(Received for publication, September 3, 1996, and in revised form, January 16, 1996)
From the Department of Medicine, San Francisco Veterans Affairs Medical Center and University of California, San Francisco, California 94121
Previous studies showed that the Hox homeodomain proteins from paralog groups 1-8 display cooperative DNA binding with the non-Hox homeodomain protein Pbx, mediated by a canonical YPWM. Although the Abd-B-like Hox proteins in paralogs 9-13 lack this sequence, Hoxb-9 and Hoxa-10 were reported to bind with Pbx1a to DNA. We show that these interactions require a tryptophan 6 amino acids N-terminal to the homeodomain. Binding site selection for Hoxb-9 with Pbx1a yielded ATGATGAC, containing a novel TTAC Hox-binding site adjacent to a Pbx site. In the presence of Pbx1a, Hoxb-9 and Hoxa-10 bound to targets containing either TTAC or TTAT. These data extend previous findings that interactions with Pbx define a Hox protein binding code for different DNA sequences across paralog groups 1 through 10. Members of the 11, 12, and 13 paralogs do not cooperatively bind DNA with Pbx1a, despite the presence of tryptophan residues N-terminal to the homeodomain in Hoxd-12 and Hoxd-13. Hoxa-11, Hoxd-12, or Hoxd-13, in the presence of Pbx1a, selected a TTAC Hox site but lacking a Pbx1a site. These data suggest that Abd-B-like Hox proteins bind to a novel TTAC site and can be divided by their cooperative binding to DNA with Pbx1a.
The Drosophila HOM-C genes are master developmental regulatory genes which share a conserved 183-nucleotide homeobox sequence (1). The 39-vertebrate Hox homeobox genes are arranged in four parallel loci (A, B, C, and D) such that the genes in each cluster can be aligned on the basis of homology within the homeobox to form so-called paralog groups (2). Although the Hox genes from paralogs 1 through 8 can be related to specific HOM-C genes on the basis of sequence homology within the homeobox, paralogs 9 through 13 appear to be equally related to the Drosophila Abd-B gene (3). Thus the homeobox sequences of human, murine, or chicken Hox genes from paralogs 9-13 are equally similar to the Abd-B homeobox. Hox-d cluster genes from paralogs 9 to 13 are expressed in spatially and temporally distinct patterns in the developing limb (4), suggesting that the Abd-B-like gene products play specific developmental regulatory roles.
The Hox homeodomain proteins are thought to function as transcription factors (5). X-ray crystal structure analysis has shown that the most conserved portion of the homeodomain, helix three, forms a portion of the DNA recognition surface (6, 7). This conservation is reflected by the fact that the homeodomains of many Hom-c and Hox proteins appear to bind preferentially to DNA oligomers containing a TAAT core recognition sequence. This observation of a shared DNA consensus binding site has been puzzling, since different homeodomain proteins have distinct biologic functions, as judged by the observed phenotypic differences caused by over-expression or targeted disruption of specific homeobox genes (1).
One mechanism for increasing functional specificity would be the interaction of Hox proteins with protein partners which might provide enhanced DNA specificity or differential binding affinity. We and others have reported that Hox and Hom-c proteins cooperatively bind to DNA with Pbx and Exd, respectively (8-12). These interactions appeared to be mediated by a conserved N-terminal YPWM sequence in Hox proteins from paralogs 1 to 8 (10, 13-16). We demonstrated that the Hoxb-4 protein requires at least the tryptophan and methionine residues from this tetrapeptide for complex formation with Pbx1a and DNA (13). These studies also revealed that complex formation occurred with representative proteins from paralogs 1 to 8, even though the YPWM motif is variably spaced, occurring 5 to 53 residues N-terminal to the homeodomain. We and others initially reported that Abd-B and the Abd-B-like Hox proteins did not cooperatively bind DNA with Exd or Pbx (8, 9, 11). However, we have recently shown that the lack of cooperative binding originally observed for Hoxa-10 was due to an inappropriate target DNA. Hoxa-10, which lacks a YPWM motif, formed a strong DNA binding complex with Pbx1a, mediated by an N-terminal ANW motif, on an ATGATGA1 target (16). In addition, Peltenburg and Murre (17) have recently demonstrated that the engrailed homeodomain protein interacts with Pbx or Exd via tryptophan residues located N-terminal to the homeodomain. The current study was initiated to determine whether members of the other Abd-B-like Hox paralogs (9, 11, 12, and 13), three of which contain tryptophan residues located N-terminal to the homeodomain, are capable of cooperative binding with Pbx1a to an appropriate DNA target, and to determine whether Pbx1a also provides DNA selectivity to these homeodomain proteins.
