Evolution of Key Cell Signaling
and Adhesion Protein Families
Predates Animal Origins
Nicole King, Christopher T. Hittinger, Sean B. Carroll*
The evolution of animals from a unicellular ancestor involved many innovations.
Choanoflagellates, unicellular and colonial protozoa closely related to Metazoa,
provide a potential window into early animal evolution. We have found that
choanoflagellates express representatives of a surprising number of cell signaling and adhesion protein families that have not previously been isolated
from nonmetazoans, including cadherins, C-type lectins, several tyrosine kinases, and tyrosine kinase signaling pathway components. Choanoflagellates
have a complex and dynamic tyrosine phosphoprotein profile, and cell proliferation is selectively affected by tyrosine kinase inhibitors. The expression in
choanoflagellates of proteins involved in cell interactions in Metazoa demonstrates that these proteins evolved before the origin of animals and were later
co-opted for development.
A central question in animal evolution is how
multicellular animals evolved from a protozoan
ancestor. One approach to animal origins is to
determine which developmental proteins predated the origin of animals and were subsequently co-opted for animal development.
Comparative genomics can identify the minimal set of genes in place at the outset of animal
evolution by revealing those shared by all animals and their nearest relatives (1). The choanoflagellates, a group of unicellular and colonial flagellates that resemble cells found only in
Metazoa, have emerged as an important model
for studies of early animal evolution (2, 3).
Analyses of nuclear and mitochondrial genes
consistently support a close phylogenetic relationship between choanoflagellates and Metazoa (4–7). Specifically, the presence in a choanoflagellate mitochondrial genome of multiple
genes lost from a conserved portion of animal
(including sponge and cnidarian) mitochondrial
genomes suggests that choanoflagellates are an
outgroup of animals (8–11) (fig. S1 and tables
S1 and S2). Therefore, genes expressed by choanoflagellates and animals to the exclusion of
other eukaryotes define the minimal complexity
of the eukaryotic genome before the emergence
of multicellular animals.
To sample the diversity of genes expressed
by choanoflagellates, we collected more than
5000 expressed sequence tags (ESTs) from two
choanoflagellate species, Monosiga brevicollis
and Proterospongia-like sp. ATCC50818 (11,
12). Under laboratory conditions, Monosiga is
strictly unicellular, whereas ATCC50818 occasionally forms small colonies of apparently unHoward Hughes Medical Institute (HHMI), University
of Wisconsin, 1525 Linden Drive, Madison, WI 53706,
USA.
*To whom correspondence should be addressed. Email: [email protected]
differentiated cells; both may exhibit additional
behaviors in the natural environment. In marked
contrast to their simple lifestyle, choanoflagellates express members of a wide variety of protein families involved in animal cell interactions,
including cadherins, C-type lectins, tyrosine kinases (TKs), and a G protein–coupled receptor
(GPCR), as well as several multidomain
polypeptides that contain protein-protein interaction domains involved in signaling and adhesion
in animals [such as the epidermal growth factor
(EGF) motif, Src homology 2 (SH2) domain,
tumor necrosis factor receptor (TNFR) domain,
and sushi or complement control protein (CCP)
domain (Fig. 1)].
The phylogenetic distribution of each protein
or domain of interest was determined by a combination of strategies. First, the occurrence of
each domain in eukaryotes, Bacteria, Archaea,
and viruses was examined with two protein
domain annotation programs, SMART and
PFAM (13, 14). Second, we queried all available sequences from representative nonanimals, including Fungi, Plantae, Dictyostelium,
and diverse bacteria, using BlastX and keyword searches. Third, we examined certain
sequences more closely for adherence to the
conserved traits of the protein domain family
(fig. S2). Proteins containing cadherin, fibrinogen, GPCR proteolytic site (GPS), somatomedin, and CCP domains were not detected in
Fungi, Plantae, or other eukaryotes (11). Although putative TK and C-type lectin domains
were detected in plant genomes by the
SMART and PFAM annotation sites, further
scrutiny revealed these sequences to have been
classified incorrectly (11). We conclude that
choanoflagellates and animals share, to the
exclusion of other eukaryotes whose genomes
have been analyzed, proteins that contain secretin-like GPCR, GPS, fibrinogen, somatomedin, and CCP domains, as well as members
of the cadherin, C-type lectin, and TK protein
families.
