(Received for publication, December 30, 1996, and in revised form, March 4, 1997)
From the Department of Molecular Pathology,
University College London Medical School, London W1P 6DB,
United Kingdom and the § Division of Neuropathology,
University of Pennsylvania Medical School,
Philadelphia, Pennsylvania 19104
The POU domain transcription factor Brn-3a is able to stimulate neurite outgrowth when overexpressed in the neuronal ND7 cell line, whereas the closely related Brn-3b factor does not have this effect. We show that Brn-3a overexpression also enhances the expression of the three neurofilament genes at both the mRNA and protein levels, whereas Brn-3b overexpression has no effect. In addition Brn-3a activates the three neurofilament gene promoters in co-transfection assays in both neuronal and non-neuronal cells. As observed for enhanced neurite outgrowth, the stimulation of neurofilament gene expression and activation of the neurofilament gene promoters is observed with the isolated POU domain of Brn-3a. A single amino acid change in the POU homeodomain of Brn-3a to the equivalent amino acid in Brn-3b abolishes its ability to activate the neurofilament promoters, whereas the reciprocal change converts Brn-3b to an activator of these promoters.
The POU (Pit-Oct-Unc) family of transcription factors (for reviews see Refs. 1 and 2) was originally defined on the basis of a common 150-160-amino acid DNA binding domain held in common between the mammalian transcription factors Pit-1, Oct-1, and Oct-2 and the regulatory protein Unc-86 which plays a key role in the development of neuronal cell types, particularly sensory neurons in the nematode (3, 4). A number of subsequent studies (see for example Ref. 5) have defined a large number of other POU family transcription factors in a variety of organisms that have been classified into subgroups according to their relationships to one another (for review see Ref. 2). Among the mammalian POU factors identified in this way, the Brn-3 factors show the closest homology to the nematode protein Unc-86 and together with the Drosophila factors I-POU and tI-POU (6, 7) these factors constitute the POU IV subfamily within the POU family (2). Indeed it has been proposed (8) that the Brn-3 factors constitute the mammalian homologues of Unc-86 (8).
Unlike the situation in the nematode, however, three distinct Brn-3 factors have been identified in mammals that show close homology in the DNA binding POU domain but are considerably less homologous outside it and are encoded by three different genes (9, 10). These factors are Brn-3a (also known as Brn-3 or Brn-3.0) (5, 11, 12), Brn-3b (also known as Brn-3.2) (12, 13), and Brn-3c (also known as Brn-3.1) (11, 14). Each of these three factors is expressed in distinct but overlapping sets of neurons in the developing and adult nervous system (5, 11, 13-15) with Brn-3a for example defining the earliest post-mitotic neurons to appear in the developing central nervous system (15). This suggests that, like Unc-86, the mammalian Brn-3 factors may play a critical role in the development of specific neuronal cells within the nervous system with each factor having distinct but possibly related roles. In agreement with this, recent experiments (8, 16) have shown that inactivation of the Brn-3b gene results in considerable loss of retinal neurons, whereas mice in which the Brn-3c gene has been inactivated show loss of olfactory neurons, and Brn-3a inactivation results in the loss of neurons in the brain stem and trigeminal ganglia (16).
These distinct functional roles and expression patterns of the Brn-3
factors are paralleled by their distinct effects on target genes. Thus
it has previously been shown that the Brn-3a factor can activate a wide
variety of target genes expressed in neuronal cells including those
encoding pro-opiomelanocortin (11), the neuronal intermediate filament
-internexin (17), and the synaptic vesicle protein SNAP-25 (18, 19).
In contrast Brn-3b represses the promoters of the
-internexin and
the gene encoding SNAP-25 and also interferes with their activation by
Brn-3a (17, 19), although it can stimulate the promoter of the neuronal
nicotinic acetylcholine receptor
2 subunit gene which is not
affected by Brn-3a (20).
