![[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}
|
|
|
|
|
||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Vol. 12, No. 5, pp. 607-620, March 1, 1998
Department of Biochemistry and Biophysics, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-7260 USA
Introduction
Top
Introduction
References
We dance round in a ring and
suppose
But the secret sits in the middle and
knows
Robert Frost
The Secret Sits
The basic-helix-loop-helix-PAS (bHLH-PAS) proteins
comprise a prominent class of transcriptional regulators that control a variety of developmental and physiological events including
neurogenesis, tracheal and salivary duct formation, toxin metabolism,
circadian rhythms, response to hypoxia, and hormone receptor function.
The bHLH-PAS proteins have a number of similarities with other bHLH protein subfamilies (Littlewood and Evan 1995
). bHLH-PAS proteins usually function as dimeric DNA-binding protein complexes; although some bHLH-PAS proteins can form homodimers, the most common functional unit is comprised of heterodimers. These heterodimers consist of one
partner that is broadly expressed, and another whose expression or
function is restricted spatially, temporally, or by the presence of
inducers. Just as other vertebrate and invertebrate bHLH proteins control cell lineage specification (Weintraub et al. 1991; Jan and Jan
1993), bHLH-PAS proteins are also important cell lineage regulators
(Thomas et al. 1988, Isaacs and Andrew 1996; Wilk et al. 1996). The
combinatorial and interactive properties of bHLH-PAS proteins provide
a variety of potential mechanisms to control their function as
transcriptional regulators, which may help explain their widespread use
in complex biological events. The purpose of this review is to describe
characteristics of the bHLH-PAS protein subfamily, in particular, how
bHLH-PAS proteins control lineage-specific gene transcription and
development of the Drosophila CNS midline cells and
respiratory system, and to discuss the evolutionary implications of the
bHLH-PAS/Arnt regulatory cassette. The underlying mechanisms employed
by the bHLH-PAS developmental regulatory proteins discussed here may
prove to be common in both vertebrates and invertebrates, and provide a
general understanding into how regulatory proteins control the
formation of cell lineages.
| |
bHLH-PAS proteins share a conserved sequence structure |
|---|
The sequence organization of bHLH-PAS proteins is remarkably
similar (Fig. 1). The bHLH domain is located near the
amino terminus. The basic region binds DNA and the HLH domain promotes
dimerization. These residues are followed closely by the PAS domain.
The carboxy-terminal residues contain transcriptional activation
domains (Franks and Crews 1994; Jain et al. 1994; Li et al. 1994) or
repression domains (Moffett et al. 1997). The unique feature of
bHLH-PAS proteins is the PAS domain, named for the first three
proteins identified with this motif: the Drosophila Period
(Per), human Arnt, and Drosophila Single-minded (Sim) (Nambu
et al. 1991). The PAS domain found in bHLH-PAS proteins is
~260-310 amino acids long (Crews et al. 1988) (Fig. 1); it is
subdivided into two well-conserved regions, PAS-A and PAS-B, separated
by a poorly conserved spacer. Within both the A and B regions lies a
copy of a 44-amino acid repeat referred to as the PAS repeat (Crews et
al. 1988; Nambu et al. 1996). The repeat begins with a nearly invariant
Phe residue and terminates with a His X X Asp motif (Wang et
al. 1995; Nambu et al. 1996). Overall, the PAS domain is not
well-conserved; nonorthologous family members are often <25%
identical in amino acid sequence. It is not surprising, given its size
and diversity in sequence, that the PAS domain can mediate a number of
biochemical functions. It is used for dimerization between PAS proteins
(Huang et al. 1993), small molecule binding (Dolwick et al. 1993;
Coumailleau et al. 1995), and interactions with non-PAS proteins
(Coumailleau et al. 1995; Gekakis et al. 1995).
|
| |
Drosophila sim and mammalian aryl hydrocarbon receptor: paradigms for bHLH-PAS protein function |
|---|
The first two bHLH-PAS proteins extensively studied were the
Drosophila sim gene and the mammalian Ahr. Genetic and
cellular analysis of sim provided the initial evidence that
bHLH-PAS proteins could act as lineage-specific developmental
regulatory proteins. These experiments showed that sim
function is required for all midline transcription and development
(Thomas et al. 1988; Nambu et al. 1990, 1991). Numerous target genes of
sim were identified and transgenic experiments identified a
regulatory element that acts as a Sim-binding site and is required for
CNS midline cell transcription (Wharton and Crews 1993; Wharton et al.
1994). This established a foundation for further molecular genetic
analysis of bHLH-PAS protein control of developmental processes.
The biochemistry of bHLH-PAS protein function has been described in
greatest detail for the mammalian aryl hydrocarbon receptor complex
(AHRC; also referred to as the dioxin receptor) (Fig. 2) (for review, see Swanson and Bradfield 1993;
Harkinson 1995; Whitlock et al. 1996; Rowlands and Gustafsson 1997).