In animals, cadherins mediate cell-cell adhesion and signaling through homophilic interactions (15). Both Monosiga and ATCC50818
were found to express a cadherin-encoding
gene (Fig. 1). MBCDH1 from Monosiga encodes at least two cadherin repeats and
PRCDH1 from ATCC50818 encodes at least
four. Comparisons of cadherin repeat sequences and the domain architectures of
cadherin-containing proteins have revealed at
least six subfamilies within the cadherin family (16). Phylogenetic analysis revealed the
choanoflagellate cadherins to be most similar
to protocadherins and to the Flamingo class
of cadherins (11) (fig. S3).
Fig. 1. Choanoflagellates
express a variety of multidomain proteins that
contain animal-type protein motifs (11). Cadherin domains are found
as tandem repeats in
predicted proteins from
ATCC50818 (PRCDH1)
and Monosiga(MBCDH1).
Choanoflagellates also
express members of the
C-type lectin (MBCTL1
and MBCTL2) and TK
(MBSRC1, MBRTK2, and
MBRTK1) protein families. Additional multidomain proteins from choanoflagellates (such as
MB3515) contain previously uncharacterized
combinations of protein domains that are commonly found in animal proteins that mediate cell adhesion and
signaling. MB7TM1 resembles secretin-like GPCRs, and MBSPEC1 resembles spectrin-containing proteins,
both families of which are used in animals for signaling and cytoskeletal structure respectively. The source
species is indicated by the first two letters of each gene name; “PR” indicates predicted proteins from
ATCC50818 and “MB” indicates predicted proteins from Monosiga.
www.sciencemag.org SCIENCE VOL 301 18 JULY 2003
Downloaded from www.sciencemag.org on December 6, 2012
REPORTS
361
C-type lectins mediate cell adhesion and signaling through calcium-dependent recognition
and binding of specific sugars. Within the superfamily that contains C-type lectin–like domains,
those containing carbohydrate recognition domains (CRDs) have previously been identified
only within animals (17). We found two CRDcontaining C-type lectin genes in Monosiga, one
containing three C-type lectin domains
(MBCTL1) and a second containing a C-type
lectin domain fused to three CCP domains
(MBCTL2) (Fig. 1 and fig. S2).
We have also isolated five homologs of
animal TKs, four from Monosiga and one
from ATCC50818, including MBRTK1,
which we have previously reported (5) (fig.
S2). Two choanoflagellate TKs, MBRTK1
and MBRTK2, conform to the general structure of receptor TKs, with a C-terminal TK
enzymatic core connected through a transmembrane span to an N-terminal extracellular domain (Fig. 1). The remaining three choanoflagellate TKs (MBSRC1, MBTK3, and
PRTK1) lack transmembrane domains and
are predicted to operate downstream in TK
signaling pathways. For instance, MBSRC1
contains an N-terminal SH2 domain linked to
a TK domain and resembles Src TKs from
animals (Fig. 1 and fig. S3).
Because components of TK signaling
pathways appear highly conserved from
sponges to vertebrates, we sought to determine whether TK signaling, although absent from fungi and plants, evolved before
the origin of animals. We identified ESTs
for many components of TK signaling pathways in Monosiga and ATCC50818, including phosphoinositide 3-kinase, growth
factor receptor– bound protein 2, Vav proteins, PDZ-containing sequences, tyrosine
phosphatases, Raf kinase, and Ras, in addition to mitogen-activated protein kinase
and Src kinase (18–20). The expression of
these signal transduction components suggests that choanoflagellates may use tyrosine phosphorylation for signaling in a
manner similar to animals.