To analyze the functional effects of the Brn-3 factors in more detail, we have used the ND7 neuronal cell line which was generated by the fusion of a mouse neuroblastoma cell line and primary rat dorsal root ganglion neurons (21). We have previously shown (12, 22) that when these cells are induced to differentiate to a non-dividing phenotype bearing numerous neurite processes (23), the levels of Brn-3a rise dramatically from a low level in the proliferating cells to a much higher level in the non-dividing cells. Conversely, the levels of Brn-3b fall from a high level in the dividing cells to a much lower level in the differentiated cells, whereas the levels of Brn-3c remain unchanged. Moreover, such changes in the expression of Brn-3a and Brn-3b appear to have functional effects. Thus as would be expected, the rise in the level of the Brn-3a activator together with the fall in the level of the Brn-3b repressor which occur during ND7 cell differentiation result in the activation of a transfected synthetic target promoter containing a binding site for these factors, when ND7 cells are induced to differentiate (22, 24). Similarly, the expression of the endogenous gene encoding SNAP-25 (whose promoter is activated by Brn-3a and repressed by Brn-3b in co-transfection experiments) (18, 19) is considerably enhanced during ND7 cell differentiation. Indeed we have recently obtained evidence that the rise in Brn-3a and the fall in Brn-3b occurring during ND7 cell differentiation play a critical role in the differentiation process itself. Thus preventing the rise in Brn-3a using an antisense approach prevents ND7 cell differentiation and the rise in SNAP-25 expression which normally occurs in response to differentiation-inducing stimuli such as serum removal (18). Moreover, artificial overexpression of Brn-3a in ND7 cells results in clonal cell lines which extend neurite processes when grown in full serum-containing medium which does not normally induce the differentiation of these cells (25). Conversely ND7-derived cell lines overexpressing Brn-3b fail to show neurite outgrowth when removed from serum which would normally induce such outgrowth in parental ND7 cells (26).
These findings indicate therefore that the balance between Brn-3a and Brn-3b is likely to play a critical role in the differentiation of ND7 cells, and they allow the use of these cells as a model system to understand the role of these factors in neuronal differentiation and to elucidate the mechanism by which they act. In view of the finding that SNAP-25 is essential for neurite outgrowth by a variety of different neuronal cells in vitro and in vivo (27), it is likely that the ability of Brn-3a to induce SNAP-25 expression plays a role in its ability to induce neurite outgrowth. However, it is likely that this dramatic effect of Brn-3a on the phenotype of ND7 cells involves more than its ability to induce SNAP-25 and that other target genes encoding proteins involved in neurite process formation are also likely to be activated by Brn-3a. We previously found, however, that the growth-associated protein GAP-43 shows no change in its expression in the Brn-3a or Brn-3b overexpressing ND7 cells, in agreement with our finding that its promoter is unaffected by co-transfection with Brn-3a (25). In view of the importance of the neurofilament proteins in neuronal process formation and particularly in specifying axonal calibre (for review see Ref. 28) and the simultaneous appearance of neurofilament and Brn-3a expression during early post-mitotic neuronal differentiation (15), we have investigated the expression of the neurofilament heavy, medium, and light chain genes in our ND7 cells overexpressing different Brn-3 transcription factors and have also studied the effect of the Brn-3 factors on the promoters of these genes in co-transfection experiments carried out in both neuronal and non-neuronal cells.
BHK-21 cells (29) and ND7 cells (21) were routinely cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and L15 medium containing 10% fetal calf serum, respectively. The ND7-derived cell lines overexpressing Brn-3a, Brn-3b, or Brn-3c were constructed as described previously (25, 26) by stable transfection of ND7 cells with cDNA clones for Brn-3a, Brn-3b, or Brn-3c under the control of the glucocorticoid-inducible mouse mammary tumor virus promoter in the vector PJ5 (30). These cells were grown in L15 medium containing 10% fetal calf serum supplemented with G418 to a final concentration of 800 µg/ml to maintain expression of the transgene. Treatment with dexamethasone at a final concentration of 1 µM was used to induced expression of the murine mammary tumor virus promoter.
Plasmid DNAThe neurofilament reporter gene constructs containing various regions of the NF-H, NF-M, and NF-L promoter linked to the readily assayable chloramphenicol acetyltransferase gene have been described previously (31, 32). In the transient transfections, the Brn-3a and Brn-3b expression vectors contain full-length cDNA or cDNAs encoding various regions of the molecule cloned under the control of the Moloney murine leukemia virus promoter in the vector pJ4 (30). The mutant forms of Brn-3a and Brn-3b containing an altered amino acid at position 22 in the homeodomain were constructed as described previously using the pALTER-1 vector (Promega) and were then subcloned into the pJ4 vector for use in transient transfection experiments (33).