This complex activates transcription of genes that encode proteins
involved in toxin metabolism, such as cytochrome P450IA1 and
glutathione S-transferase (GST). The functional DNA-binding
complex consists of the Ahr ligand-binding bHLH-PAS protein (Burbach
et al. 1992; Ema et al. 1992) and another bHLH-PAS protein, Arnt
(Hoffman et al. 1991). Transcriptional control involves AHRC binding to
the xenobiotic response element (XRE) that contains a GCGTG core
binding sequence. The induction of AHRC function is controlled by
ligand (e.g., dioxin) binding to Ahr, and thus AHRC constitutes a
regulated signaling pathway. Ahr is found in the unliganded state in
the cytoplasm complexed to heat shock protein 90 (Hsp90) and Ahr
interaction factor (AIF) (Ma and Whitlock 1997). These proteins are
thought to keep the unliganded Ahr in a state responsive to ligand
binding and interaction with Arnt. Ligand passes through the plasma
membrane and binds to a site in the Ahr PAS domain. Although the
sequence of events is controversial, Ahr dissociates from Hsp90 and
AIF, binds to Arnt, and the Ahr::Arnt complex enters nuclei, where it
activates transcription. Analysis of AHRC function has established a
paradigm for bHLH-PAS protein function: signal transduction by small
molecule binding, control of nuclear localization, bHLH-PAS protein
heterodimerization with Arnt, DNA binding to XRE-related sequences, and
transcriptional activation. This paradigm was instrumental in
investigating the molecular genetics of how Drosophila sim controls CNS midline development and transcription.
|
| |
Development and function of the Drosophila CNS midline cells |
|---|
The Drosophila embryonic CNS (for review, see Goodman and
Doe 1993) consists of a brain and ventral nerve cord (vnc). The vnc is
comprised of 14 fused ganglia, each consisting of 400 neurons and
additional glia. CNS neurons extend axons that join together to form
axon bundles. Longitudinal axon bundles connect the ganglia and run
along the anterior-posterior axis of the vnc, whereas within each
ganglion two commissural axon bundles cross the midline and connect
each side of the vnc. Each hemiganglion is separated by a set of CNS
midline cells that are best considered as a discrete tissue distinct
from the rest of the CNS (Nambu et al. 1993). They have a different
developmental origin, are specified by distinct regulatory genes, and
play important roles in controlling the formation of adjacent tissues
and guiding commissure formation.
The existence of the insect CNS midline cells was recognized over a
century ago (Wheeler 1893), and the development and function of these
cells first characterized using modern techniques in the grasshopper
embryo (Bate and Grunewald 1981; Goodman 1982). More recently, the
development and function of the CNS midline cells (Fig.
3) have been extensively studied in
Drosophila by a number of laboratories (for review, see Nambu
et al. 1993; Bossing and Technau 1994). The mature midline cells consist
of two to six midline glia, two midline precursor 1 (MP1) interneurons, two unpaired median interneurons (UMI), six ventral unpaired median (VUM) motorneurons and interneurons, and five to eight interneuronal and motorneuronal progeny of the median neuroblast (MNB). Midline cells
have an unusual origin: They are derived from cells that are initially
separate in the embryo. In the blastoderm embryo, precursors to the CNS
midline cells form two single-cell-wide groups of cells (approximately
four cells per hemisegment) that lie between the presumptive mesoderm
and lateral neuroectoderm (Fig. 4; see also Thomas et
al. 1988). These cells, referred to as the "mesectoderm," join
together at the end of gastrulation to form seven to eight midline
precursor cells per segment.
|
|
The midline precursor cells undergo a synchronous cell division and
then a cell shape change, in which the nuclei migrate internally and
leave a cytoplasmic projection joined to the surface of the embryo
(Nambu et al. 1991). Most precursor cells will not divide again; they
differentiate into neurons and glia (Bossing and Technau 1994). The
current view is that two precursor cells give rise to the midline glia,
one to the pair of MP1s, one to the UMIs, three to the six VUMs, and
one to the median neuroblast and its progeny (Bossing and Technau
1994). The cellular simplicity of the 22-26 nerve cells and glia that
reside at the midline of each ganglion has made them an attractive
system for studying the molecular genetics of neural development and
function. However, the truly remarkable aspect of the CNS midline cells
are their additional developmental roles.
Early in embryonic development when cells are acquiring their
tissue-specific fates, mesectodermal cells are in contact with the
adjacent lateral neuroectoderm and mesoderm. The lateral
neuroectodermal cells give rise to both ventral epidermis and the
lateral CNS. Genetic studies have shown that the mesectoderm instructs
the adjacent ectoderm to form the ventral epidermis (Kim and Crews 1993). In addition, some ectodermal cells give rise to lateral neuroblasts whose proper development is dependent on a signal from the
mesectoderm (Menne et al. 1997; Y. Lee, S.T. Crews, and S.H. Kim, in
prep.). The adjacent mesoderm also requires an influence from the
mesectoderm for proper development (Lüer et al. 1997; Zhou et al.
1997). All of these effects are mediated by a signal emanating from the
midline (Mayer and Nüsslein-Volhard 1988; Kim and Crews 1993;
Golembo et al. 1996; Xiao et al. 1996). The mesectodermal cells secrete
the Spitz protein, which is related to vertebrate transforming growth
factor-
. Spitz acts a ligand for the Drosophila ortholog
of the epidermal growth factor receptor (DER), which is present on the
adjacent ectoderm and mesoderm (Raz and Shilo 1992). Midline cells also
influence the migration of a subset of muscle precursor cells (Lewis
and Crews 1994), although it is not known how this is accomplished.
Another important function of the insect CNS midline cells, which is
shared with vertebrate ventral midline or floor plate cells, is
attraction of commissural axons to the midline (Goodman 1996).
Approximately 90% of Drosophila CNS neurons extend axons across the midline to the contralateral side of the CNS, where they
join with other axons and migrate to their synaptic targets. The
midline cells secrete Netrin proteins (Harris et al. 1996; Mitchell et
al. 1996) that attract axons expressing the Netrin receptor, the
product of the frazzled gene (Kolodziej et al. 1996), to the
midline. The midline cells also act as a barrier, repelling axons that
are either not programmed to cross the midline or to prevent those that
have crossed from migrating back (Seeger et al. 1993; Tear et al.
