- Email: [email protected]
Measuring inputs to a common function: The case of Dlx5 and Dlx6
Biochemical and Biophysical Research Communications 478 (2016) 371e377
Contents lists available at ScienceDirect
Biochemical and Biophysical Research Communications journal homepage: www.elsevier.com/locate/ybbrc
Measuring inputs to a common function: The case of Dlx5 and Dlx6 Anna Quach, Rachel K. MacKenzie, Andrew J. Bendall* Department of Molecular and Cellular Biology, University of Guelph, Guelph, Ontario, N1G 2W1, Canada
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 June 2016 Accepted 8 July 2016 Available online 11 July 2016
Physically linked Dlx5 and Dlx6 paralogs are co-expressed in vertebrates and various combinations of null alleles in mice demonstrate not only functional redundancy between the paralogous factors but a similar quantitative contribution to craniofacial functions during development. While it is not possible to rule out that the bigene pair contributes some paralog-specific functions it is clear that, for many functions in the head, Dlx5 and Dlx6 are interchangeable. To assess the relative quantitative contribution made by each paralog to bigene function, we have made comparisons of the expression of Dlx5 and Dlx6 in chick embryos and quantitated the transcriptional properties of the encoded proteins in a variety of regulatory and cellular contexts. Our data indicate that the transcriptional activities of both Dlx5 and Dlx6 are very much context dependent; isolated domains fused to a heterologous DNA binding domain have little intrinsic activity, while individual domains are more active when contiguous with their own homeodomain. We find Dlx5 and Dlx6 to be quantitatively indistinguishable on a variety of natural cis-regulatory sequences in a heterologous cellular context but observed quantitatively different transcriptional outputs in cells that normally express these genes, suggesting differential interactions with co-evolved co-activators. © 2016 Elsevier Inc. All rights reserved.
Keywords: Dlx5 Dlx6 Functional equivalence Paralog Transcription factor
1. Introduction The Dlx family of transcriptional regulators execute a variety of developmental functions [1e4]. The six-gene vertebrate family is arranged in three bigene clusters such that each first-order paralog shares tissue-specific enhancers with its cis-linked sister gene [5]. Physically linked Dlx genes are therefore co-expressed in the cranial ectomesoderm of the first pharyngeal arch where they regulate upper and lower jaw patterning [6]. Co-expression of the first-order paralogs Dlx5 and Dlx6 and observations of various null allele combinations of the Dlx5/6 bigene have revealed a significant functional overlap between Dlx5 and Dlx6 [6e9] that is consistent with the idea that, at least in pharyngeal tissues, functional inputs are collectively measured from the bigene pair rather than from individual paralogs. This support for allele equivalency and a quantitative model of Dlx function in the pharyngeal arches prompted us to consider the relative quantitative contribution made by each protein. While morphological and molecular Dlx5/ and Dlx6/ phenotypes are very similar, consistently, perturbations in Dlx5/ embryos are greater [9] suggesting that Dlx5 alleles
* Corresponding author. E-mail address: [email protected] (A.J. Bendall). http://dx.doi.org/10.1016/j.bbrc.2016.07.044 0006-291X/© 2016 Elsevier Inc. All rights reserved.
contribute more to the functional pool. To better understand the quantitative contribution made by either gene product we compared the expression levels of the two genes in early chick development and directly compared the transcription activity of the Dlx5 and Dlx6 proteins, using a variety of target regulatory regions and cell contexts. 2. Materials and methods 2.1. Embryos Fertile eggs from Barred Rock chickens were obtained from a flock maintained at the Arkell poultry barn (Guelph, ON) and incubated at 38 C. Embryos were staged according to Hamburger and Hamilton [10]. 2.2. Plasmids 2.2.1. Expression plasmids pcDNA3-mycGAL4 was made by amplifying the 50 end of the S. cerevisiae GAL4 coding sequence corresponding to the first 148 amino acids (DNA binding domain) as a BamHI-XbaI fragment and cloning downstream of a 2x Myc epitope sequence in pcDNA3, destroying the BamHI recognition site in the process. The reverse