Transient TransfectionTransient transfection of DNA was
carried out according to the method of Gorman (34). Routinely 1 × 106 BHK-21 cells (29) or ND7 cells (21) were transfected
with 10 µg of the reporter plasmid and 10 µg of the Brn-3
expression vectors. In all cases cells were harvested 72 h later.
The amount of DNA taken up by these cells in each case was measured by
slot blotting the extract and hybridization with a probe derived from the ampicillin resistance gene in the plasmid vector (35). This value
was then used to equalize the values obtained in the
CAT1 assay as a control for
differences in uptake of plasmid DNA in each sample. In parallel
experiments, we also included an internal control plasmid in which the
cytomegalovirus promoter drives -galactosidase expression allowing
equalization of the samples on the basis of
-galactosidase activity
prior to assay of CAT activity.
Assays of chloramphenicol acetyltransferase activity were carried out according to the method of Gorman (34) using samples that had been equalized for protein content as determined by the method of Bradford (36).
RNA Isolation and QuantitationRNA was isolated from stably
transfected cells by the guanidinium thiocyanate method (37). The
expression of the neurofilament RNAs was then quantitated using a
reverse transcriptase/polymerase chain reaction assay. Initially all
samples were amplified using primers for the constitutively expressed
mRNA encoding the L6 ribosomal protein. Subsequently the
neurofilament mRNAs were amplified using specific primers for each
mRNA. Primers were as follows: NFL 5-CCGTACTTTTCGACCTCCTA-3
and
5
-TCTCTTGTGTGCGGATAGAC-3
; NFM 5
-ACACCATCCAGCAGTTGGAA-3
and
5
-TGTGTTGGACCTTGAGCTTG-3
; NFH 5
-CCCAAGAGGAGATAACTGAG-3
and
5
-TCAATGTCCAGGGCCATCTT-3
; L6 5
-ATCGCTCCTCAAACTTGACC-3
and
5
-AACTACAACCACCTCATGCC-3
. All amplifications were carried out using a
number of cycles and amount of mRNA that had been determined in
preliminary experiments to be in the range of the assay where the
amount of product formed was linearly related to the amount of input
RNA.
Protein was isolated from cells by freeze-thaw extraction in protein buffer (20 mM HEPES, pH 7.9, 0.45 M NaCl, 25% glycerol, 0.2 mM EDTA contains 1 µg/ml leupeptin, 1 µg/ml pepstatin, and 1 mM dithiothreitol) and submitted to analysis by SDS-polyacrylamide gel electrophoresis. Gels were transferred to nitrocellulose filter by Western blotting, and the neurofilament proteins were detected by probing with the mouse monoclonal antibodies NM-08-304, NM-08-307, and NM-08-306 (Genosys). To ensure equal loading of protein samples, gels were stripped and reprobed using a control antibody, in this case pGp9.5.
The ND7-derived cell lines overexpressing individual forms of Brn-3 were constructed (25, 26) by stably transfecting ND7 cells with expression vectors containing each form of Brn-3 under the control of the dexamethasone-inducible murine mammary tumor virus promoter in the vector pJ5 (30). Both by RNA analysis (25, 26) and DNA mobility shift assays using an oligonucleotide containing a binding site for the Brn-3 factors (data not shown), all the cell lines showed some overexpression of the appropriate form of Brn-3 in the absence of dexamethasone with enhanced expression being inducible upon addition of 1 µM dexamethasone. Although expression is therefore somewhat leaky and induced severalfold by hormone, it is possible to use these cells to examine the effects of overexpressing different forms of Brn-3 by studying these effects both in the absence or presence of dexamethasone and comparing the results to similarly treated control cells. We therefore prepared mRNA and protein from the various ND7-derived cell lines proliferating in full serum-containing medium either in the presence or absence of dexamethasone. In each case mRNA and protein was prepared from three independent cell lines of each type which had previously been shown to overexpress Brn-3a, Brn-3b, or Brn-3c (25, 26). Initially, the expression of each of the neurofilament proteins was quantitated using a Western blotting assay.