1996). Midline glia, which ensheath the commissural axon bundles
(Jacobs and Goodman 1989), physically separate the anterior and
posterior commissures as they migrate to their final positions
(Klämbt et al. 1991). In summary, the CNS midline cells are a
functionally rich set of cells that not only act as motorneurons,
interneurons, and glia, but also influence axon guidance and the
development of the epidermis, mesoderm, and the lateral CNS.
| |
Mesectodermal specification results from the initial activation of sim transcription by dorsal-ventral patterning genes |
|---|
The sim gene acts as a simple genetic switch for midline
cell development. When the gene is activated in ventral-lateral
ectodermal cells around the time of gastrulation, it drives those cells
into the CNS midline cell lineage. Specification of the CNS midline lineage is dependent on precise expression of sim in the
mesectodermal precursor cells. Initial sim transcription is
restricted to these two single-cell-wide rows of ectodermal cells that
separate mesoderm from lateral neuroectoderm; there is no refinement
from an initial broader domain of expression (Thomas et al. 1988). This
represents the most extreme example of initial dorsal-ventral
patterning in that high levels of sim expression occur in a
single row of cells while it is undetectable in the adjacent cells.
Biochemical, genetic, and molecular studies suggest how this is
achieved (Fig. 5). Genetic studies implicate the
dorsal, snail (sna), twist (twi),
scute (sc), daughterless (da), and
Notch genes in sim activation (Fig. 5A) (Kosman et
al. 1991; Leptin 1991; Rao et al. 1991; Kasai et al. 1992; Lewis 1994).
All are transcription factors or, in the case of Notch, presumably
function through transcription factors. The Dorsal protein, an
NF-
b relative, forms a morphogenetic gradient that is the key
regulator of tissue specification along the dorsal-ventral axis of the
embryo (for review, see Rusch and Levine 1996). Dorsal forms a nuclear
gradient with highest concentrations at the ventral surface of the
embryo (Roth et al. 1989; Rushlow et al. 1989; Steward 1989). Dorsal, in turn, activates twi and sna expression. The Twi
bHLH protein also forms a gradient along the ventral side of the
embryo. Both Dorsal and Twi proteins are found in the mesectodermal
cell anlage, and in both cases these cells lie in a steep region of
their gradients (Leptin and Grunewald 1990; Kosman et al. 1991). The
distribution of the Sna zinc finger protein is more highly restricted;
it is found at high concentrations in the mesoderm but is absent in the
adjacent mesectoderm (Kosman et al. 1991; Leptin 1991). Da and Sc bHLH
proteins are expressed at this time throughout the embryo, forming an
E-box (ACNNGT) binding heterodimer (Jiang and Levine 1993). Thus,
Dorsal, Twi, and Da::Sc act together to activate sim ventrally
in the mesoderm and mesectoderm. Sna, which is restricted to the
mesoderm, represses sim in those cells (Nambu et al. 1990; Rushlow and Arora 1990), leading to sim activation in the
mesectoderm. In addition, the Notch signaling pathway
positively regulates sim transcription (Lewis 1994; Menne and
Klämbt 1994; Martin-Bermudo et al. 1995).
|
Biochemical and germ-line transformation studies indicate that the
dorsal-ventral patterning proteins directly control sim transcription (Fig. 5B) (Kasai et al. 1992; Y. Kasai, M. Sonnenfeld, J. Lewis, S. Stahl, and S. Crews, in prep.). The sim gene
contains two promoters, one (PE) that controls early midline
transcription and another (PL) that controls late midline
transcription (Nambu et al. 1991; Muralidhar et al. 1993). Comparing
the sequence of the sim PE regulatory region between
two different Drosophila species revealed a series of
conserved sequence elements (Wharton et al. 1994), which include
predicted binding sites for Dorsal, Twi, Da::Sc, and Sna. These factors
bind to the sim PE regulatory region in vitro (Fig.
5B; Kasai et al. 1991; Y. Kasai, M. Sonnenfeld, J. Lewis, S. Stahl, and
S. Crews, in prep.). Mutagenesis and analysis by germ-line
transformation indicate that the binding sites are used in vivo:
elimination of Dorsal, Twi, and Da::Sc binding sites results in an
absence of initial mesectodermal transcription. These observations
suggest a model in which the sim early regulatory region
employs binding sites for the cooperatively-acting Dorsal, Twi, and
Da::Sc transcription factors. Presumably, the Dorsal nuclear gradient
is insufficient to establish on-off transcription with single cell
resolution, and thus, additional proteins are required. Sna sets the
ventral boundary of initial sim transcription by repressing
sim in the adjacent mesoderm. As several Da::Sc binding sites
are embedded within a subset of Sna binding sites (Kasai et al. 1992),
one attractive model of Sna repression is that it directly competes
with Da::Sc binding sites in the mesoderm (Ip et al. 1992). Is there a
similar repressor designed to limit the dorsal boundary? This remains
possible. Alternatively, the dorsal boundary may be attributable
strictly to the steep concentration gradients of Dorsal and Twi,
allowing activation of sim transcription in the mesectoderm
but not in more dorsal cells. In summary, the sim gene is
designed to respond directly to multiple positively and
negatively-acting regulatory proteins that are expressed in the early
embryo. These proteins direct sim spatial expression, and also
dictate that sim is first expressed at gastrulation when cell
lineage specification is established.
| |
sim controls CNS midline cell specification |
|---|
Sim protein specifically accumulates in mesectodermal cell nuclei
during gastrulation (Fig. 6A) at the time when
ectodermal cells are acquiring their fates (Crews et al. 1988).
sim is not expressed in other ectodermal cells. Sim protein is
expressed in the midline cells throughout neurogenesis and is present
in the differentiated midline neurons and glia (Crews et al. 1988). Null mutants of sim have a complete absence of midline cell
development: midline precursor cells fail to divide, and subsequently
do not undergo their characteristic cell shape changes or differentiate into neurons and glia (Thomas et al. 1988; Nambu et al. 1991). Sim
exerts its affect by controlling target gene transcription (Nambu et
al. 1990). Midline expression of >20 genes is abolished in
sim mutant embryos (Table 1), and probably
all midline transcription is directly or indirectly dependent on
sim function. The master regulatory role of sim is
reinforced by experiments in which sim is ectopically
expressed using a heat shock-sim transgene (Nambu et al.
