Woods Hole, Mass. talk Oct. 4, 2002
2.7 Billion years of eukaryotic P450 evolution.
David Nelson
Determining the age of eukaryotic life on earth had been limited in the
past to finding eukaryotic microfossils. Algal microfossils called acritarchs
[slide 1] are seen around 1.8 billion years ago and more tentative eukaryotic
microfossils have been claimed to be 2.1 billion years old. That was the
oldest evidence for eukaryotic life until 1999. Then, it was reported in a
Science article that eukaryotic lipid derivatives called steranes had been
found in Australian shales dated to 2.7 billion years. [slide 2] Steranes
derive from sterols, that are uniquely the products of eukaryotes so steranes
are eukaryotic biomarkers. These molecules have the 4 fused ring structure
of cholesterol and related lipids but they are lacking all double bonds and the
3 hydroxyl is gone. This finding pushed back the standard date of the first
appearance of eukaryotes by almost a billion years.
Cytochrome P450 evolution is tied to this discovery also because an
early step in the synthesis of sterols is the removal of a 14 methyl group.
This oxygen requiring step is catalyzed by the P450 CYP51, the sterol 14
alpha demethylase. Another P450, CYP61 is the C-22 desaturase of fungi
that creates this double bond. CYP51 is found in plants, animals, fungi,
trypanosomes, diatoms and even Mycobacterium tuberculosis, though what
it is doing in Mycobacterium is unknown, because Mycobacterium is not
known to have any sterols. Because CYP51 is the only P450 found in all
main branches of the eukaryotic tree (except anaerobes), it has been
proposed to be the original eukaryotic P450, with all others deriving from it.
The origin of CYP51 is lost. There is no clear link to bacterial enzymes that
have the same function. Though it woud be desirable to trace the P450
superfamily into the bacteria, this does not seem possible now. That is why
I limited this talk to eukaryotic p450 evolution.
In 1987 I published my first paper on P450 with Henry Strobel.
[slide 3] This was in the days before the dash was removed from the name and
before the nomenclature system was started. This paper showed a P450 tree
with 34 sequences, all that were known at the time. Today there are about
2400 named P450 sequences. [slide 4] These are a result of the many
genome projects that have been carried out since 1996 when yeast was
finished. This slide [slide 5] summarizes the genome project coverage of
eukaryotes. Red names indicate completed or nearly completed genomes,
blue names have genome projects proposed and perhaps some EST or BAC
end sequencing has been done, but they are not so far along.
Fungi have four genomes listed, Aspergillus, Fusarium and Candida
albicans could probably be added to these, but the data may not be publically
available. Animals have six genomes shown, mouse and zebrafish will
make eight. Dictyostelium is just below animals and fungi. This genome is
almost done.
You may notice that I link plants to alvelolates and stramenopiles, instead of
closer to animals and fungi. I do this because of a unique 5 aa insertion
[slide 6] in the enolase gene that is only shared among these three groups.
There is actually another 1 aa insert just upstream of this that is also limited
to these organisms. The Stramenopiles include kelp, [slide 7] which looks
like a plant, but is not. [slide 8] Tetrahymena and Paramecium are free living
ciliates in the alveolates with genome projects that are just beginning.
Tetrahymena has about 10,000 ESTs in Genbank, mostly from this year.
Paramecium has about 3200 Genome Survey Sequences. [slide 9]
Plasmodium falciparum, a parasitic alveolate and the malaria causing
organism, has a finished genome. Plants now have two complete genomes,
rice and Arabidopsis, a monocot and a dicot. I have just finished naming the
rice p450s. Rice is the current record holder for the most P450s at 458
sequences. At least 309 of these are full length and that number will rise to
about 325 as the rice genome is finished. Outside the crown group of
eukaryotes there are projects mostly on pathogens like Trypanosoma and
Leishmania. And of course here at Woods Hole, Giardia has been done.
The genomes that have P450 sequences detected are maked with a red X. [slide 10]
Notice that Plasmodium has no P450s detected by blast searches. This is a
parasite that lives in a rich environment, so it has stripped down its genome
and jettisoned all its P450s. Its cousin, the free living Tetrahymena does not
have this luxury, and so far, one p450 has been found in Tetrahymena that
most resembles a CYP4V sequence. One warning to those who want to
blast search for genes in Tetrahymena. Ciliates use a different genetic code
so when translating and blast searching you must be aware that TAA and
TAG will be glutamines instead of stop codons.