As shown in Fig. 1, each of the
neurofilament proteins showed a somewhat enhanced level in the Brn-3
overexpressing cells in the absence of steroid compared with control
cells stably transfected with the pJ5 vector alone. Moreover, upon
steroid induction of exogenous Brn-3a gene expression, the levels of
each of the neurofilament proteins increased in the Brn-3a cell lines
but not in the control cells. This resulted in overexpression in the
Brn-3a cell lines of 3.7-fold for NF-H (p < 0.01)
2.2-fold for NF-M (p < 0.05), and 3-fold for NF-L
(p < 0.01) relative to control cells. Hence, the
expression of all three neurofilament proteins is enhanced in the
Brn-3a expressing cells. In contrast no increase in expression of any
of the neurofilament proteins was observed in the Brn-3b expressing
cells, whereas only small (less than 2-fold) increases were observed in
the Brn-3c expressing cells.
Having established that Brn-3a can enhance neurofilament protein levels, we wished to determine whether this was paralleled by increased neurofilament mRNA levels as would be expected if Brn-3a directly activates the neurofilament gene promoters. The relative expression of each of the neurofilament mRNAs was therefore quantified in the cells using a reverse transcriptase/polymerase chain reaction assay.
In these experiments (Fig. 2) the
neurofilament heavy chain mRNA was overexpressed approximately
2-fold in the Brn-3a overexpressing cells compared with control cells
stably transfected with the pJ5 vector alone (p < 0.01). Upon steroid induction this overexpression of the neurofilament
heavy chain gene rose to approximately 6-fold over that observed in
similarly treated control cells (p < 0.01). Similar
overexpression of the neurofilament light chain of approximately 2-fold
before steroid treatment (p < 0.01) and of
approximately 5-fold after steroid induction (p < 0.01) was also observed in the Brn-3a overexpressing cells, whereas the
neurofilament medium chain mRNA showed only marginal overexpression
in the Brn-3a cells before steroid treatment but was overexpressed
approximately 3-fold (p < 0.01) after steroid
treatment (Fig. 2). In contrast no significant overexpression of any of
the neurofilament genes was observed in the Brn-3b or Brn-3c
overexpressing cells either before or after steroid treatment.
These findings indicate therefore that the overexpression of Brn-3a in the ND7 cells can induce the overexpression of all three of the neurofilament genes at both the mRNA and protein levels. Thus prior to dexamethasone treatment when some overexpression of Brn-3a is observed in the ND7 cells (25), there is also some overexpression of the different neurofilament genes. Moreover, when enhanced expression of Brn-3a is induced by steroid treatment, this results in significant overexpression of all three neurofilament genes above the level observed in comparably treated control cells with the neurofilament heavy chain showing the greatest effect at both the mRNA and protein levels, followed by the light chain and then the medium chain.
In previous experiments we have shown that the ability of Brn-3a to induce target genes in co-transfection experiments is dependent upon two activation domains, one of which is located at the N terminus of the molecule (17), and the other is located at the C terminus coincident with the POU domain. The relative importance of these different activation domains varies with the target promoter tested (17, 22, 24). In our previous experiments (25) we showed that transfection of ND7 cells with a naturally occurring form of Brn-3a (Brn-3a short) that lacks the N-terminal activation domain (9, 38) still results in the promotion of neurite outgrowth, although to a somewhat lesser extent than that observed with the full-length Brn-3a, suggesting therefore that the N-terminal activation domain plays only a small role in stimulating the expression of the genes required for neurite outgrowth. In contrast, overexpression of the POU domain of Brn-3a alone is able to promote neurite outgrowth, although to a somewhat lesser extent than that observed with full-length Brn-3a (25). This suggests that the POU domain alone can stimulate the expression of the genes that are required for neurite outgrowth.
