In Atlas of protein
sequence and structure 5, Suppl. 3. pp.
345-352. National Biomedical Research Foundation,
Washington, DC.
Dear, P.H. 1997. HAPPY mapping. In Genome
mapping: A practical approach (ed. P.H. Dear). pp.
95-123. IRL Press, Oxford, UK.
Dear, P.H. and Cook, P.R. 1993. HAPPY
mapping—Linkage mapping using a physical analog of meiosis.
Nucleic Acids Res. 21:
13-20.[Abstract/Free Full Text]
Dear, P.H., Bankier, A.T., and Piper, M.B. 1998. A
high-resolution metric HAPPY map of human chromosome 14
Genomics 48:
232-241.[CrossRef][Medline]
Fayer, R., Speer, C.A., and Dubey, J.P. 1997. The
general biology of Cryptosporidium. In Cryptosporidium and
cryptosporidiosis (ed. R. Fayer), pp.
1-42. CRC Press, Boca Raton, FL.
Fichera, M.E. and Roos, D.S. 1997. A plastid organelle
as a drug target in apicomplexan parasites. Nature
390:
407-409.[CrossRef][Medline]
Frengen E., Weichenhan D., Zhao B., Osoegawa K., van Geel M., and
de Jong P.J. 1999. A modular, positive selection bacterial
artificial chromosome vector with multiple cloning sites.
Genomics 58:
250-253.[CrossRef][Medline]
Galtier, N., Gouy, M., and Gautier, C. 1996. SEAVIEW
and PHYLO_WIN: Two graphic tools for sequence alignment and molecular
phylogeny. Comput. Appl. Biosci.
12:
543-548.[Abstract/Free Full Text]
Gardner, M.J., Hall, N., Fung, E., White, O., Berriman, M., Hyman,
R.W., Carlton, J.M., Pain, A., Nelson, K.E., Bowman, S., et al.
2002. Genome sequence of the human malaria parasite
Plasmodium falciparum. Nature
419:
498-511.[CrossRef][Medline]
Glaberman, S., Moore, J.E., Lowery, C.J., Chalmers, R.M., Sulaiman,
I., Elwin, K., Rooney, P.J., Millar, B.C., Dooley, J.S., Lal, A.A., et al.
2002. Three drinking-water-associated cryptosporidiosis
outbreaks, Northern Ireland. Emerg. Infect. Dis.
8: 631-633.[Medline]
Goffeau, A., Barrell, B.G., Bussey, H., Davis, R.W., Dujon, B.,
Feldmann H., Galibert, F., Hoheisel, J.D., Jacq, C., Johnston, M., et al.
1996. Life with 6000 genes. Science
274:
546-567.[Abstract/Free Full Text]
Gough, J. and Chothia, C. 2002. SUPERFAMILY: HMMs
representing all proteins of known structure. SCOP sequence searches,
alignments and genome assignments. Nucleic Acids Res.
30:
268-272.[Abstract/Free Full Text] |
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
ABSTRACT
RESULTS