Glaucophyta and Heterolobosea have no detected p450s, but that is probably
a sampling problem. I expect they will have p450s. Giardia does not seem
to have a p450, but it is anaerobic and p450s usually use oxygen. No
anaerobe to date has had a p450 found in its genome. Both trypanosoma and
Leishmania have CYP51 genes. That is why CYP51 has been placed at
almost the deepest branch on the eukaryotic tree, because it has members in
most of these branches except for parasites and poorly sampled genomes.
CYP97 is beginning to emerge as another cross-kingdom P450. It is seen in
plants and diatoms, but not outside of this clade. A third p450 CYP61 acts
after CYP51 in the ergosterol biosynthetic pathway. It is seen in all fungi,
but not outside fungi.
To follow the evolution of the P450s in Deep time, we will need more
completed genomes from some of these branches. Right now the only
branches that are covered are anaerobes and parasites that have no P450s
and plants. We will have to wait on Tetrahymena, Phytophthora and
Trypanosoma to fill in some of these gaps. In the meantime, the top of this
tree is better sampled. By looking at Dictyostelium and Fungal
genomes, compared to plant and animal genomes, we begin to see some
patterns. As the early lineages branched off from one another, there were
very few P450s in common that have been retained as recognizable families.
CYP51 is the only one that crosses all main divisions and CYP97 seems to
be limited to the plants/alveolates/stramenopiles clade. Aside from these,
each major lineage has done its own independent evolution of p450s.
Dictyostelium has at least 46 P450s. Only CYP51 is found in common with
other lineages.
Fungi are very diverse. The first eukaryotic genome Baker's yeast (shown
here) had only three P450s a CYP51, a CYP61 and CYP56. The first two
are in the ergosterol pathway of all fungi, the last is a spore wall maturation
gene specific to yeast and some close relatives (probably including candida).
So yeast could be considered a minimalist P450 genome. S. pombe is even
more minimalist, since it has only CYP51 and CYP61. Neurospora on the
other hand has 38 P450s. This is a lot for a single celled organism and it
blew the assumption that fungi would have very few P450s. However, the
real surprise in fungi was the white rot genome. [slide 11] Phanerochaete
chysosporium appears to have about 150 p450s. White rot is a group of
wood degrading fungi. They break down lignin, which is brown leaving
behind cellulose which is white, thus the name white rot. Without this type
of fungi, the planet would be covered with lignin, a very complex aromatic
ring based polymer. This genome is being sequenced at the Dept. of Energy
Joint Genome Institute in Walnut Hill California as part of a microbial
genome initiative. The fungus can grow at the high temperatures found in
wood chip piles, making it a potential industrial agent for bleaching paper
pulp instead of the polluting acid or base chemistries that are currently used.
The fungus secretes many oxidative enzymes including peroxidases, and
may be useful in bioremediation of toxic waste sites. The concept of fungi
as environmental clean up agents is shown here. [slide 12] This slide shows
a mushroom taking a bite out of a chlorinated and hydroxylated ring
compound. The picture was taken from Thom Volk's Fungus of the Month
web site, which has detailed write ups on dozens of fungi along with some
exceptional photos.
I began searching the white rot genome expecting it would have a few
P450s, maybe 10-15, but I was very surprised by the large number of hits.
After doing multiple searches with a variety of P450s and assembly of
overlapping fragments, I was left with 167 contigs. So far I have assembled
103 genes with all intron-exon boundaries identified and I have 64 more to
do. 96 of the 103 sequences are full length P450 genes. I expect when all
assemblies are done that white rot fungus will have between 130 and 150
P450s.
The white rot genes have many exons with short introns separating them.
There are also some unexpected features. This slide [slide 13] shows the
structure of one P450. This gene has 12 exons. Please notice the red ones
are very short. What is the evidence that these are real? First, the end of
exon 6 is phase 1, while the beginning of exon 8 is phase 0. You cannot join
these two exons together without an intervening exon with phase 1 and
phase 0 ends. Exon 8 and 10 have a similar problem. Second, there are 28
P450 genes in white rot that have this same exon structure. Third, the
AGSDT sequence is the highly conserved part of the I helix oxygen binding
pocket. Without this exon, this five amino acid motif is clearly missing from
sequence alignments with other P450s, so it must be there in the gene
someplace. The short exon 9 is also missing in alignments right after the
EXXR motif.