To investigate the extent of TK signaling in
choanoflagellates, we probed Monosiga cell lysates with antibodies specific to phosphotyrosine. In cells maintained under conditions of
nutrient depletion, approximately 15 to 18 specific phosphotyrosine-containing proteins were
detected (Fig. 2, A and B), a similar number to
that seen in unstimulated animal cells (21, 22). In
animal cells, TK signaling is rapidly stimulated
by the addition of serum to cultures of serumstarved cells (21, 23). We compared the tyrosine
phosphoprotein profile of nutrient-depleted cells
cultured in seawater with the profile in cells
exposed to seawater enriched with either a cereal
grass infusion (1525) or 1525 supplemented with
Enterobacter (1525/Ea). No change was seen in
the relative abundance of tyrosine-phosphorylated proteins from starved cells treated with
fresh seawater compared to untreated starved
cells (data not shown). In contrast, treatment of
starved cells with either 1525 or 1525/Ea resulted in rapid modifications to the tyrosine phosphorylation status of at least four proteins (Fig.
Fig. 2. The spectrum of tyrosine phosphoproteins in Monosiga. Monosiga cell lysates were
electrophoresed, blotted onto membranes, and probed with antibodies specific to phosphotyrosine
(11). (A) The profile of tyrosine-phosphorylated proteins in nutrient-deprived cells. About 15 to 18
tyrosine-phosphorylated protein species, ranging in size from 40 kD to ⬎200 kD, were detected in
lysates from Monosiga cultured in seawater. (B) All but one of the detected tyrosine phosphoproteins in (A) derive from Monosiga. We compared the tyrosine phosphoproteins in protein extracts
from Monosiga cultured with Enterobacter (M.b. ⫹ E.a.) with those from Enterobacter (E.a.) protein
extracts (left panel). Two bacterial protein species reacted with the antibodies to phosphotyrosine,
but only the larger molecular weight band (approximately 78 kD) is specific as determined by
inhibition with the hapten p-nitrophenyl phosphate (PNPP) (right panel). (C) Protein tyrosine
phosphorylation status is affected by environmental conditions. Cells were incubated in seawater
as in (A) and (B), then treated for 5, 10, or 30 min with either seawater (S.W.), a cereal grass
infusion of seawater (1525), or 1525 supplemented with Enterobacter at 108 cells/mL (1525/Ea).
Within 5 min, a 95-kD protein species (P95) was detected in lysates treated with cereal grass
infusion either with or without Enterobacter, but not in lysates treated with seawater. In contrast,
the antibodies to phosphotyrosine showed diminished reactivity to protein species of 38, 80, and
83 kD (P38 and P80/83) in the presence of cereal grass infusion.
362
2C). An apparently phosphotyrosine-free 95-kD
protein (P95) became tyrosine-phosphorylated
within 5 min of treatment, indicating the probable activation of one or more tyrosine kinases,
and the tyrosine phosphorylation levels of three
protein species (P38, P80, and P83) diminished
in response to stimulation with enriched media
(Fig. 2C).
To examine the functional significance of
TK signaling in choanoflagellate biology, we
measured the proliferation and maximal cell
density of Monosiga cultures treated with specific inhibitors of TKs at concentrations previously shown to be effective in cultured animal
cells (11, 20, 24). Control cultures of Monosiga
grown in 1525/Ea conformed to the typical
growth characteristics of cultured microbes, exhibiting lag, log, and stationary phases (Fig. 3).
The doubling time of untreated cells during log
phase was approximately 6 hours and cells
reached a maximum density of ⬃107 cells/mL.
In contrast, cultures treated with genistein, a
general inhibitor of TKs, were delayed in their
entry into log phase and reached a maximum
density of 2 ⫻ 106 cells/mL, one-fifth that of
untreated cells. In addition, treatment of
Monosiga cultures with PP2, an inhibitor of Src
kinases, entirely blocked cell proliferation.
Treatment of Monosiga cultures with inactive
analogs of genistein and PP2, the control compounds genistin and PP3, had no significant effect on cell density (not shown). Importantly,
genistein- and PP2-treated cells displayed normal levels of motility and feeding behavior. The
dependence of Monosiga cells on TK activity
indicates a functional link between TK signaling
and the cell cycle or cell survival.
The discovery of multiple signaling and adhesion gene family members in choanoflagellates demonstrates that key proteins required for
animal development evolved before the origin of
animals. Other multicellular eukaryotes, and
even those with complex modes of development
Fig. 3. TK activity is required for proliferation of
Monosiga cultures. In contrast with cultures
grown in the presence of a control solvent
[dimethyl sulfoxide (DMSO), diamonds], the increase in cell density of cultures treated with
TK inhibitors [25 ␮M genistein (circles) or 500
nM PP2 (squares) in DMSO] is significantly
reduced (11).