372
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
GAL4 primer included an in-frame BamHI site upstream of a stop codon such that additional domain-encoding sequences could be sub-cloned as BamHI-XbaI fragments to generate pcDNA3mycGAL4:Dlx5(C), pcDNA3-mycGAL4:Dlx6(C), and pcDNA3mycGAL4:VP16. To preserve the normal spatial arrangement of amino-terminal Dlx domains with respect to the DNA binding domain, a Dlx5(N):GAL4 encoding sequence was made by amplifying each individual domain with PCR using overlapping primers, re-amplifying as a single product, and cloning into pM2 [11] as a BglII-XbaI fragment. This sequence was subsequently tagged with an in-frame 2x Myc epitope sequence at the BglII site. This entire coding sequence was then sub-cloned into pcDNA3 (Life Technologies) as an EcoRI-XbaI fragment. In direct comparisons, we observed more reliable and robust protein expression in transfected cell extracts from the pcDNA3 backbone, compared to pM2 (AQ & AJB, unpublished data). pcDNA3-mycDlx6(N):GAL4 was made by swapping the mycDlx5(N) coding sequence with that of mycDlx6(N) after amplifying the latter with HindIII site-containing primers and screening for orientation. pcDNA3-VP16:GDlx5(eH þ C) and pcDNA3-Dlx5HDm were made from existing plasmids [12] by sub-cloning the appropriate sequence into pcDNA3 as an XbaI-ApaI or EcoRI-XbaI fragment respectively. pcDNA3-Dlx5, pcDNA3-Dlx5DC, pcDNA3-Dlx5DN, pcDNA3-Dlx6, pcDNA3-Dlx6DC, pcDNA3-Dlx6DN have been previously described [12]. 2.2.2. Reporter plasmids Approximately 1 kb of murine Gbx2 genomic sequence was amplified with primers 50 -ACACCTCGAGAGAGGATGACAGCGAGCTTCG-30 and 50 -GTGTAAGCTTGAGCAAACATTCCAGTTTTAATGC-30 and cloned as a XhoI-HindIII fragment into a pGL3-Basic vector that had been modified to contain the minimal promoter sequence present in the pGL4.2x series of vectors (Promega). Approximately 1 kb of murine Bglap1 (Osteocalcin) genomic sequence was amplified with primers 50 -GCAGCTCGAGGCGCTAGGTTACTTT-30 and 50 -TGTGAAGCTTGTCTGTTCTGCACCC-30 and cloned as a XhoI-HindIII fragment into pGL3-Basic. Approximately 2.5 kb of murine Ibsp (Bone sialoprotein) genomic sequence was amplified with primers 50 -CTGCGGTACCGTGTCTAGAAAGCACTGT-30 and 50 -CTGGCTCGAGTGAGTGGCACGGATT-30 and cloned as a KpnI-XhoI fragment into pGL3-Basic. pSNM-luc, containing approximately 0.5 kb of the human c-MYC proximal promoters, is described in Ref. [13]. 2.3. In situ hybridization An 867 bp sequence corresponding to the CDS of chicken Dlx5 was cloned into the BamHI and HindIII sites of pBlueIISK. Antisense riboprobe was synthesised with T7 RNA polymerase from a template linearized with BamHI. An 804 bp sequence corresponding to the CDS of chicken Dlx6 was cloned into the BamHI and HindIII sites of pGEM11Zf(). Antisense riboprobe was synthesised with SP6 RNA polymerase from a template linearized with BamHI. Whole embryos or adjacent cryosections were hybridized with digoxygenin-labelled probes, in parallel, as described in Ref. [14]. 2.4. Transcription assays Subconfluent HEK293T cells (ATCC CRL-11268) were maintained in high glucose DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 mg/ml streptomycin, 2 mM L-glutamine at 37 C and 5% CO2. Unless noted otherwise in the figure legend, cells were transfected in 12-well dishes with 16 ng pRL-SV40-Renilla (Promega), 400 ng of firefly luciferase reporter plasmid, and varying amounts of pcDNA3 plasmids encoding Dlx polypeptides. Total pcDNA3 backbone levels were kept constant with empty vector; for
experiments involving large ranges of expression plasmid, basal activity of reporter plasmids was measured at two or three different levels of empty pcDNA3 for comparison with Dlx-encoding plasmids. For experiments involving GAL4 fusions, reporter activities were compared to basal levels measured in the presence of equal amounts of plasmid encoding the GAL4 DNA binding domain alone. Twenty-four hours post transfection (or 48 h for primary cells), cells were scraped into chilled PBS and centrifuged for 1 min at 3000 rpm. Cells were resuspended in passive lysis buffer (Promega) and incubated on ice for 30 min. Luciferase levels were detected using the Dual-Luciferase Reporter Assay System (Promega) in a Turner TD 20e luminometer. 