There are two other 5 amino acid exons at different locations in some of the
white rot P450 genes, but those are not the shortest exons. [slide 14] Here is
a gene with a three amino acid exon (actually 8 nucleotides long). Again,
the phases are incompatible from exon 9 to 10, requiring an intermediate
exon, The sequence is in the same region as before at the AGHETT
conserved site in the I-helix and there are six P450s with this same intron-
exon structure. As further evidence I offer this sequence [slide 15] from an
adjacent gene where exon 9 is not split and the GHE sequence is on the end
of exon 9. These short exons make the gene assembly process difficult.
They also adversely affect automated gene assembly programs that probably
cannot detect them.
Our previous eukaryotic tree [slide 16] showed animals as a single branch
on the top of the tree. However, animals are quite complex as shown here
[slide 17]. At the bottom, there are the sponges, followed by the radial
animals. We have no genomes for any of these, but I am sure they will be
done. Above the radial animals are the bilaterians, which get split into two
main groups, the protostomes (mouth first) and the deuterostomes (mouth
second), which is a major distinction in development of the embryo.
Drosophila and C. elegans are in the protosome groups, but for now we will
concentrate on the deuterostomes. The lowest branch includes echinoderms,
the sea urchins and sea stars. EST projects are underway on sea urchins, but
they are not very extensive yet. Above the echinoderms are the tunicates
also called urochordates. Above the urochodates are the chordates proper.
These include the ray-finned fish and tetrapods (that includes us) that
diverged about 420 million years ago. The urochordate group is the sister
group to chordates, branching about the time of the Cambrian explosion 540
million years ago. Two genomes of the genus Ciona have been sequenced.
Ciona is a sea squirt. Sea squirts are filter feeders that have a larval stage
that looks like a tadpole, [slide 18] with a visible notochord. These larvae
swim for a while then settle on the sea floor and assume the adult form. This
is Ciona intestinalis, [slide 19] sequenced by the Joint Genome Institute (the
same people who did white rot). A second species, Ciona savignyi has been
sequenced by the Whitehead Institute. Their proteins are about 70%
sequence identical, less than mouse and human, so they are quite different. I
have used the Ciona genomes in teaching a bioinformatics class. We
collected about 800 sequence reads containing P450 sequence from the JGI
genome data by blast searches. These were translated and assembled into
about 200 contigs. The last assignment in the class was to assemble a P450
gene from these contigs and the raw sequence reads at JGI. I must say the
students struggled with that assignment. Not one student was able to
assemble a complete gene. Only 24 Ciona p450 genes are assembled so far,
some from each species. I can say from sequence alignments of the heme
signature region that there are at least 60 different P450 genes in Ciona.
When I was in Japan in July for the MDO2002 meeting, we traveled to Ise
and visited Mikimoto Pearl Island, where the famous pearl culturing
company is located. After just spending weeks on looking at the Ciona
genome with my students you can imagine how surprised I was to see this in
the pearl museum. [slide 20] Ciona intestinalis is a pest to the pearl industry
because they like to grow on the oyster shells. They have to be cleaned off
once a year.
It is still too early to do much comparative genomics on the Ciona P450
sequences, because we do not have a complete set assembled yet. They have
been much harder to do than expected. We do have a nearly complete set
for the Fugu genome. [slide 21] Fugu is the Japanese pufferfish. It has been
sequenced by several labs including JGI again. Last October JGI assembled
the genome. I have done an extensive anlaysis of the Fugu P450 genes by
blast searching with members of every mammalian P450 family and some of
the subfamilies. After all the searches were done the resulting list of
accession numbers is nearly comprehensive, at least for the most conserved
regions of the P450s, especially the heme signature region.
The alpha list of accession numbers had 332 hits after this process. This is
smaller than Ciona's 800 because the genome sequence had been assembled.
The contigs are larger and there is not the redundancy of coverage. The
translated protein sequences were put in a blast server and each sequence
was compared against each other sequence for exact matches. These were
then combined into contigs. Right now there are 71 contigs from Fugu
P450s. 35 of these are full length genes and 12 are missing only small
pieces. These 47 sequences were put in a sequence alignment with all the
human P450s and 8 other fish sequences. A phylogenetic tree was
constructed to compare the Fugu and human proteins. [slide 22]. This is the
result.