18 JULY 2003 VOL 301 SCIENCE www.sciencemag.org
Downloaded from www.sciencemag.org on December 6, 2012
REPORTS
REPORTS
References and Notes
1. Comparisons of yeast, Caenorhabditis, Drosophila,
and human genomes have revealed a set of protein
domains conserved among Bilateria and absent
from yeast (25, 26). However, because the Bilateria
represent a recent derivation within the Metazoa
and the Fungi diverged from the animal lineage
long before the transition to multicellularity, comparisons limited to fungal and bilaterian animal
genomes do not reveal the complexity of the ancestral animal genome.
2. W. Saville Kent, A Manual of the Infusoria (David
Bogue, London, 1880 –1882).
3. D. J. Hibberd, J. Cell Sci. 17, 191 (1975).
4. P. O. Wainright, G. Hinkle, M. L. Sogin, S. K. Stickel,
Science 260, 340 (1993).
5. N. King, S. B. Carroll, Proc. Natl. Acad. Sci. U.S.A. 98,
15032 (2001).
6. E. A. Snell, R. F. Furlong, P. W. H. Holland, Curr. Biol.
11, 967 (2001).
7. B. F. Lang, C. O’Kelly, T. Nerad, M. W. Gray, G. Burger,
Curr. Biol. 12, 1773 (2002).
8. G. Burger, L. Forget, Y. Zhu, M. W. Gray, B. F. Lang,
Proc. Natl. Acad. Sci. U.S.A. 100, 892 (2003).
9. R. F. Watkins, A. T. Beckenbach, J. Mol. Evol. 48, 542
(1999).
10. J. L. Boore, Nucleic Acids Res. 27, 1767 (1999).
11. Materials and methods are available as supporting
material on Science Online.
12. Searchable choanoflagellate EST databases are
available at http://projects.bocklabs.wisc.edu/
carroll/choano/.
13. I. Letunic et al., Nucleic Acids Res. 30, 242 (2002).
14. A. Bateman et al., Nucleic Acids Res. 30, 276 (2002).
15. L. Shapiro, P. D. Kwong, A. M. Fannon, D. R. Colman,
W. A. Hendrickson, Proc. Natl. Acad. Sci. U.S.A. 92,
6793 (1995).
16. F. Nollet, P. Kools, F. van Roy, J. Mol. Biol. 299, 551
(2000).
17. R. B. Dodd, K. Drickamer, Glycobiology 11, 71R
(2001).
18. X. R. Bustelo, Mol. Cell. Biol. 20, 1461 (2000).
19. A. C. Porter, R. R. Vaillancourt, Oncogene 17, 1343
(1998).
20. A. Levitzki, A. Gazit, Science 267, 1782 (1995).
21. E. B. Pasquale, P. A. Maher, S. J. Singer, J. Cell. Physiol.
137, 146 (1988).
22. K. Miyazawa et al., Blood 80, 1685 (1992).
23. J. A. Cooper, B. M. Sefton, T. Hunter, Mol. Cell. Biol.
4, 30 (1984).
24. P. Kumar, S. Hosaka, A. E. Koch, J. Biol. Chem. 276,
21039 (2001).
25. G. M. Rubin et al., Science 287, 2204 (2000).
26. V. Wood et al., Nature 415, 871 (2002).
27. We thank P. Bertics, A. Guadarrama, B. Leadbeater,
and T. Hunter for advice and technical assistance; L.
Olds for assistance with graphics; J. Holt for computing support; and B. Hersh and C. Malone for critical
reading of the manuscript. N.K. is supported by an
NIH postdoctoral fellowship (GM-20734) and C.T.H.
is an HHMI predoctoral fellow. This work was supported by the HHMI (S.B.C.).