2.5. Immunoblotting Cells were transfected with 4 mg plasmid in 35 mm dishes, collected 48 h post-transfection in chilled PBS and lysed with sonication in high salt lysis buffer (50 mM Tris, pH 8; 500 mM NaCl; 1% Triton X-100), supplemented with Complete Mini Protease Inhibitor (Roche). Protein concentration was quantified using a BCA protein assay kit (Pierce) and 80 mg total protein was separated by SDS-PAGE, and transferred to PVDF membrane. Antibodies: anti-cMYC (9E10 monoclonal hybridoma supernatant), anti-b-actin (Sigma, A5441), HRP-conjugated goat-a-mouse (Pierce, 31,430). Immunoreactive proteins were visualized with Western Lightning Chemiluminescence Reagent Plus (PerkinElmer Life Sciences, NEL105001EA) and imaged using a Bio-Rad molecular Imager Chemi-Doc XRSþ and ImageLab software. 3. Results & discussion In knockout mice, where differences were apparent, loss of two Dlx5 alleles had a greater effect on morphology and downstream gene expression compared to loss of two Dlx6 alleles [9], suggesting that Dlx5 makes a greater quantitative contribution to the shared function(s). Additionally, where published studies permit a direct comparison, Dlx5 was expressed at higher levels than Dlx6 [9,15]. Thus, expression levels could account for the observation that Dlx5 makes a quantitatively larger contribution to pharyngeal arch development than Dlx6. We asked whether this was a conserved feature in avian embryos. Similarly sized antisense riboprobes to GDlx5 and GDlx6 were used in parallel on whole embryos and tissue sections to measure relative transcript abundance in a variety of developing tissues (Fig. 1). At the earliest developmental stage examined (HH8), Dlx5 and Dlx6 expression appeared to be very similar in the anterior neural folds, where cranial neural crest are undergoing delamination from the neural epithelium (Fig. 1 A/A0 ). At later times, and in most other tissues examined, stronger signals were consistently detected with Dlx5 riboprobes. Differences were particularly obvious in the ventral telencephalon (Fig. 1 B/B0 ), pharyngeal arches (Fig. 1 B/B0 and C/C0 ), and limbs (Fig. 1D/D0 ). Given the similarity between Dlx5 and Dlx6 expression in the cranial neural folds, we interpreted the disparities in signal accumulation in other tissues and at later stages as being representative of differences in transcript abundance, and not due to variation in hybridization or stability of the two riboprobes. Thus, in tissues where they are co-expressed, Dlx5 was typically expressed at higher levels than Dlx6. This is consistent with studies in mouse embryos where Dlx5 was seen to be expressed at higher levels in cranial tissues [9,15]. Such differential expression of Dlx genes may result from orientation-dependent activity of intergenic enhancers. Alternatively, cis-regulatory elements like those found upstream of Dlx1 [16] may account for the higher levels of transcription activity of one gene of the bigene cluster. The quantitative inputs to gene regulatory networks depend on
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
373
Fig. 1. Comparative expression of Dlx5 and Dlx6 in chick embryos. Whole-mount or adjacent section in situ hybridization of chicken embryos at various stages with antisense riboprobes against Dlx5 or Dlx6 as shown. (A, A0 ) Dorsal view of the anterior end of Hamburger and Hamilton stage (HH) 8 embryos; anterior is up. (B, B0 ) Adjacent coronal sections through the head and pharyngeal region of a HH27 embryo. (C, C0 ) Lateral view of the head and pharyngeal arches of HH15 embryos. (D, D0 ) Dorsal view of dissected limbs from the left side of HH26 embryos; anterior is up, distal is to the left. Each pair of embryos or adjacent sections come from the same experiment and were treated equivalently. Abbreviations: anp, anterior neuropore; cnc, cranial neural crest; ey, eye; fb, forebrain; fl, forelimb; fnp, frontonasal process; hl, hindlimb; md, mandibular branch of the first pharyngeal arch; nf, neural fold; ov, otic vesicle; pa1, first pharyngeal arch etc.; vt, ventral thalamus. Scale bars: 0.25 mm (A0 ), 0.5 mm (B0 , C0 ), 1.0 mm (D0 ). Panel B is reproduced from Ref. [14] with permission from the International Journal of Developmental Biology.