1991). If sim is induced in neuroectodermal cells as they are
adopting their fates, they are transformed from lateral CNS into CNS
midline cells. This is accompanied by ectopic gene transcription of
midline-expressed genes.
|
|
Midline cells in sim mutant embryos take on a lateral
neuroectodermal cell fate and misexpress genes that correspond to this lineage (Chang et al. 1993; Mellerick and Nirenberg 1995; Xiao et al.
1996). Thus, it appears that the default state of all neuroectodermal cells is lateral CNS. When sim is turned on in these cells, it activates midline transcription and represses lateral CNS
transcription. The combination of these two activities results in CNS
midline cell development.
| |
Sim controls target gene transcription through a midline enhancer element |
|---|
Once activated in CNS midline cells, Sim controls midline
transcription, and also maintains its own expression by positive autoregulation (Nambu et al. 1991; Muralidhar et al. 1993). Progress in
understanding how sim controls midline transcription has been achieved by identifying target genes and their midline enhancer elements and identifying the bHLH-PAS dimerization partner of Sim.
Numerous genes are expressed in the CNS midline precursor cells soon
after the initial appearance of Sim in cell nuclei (Crews et al. 1992)
that are likely to be directly regulated by sim. In many
cases, the genes were cloned without prior knowledge of their midline
expression, and subsequently shown to be expressed in the CNS midline
cells. However, two enhancer trap screens have identified additional
CNS midline-expressed genes (Klämbt et al. 1991; Crews et al. 1992).
Four sim target genes, breathless (btl),
sim, slit, and Toll (Tl), have been
characterized in detail at the molecular level (Wharton and Crews 1993;
Wharton et al. 1994; Ohshiro and Saigo 1997). Each of these genes
represents a distinct mode of midline regulation: (1) Tl is
expressed in midline precursor cells; (2) sim is an
autoregulatory target; (3) slit is expressed in differentiated midline glial cells; and (4) btl is expressed in both midline and tracheal cells. Each regulatory region was assayed for the ability
to drive lacZ in the midline cells. Deletional analysis and
site-directed mutagenesis identified a CNS
midline enhancer element (CME), with a core
ACGTG sequence (Wharton et al. 1994). This element is found in multiple
copies in btl, sim, and Tl and as a single copy in
slit. The CME is required for midline transcription in all
four genes (Wharton et al. 1994; Ohshiro and Saigo 1997) and, when
multimerized, Tl site 4 is sufficient to drive transcription from a heterologous promoter in the CNS midline precursor cells and
differentiated midline neurons and glia (Fig. 6B) (Wharton et al. 1994;
Sonnenfeld et al. 1997).
| |
Sim::Tango heterodimers activate CNS midline transcription |
|---|
Theoretical (Wharton et al. 1994) and experimental (Swanson et al.
1995) considerations predicted that the ACGTG core sequence was a
binding site for heterodimers between Sim and a Drosophila Arnt-like protein. The Drosophila tango (tgo) gene
was cloned and shown to be highly related to mammalian Arnt (Ohshiro
and Saigo 1997; Sonnenfeld et al. 1997). Several lines of evidence indicate that Sim forms heterodimers with Tgo to bind the CME in vivo
and activate CNS midline gene transcription (Sonnenfeld et al. 1997).
Both proteins are found in CNS midline cells during embryonic
development. Sim and Tgo can form dimers and activate transcription of
a multimerized CME when cotransfected into Drosophila cell
culture. Mutations in both sim and tgo affect CNS
midline transcription and development, including transcription of the multimerized CME transgene. The phenotype of tgo mutants is
less severe than null sim or trh mutations, although
this is most likely because of a maternal contribution of tgo
and the use of hypomorphic tgo alleles. Gene dosage
experiments reveal that loss of a single copy of sim enhances
the CNS midline phenotype of tgo mutations resulting in a
severe "sim-like" collapsed CNS phenotype. Activation of
transcription in vivo by Sim::Tgo heterodimers is likely to be direct
since both proteins possess potent transcriptional activation domains
(Franks and Crews 1994; Sonnenfeld et al. 1997).
| |
Sim autoregulation and early and late phase transcription |
|---|
Once sim is localized to cell nuclei, it activates
transcription of target genes in CNS midline precursor cells. In
addition, sim transcription is controlled by a positively
acting autoregulatory feedback loop (Nambu et al. 1991). This may be a
mechanism in which the midline lineage-conferring properties of
sim are maintained throughout embryonic development.