Since in previous studies, proteins from the same paralogs appeared to have similar DNA binding preferences (13), cDNAs encoding representative full-length Hox proteins from each paralog and Pbx1 were subcloned into either an sp65 vector containing an SP6 promoter (Promega, Madison, WI) engineered to express proteins containing an N-terminal FLAG epitope tag (MDYKDDDDK) (Pbx1a, Hoxb-7, and Hoxa-10); or into a pET vector (Novagen, Madison, WI) containing a T7 promoter, which produces proteins with an N-terminal T7 epitope tag (Pbx1a, Hoxb-8, Hoxb-9, Hoxa-11, Hoxd-12, and Hoxd-13). The identity of each Hox protein was confirmed by Western blot analysis of bacterially expressed proteins using specific polyclonal antisera. For gel shift and DNA target selection assays, proteins were synthesized containing the full-length homeodomain protein fused to the respective epitope tag using the TNT-coupled in vitro transcription-translation system (Promega), in parallel reactions in the presence and absence of [35S]methionine. Electrophoresis of the labeled proteins demonstrated synthesis of the appropriate full-length products (data not shown). Using autoradiography and densitometry of the 35S-labeled proteins, and calculating the incorporation of labeled methionine of known specific activity into each protein, we estimated that the relative protein concentrations used were within a 2-fold range. Hoxb-9, Hoxa-10, and Pbx1a were cloned in Bluescript (Stratagene, La Jolla, CA), for synthesis of non-epitope-tagged proteins, in parallel reactions with and without [35S]methionine to check protein size and to estimate relative concentrations. Each of the epitope-tagged Abd-B-like proteins was also shown to be functional in DNA site selection assays (see "Results").
Human Hoxb-7 and Hoxa-10 were cloned previously (18, 19). A full-length Hoxd-12 cDNA was cloned from 12-day mouse embryo RNA by standard reverse transcriptase-polymerase chain reaction using primers from the published sequence (3), and checked by DNA sequencing. Other full-length cDNA clones were: murine Hoxb-9 (20), murine Hoxa-11 (21), murine Hoxb-8 (22), and chicken Hoxd-13 (23). A full-length cDNA encoding human Pbx1a was kindly by Dr. Michael Cleary (24). The codon encoding the tryptophan residue located 6 amino acids N-terminal to the homeodomain in the Hoxb-9 cDNA was changed to encode glutamine using a Muta-Gene M13 in vitro mutagenesis kit (Bio-Rad).
Electrophoretic Mobility Shift AssaysComplementary
oligonucleotides containing consensus binding sites determined by site
selection for Hoxb-9 with Pbx1a (CTGCGATGATGACCGC) and
Hoxa-10 with Pbx1a (CTGCGATGATGACCGC) were synthesized (Operon Technologies, Alameda, CA). Standard conditions used were similar to those previously described (16). Double-stranded, end-labeled DNA (50,000 cpm/binding reaction, 10 nM) was
incubated with 2 µl of reticulocyte lysate mixture containing the Hox
protein (1 nM) either in the presence of 2 µl of
reticulocyte lysate mixture containing Pbx1a (1 nM) or with
2 µl of the lysate control, in 75 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 10 mM
Tris-HCl (pH 7.5), 6% glycerol, 2 µg of bovine serum albumin, and 2 µg of poly(dI·dC) as nonspecific competitor, in a final reaction
volume of 15 µl. Experiments designed to detect complex formation
(Figs. 1, 2, 3) were performed with a 30-min incubation at 4 °C.