both the steady state levels of individual regulatory factors in the nucleus and the strength of their transcriptional activities. To the extent that quantitative differences between paralogous genes are considered, expression levels are typically emphasized while the contribution made to the collective functional protein pool by the activity of individual paralogs is rarely considered and even more rarely tested. We therefore sought to directly compare the transcription activities of the paralogous Dlx5 and Dlx6 proteins. This is especially relevant for these two factors, whose amino- and carboxy-terminal domains (NTD and CTD respectively) have diverged to the point where it is difficult to appreciate their familial relationship and shared gene ancestry from protein sequence alignments [12]. To make a meaningful comparison we chose to measure Tmax for each protein under a variety of conditions. We define Tmax as the maximum response of a given reporter in a given cell type. To find Tmax, we titrated each expression vector against a fixed amount of reporter to identify the maximal effectordependent reporter activity before further levels of exogenous effector started sequestering limiting endogenous basal transcription factors or co-factors and suppressing the reporter response (squelching). In this way, we could directly compare the maximal activity of the Dlx5 and Dlx6 proteins in a way that does not rely on equal transfection efficiencies or precise measurements of protein abundance or stability nor without identifying the underlying mechanisms responsible for that activity, like differential modification of the Dlx5 and Dlx6 proteins in a particular cell type or their ability to occupy binding sites on the reporter. We first asked whether the amino acid sequence divergence between the paralogous Dlx5 and Dlx6 proteins had resulted in different intrinsic transcriptional potencies. As a first step, we fused
the individual amino- or carboxy-terminal domains to the heterologous DNA binding domain (DBD) from the S. cerevisiae transcription factor GAL4, maintaining the native arrangement of each Dlx domain relative to the DBD. We chose the heterologous cell line HEK293T for these experiments to represent a suitable generic mammalian cell in which to measure intrinsic Dlx transcriptional activities independently of associations with co-evolved protein partners. Expression plasmids were co-transfected with a luciferase reporter plasmid containing five tandem copies of the GAL4 upstream activation sequence and a Renilla luciferase-encoding plasmid as a normalizer for variations in cell number, transfection efficiency, and/or protein extraction. To calibrate the sensitivity of this reporter, we first cotransfected the effector plasmid pcDNA3-GAL4:VP16. We observed a robust dose-dependent activation over a 30-fold range of effector plasmid, up to a maximum stimulation of 800-fold over the basal reporter activity in HEK293T cells (Fig. S1). Such robust stimulation was likely a result of the tandem array of five UAS binding sites present in the reporter. Further increases in expression plasmid resulted in a loss of activation (not shown), likely due to the squelching of general transcription factors from the episomal reporter [17]. In contrast, titration of a Dlx5(N):GAL4 expression plasmid showed a dose-dependent stimulation of the reporter by the Dlx5 NTD, to a maximum of 7-fold activation; thereafter, increasing the amount of co-transfected plasmids, even over two orders of magnitude, did not greatly perturb reporter activity (Fig. 2A). The Dlx6 NTD activity peaked at 3-fold activation, and behaved similarly when overexpressed at higher levels (Fig. 2A). Thus, the NTD’s of Dlx5 and Dlx6 showed modest transcriptional activation, with the Dlx5 NTD being twice as active in this assay
374
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
Fig. 2. The amino-terminal and carboxy-terminal domains of Dlx5 and Dlx6 have weak intrinsic transcriptional activation activity. (A, C) Transcription assays with a 5x GAL4 UAS-luciferase reporter co-transfected with the indicated amounts of pcDNA3 plasmids encoding GAL4 DNA binding domain fusions to Dlx5 or Dlx6 amino-terminal domains (N) or Dlx5 or Dlx6 carboxy-terminal domains (C) respectively. Graphs represent the average fold activation of the reporter for a minimum of three independent transfections, each done in triplicate, ± SEM. *, P < 0.05; ns, P > 0.05 (Mann-Whitney U test). (B, D) Immunoblots of expressed proteins in HEK293T cells probed with monoclonal antibodies against the cMyc epitope tag (a-Myc) or b-actin (a-Actin). Note that the Dlx polypeptides were re-detected with the anti-mouse secondary antibody to the a-Actin in panel B. Migration of selected protein standards is indicated at right.