Autoregulation has an interesting twist in that both sim
promoters, PE and PL, are autoregulated. PE continues to be transcribed in the midline because of
Sim::Tgo function (Nambu et al. 1991). Eventually, this transcription
is extinguished by an unknown mechanism as the midline precursors differentiate into neurons and glia. PL is also activated by
Sim autoregulation (Nambu et al. 1991; Muralidhar et al. 1993). This promoter drives transcription in the midline precursor cells, later in
the differentiated midline cells, and in a subset of muscle precursor
cells (Lewis and Crews 1994). Although it is clear that the maintenance
of sim transcription is dependent on sim, the
developmental significance of later sim transcription is
unknown, as genetic studies that completely eliminate only late
sim function have not been carried out. However, the midline glial enhancer of the slit gene contains a CME that is required for
midline transcription suggesting that Sim::Tgo or related proteins function in
late phases of midline transcription (Wharton et al. 1994).
| |
Mammalian sim and Down Syndrome |
|---|
Two mammalian sim orthologs (Sim1 and
Sim2) have been discovered that share a number of functional
similarities with Drosophila sim (for review, see Michaud and
Fan 1997). Both Sim1 and Sim2 proteins dimerize with
mammalian Arnt and bind the CME in vitro (Ema et al. 1997a; Probst et
al. 1997). The mammalian and Drosophila genes are both
expressed in the developing CNS and mesoderm (Lewis and Crews 1994;
Dahmane et al. 1995; Fan et al. 1996; Ema et al. 1997a). Within the
CNS, Drosophila sim expression is restricted to the CNS
midline cells. The floor plate of the vertebrate spinal cord is thought
to be the analogous cell type. Neither Sim1 nor Sim2
is expressed in the floor plate, but both are expressed in the ventral
diencephalon and Sim1 is present in the spinal cord cells
adjacent to the floor plate. The Sim genes are expressed early
in brain development suggesting that they might play roles in
neurogenesis analogous to Drosophila sim.
The roles of Sim1 and Sim2 in embryonic development
should be soon forthcoming since both genes have been knocked out in
mice (Michaud and Fan 1997). One other distinction is that cell culture transfection experiments suggest that both mammalian Sim
proteins function as transcriptional repressors (Ema et al. 1997a;
Moffett et al. 1997). While both cell culture and in vivo experiments have established the ability of Drosophila Sim to activate
transcription (Franks and Crews 1994), genetic experiments have
established that sim can also repress transcription (Chang et
al. 1993; Mellerick and Nirenberg 1995; Xiao et al. 1996). It will be
interesting to see if Drosophila sim-mediated midline repression is
mechanistically similar to mammalian Sim repression.
The most intriguing aspect of mammalian Sim is that
Sim2 maps to Chromosome 21 in the region responsible for Down
Syndrome (DS) (Chen et al. 1995; Dahmane et al. 1995; Muenke et al.
1995; Chrast et al. 1997). Given the important role of sim in
Drosophila development and the expression of Sim2 in
cell types that are affected in DS individuals, it was proposed that
Sim2 may play a causative role in DS. This remains
speculative, however, as evidence is lacking and other candidate DS
genes exist. However, the existence of mouse models of DS (Reeves et
al. 1995) and systematic approaches to uncover the genetic basis for DS
(Lamb and Gearhart 1995) will hopefully provide answers to this
question. Since DS is a trisomy of chromosome 21 (and Sim2),
if Sim2 does play a role in DS, one possibility is that
surplus Sim2 protein may act by excessively binding Arnt,
leaving Arnt unable to interact with other bHLH-PAS proteins that are
critical for proper development or cellular function.
| |
Trachealess::Tgo heterodimers control formation of the trachea and salivary ducts |
|---|
The trachealess (trh) gene encodes a bHLH-PAS protein
that is specifically expressed in the developing trachea plus posterior spiracle, salivary gland placode, and salivary ducts (Isaac and Andrew
1996; Wilk et al. 1996). Trh protein is found in the developing tracheal pits and the later-formed tracheal tubules (Fig. 6C; see also
Wilk et al. 1996). Genetic analysis indicates that trh is
required for the formation of trachea and controls the transcription of
genes involved in this process (Younossi-Hartenstein and Hartenstein 1993; Isaac and Andrew 1996; Wilk et al. 1996). Ectopic expression of
trh results in formation of ectopic trachea (Wilk et al.
1996). trh is also required for formation of the posterior
spiracle and salivary duct (Isaac and Andrew 1996). These results
indicate that trh function is required for the specification
and invagination of the trachea, and probably acts similarly in the
development of the posterior spiracle and salivary duct.
The functional similarity between Sim and Trh is remarkable; both are
lineage-specific regulators, autoregulatory, and bHLH-PAS transcriptional activators. Work described below indicates that Sim and
Trh control transcription in a similar fashion by binding the same DNA
sequence element using Tgo as a dimerization partner (Ohshiro and Saigo
1997; Sonnenfeld et al. 1997; Zelzer et al. 1997). This is particularly
interesting since it is commonly observed that many genes expressed in
the CNS midline cells are also expressed in trachea (Manning and
Krasnow 1993).
Numerous experiments indicate that Tgo is a dimerization partner for
Trh, and that together they bind the CME and activate tracheal
transcription in vivo. Tgo can form dimers with Trh assayed by
two-hybrid analysis and co-immunoprecipitation (Sonnenfeld et al.
1997), Trh::Tgo binds CME-containing DNA in vitro (Ohshiro and Saigo
1997), and Trh::Tgo can activate transcription from CME-bearing
promoters in Drosophila cell culture (Sonnenfeld et al. 1997).
In addition, tgo mutants show tracheal defects, and double
mutant analysis reveals genetic interactions between trh and
tgo, further indicating in vivo associations (Sonnenfeld et al. 1997). Analysis of embryos harboring a multimerized Tl
site 4 CME shows that this transgene is expressed not only in the CNS midline cells, but in the trachea and salivary duct, those cells in
which sim and trh function (Fig. 6B) (Sonnenfeld et
al. 1997; Zelzer et al. 1997). Consistent with the idea that both
sim and trh act through the CME, mutants in
sim specifically abolish CME midline expression, whereas
mutants in trh abolish CME tracheal and salivary duct
expression (Sonnenfeld et al. 1997). Additional evidence that Trh::Tgo
functions through the CME comes from work on btl (Ohshiro and
Saigo 1997), which is expressed in both trachea and CNS midline cells.