Reaction mixtures were run on a 6% polyacrylamide gel to visualize
complex formation by retardation of the 32P-labeled target
DNA. In some experiments, polyclonal antisera to the appropriate
epitope tag was incubated with aliquots of the reaction mixture for an
additional 30 min. The Hox protein was fused to one epitope tag while
the Pbx1a molecule was fused to a different epitope tag, such that it
was possible to use specific antisera to identify the presence of the
Hox protein or the Pbx1a protein in the complex by supershifting the
retarded complex band. In experiments designed to measure dissociation
rate constants, reaction mixtures were incubated at 30 °C for 30 min
and either applied directly to a 6% polyacrylamide gel (zero time
sample) or mixed with a 100-fold excess of unlabeled oligonucleotide
followed by incubation for fixed times (1-30 min) prior to application to the running polyacrylamide gel. Gel electrophoresis was performed in
0.25 × TBE buffer as described previously. For each gel shift reaction, a control containing the reticulocyte lysate and appropriate viral polymerase was used to detect possible DNA binding by endogenous factors. Lysate controls showed variable intensity gel shift bands with
the DNA target. These bands varied with both the lysate batch and the
batch of poly(dI·dC) used as non-competitive inhibitor.
Calculation of Complex Half-lives
Electrophoretic mobility
shift assay gels were autoradiographed for densitometric quantitation
of complex band using a MacIntosh 8500 Power PC computer and the
NIH-Image software program. Each gel was autoradiographed for various
times to ensure that the densities measured were within the linear
range of the scanner and software program. A dissociation rate was
calculated for each Hox-Pbx1a-DNA complex from the slope of the
regression line generated by plotting the log of the complex band
intensities versus time (Fig. 4C). For each
dissociation experiment, the correlation coefficient for the line was
>0.96. For each complex, the half-life was calculated using the
equation, T1/2 = log
(0.5)/kd.
DNA Site Selection Protocol
Site selection was performed following the basic protocol described by Blackwell (25). The T7-epitope tag Hox fusion protein of interest and native Pbx1a were synthesized in vitro and incubated at 4 °C for 30 min with a 59-mer containing a random 18-mer core flanked by arms which contained cloning sites (GCTCGAATTCAAGCTTCTN18CATGGATCCTGCAGAATTCAGT). Bound DNA was immunoprecipitated using an antisera to the T7 tag sequence. Following extensive washing steps, the DNA was amplified by 15-20 cycles of polymerase chain reaction (94 °C, 1 min; 54 °C, 1 min; 72 °C, 1 min), using primers designed against the flanking arms. After six selection cycles, the amplified DNA was subcloned into M13mp19 (New England BioLabs, Beverly MA) and sequenced using the dideoxy method with 35S-labeled adenosine triphosphate. Consensus sequences were determined by visual alignment of sequences from unique clones. The number of independent clones used to define each consensus are given in parentheses: Hoxb-9 plus Pbx1a (11); Hoxb-9 alone (10); Hoxa-11 (34); Hoxd-12 (17); and Hoxd-13 (18).
Previous studies presented conflicting data concerning the capability of Abd-B-like Hox proteins to cooperatively bind DNA with Pbx1a. We initially used an oligonucleotide (ATGATGA), which was identified in a DNA site selection protocol using Hoxa-10 with Pbx1a (16), to examine the ability of representative Abd-B-like Hox proteins to cooperatively bind to DNA with Pbx1a. The first five nucleotides, ATGAT, comprise the Pbx consensus binding site (26-28). As described below, the Hox site overlaps the Pbx site and consists of the TGA sequence. Electrophoretic mobility shift assays (EMSA)2 were used to detect complex formation between the labeled oligonucleotide and Hox and Pbx1a proteins. Since we have previously observed that full-length Hox proteins behave differently from the truncated homeodomain fragments used in many experiments (13), all of these studies have been performed using full-length proteins. In most experiments the Hox and Pbx1a proteins were synthesized as fusion molecules containing short N-terminal T7 or FLAG epitope sequences to permit identification using epitope-specific antisera.