Supporting Online Material
www.sciencemag.org/cgi/content/full/301/5631/361/
DC1
Materials and Methods
SOM Text
Figs. S1 to S4
Tables S1 and S2
References and Notes
25 February 2003; accepted 28 May 2003
Hox10 and Hox11 Genes Are
Required to Globally Pattern the
Mammalian Skeleton
Deneen M. Wellik and Mario R. Capecchi*
Mice in which all members of the Hox10 or Hox11 paralogous group are
disrupted provide evidence that these Hox genes are involved in global patterning of the axial and appendicular skeleton. In the absence of Hox10 function,
no lumbar vertebrae are formed. Instead, ribs project from all posterior vertebrae, extending caudally from the last thoracic vertebrae to beyond the sacral
region. In the absence of Hox11 function, sacral vertebrae are not formed and
instead these vertebrae assume a lumbar identity. The redundancy among these
paralogous family members is so great that this global aspect of Hox patterning
is not apparent in mice that are mutant for five of the six paralogous alleles.
Hox genes have long been recognized as
important transcriptional regulators of embryonic development. In mammals, this complex of 39 genes resides on four separate
chromosomal linkage groups designated A,
B, C, and D, which arose early in the evolution of vertebrates from successive duplications of a single ancestral complex. Homologous members within the separate linkage
groups are divided into 13 sets of paralogous
genes, each having two to four members.
During development, paralogous sets of
genes are activated sequentially, with Hox1
and Hox2 paralogous genes being expressed
earlier and more anteriorly in the embryo and
successive genes through paralogous group
Hox13 appearing later and more posteriorly.
Howard Hughes Medical Institute and University of
Utah, Salt Lake City, UT 84112, USA.
*To whom correspondence should be addressed. Email: [email protected]
The spectrum of perturbations of the
mammalian skeleton resulting from either
gain- or loss-of-function mutations in individual Hox genes has been difficult to interpret in terms of a coherent model of how
these genes participate in the patterning of the
axial skeleton. Loss-of-function Hox mutations have yielded changes in vertebral morphology along the anteroposterior (AP) axis
that have been interpreted as anterior homeotic transformations as well as posterior
homeotic transformations. Typically, these
morphological changes involve perturbations
in one or a small number of vertebrae.
Among different vertebrate species, axial
skeletal patterns have diverged considerably. A
comparative survey of Hox gene expression in
mice and chicks showed that Hox gene expression boundaries along the rostrocaudal axis shift
in accordance with changes in the class of vertebrae produced at a given axial level (1). This
observation suggests that Hox genes play a crit-
www.sciencemag.org SCIENCE VOL 301 18 JULY 2003
Downloaded from www.sciencemag.org on December 6, 2012
(e.g., Dictyostelium, fungi, and plants), apparently lack many proteins used by animals for signaling and adhesion. Therefore, choanoflagellates, with their closer evolutionary relationship
to animals (4–7) and their expression of signaling and cell adhesion protein homologs, are
more informative for studies of animal origins.
The existence in unicellular choanoflagellates of proteins used for cell adhesion and
signal transduction in animals prompts the
question of their ancestral function in the
progenitor of animals and choanoflagellates.
Despite the apparent simplicity of the choanoflagellate lifestyle, it is possible that choanoflagellate homologs of animal proteins
perform similar biochemical functions within
a unicellular context. For instance, TKs may
act in choanoflagellates to detect changes in
the extracellular environment, as we have
demonstrated through their response to nutrient availability. In addition, animal cell adhesion proteins, such as the cadherins, may
derive from ancestral proteins that stabilized
the interactions between protozoan cells during conjugation or colony formation. Proteins
that mediate cell attachment or defense
against pathogens in animals may have
evolved from proteins required for prey recognition and capture. C-type lectins might
allow choanoflagellates to distinguish between and capture different bacterial species
by binding specific sugar groups displayed on
bacterial cell walls. Targeted manipulations
of gene function in choanoflagellates will be
necessary to test hypotheses about the ancestral roles of these conserved molecules.
We have sampled just a fraction of the
choanoflagellate proteome. The diversity of
choanoflagellate proteins predicted to function in cell interactions suggests that additional proteins shared exclusively with animals will be discovered through sequencing
the entire choanoflagellate genome. Of particular interest will be the repertoire of transcription factors and the potential representation of families of proteins that regulate cell
differentiation and development in animals. It
may then be possible to determine whether
entire regulatory pathways linking receptorbased signaling inputs to gene regulation and
cell behavior predate the origin of animals.
363