The btl gene contains three CMEs upstream of the promoter and
mutational analysis indicates that they are required for both CNS
midline and tracheal expression. The tgo gene is also
expressed at elevated levels in tracheal cells (Ohshiro and Saigo 1997;
Sonnenfeld et al. 1997), raising the possibility that tgo is
autoregulated by Trh::Tgo heterodimers.
| |
Control of bHLH-PAS protein nuclear localization: Sim and Trh direct the nuclear accumulation of their respective Sim::Tgo and Trh::Tgo heterodimers |
|---|
The close biochemical relationship between Sim, Trh and the
ligand-binding Ahr has engendered speculation that nuclear localization of Sim and Trh are controlled by small molecule binding. This view is
reinforced by the binding of Hsp90 to Sim in vitro (McGuire et al.
1996; Probst et al. 1997), suggesting that Sim may be held in a
ligand-responsive state as proposed for the Ahr-Hsp90 complex. Examination of the subcellular localization of Sim, Trh, and Tgo in
wild-type, mutant, and transgenic Drosophila embryos has provided insight into whether their nuclear localization is regulated by ligand.
Sim protein is first detected as the mesectodermal cells move towards
the midline at gastrulation, and imediately accumulates in cell nuclei
(Fig. 6A; see also Crews et al. 1988). Sim protein remains
predominantly nuclear in the CNS midline cells throughout embryonic
development. When sim is ectopically expressed in the embryo,
Sim protein also rapidly accumulates in nuclei (Ward and Crews 1998).
Thus, the localization of Sim during normal development does not
provide positive evidence for regulated nuclear localization. If
nuclear localization of Sim is dependent on binding to an unknown ligand, then the ligand is not spatially or temporally localized (and
thus, not developmentally significant). In a similar fashion, Trh
protein appears in tracheal pit cell nuclei soon after it can be
detected, and remains in tracheal cell nuclei throughout embryogenesis
(Fig. 6C; see also Wilk et al. 1996). Ectopic expression experiments
also reveal that Trh is localized to nuclei in all embryonic cells
assayed (Ward and Crews 1998).
In contrast, Tgo protein localization is more dynamic. Tgo protein is
found in all embryonic cells (Sonnenfeld et al. 1997), but its nuclear
localization correlates with sites of function (Ward and Crews 1998).
Tgo is localized to the cytoplasm in many cells, but is nuclear in the
CNS midline cells and trachea (Fig. 6D). Ectopic expression experiments
show that Tgo accumulates in cell nuclei in all cells in which Sim and
Trh are expressed.
These results suggest a model (Fig. 7), in which Tgo
is retained in the cytoplasm in the absence of a dimerizing bHLH-PAS protein. When Sim or Trh protein appears, it forms dimers with Tgo, and
the Sim::Tgo or Trh::Tgo dimer complex enters the nucleus. Although it
remains possible that Sim or Trh binds and responds to a ligand during
either embryogenesis or postembryonically, these results suggest that
it is more likely that their nuclear localization is not ligand
responsive. The role of Hsp90 in binding to Sim may be to facilitate
dimerization of Sim to Tgo, as has been postulated for other bHLH
proteins (Shue and Kohtz 1994), rather than promote ligand
interactions. Thus, ligand-mediated control of nuclear transport may
not be a feature common to all bHLH-PAS proteins and specificity of
Sim::Tgo and Trh::Tgo function is instead dependent on expression of
sim in mesectodermal cells and trh in ectodermal
cells by dorsal/ventral and
anterior/posterior patterning proteins. The situation may
be similar in mammals since Arnt subcellular localization varies
spatially and temporally during embryonic development (Abbott and
Probst 1995). However, differences exist, since mammalian Arnt
possesses nuclear localization sequences absent in Drosophila
Tgo (Eguchi et al. 1997), and Arnt is localized to cultured cell nuclei
in the absence of any obvious bHLH-PAS protein partner (Pollenz et al.
1994; Eguchi et al. 1997).
|
| |
bHLH-PAS proteins regulate hypoxia responsiveness |
|---|
The cellular response to oxygen deprivation can trigger a number
of important physiological responses that are controlled, in part, at
the level of transcription. In mammals, depending on the cell type,
these can include induction of glycolytic pathway enzymes,
erythopoiesis, and angiogenesis. The breakthrough concerning how these
responses are controlled was the identification of the key regulatory
protein, HIF (Wang et al. 1995). HIF was shown to consist of two
subunits, HIF-1, a Sim-related bHLH-PAS protein, and HIF-1
,
which is Arnt. Interestingly, the binding site for HIF, the
hypoxia response element
(HRE), contains a core ACGTG sequence, identical to the CME (Firth et
al. 1994; Semenza et al. 1994). New mammalian bHLH-PAS proteins, such
as endothelial PAS domain protein 1 (EPAS1), (Ema et al. 1997b;
Hogenesch et al. 1997; Tian et al. 1997) that are related to HIF-1
and also form dimers with Arnt, are also likely to play roles in
controlling the physiological response to oxygen levels.
Insects also respond to oxygen deprivation by transcriptional
up-regulation of glycolytic pathway genes (Nagao et al. 1996). Biochemical studies in Drosophila cell culture have identified a hypoxia-inducible factor that can bind a HRE (Nagao et al. 1996). Although biochemical identification of the protein factors is lacking,
Tgo may be a constituent of the HRE binding activity. If so, one
candidate for Tgo's partner is the Sima bHLH-PAS protein (Nambu et
al. 1996). Sima is most related to HIF-1, is
ubiquitously-expressed in the embryo as is HIF-1, and can form
stable dimers with Tgo (Nambu et al. 1996; Sonnenfeld et al. 1997).