Hoxb-9 and Hoxa-10 formed strong cooperative complexes with Pbx1a on this target DNA, under conditions in which the Hox proteins alone bound very weakly and Pbx1a binding alone was undetectable (Fig. 1A). For comparison, the neighboring paralog proteins, Hoxb-8 and Hoxb-7, formed weak complexes with Pbx1a and this oligonucleotide. Since uncharacterized DNA-binding proteins present in the reticulocyte produced a variable gel shift band in the same position as some of the specific complex bands, supershift experiments using antibodies to the epitope tags were used to show that retarded bands ascribed to the Hox-Pbx1a-DNA complex contained both Pbx1a and Hox protein. Although each of the other three Abd-B-like paralog members, Hoxa-11, Hoxd-12, or Hoxd-13, were capable of binding this oligonucleotide probe, none was able to form a detectable cooperative complex with Pbx1a and this DNA target.
To demonstrate that the epitope tag does not alter complex formation, DNA binding reactions were performed using one tagged protein with the other protein being synthesized without an epitope tag. Both native Hoxb-9 and Hoxa-10 behaved similarly to the epitope-tagged proteins in DNA binding reactions with epitope-tagged Pbx1a (Fig. 1B). Native Pbx1a formed a complex with epitope-tagged Hoxa-10 which migrated with the same mobility as the complex formed with tagged Pbx1a.
A Conserved Tryptophan in Hoxb-9 Confers Complex Forming CapabilityAlthough we reported that the interaction of Hoxa-10 with Pbx1a is mediated by a conserved ANW sequence located N-terminal to the homeodomain in Hoxb-9 and Hoxa-10 (Table I) (16), the importance of individual amino acids for complex formation was not defined. Since we and others had previously shown that Pbx1a interaction with other Hox proteins required a tryptophan residue, we focused our studies on the invariant tryptophan within this amino acid triplet. A mutant Hoxb-9 protein containing a glutamine in place of this tryptophan was unable to form a complex with Pbx1a on the Hoxa-10 DNA target, under conditions in which the wild type Hoxb-9 formed a very strong complex (Fig. 2, compare lanes 4 and 5). This mutation did not prevent DNA binding since the mutant protein was still capable of shifting DNA in the absence of Pbx1a (Fig. 2, compare lanes 2 and 3).
|
Since Hoxa-10 complex formation with Pbx1a is highly
dependent on the DNA target sequence (16), we performed DNA site
selection experiments to determine the preferred binding sites of each
of the other Abd-B-like Hox proteins in the presence of Pbx1a (see also
below). We initially used an epitope-tagged Hoxb-9 in the presence of
Pbx1a to select a very highly conserved 12-nucleotide sequence:
ATGATGAC (Table II, part A). This
sequence was identical to that previously selected for Hoxa-10 with the
important exception of the occurrence of a C in place of a T at
position 9 in the putative Hox homeodomain core recognition site
(underlined region), as well as being extended by one extra
3-nucleotide (see Table IV). The first five nucleotides of this
sequence (ATGAT) correspond to that obtained previously for Pbx1
(26-28), demonstrating that the Pbx1a protein binds cooperatively with
the Hoxb-9 protein during DNA-protein complex formation.
|
|
Since these data suggested that the Hoxb-9 protein was binding to the
other seven bases GAC, which contained a novel TTAC Hox
recognition site, a site selection protocol was performed for Hoxb-9 in
the absence of Pbx1a. As shown in Table II, part B, the Hoxb-9 protein
alone selected DNA targets which contained a similar core sequence
(TTAC) with an additional T residue at the 5 end, but lacking the ATGA
portion of the Pbx1a-binding site. The first T of the Hoxb-9 site could
not be distinguished from the last T of the Pbx1a-binding site,
suggesting that both proteins may form specific interactions with this
base pair. The high specificity obtained in the presence of Pbx1a was
relaxed to some degree for Hoxb-9 alone. In particular, the T at
position 7 of the consensus sequence selected with Pbx1a showed a
substantial relaxation to a mixture of T and A in targets selected with
Hoxb-9 alone. This position is equivalent to the second position of the canonical TAT sequence defined for many Hox/Hom-c
proteins. However, only a single sequence containing a TAAT in the core
positions (6 to 9) was obtained, with 80% of the sequences containing
a Cys as the final nucleotide of the core recognition site.