In view of this we therefore investigated the expression of the various neurofilament mRNA in cells overexpressing either the POU domains of Brn-3a or Brn-3b or the short form of Brn-3 lacking the N-terminal domain. In these experiments (Fig. 2) the short form of Brn-3a was able to promote significant overexpression of the different neurofilament genes particularly following steroid treatment to induce the expression of Brn-3a (p < 0.01 for NF-H and NF-L and p = 0.01 for NF-M when cells transfected with the short form of Brn-3a are compared with control cells following steroid treatment). This suggests that the N-terminal domain is dispensable for induction of the neurofilament genes. Interestingly, cells transfected with the POU domain of Brn-3a alone showed overexpression of the different neurofilament genes of between 2- and 3-fold following steroid treatment (p < 0.01 for NF-L, NF-M, and NF-H when Brn-3a POU overexpressing cells are compared with control cells following steroid treatment). In contrast no overexpression of any of the neurofilament genes was observed in the cells transfected with the Brn-3b POU domain. This indicates therefore that overexpression of the POU domain alone is sufficient to induce overexpression of all three neurofilament genes, whereas Brn-3b POU does not have this effect.
These findings indicate therefore that the ability of full-length Brn-3a to induce the overexpression of the neurofilament genes is very little impaired by the deletion of its N-terminal activation domain and that a significant effect on these genes can be obtained with the POU domain alone. These findings closely parallel our previous findings (25) which suggested that neurite outgrowth could be induced by full-length Brn-3a, the short form of Brn-3a lacking the N-terminal domain, and by the isolated POU domain alone suggesting that the POU domain had the prime role in this effect.
Although our data thus clearly indicate the overexpression of the different neurofilament genes in the Brn-3a expression cells, they do not indicate whether this is a direct effect of Brn-3a on the promoters of the various genes or simply represents a downstream effect of the ability of Brn-3a to induce neurite outgrowth which in turn somehow induces overexpression of the neurofilament genes. To distinguish these possibilities, we therefore investigated whether Brn-3a could have a direct effect on the activity of the neurofilament gene promoters in transiently transfected cells. Thus we co-transfected neurofilament reporter gene constructs containing the promoters of each of the neurofilament genes (31, 32) with expression vectors containing the different forms of Brn-3 under the control of the constitutively active Moloney murine leukemia virus promoter in the vector pJ4 (30). These experiments were carried out both in ND7 cells which naturally express the various forms of Brn-3 (12, 22) and in BHK 21 fibroblast cells which do not express the different forms of Brn-3 (12, 24) and in which overexpression of any of the forms of Brn-3 would not be expected to induce neuronal differentiation.
In view of our previous findings (see Figs. 1, 2) that the
neurofilament heavy gene was the most strongly stimulated by Brn-3a, we
carried out our initial transfection experiments using a construct (31)
that contains 4.0 kilobases of upstream sequence derived from the NF-H
promoter. This construct was transfected into both BHK-21 and ND7 cells
together with Brn-3a or Brn-3b expression vectors. In these experiments
Brn-3a stimulated the NF-H promoter approximately 8-9-fold in both
BHK-21 cells (Fig. 3a) or ND7
cells (Fig. 3b). In contrast Brn-3b had no significant
effect on the promoter. Although the samples in these experiments were
equalized on the basis of plasmid uptake in each sample, similar
results were obtained when the samples were equalized on the basis of the activity of a co-transfected cytomegalovirus--galactosidase plasmid (data not shown). These experiments indicate therefore that
Brn-3a but not Brn-3b can directly stimulate the NF-H promoter in both
neuronal cells and in fibroblast cells where any indirect effect
exerted via the stimulation of neuronal differentiation can be
excluded. These results indicate therefore that Brn-3a is likely to
have a direct stimulatory effect on the NF-H promoter.
To investigate the regions of the NF-H promoter that were responsible
for this effect, we utilized reporter constructs that contain
progressively smaller regions of the NF-H promoter linked to the
CAT gene. In these experiments a construct containing the region of the NF-H promoter from 385 to +46 relative to the
transcriptional start site was stimulated by Brn-3a in a similar manner
to the longer construct (data not shown). Similarly, a construct that contains only 63 bases upstream of the start site of transcription was
also stimulated by Brn-3a but not by Brn-3b in both BHK cells (data not
shown) and ND7 cells (Fig.