compared to the Dlx6 NTD. In contrast, the Dlx5 and Dlx6 CTD’s were indistinguishable in their behaviour on this reporter; both peaked at 3- to 4-fold activation, but without a clear dose-response of the reporter and large excesses of the effectors did not result in squelching (Fig. 2C). Given the clustered DNA binding sites present in this reporter and the stable expression of all polypeptides in HEK293T cells (Fig. 2B, D) the measured activities of individual Dlx5 and Dlx6 domains in this heterologous context were low; only the Dlx5 NTD showed a clear dose-dependent activation of the reporter. The absence of a strong squelching response, even in the presence of a large excess of fusion protein, also points to a weak interaction with core transcription factors. Thus, in a heterologous DNA binding context, Dlx domains were unable to interact effectively with the transcriptional machinery. The Dlx5 NTD has previously been reported to be a strong activator of a 5x UASeluc reporter [18] but, in that case, the NTD was fused carboxy-terminal to the GAL4 DBD and so is not directly comparable to our experiments or to the wild type Dlx5 protein. We next asked whether the Dlx5 and Dlx6 domains would function differently in the context of their native homeodomain. We selected a 1 kb enhancer sequence from the murine Gbx2 locus for this test since Gbx2 appears to require a threshold of Dlx function for mandibular arch expression, and Gbx2 expression was minimally affected in Dlx5/ and Dlx6/ embryos but was lost or considerably reduced in compound mutants [9]. A Dlx-responsive enhancer has been identified at the Gbx2 locus [9], and this region contains four TAATT core binding sites for Dlx proteins (Fig. 3A). When co-transfected with a Gbx2enh-luc reporter plasmid, both Dlx5(N þ H) and Dlx6(N þ H) polypeptides stimulated in a doseresponsive manner, having a maximal activity of 10- to 16-fold
over basal reporter activity in HEK293T cells (Fig. 3B). The Dlx5 and Dlx6 amino-terminal domains were not distinguishable statistically. Conversely, the Dlx6 CTD was almost three times more active on the Gbx2 enhancer compared to the Dlx5 CTD, stimulating reporter transcription an average of 44-fold, compared to 16-fold for the Dlx5 CTD (Fig. 3C). All polypeptides were stably expressed in HEK293T cells (Fig. 3D). Given that Dlx6 domains were more active than the corresponding domains of Dlx5, we asked whether this would be reflected in the activities of the full-length proteins. We therefore co-transfected pcDNA3 constructs encoding wildtype Dlx5 and Dlx6 (Fig. 3F) with the same Gbx2-dependent reporter in HEK293T cells. Dlx5 and Dlx6 were indistinguishable on this reporter, having a dose-responsive stimulation to a maximum of 18- to 19-fold (Fig. 3E). To compare this level of Dlx5 and Dlx6 activity with that of VP16 on this reporter, we substituted the amino-terminal domain of Dlx5 with the VP16 activation domain and co-transfected the resulting pcDNA3 expression plasmid with the Gbx2-dependent reporter. The VP16:Dlx5 chimeric protein showed a dose-responsive stimulation to a maximum (average) of 50-fold (Fig. S2A). While VP16:Dlx5(eH þ C) showed higher activity on the Gbx2enh-luc reporter (approximately 2-fold more active compared to Dlx5), the differential in activities was much smaller than in the heterologous context of the GAL4 DBD (VP16 > 100-fold more active compared to Dlx5 NTD). We also confirmed that the trans-activation activity of Dlx5 was dependent on binding to the reporter by comparing the activity of a mutated form of Dlx5 in which three amino acids in the amino-terminal arm of the homeodomain were substituted with alanine. This Dlx5HDm protein showed weak stimulation of the Gbx2 reporter at the highest level of co-transfected plasmid tested (Fig. S2B). Thus, a Dlx5 protein
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
375
Fig. 3. Context-dependent differences in Dlx5 and Dlx6 transcription activities. (A) Schematic of the distal Gbx2 enhancer used in the reporter construct. TAATT-containing motifs that represent likely Dlx binding sites are indicated with asterisks above or below the line, depending on orientation. Relative binding site spacing within the enhancer is to scale, distance from the Gbx2 transcription start site (arrow) is indicated in base pairs. (B, C, E) Transcription assays in HEK293T cells with a Gbx2enh-luciferase reporter cotransfected with the indicated amounts of pcDNA3 plasmids encoding Dlx5DC or Dlx6DC (B), Dlx5DN or Dlx6DN (C), or Dlx5 or Dlx6 (E). (D, F) Immunoblots of expressed proteins in HEK293T cells probed with monoclonal antibodies against the c-Myc epitope tag (a-Myc) or b-actin (a-Actin). Migration of selected protein standards is indicated at right. (G) Transcription assays in primary HH stage 22 pharyngeal arch cells with the Gbx2enh-luciferase reporter co-transfected with the indicated amounts of pcDNA3 plasmids encoding Dlx5 or Dlx6. Graphs represent the average fold activation of the reporter for a minimum of three independent transfections, each done in triplicate, ± SEM. *, P < 0.05; ns, P > 0.05 (Mann-Whitney U test).