Gel shift assays were performed to confirm the site selection data for Hoxb-9 with Pbx1a (Fig. 3A). Both Hoxb-9 and Hoxa-10 cooperatively bound with Pbx1a to the target DNA sequence ATGATGAC. In contrast, the neighboring Hoxb-7 and Hoxb-8 proteins appeared to form relatively weak complexes with Pbx1a and this target. Since the site selection experiments yielded more stringently conserved DNA binding sequences for Hoxb-9 or Hoxa-10 than were observed by gel shift, an oligonucleotide containing a TAAT core sequence (ATGATGAC) was also tested for cooperative DNA binding in gel shift experiments (Fig. 3B). Complex formation by Hoxb-9 or Hoxa-10 with Pbx1a on this target was clearly reduced compared with the targets selected by either of these proteins (compare with Figs. 1 or 3A). In contrast, Hoxb-7 and Hoxb-8 appeared to bind to the TAAT containing sequence somewhat more strongly, reflecting an apparent greater preference for this core recognition sequence by the Hox proteins in the middle of the locus (see below).
Hox Proteins from Paralog Groups 11, 12, and 13 do Not Cooperatively Bind DNA with Pbx1aProteins representing the three
paralog groups located at the extreme 5 end of the loci, Hoxa-11,
Hoxd-12, and Hoxd-13, did not form detectable complexes with Pbx1a on
DNA targets containing either the core consensus sequence for the
Hoxb-9 protein (TTAC), the core consensus for Hoxa-10 (TTAT), or an
oligonucleotide containing a TAAT core sequence (Figs. 1 and 3). The
fact that each of these proteins bound DNA in the absence of Pbx1a
suggested that the lack of cooperativity was not due to denatured
proteins. Since Hoxa-11 does not contain a tryptophan residue within
the 50 amino acids N-terminal to the homeodomain (21), it seemed likely
that this protein might not cooperatively bind DNA with Pbx1a. However, Hox proteins from both the 12 and 13 paralogs contain tryptophan residues which are 9 and 21 residues N-terminal to the homeodomain, respectively (Table I), suggesting that given a different DNA binding
target they might form complexes with Pbx1a. In this regard, it should
be noted that restrictions on the distance between the tryptophan
residue which mediates cooperative binding and the homeodomain appear
to be relatively modest for the Hox proteins since linker arms from 5 to 53 amino acids are tolerated (13).
To search for putative DNA targets on which Hoxa-11, Hoxd-12, and Hoxd-13 might form cooperative complexes with Pbx1a, each protein was used for site selection in the presence of Pbx1a. After six rounds of selection, there was a clear consensus Hoxa-11 binding site consisting of TGAC (Tables III and IV), but there was no apparent binding site for Pbx1a. In a similar manner, site selection experiments using Hoxd-12 and Hoxd-13 yielded clear consensus sequences containing Hox but not Pbx1a-binding sites (Table IV). Thus there do not appear to be unique DNA sequences on which these Hox proteins will cooperatively bind with Pbx1a. These data also confirmed that each of these Abd-B-like proteins was capable of binding DNA. Taken together with the lack of gel shifting seen with an oligonucleotide target which is extremely similar to that selected by the Hoxa-11, Hoxd-12, and Hoxd-13 proteins (Fig. 3A), these data demonstrate that these three Abd-B-like Hox proteins do not cooperatively bind DNA with Pbx1a, under conditions where members of the other Hox paralogs all cooperatively interact with Pbx1a to bind DNA.
|
It is of interest that both Hoxa-11 and Hoxd-12 showed a DNA binding consensus of TGAC, which was identical to that found for Hoxb-9 in the absence of Pbx1a (Table IV). The core region (TTAC) within the consensus sequence for Hoxd-13 was the same as that found in the other three consensus sequences which we determined. However, the Hoxd-13 consensus was unique in having a specificity for G at position 12, while the other Abd-B-like proteins appeared to prefer a C in this position. Hoxd-13 also had a higher selectivity for a T at position 4 than was observed for the other proteins. These results differ from those previously obtained for the Hoxa-10 and Abd-B proteins, both of which appear to prefer a TTAT core sequence (16, 29). However, as shown in Fig. 3A, Hoxa-10 formed a strong complex with the sequence containing a TTAC core (see also below). In addition, the site selection protocol which identified a TTAT core recognition site for Hoxa-10 with Pbx (16), also yielded a significant number of sequences containing a TTAC core site.3 Taken together, the site selection and gel shift data suggest that the Abd-B-like Hox proteins bind to both a novel TTAC core DNA sequence, as well as to a sequence containing a TTAT core recognition site.