Human sequences are in red, Fugu in blue. If you look at the bottom of the
tree first, you will notice an alternation of red and blue branches. This is
equivalent to saying there is nearly a one to one correspondence between
Fugu and human in their P450s. In the middle region of the tree this one to
one arrangement starts to break down at the subfamily level. Here there are
new subfamilies in one species that are not in the other. Notice the 4A, 4B,
4X and 4Z in humans and only 4T in fish. Also see the six 4Fs in humans
and only one 4F28 in Fugu. At the top of the tree, the relationship breaks
down almost completely. This is the CYP2 family. Many CYP2 sequences
in humans are involved in metabolizing drugs and foreign chemicals. This
has allowed diversification of the CYP2 subfamilies.
To make the relationships easier to see, I have drawn some figures that link
the families and subfamilies together. The first of these shows just the
CYP2 family. [slide 23] Humans have 13 subfamilies, while Fugu has 8.
Only 2 of them are preserved across the 420 MY of evolution. These are the
CYP2R1 and CYP2U1. There are no publications on these P450s yet, so we
do not know what they do. Because they are conserved between fish and
man, I would predict that they metabolize endogenous substrates rather than
foreign compounds. This suggests they might also be implicated in disease.
The next figure [slide 24] covers the middle region of the tree. Here we see
the 4 family again with 4A, B, X and Z as possible diversified subfamiles
coming from the 4T subfamily in fish. There are also 1C and 3B subfamilies
in Fugu that are not seen in human. One other item in fish that comes up
several times is expansion of familes that only have one member in humans.
This is seen here in the CYP17 sequences. The third figure [slide 25]
covers the bottom of the tree and we see very clearly the one to one
relationship between families and even subfamiles that goes all the way
through the two species P450s. There is only one exception at the family
level and that is CYP39. This sequence cannot be found in Fugu. You
might say that the genome is not quite complete and it might be found. That
is possible, but I have also looked for CYP39 in zebrafish and every other
fish cDNA and genomic sequence in genbank with no luck. I suspect that
this is the one innovation that is new to mammals.
[Note added in 2004, CYP39 has been found in fish now]
Note that CYP19, CYP8A and CYP8B have two sequences in fish and only
one in human. Also, CYP27A has three fish sequences and one in human.
Combined with the CYP17s seen before, these may be relics of a complete
genome duplication in fish that did not happen in tetrapods. Evidence for a
fish genome duplication is found in the Hox genes. Fish have 7 clusters of
hox genes, while mammals have only 4. These are all on different
chromosomes, so it looks like the whole genome duplicated and then one
hox gene cluster was lost. P450s that were present at the time of this
duplication should also have been duplicated. That, in combination with
gene loss might explain the odd sets of duplicate fish P450s that have only
one copy in humans. It will be useful to compare Fugu with zebrafish when
that genome is more complete to see if the gene duplications are in both fish
genomes.
Fish and humans are remarkably similar even after 420 million years, but
this is not true for Ciona. With 24 complete sequences to work with I made
this phylogenetic tree on Tuesday. [slide 26] This includes 13 Ciona
intestinalis sequences and 11 Ciona savignyi sequences. A mixture of
human, Fugu, Drosophila and C. elegans sequences were added to sort the
new sequences into clans. Clans are groups of P450 families that always
cluster together on trees, so they are like gene clades. The genes have color
coded species symbols on the right. The four major clans are represented on
this tree. After looking fairly hard for a CYP51, no CYP51 like sequence
could be found in Ciona. It seems to have lost that gene. Drosophila and C.
elegans have also lost CYP51. Apparently it is not required if you can eat
sterols in your diet.
What this tree shows is that the common ancestor to the bilateral animals
[slide 27] had at least these four clans of P450s represented, plus we know
CYP51 had to be present, since it is found in both lower and higher
eukaryotes. That is a minimum of 5 P450s at the time of the protostome
deuterostome divergence about 670 million years ago.
We can also see [slide 28] that Ciona is expanding the CYP2 Clan. It is
common for one lineage to expand one or two clans to meet its P450 needs.
This is shown here [slide 29] for C. elegans, where half of the C. elegans
P450s also are in a single clan that is related to the CYP2 clan.
My last slide [slide 30] is just a pretty image of some more sea squirts.
We have to do more work to finish assembling about 100 more P450 genes
from the two Ciona species. Only then can we know about the relationships
to other P450 families found in the vertebrates. This may shed some light on
the origins of steroidogenic and retinoid metabolic pathways in these
organisms.