376
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
whose DNA binding is compromised showed much reduced transactivation activity. We next sought to compare the transcriptional properties of Dlx5 and Dlx6 on other target sequences to ask whether their quantitatively indistinguishable behaviour was a general feature. Based on published studies showing Dlx-responsiveness, we selected two cell type-specific Dlx-regulated targets, Bglap1 (Osteocalcin) [19] and Ibsp (Bone sialoprotein) [20], and a cell cycle regulator, c-Myc [21]. When co-transfected with a luciferase reporter containing the proximal 1008 bp of the Bglap1 gene (Fig. S3A), both Dlx5 and Dlx6 stimulated with the same activities, to a maximum of approximately 5-fold (Fig. S3B). Similarly, when co-transfected with a luciferase reporter containing the proximal 2472 bp of the Ibsp gene (Fig. S3C), both Dlx5 and Dlx6 stimulated with the same activities, to a maximum of approximately 3-fold (Fig. S3D). Finally, when co-transfected with a luciferase reporter containing the proximal 454 bp of the c-Myc gene, and incorporating both transcription start sites (Fig. S3E), both Dlx5 and Dlx6 repressed with the same activities, to a maximum of approximately 3-fold (Fig. S3F). Taken together, our results show that, in a heterologous cellular context, Dlx5 and Dlx6 behave in a quantitatively indistinguishable manner on a variety of natural cis-regulatory sequences containing dispersed DNA binding sites. Finally, we asked whether these two regulatory proteins were differentially adapted to the cell types in which they are normally expressed. To this end we re-examined the transcriptional activities of Dlx5 and Dx6 on the Gbx2enh-luc reporter in primary pharyngeal arch cells; a cell population that normally expresses Dlx5, Dlx6, and Gbx2. Both proteins stimulated the transcription of the Gbx2dependent reporter to higher levels in primary pharyngeal arch cells, compared to HEK293T (compare Fig. 3E and G). We also saw, for the first time, a clear quantitative difference in the activities of these two proteins. Dlx6 activated reporter transcription to approximately 120-fold over basal levels, while Dlx5 peaked at approximately 45-fold (Fig. 3G). Thus, in a nuclear environment for which Dlx5 and Dlx6 proteins are naturally adapted, quantitative differences in transcription activities become apparent, likely reflecting differential recruitment or interaction with co-evolved cofactors. In evolutionary terms then, downstream gene regulatory networks likely responded to the collective input from the Dlx5/6 bigene, influenced by the expression level of each paralog and the potency of each protein as a transcriptional regulator. Dlx5 and Dlx6 proteins appear to be highly adapted to discrete cellular and regulatory contexts and function sub-optimally in other cell types. Consistent with the notion that Dlx5 and Dlx6 proteins are not sufficient for threshold levels of target gene activation in heterologous cellular contexts, Dlx5 was unable to induce ectopic mineralization in Col2-Dlx5 transgenic mice [22]. Additionally, and in contrast to Sox9 [23], neither Dlx5 nor Dlx6 could overcome the cell-crowding requirement for chondrogenesis in vitro in otherwise competent mesenchymal limb bud precursors [12]. Finally, widespread misexpression of Dlx5 or Dlx2 in the early chick head produced stereotypical phenotypes of ectopic bone and cartilage [24], suggesting that not all cranial mesenchyme is competent to respond to misexpressed Dlx proteins. Taken together, our observations suggest that Dlx proteins are not “master” regulatory factors of skeletogenesis, but rather that they exert their pro-differentiation effects in cell types that are permissive to respond. More generally, our data emphasize the need to measure transcriptional activities in an appropriate cellular context when comparing the effects of any two proteins on a specific regulatory sequence. The kind of quantitative approach taken here will also likely have general relevance when it comes to understanding axial