Another relevant aspect of insect respiratory physiology concerns the
regulation of tracheal terminal branching. Studies in the blood-sucking
insect, Rhodnius, demonstrated that branching is dependent on
the oxygen levels of the surrounding tissues (Wigglesworth 1954). It is
tempting to speculate that a HIF-like activity operating in the cells
adjacent to the tracheoles may regulate terminal branching (Guillemin
and Krasnow 1997).
| |
Coregulators influence the specificity of bHLH-PAS target gene transcription |
|---|
There exist a number of genes expressed in both CNS midline cells
and trachea that are likely targets of both Sim and Trh. However, many
genes regulated by Sim and Trh are expressed in only one of the two
cell types. As Sim::Tgo and Trh::Tgo (as well as HIF-1::Arnt)
heterodimers bind the same core ACGTG sequence in vivo to regulate
target gene transcription, additional regulatory elements and factors
are required to generate transcriptional specificity. Experiments using
the multimerized Tl site 4 CME in vivo and in cell culture
suggest this element is sufficient for transcription in midline
precursors, midline neurons and glia, trachea, and SL2 cells
(Sonnenfeld et al. 1997; Zelzer et al. 1997). However, tracheal
expression is weak compared to midline transcription (Fig. 6B).
Misexpression studies suggest that Trh requires an additional factor
restricted to dorsal ectoderm for activation of the CME, whereas Sim
can activate CME transcription throughout the ectoderm and other cell
types (Zelzer et al. 1997; Ward and Crews 1998). This restriction does
not function at the level of Sim, Trh, and Tgo nuclear localization,
since ectopic expression experiments indicate that the proteins are
localized to nuclei in all cell types examined.
Additional factors besides the CME are necessary for transcriptional
activation by Trh. Mutational analysis of the btl gene has
shown that the CME and adjacent sequences are both required for
tracheal transcription (Ohshiro and Saigo 1997). The rhomboid gene is expressed in both CNS midline and tracheal cells, and an 0.7-kb
fragment containing 4 CMEs is expressed in both tissues (Ip et al.
1992; S.T. Crews, unpubl.). However, an 0.3-kb subfragment containing 2 CMEs is strongly expressed in the midline, but is greatly reduced in
the trachea (Ip et al. 1992), indicating the presence of elements
required for tracheal expression distinct from those required for
midline expression. Ectopic expression studies also suggest the
existence of tracheal-specific elements distinct from the CME (Zelzer
et al. 1997). In a model consistent with existing data (Zelzer et al.
1997), it is proposed that midline-specific target genes cannot be
activated by Trh::Tgo because they lack tracheal-specific control
elements in addition to the CME. Presumably, tracheal-specific target
genes cannot be activated by Sim::Tgo in the midline because of the
existence of positive or negative control elements in addition to the CME.
Elegant work on bHLH proteins that control myogenesis and neurogenesis
have demonstrated that specific basic region residues are necessary for
transcriptional specificity. In the case of the vertebrate myogenic
bHLH proteins, including MyoD and Myogenin, it has been shown that two
adjacent basic region residues are required for muscle-specific
transcription (Davis et al. 1990; Brennan et al. 1991, Davis and
Weintraub 1992). Biochemical experiments have shown that these residues
are required for interaction of the MEF2 coregulatory MADS-box protein
with the MyoD::E12 bHLH heterodimer (Molkentin et al. 1995). Mutational
and chimeric protein studies of the Drosophila neurogenic bHLH
proteins Atonal and Scute also suggest that residues within their basic
regions are involved in the ability of these proteins to activate
transcription in different classes of nerve cells (Chien et al. 1996).
Similar experiments carried out with Sim and Trh indicate that
transcriptional specificity resides not within the basic region, but
within the PAS domain (Zelzer et al. 1997). It was proposed that the
PAS domain mediates interactions with the additional factors
hypothesized to impart tissue specificity.
Genetic experiments indicate that sim can act as a midline
repressor, as well as activator. The biochemical mechanism of midline repression by sim is unknown. It could involve binding of
Sim::Tgo heterodimers to CMEs on target genes, in which case, the
presence of adjacent corepressor sites would dictate repression instead of activation. This mechanism is analogous to how Dorsal can both activate transcription ventrally in the blastoderm embryo and repress
ventral transcription in combination with sites of corepression (Jiang
et al. 1993; Kirov et al. 1993; Huang et al. 1995). An alternative
mechanism postulates Sim disrupting a positively acting transcription
complex. If so, it is unlikely to be inhibiting a bHLH-PAS::Tgo
heterodimer since cellular studies show Tgo to be cytoplasmic (and
presumably transcriptionally inert) in the lateral CNS. Answers to
questions of transcriptional specificity and mode of action await a
concerted analysis using germline transformation, biochemistry, and
genetic approaches.
| |
Evolutionary conservation and functional diversity of the bHLH-PAS regulatory cassette |
|---|
bHLH-PAS proteins mediate a wide variety of biological processes,
which raises two issues. (1) Does the PAS domain carry out related
biochemical functions in these disparate developmental and
physiological events? (2) Is there a common origin to these biological
events? The PAS domain clearly represents a polyfunctional interaction
domain. Its large size allows a variety of interactions facilitating
complex regulation of protein function. There are common functions to
PAS domains: most bHLH-PAS proteins require the PAS domain for
interaction with Arnt and Hsp90. In contrast, the relative lack of
sequence conservation within the PAS domain suggests that different PAS
proteins can mediate distinct molecular interactions. For example, Per
interacts through its PAS domain with Timeless (Tim), a non-PAS protein
(Gekakis et al. 1995). Yet, other bHLH-PAS proteins do not interact
with Tim (G. Nystrom and S.T. Crews, unpubl.). Ahr interacts with
halogenated aromatic hydrocarbons such as TCDD (dioxin) through its PAS
domain to control nuclear localization, yet Sim, Per, and Arnt do not
bind TCDD (Swanson and Bradfield 1993; Coumailleau et al. 1995). In all cases, the bHLH-PAS protein PAS domain mediates protein-protein interactions. These can be regulated by interactions with other molecules or unregulated.