4a). Hence Brn-3a can
stimulate a minimal NF-H promoter containing the basic sequences
required for transcription of the neurofilament heavy chain gene.
In view of this we tested whether similar stimulation of the NF-M and
NF-L promoters could also be observed with Brn-3a. In these experiments
a minimal NF-M (Fig. 4b) promoter containing the region from
98 to +27 relative to the transcriptional start site was stimulated
by Brn-3a and not by Brn-3b. Similarly a minimal NF-L (Fig.
4c) promoter containing the region from
60 to +82 relative
to the transcriptional start site was also stimulated by Brn-3a and not
by Brn-3b. Hence constructs containing relatively short regions of the
three neurofilament gene promoters around the transcriptional start
site can be directly stimulated by Brn-3a but not by Brn-3b in
co-transfection assays.
In our previous experiments (Fig. 2) using stably transfected ND7 clonal cell lines, the isolated Brn-3a POU domain was also able to stimulate the expression of the endogenous neurofilament genes, although not as strongly as full-length Brn-3a. We therefore tested whether the isolated Brn-3a POU domain could similarly stimulate the minimal neurofilament gene promoters in co-transfection assays. As illustrated in Fig. 4, clear stimulation of the NF-H, NF-M, and NF-L promoters was observed in co-transfections carried out in ND7 cells, although the extent of stimulation of each of the promoters was less than observed with the full-length Brn-3a construct paralleling the results in stably transfected cell lines. In contrast no significant stimulatory effect was observed with Brn-3b POU domain confirming that this effect is specific for Brn-3a (Fig. 4). Similar results were also observed when co-transfections were carried out in the BHK-21 fibroblast cells (data not shown).
These experiments indicate therefore that the isolated POU domain of
Brn-3a can directly activate the minimal neurofilament promoters in
co-transfection experiments and that such a direct effect on the
promoters is likely to be responsible for the observed activation of
the endogenous neurofilament genes in ND7 cells stably transfected with
the Brn-3a POU domain construct. The fact that the Brn-3b POU domain
construct cannot reproduce these effects in either stably transfected
cell lines or in transient transfection experiments indicates that one
of the seven amino acid differences between the POU domains of Brn-3
and Brn-3b (9, 12) must play a critical role in this effect. The POU
domain consists of two distinct subdomains, the POU-specific domain and
the POU homeodomain separated by a short linker region (for review see
Refs. 1 and 2). Although the POU-specific domain of Brn-3a and Brn-3b is identical, the POU homeodomain contains a single amino acid difference at position 22 in the first -helix, whereas the linker domain contains six amino acid differences. Since the linker region is
very variable between different POU factors and is believed to allow
the two major domains to move relative to one another (1, 2), we have
concentrated our attention on the single amino acid difference at
position 22 in the first
-helix.
We have previously shown that altering the isoleucine residue at this position in Brn-3b to the valine residue located at the equivalent position in Brn-3 converts Brn-3b from a repressor to an activator when it is co-transfected with an artificial test promoter containing a binding site for both Brn-3a and Brn-3b (33). The effect of this mutation on a naturally occurring promoter which is normally regulated by Brn-3a/Brn-3b has not, however, been tested. Similarly, we have not previously investigated on any promoter the effects of the converse mutation in which the valine at this position in Brn-3a is mutated to the isoleucine residue in Brn-3b.
The effect of these mutant constructs Brn-3a(I) and Brn-3b(V) on the
three minimal neurofilament promoters was therefore tested in
co-transfection assays. As illustrated in Fig.
5, the Brn-3b(V) mutant strongly
transactivated all three minimal neurofilament promoters to a degree
greater or equal to that observed with intact Brn-3a. Hence as with the
artificial test promoter this single amino acid difference within
otherwise intact Brn-3b suffices to convert Brn-3b to an activator of
the neurofilament promoters. Conversely the Brn-3a(I) mutant was now
unable to activate any of the neurofilament promoters and indeed
appeared to inhibit the activity of these promoters below the basal
level observed upon co-transfection of plasmid vector alone. Hence this
single amino acid difference at position 22 of the homeodomain
critically regulates the ability of Brn-3a to stimulate the
neurofilament promoters, whereas Brn-3b does not normally do so.