patterning via combinatorial transcription factor action. Acknowledgements We thank Scott Baynton, Robin Hallett, and Hui Zhu for technical help, and Mark Mindon (University of Toronto) for pSNM-luc. This work was supported by the Natural Sciences and Engineering Research Council of Canada (Discovery Grant 249773-2013 to AJB and a Canadian Graduate Scholarship-MSc to RKM). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.bbrc.2016.07.044. Transparency document Transparency document related to this article can be found online at http://dx.doi.org/10.1016/j.bbrc.2016.07.044. References [1] A.J. Bendall, C. Abate-Shen, Roles for Msx and Dlx homeoproteins in vertebrate development, Gene 247 (2000) 17e31. [2] P. Kraus, T. Lufkin, Dlx homeobox gene control of mammalian limb and craniofacial development, Am. J. Med. Genet. A 140 (2006) 1366e1374. [3] G.R. Merlo, B. Zerega, L. Paleari, S. Trombino, S. Mantero, G. Levi, Multiple functions of Dlx genes, Int. J. Dev. Biol. 44 (2000) 619e626. [4] G. Panganiban, J.L. Rubenstein, Developmental functions of the Distal-less/Dlx homeobox genes, Development 129 (2002) 4371e4386. [5] N. Ghanem, O. Jarinova, A. Amores, Q. Long, G. Hatch, B.K. Park, J.L. Rubenstein, M. Ekker, Regulatory roles of conserved intergenic domains in vertebrate Dlx bigene clusters, Genome Res. 13 (2003) 533e543. [6] M.J. Depew, C.A. Simpson, M. Morasso, J.L. Rubenstein, Reassessing the Dlx code: the genetic regulation of branchial arch skeletal pattern and development, J. Anat. 207 (2005) 501e561. [7] A.J. Bendall, Direct evidence of allele equivalency at the Dlx5/6 locus, Genesis 54 (2016) 272e276. [8] A. Beverdam, G.R. Merlo, L. Paleari, S. Mantero, F. Genova, O. Barbieri, P. Janvier, G. Levi, Jaw transformation with gain of symmetry after Dlx5/Dlx6 inactivation: mirror of the past? Genesis 34 (2002) 221e227. [9] J. Jeong, X. Li, R.J. McEvilly, M.G. Rosenfeld, T. Lufkin, J.L. Rubenstein, Dlx genes pattern mammalian jaw primordium by regulating both lower jaw-specific and upper jaw-specific genetic programs, Development 135 (2008) 2905e2916. [10] V. Hamburger, H.L. Hamilton, A series of normal stages in the development of the chick embryo, J. Morphol. 88 (1951) 49e92. [11] I. Sadowski, B. Bell, P. Broad, M. Hollis, GAL4 fusion vectors for expression in yeast or mammalian cells, Gene 118 (1992) 137e141. [12] S.-h.C. Hsu, B. Noamani, D.E. Abernethy, H. Zhu, G. Levi, A.J. Bendall, Dlx5- and Dlx6-mediated chondrogenesis: differential domain requirements for a conserved function, Mech. Dev. 123 (2006) 819e830. [13] L.M. Facchini, S. Chen, W.W. Marhin, J.N. Lear, L.Z. Penn, The Myc negative autoregulation mechanism requires Myc-Max association and involves the cmyc P2 minimal promoter, Mol. Cell. Biol. 17 (1997) 100e114. [14] H. Zhu, A.J. Bendall, Dlx3 is expressed in the ventral forebrain of chicken embryos: implications for the evolution of the Dlx gene family, Int. J. Dev. Biol. 50 (2006) 71e75. [15] G.R. Merlo, S. Mantero, A.A. Zaghetto, P. Peretto, S. Paina, M. Gozzo, The role of Dlx homeogenes in early development of the olfactory pathway, J. Mol. Histol. 38 (2007) 347e358. [16] N. Ghanem, M. Yu, J. Long, G. Hatch, J.L. Rubenstein, M. Ekker, Distinct cisregulatory elements from the Dlx1/Dlx2 locus mark different progenitor cell populations in the ganglionic eminences and different subtypes of adult cortical interneurons, J. Neurosci. 27 (2007) 5012e5022. [17] S. Natesan, V.M. Rivera, E. Molinari, M. Gilman, Transcriptional squelching reexamined, Nature 390 (1997) 349e350. [18] Y. Masuda, A. Sasaki, H. Shibuya, N. Ueno, K. Ikeda, K. Watanabe, Dlxin-1, a novel protein that binds Dlx5 and regulates its transcriptional function, J. Biol. Chem. 276 (2001) 5331e5338. [19] M.Q. Hassan, A. Javed, M.I. Morasso, J. Karlin, M. Montecino, A.J. van Wijnen, G.S. Stein, J.L. Stein, J.B. Lian, Dlx3 transcriptional regulation of osteoblast differentiation: temporal recruitment of Msx2, Dlx3, and Dlx5 homeodomain proteins to chromatin of the osteocalcin gene, Mol. Cell. Biol. 24 (2004) 9248e9261. [20] M.D. Benson, J.L. Bargeon, G. Xiao, P.E. Thomas, A. Kim, Y. Cui, R.T. Franceschi,
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377 Identification of a homeodomain binding element in the bone sialoprotein gene promoter that is required for its osteoblast-selective expression, J. Biol. Chem. 275 (2000) 13907e13917. [21] J. Xu, J.R. Testa, DLX5 (Distal-less Homeobox 5) promotes tumor cell proliferation by transcriptionally regulating MYC, J. Biol. Chem. 284 (2009) 20593e20601. [22] H. Zhu, A.J. Bendall, Dlx5 is a cell autonomous regulator of chondrocyte
377
hypertrophy in mice and functionally substitutes for Dlx6 during endochondral ossification, PLoS One 4 (2009) e8097. [23] C. Healy, D. Uwanogho, P.T. Sharpe, Regulation and role of Sox9 in cartilage formation, Dev. Dyn. 215 (1999) 69e78. [24] C.T. Gordon, I.M. Brinas, F.A. Rodda, A.J. Bendall, P.G. Farlie, Role of Dlx genes in craniofacial morphogenesis: Dlx2 influences skeletal patterning by inducing ectomesenchymal aggregation in ovo, Evol. Dev. 12 (2010) 459e473.