Is there a functional connection between the different developmental
and physiological events governed by bHLH-PAS proteins? There are some
interesting similarities. Two of the basic biological processes that
bHLH-PAS proteins participate in are biological rhythms and response
to oxygen levels. Controlling gene expression in response to the
circadian light/dark cycle is widespread across phylogeny. The identification of related PAS proteins implicated in
rhythms between insects and mammals (King et al. 1997; Z.S. Sun et al.
1997; Tei et al. 1997) indicates that the mechanism of circadian
regulation is evolutionarily well conserved. More surprising is the
discovery that fungal PAS proteins mediate light-controlled rhythmic
behavior (Linden and Macino 1997), suggesting an even stronger
association between the PAS domain and regulation of rhythms.
Conceivably, this could be a primordial function of bHLH-PAS proteins.
Physiological regulation of oxygen responsiveness is another basic
organismal function necessary since the origins of the oxygen-rich
environment 1.4 billion years ago (Bunn and Poyton 1996). bHLH-PAS
proteins control both developmental and physiological aspects of oxygen
delivery. HIF and related proteins control the response to oxygen
levels, including finer aspects of vascular branching. Development of
the respiratory system including the formation of tracheal tubules is
controlled by trh. These genes could be specializations of a
more primitive respiratory system regulatory protein. More speculative
is the possible relationship between CNS midline and tracheal cell
development. Although the CNS and trachea have different functions,
numerous genes are utilized in the development of both lineages, and
both tissues are ectodermal derivatives in which bHLH-PAS::Tgo
heterodimers are activated in undifferentiated cells to form their
respective tissues. It is possible that the CNS midline cells, which
comprise a tissue distinct from the lateral CNS, and the trachea may
have a common evolutionary origin. Although uncovering ancestral
relationships from extant creatures can be problematic, hopefully
functional analysis of PAS proteins from different organisms may shed
light on these issues.
Comparative analysis of bHLH-PAS gene functions (Figs. 2 and 7)
indicates that they constitute an evolutionarily-conserved regulatory
gene system (Sonnenfeld et al. 1997). Organisms as diverse as
Caenorhabditis elegans, Drosophila, and mammals have Arnt
proteins that can form heterodimers with a variety of
Sim/Ahr-related bHLH-PAS proteins. These heterodimers
bind a sequence element related to the XRE (core GCGTG) or CME (core
ACGTG). From this original regulatory cassette has emerged gene
combinations that control developmental processes including
neurogenesis and tracheal formation, and physiological processes
including toxin metabolism, the response to oxygen deprivation and
probably circadian rhythms (Fig. 8). Analysis of Ahr,
Sim, and Trh have established that there are at least two fundamental
modes of bHLH-PAS protein function. Ahr is a broadly distributed
protein whose function is controlled by ligand dependent nuclear
transport. In contrast, the Sim and Trh developmental regulators are
not controlled at the level of nuclear transport but achieve
specificity of function by virtue of their restricted expression. It
will be interesting to see whether other bHLH-PAS proteins fit these
modes or are regulated in novel ways.
|
| |
Acknowledgments |
|---|
I thank the scientists in my laboratory, Wei Chen, Robert Franks, Song Hu, Yumi Kasai, Sang Hee Kim, Josette Lewis, Beverly Matthews, Jack Mosher, John Nambu, Jay Nystrom, Margaret Sonnenfeld, Stephanie Stahl, Mary Ward, and Keith Wharton, who have worked on bHLH-PAS proteins and midline development for many stimulating discussions and collaborations. My initial interest in this area began with a productive and enjoyable collaboration with John Thomas and Corey Goodman. I also thank Chen-Ming Fan, Oliver Hankinson, Michael Levine, Lorenz Poellinger, Michael Rosbash, and Greg Semenza for useful discussions, Mark Peifer for critically reading the manuscript, and Mary Ward for generating the images shown in Figure 6. Work from my laboratory was supported by the National Institute of Child Health and Human Development, the National Science Foundation, and the Lucille P. Markey Charitable Trust.
| |
Note added in proof |
|---|
Analysis of the cell culture-derived Ahr D mutation indicates that
the mutation is in the PAS domain and that the mutant protein is
capable of ligand binding and dimerization to Arnt, but binds DNA
weakly. This suggests that the PAS domain may directly influence DNA
binding (W. Sun et al. 1997).
| |
Footnotes |
|---|
1 E-MAIL steve_crews{at}unc.edu; FAX (919) 962-3155.
| |
References |
|---|
|
Top
Introduction
References
|
|---|
is a novel bipartite type recognized by the two components of nuclear pore-targeting complex.
J. Biol. Chem.
272:
17640-17647
) sim transcription
in those cells. (B) The 2.8-kb sim regulatory DNA
that is sufficient for initial sim transcription in
mesectodermal cells is indicated by the blue box. The arrow denotes
PE and the direction of transcription. Below are shown the
locations of binding sites for Dorsal, Twi, and Sna. The lavendar boxes
at the bottom show the subset of Sna binding sites that contain E-boxes
that bind Da::Sc. Mutation of selected Dorsal, Twi, and Sna-E
sequences results in an absence of midline transcription indicating
that these proteins function in vivo to directly regulate sim
transcription.