In our previous work (25) we showed that overexpression of the
Brn-3a transcription factor can induce ND7 cells to extend numerous
neurite processes even in full serum-containing medium, thus mimicking
the differentiation of these cells which normally occurs when they are
placed in serum-free medium. This ability of a single transcription
factor to induce the complex phenotypic changes involved in neurite
outgrowth is likely to depend on its ability to activate target genes
whose protein products are required for this purpose. Indeed we
previously demonstrated that overexpression of Brn-3a can activate the
endogenous genes encoding the synaptic vesicle proteins SNAP-25,
synapsin, synaptotagmin, and synaptophysin (25) as well as directly
stimulating the SNAP-25 promoter (18) and that of the neuronal
intermediate filament protein -internexin (17) in
co-transfection experiments. As the expression of SNAP-25 is essential
for neurite outgrowth in different neuronal cell types (27), it is
likely that its stimulation by Brn-3a is involved in the process of
neurite outgrowth. Target genes for Brn-3a are essential for diverse
aspects of early neuronal differentiation, including neurite outgrowth
(27), synaptogenesis, as well as establishing the early axonal
cytoskeleton, suggesting that Brn-3a may act as a master control
complex promoting early neuronal differentiation.
In agreement with this view, we show here that expression of the three
neurofilament genes is greatly enhanced in cell lines overexpressing
Brn-3a. Activation of neurofilament (and -internexin) genes provides
the cytoskeletal framework for neurite elongation and for the radial
growth of axons (28, 39). During development, neurofilaments became the
major cytoskeletal component in large axons, and their levels of
expression are major determinants of axonal size (for review see Ref.
40). Moreover, neurofilaments are obligate heteropolymers in
vivo (41, 42), so it is necessary for subunit expression to be
coordinated to ensure a continuous supply and delivery of
neurofilaments along the axon. Increases (or decreases) of individual
subunit expression are disruptive to assembly and/or transport, leading
to accumulations of neurofilaments and reduced axonal size
(43-48).
It is therefore likely to be of critical importance that Brn-3a can stimulate the endogenous genes encoding all three neurofilament proteins in stably transfected cell lines. As illustrated in our co-transfection experiments, this effect is likely to involve a direct stimulation of the neurofilament promoters by Brn-3a rather than being a consequence of the neuronal differentiation induced by Brn-3a. Brn-3a can stimulate the minimal promoters of all three neurofilament genes in co-transfection experiments not only in the neuronal ND7 cell line but also in BHK-21 fibroblast cells. Since such BHK cells evidently do not show neuronal differentiation in response to Brn-3a, this indicates that the changes in neurofilament gene expression which occur in response to Brn-3a are likely to be due to a direct effect on the neurofilament promoter.
Interestingly, we have previously shown that Brn-3a can also activate
the promoter of the -internexin gene (17, 49) which encodes a
closely related type IV intermediate filament protein (26, 50). In
contrast we have previously observed no transactivation by Brn-3a of
the promoter encoding the vimentin protein which belongs to a different
class of intermediate filament
proteins.2 Interestingly,
vimentin is expressed in proliferating neuronal precursor cells that
line the ventricular surface of the neural tube but is rapidly
down-regulated upon terminal neuronal differentiation.
Moreover, we previously reported that induction of the -internexin
promoter by Brn-3a could be observed using a reporter construct
containing only 77 bases of upstream promoter sequence (17), and
subsequently we identified a response element for Brn-3a between
positions
77 and
60 of the
-internexin promoter (49). Similarly,
in the experiments reported here we have shown that neurofilament
promoter constructs containing relatively small amounts of upstream
sequence are responsive to Brn-3a, suggesting that the promoters of the
genes encoding the type IV intermediate filament proteins contain
Brn-3a-responsive elements located close to the transcriptional start
site.