To further investigate the DNA-binding site selectivity of the Hoxb-9
and Hoxa-10 proteins in the presence of Pbx1a, we performed dissociation rate determinations using these proteins, along with Hoxb-7 and Hoxb-8 for comparison to the neighboring non-Abd-B-like proteins in the Hox loci. In these studies, complexes were formed at
30 °C between each of the respective epitope-tagged Hox and Pbx1a
proteins with the DNA targets. Following removal of a time 0 sample, a
100-fold excess of cold-competitor DNA was added to the pre-formed
complex, and at specified times aliquots were loaded onto the running
gel. Fig. 4 shows a representative experiment for the
dissociation of Hoxb-9 and Hoxa-10 from complexes with Pbx1a on a probe
consisting of either the Hoxa-10 consensus site containing a TTAT core
or the Hoxb-9 site containing a TTAC core. As seen in Table
V, the dissociation rates for complexes formed by either
Hoxb-9 or Hoxa-10 with Pbx1a on an oligonucleotide containing a TTAC
core were lower than those observed for the dissociation of these
proteins from an oligonucleotide containing a TTAT site. In contrast,
Hoxb-7 and Hoxb-8 exhibited higher dissociation rates for the
oligonucleotide with the TTAC core sequence. The differences in
stability of complexes formed between either Hoxb-9 or Hoxa-10 with
Pbx1a on both targets were relatively modest, reinforcing the gel shift
results which suggested that Hoxb-9 and Hoxa-10 form strong cooperative
complexes with Pbx1a on both TTAC and TTAT containing targets. Table V
also shows that a DNA target containing a TAAT core showed the highest
dissociation rate with all four proteins. Assuming that the association
rate constants are the same, the observed high dissociation rates are
in agreement with the site selection and gel shift results showing
that, in the presence of Pbx1a, the conventional TAAT core sequence was not preferred by either the Abd-B-like proteins or the neighboring proteins from the 5 side of the Hox-b locus.
|
We show that the Abd-B-like Hox homeodomain proteins can be divided into those from the 9 and 10 paralogs which cooperatively interact with Pbx1a to bind DNA and the 11 to 13 paralog proteins which do not bind cooperatively to DNA with Pbx1a.4 A number of studies have shown that an N-terminal tryptophan appears to mediate the interaction of the Hox homeodomain proteins with the Pbx/Exd homeodomain proteins (13-17). However, the structural context for the tryptophan residue within the Hox proteins is not clear. In Hoxb-1 through Hoxb-8, the tryptophan resides within a relatively conserved, but variably spaced hexapeptide motif (30). In the 9 and 10 paralog proteins the tryptophan is located 6 residues N-terminal to the homeodomain within a conserved ANW. Furthermore, Peltenburg and Murre (17) have recently demonstrated that cooperative DNA binding of the engrailed homeodomain protein with Pbx is mediated by N-terminal tryptophan residues which show no homology to either the YPWM or ANW motifs found in Hox proteins. Our data show that the presence of tryptophan residues in Hoxd-12 and Hoxd-13, which are 9 and 21 residues N-terminal to the homeodomain, respectively, are not sufficient to confer Pbx1a binding capability to these proteins. In addition, Abd-B has been reported to be unable to bind cooperatively to DNA with Exd despite the presence of a tryptophan residue 6 amino acids upstream of the homeodomain (see Table I) (11).
Abd-B-like Hox Proteins Bind a Novel TTAC DNA SiteSite selection experiments for Hoxb-9, Hoxa-11, Hoxd-12, and Hoxd-13 in the presence of Pbx1a revealed that these four Abd-B-like Hox proteins selected a consensus TGA(C/G) sequence containing a novel TTAC core-binding site. In contrast to Hoxb-9, neither the Hoxa-11, Hoxd-12, nor Hoxd-13 proteins were able to select a sequence containing a Pbx-binding site. These results confirm the gel shift data showing that these Hox proteins are unable to cooperate with Pbx1a to bind DNA. The gel shift data also show substantial cooperative binding of Hoxb-9 or Hoxa-10 with Pbx1a to an oligonucleotide containing a TTAT core Hox recognition sequence, which was previously selected by Hoxa-10 and Pbx1a (16).