As well as showing a similar location of the Brn-3a-responsive region,
the neurofilament promoters also resemble that of -internexin in
being stimulated by Brn-3a but not by the closely related transcription factor Brn-3b (this paper and Ref. 17). However, despite these similarities, the region of Brn-3a which is required for stimulation of
the promoter differs in the case of the neurofilament genes and that of
-internexin. Thus, we previously showed (17) that stimulation of the
-internexin promoter requires an activation domain located at the N
terminus of the protein and is not observed when the promoter is
co-transfected with the POU domain alone. In the experiments reported
here, we have shown, however, that the neurofilament promoters resemble
that encoding SNAP-25 (19) and a artificial promoter containing a
synthetic binding site for Brn-3a and Brn-3b (24) in that they can be
stimulated by co-transfection with the Brn-3a POU domain but not by the
closely related Brn-3b POU domain. In previous experiments (49) we
showed that transfer of the Brn-3a response element from the
-internexin promoter to the same site in the test promoter as
normally occupied by the synthetic Brn-3a/Brn-3b binding site resulted
in the
-internexin response element becoming responsive to the POU
domain of Brn-3a. Thus the difference in the responses of the
neurofilament promoters and that of
-internexin is likely to
involve a different context of the Brn-3a response element in these
various promoters.
The stimulation of the neurofilament promoters by the POU domain of
Brn-3a but not by that of Brn-3b prompted us to examine the effect on
these promoters of mutant forms of Brn-3a and Brn-3b containing a
single amino acid substitution at position 22 in the first -helix of
the POU homeodomain. We have thus demonstrated that this amino acid
plays a critical role in the ability of Brn-3a and not Brn-3b to
stimulate the neurofilament promoters. The ability of Brn-3b containing
a valine rather than an isoleucine residue at this position to activate
the neurofilament promoter parallels our previous finding (33) that
this mutant can also activate an artificial promoter containing a
synthetic binding site for the Brn-3 factors and, moreover, extends
this finding to a naturally occurring neuronal promoter. Similarly, the
data reported here show for the first time that the converse mutation
in which isoleucine is substituted for valine in Brn-3a completely
abolishes the ability of Brn-3a to activate a target promoter.
Although the POU domain of the Brn-3 factors directs their binding to
DNA, the critical role of this amino acid within the POU domain in the
different effects of Brn-3a and Brn-3b does not appear to be due to
differences in the DNA binding abilities of the various wild type and
mutant forms of Brn-3. Thus, the POU domains of both Brn-3a and Brn-3b
are able to bind to DNA (9, 13, 49), and the mutations do not appear to
have any effect on this DNA binding ability (33). Similarly, position 22 of the homeodomain is located at the C terminus of the first -helix and is thus not in contact with the DNA according to
crystallographic analysis of the POU domain of the related Oct-1
protein (51). Indeed such crystallographic analysis indicates that this
amino acid is located on the surface of the POU domain at a position that could be involved in protein-protein interactions.
Such a role for the amino acid at this position is supported by the finding that in the related POU proteins Oct-1 and Oct-2, the amino acid at this position plays a critical role in determining their ability to interact with the herpes simplex virus virion protein VMW65 (52). Thus substitution of the alanine found at position 22 of the homeodomain in Oct-2 with the glutamic acid found at the equivalent position in Oct-1 allows Oct-2 to interact with VMW65 which is normally a property of Oct-1 only. By analogy therefore, it is likely that the presence of a valine at this position in Brn-3a or the mutant Brn-3b allows the protein to interact with and thus recruit an activating molecule to the promoter, whereas Brn-3b or Brn-3a containing isoleucine cannot do so. Alternatively Brn-3b and Brn-3a containing isoleucine at this position can recruit an inhibitory molecule which does not interact with molecules containing valine at this position.
In either case, it is clear that the nature of the amino acid at this
position in Brn-3a and Brn-3b plays a critical role in the ability of
Brn-3a alone to activate the three neurofilament gene promoters,
whereas it is not involved in the activation of the -internexin
promoter which is dependent upon the N terminus of Brn-3a. The
different alternatively spliced forms of Brn-3a which either possess or
lack the N-terminal domain (9, 38) may thus be involved in the
disparate expression of the neurofilament genes and the
-internexin
gene in different subsets of neurons and their dissimilar expression
pattern over time. Conversely the similar activation mechanism of the
three neurofilament subunit genes by Brn-3a may help to explain their
co-expression in the same subset of neurons.
We thank Tarik Moroy for the gift of Brn-3 cDNA clones.