The C at position 9, the last base of the core recognition sequence, appears to be unique to the vertebrate Abd-B-like Hox proteins, since neither the Drosophila Abd-B protein (29) or other Hox homeodomain proteins (31) have been shown to bind to DNA targets containing a core. Current x-ray crystallographic studies do not provide an explanation for the preference for a C at position 9 of the recognition sequence (6, 7).
Pbx Confers Selective DNA Binding to Hox Proteins Across the LociIt has been difficult to explain Hox protein function given
the apparent lack of DNA binding specificity across the Hox loci. Chang
et al. (16) initially proposed that cooperative binding with
Pbx conferred a differential DNA binding selectivity to Hox proteins
across paralog groups 1 to 10, based on the nucleotide at position 7 of
the consensus sequence ATGATTAT (16). As shown in Fig.
5, these studies demonstrated that, in the presence of Pbx, Hox proteins from paralogs 1-5 preferentially bound
oligonucleotides containing a TAT core sequence, while
proteins from paralogs 6-10 appeared to prefer a TAT
core. Proteins from the middle of the locus (paralog groups 3 to 8)
also tolerated a TAT core sequence, although this sequence
was not observed during site selection using Hoxb-4 or Hoxb-6 with Pbx.
We have now extended these observations to show that the proteins from
paralog groups 9 and 10 bind a TTA core recognition
sequence in the presence of Pbx1a. We propose that the Abd-B-like Hox
proteins can be subdivided by their ability to bind to DNA
cooperatively with Pbx, such that proteins from groups 9 and 10 will
form much stronger complexes than the proteins from groups 11, 12, and
13. In addition, the 9 and 10 paralog proteins can, through
interactions with Pbx1a, exhibit increased selectivity for DNA targets
due to the longer recognition site bound by the combination of Hox and
Pbx proteins. Although the studies of Chang et al. (16)
focused on the role of the second nucleotide in the core sequence,
selection preferences were also noted for individual Hox proteins in
positions 10, 11, and 12. The fact that Hoxd-13 selects a G at position
13, while Hoxa-11 and Hoxd-12 select a C, suggests that this difference may provide in vivo binding selectivity between proteins
from these paralogs.
The scheme shown in Fig. 5 provides a possible rationale for DNA binding specificity for Hox proteins across the vertebrate loci. It is encouraging that one of the first in vivo DNA targets described for a Hox protein appears to conform to this concept. Thus Hoxb-1 appears to cooperatively bind with Pbx1 to recognize a consensus TGAT sequence (32). However, we note that the preferences described for positions 7 or 9 of the core Hox recognition sequence are insufficient to specify the Pbx1a-mediated Hox homeodomain protein binding to DNA. Thus we have recently shown that a DNA target containing a TAT core, which was preferred over other DNA targets by Hoxb-1 through Hoxb-3 in gel shift assays with Pbx1a (16), actually exhibited a 100-fold lower dissociation rate for Hoxb-6 and Hoxb-5 compared with Hoxb-2 or Hoxb-3 (13). While recognizing that dissociation rate data provide only one component of the equilibrium binding constant, it seems likely that the in vitro experimental systems employed in these and other studies provide only partial insights as to the in vivo binding specificity of these transcription factors. A greater understanding of the physiological interactions of Hox and Pbx homeodomain proteins with DNA targets will require identification of the natural regulatory targets of these putative transcription factors.
We acknowledge Dr. Mike Cleary for initiating our studies on the interactions between Pbx and Hox proteins and Ching-Pin Chang and Mike Cleary for initially recognizing and proposing that Pbx provides specificity for Hox protein-DNA interactions, as well as for providing us with a full-length Pbx1a cDNA clone and other useful reagents and suggestions. We thank William McGinnis, Steven Potter, Suzanne Cory, and William Upholt for kindly providing cDNAs encoding full-length proteins for Hoxb-9, Hoxa-11, Hoxb-8, and Hoxd-13, respectively.