Contents lists available at ScienceDirect
Biochemical and Biophysical Research Communications journal homepage: www.elsevier.com/locate/ybbrc
Measuring inputs to a common function: The case of Dlx5 and Dlx6 Anna Quach, Rachel K. MacKenzie, Andrew J. Bendall* Department of Molecular and Cellular Biology, University of Guelph, Guelph, Ontario, N1G 2W1, Canada
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 June 2016 Accepted 8 July 2016 Available online 11 July 2016
Physically linked Dlx5 and Dlx6 paralogs are co-expressed in vertebrates and various combinations of null alleles in mice demonstrate not only functional redundancy between the paralogous factors but a similar quantitative contribution to craniofacial functions during development. While it is not possible to rule out that the bigene pair contributes some paralog-specific functions it is clear that, for many functions in the head, Dlx5 and Dlx6 are interchangeable. To assess the relative quantitative contribution made by each paralog to bigene function, we have made comparisons of the expression of Dlx5 and Dlx6 in chick embryos and quantitated the transcriptional properties of the encoded proteins in a variety of regulatory and cellular contexts. Our data indicate that the transcriptional activities of both Dlx5 and Dlx6 are very much context dependent; isolated domains fused to a heterologous DNA binding domain have little intrinsic activity, while individual domains are more active when contiguous with their own homeodomain. We find Dlx5 and Dlx6 to be quantitatively indistinguishable on a variety of natural cis-regulatory sequences in a heterologous cellular context but observed quantitatively different transcriptional outputs in cells that normally express these genes, suggesting differential interactions with co-evolved co-activators. © 2016 Elsevier Inc. All rights reserved.
Keywords: Dlx5 Dlx6 Functional equivalence Paralog Transcription factor
1. Introduction The Dlx family of transcriptional regulators execute a variety of developmental functions [1e4]. The six-gene vertebrate family is arranged in three bigene clusters such that each first-order paralog shares tissue-specific enhancers with its cis-linked sister gene [5]. Physically linked Dlx genes are therefore co-expressed in the cranial ectomesoderm of the first pharyngeal arch where they regulate upper and lower jaw patterning [6]. Co-expression of the first-order paralogs Dlx5 and Dlx6 and observations of various null allele combinations of the Dlx5/6 bigene have revealed a significant functional overlap between Dlx5 and Dlx6 [6e9] that is consistent with the idea that, at least in pharyngeal tissues, functional inputs are collectively measured from the bigene pair rather than from individual paralogs. This support for allele equivalency and a quantitative model of Dlx function in the pharyngeal arches prompted us to consider the relative quantitative contribution made by each protein. While morphological and molecular Dlx5/ and Dlx6/ phenotypes are very similar, consistently, perturbations in Dlx5/ embryos are greater [9] suggesting that Dlx5 alleles
* Corresponding author. E-mail address: [email protected] (A.J. Bendall). http://dx.doi.org/10.1016/j.bbrc.2016.07.044 0006-291X/© 2016 Elsevier Inc. All rights reserved.
contribute more to the functional pool. To better understand the quantitative contribution made by either gene product we compared the expression levels of the two genes in early chick development and directly compared the transcription activity of the Dlx5 and Dlx6 proteins, using a variety of target regulatory regions and cell contexts. 2. Materials and methods 2.1. Embryos Fertile eggs from Barred Rock chickens were obtained from a flock maintained at the Arkell poultry barn (Guelph, ON) and incubated at 38 C. Embryos were staged according to Hamburger and Hamilton [10]. 2.2. Plasmids 2.2.1. Expression plasmids pcDNA3-mycGAL4 was made by amplifying the 50 end of the S. cerevisiae GAL4 coding sequence corresponding to the first 148 amino acids (DNA binding domain) as a BamHI-XbaI fragment and cloning downstream of a 2x Myc epitope sequence in pcDNA3, destroying the BamHI recognition site in the process. The reverse
372
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
GAL4 primer included an in-frame BamHI site upstream of a stop codon such that additional domain-encoding sequences could be sub-cloned as BamHI-XbaI fragments to generate pcDNA3mycGAL4:Dlx5(C), pcDNA3-mycGAL4:Dlx6(C), and pcDNA3mycGAL4:VP16. To preserve the normal spatial arrangement of amino-terminal Dlx domains with respect to the DNA binding domain, a Dlx5(N):GAL4 encoding sequence was made by amplifying each individual domain with PCR using overlapping primers, re-amplifying as a single product, and cloning into pM2 [11] as a BglII-XbaI fragment. This sequence was subsequently tagged with an in-frame 2x Myc epitope sequence at the BglII site. This entire coding sequence was then sub-cloned into pcDNA3 (Life Technologies) as an EcoRI-XbaI fragment. In direct comparisons, we observed more reliable and robust protein expression in transfected cell extracts from the pcDNA3 backbone, compared to pM2 (AQ & AJB, unpublished data). pcDNA3-mycDlx6(N):GAL4 was made by swapping the mycDlx5(N) coding sequence with that of mycDlx6(N) after amplifying the latter with HindIII site-containing primers and screening for orientation. pcDNA3-VP16:GDlx5(eH þ C) and pcDNA3-Dlx5HDm were made from existing plasmids [12] by sub-cloning the appropriate sequence into pcDNA3 as an XbaI-ApaI or EcoRI-XbaI fragment respectively. pcDNA3-Dlx5, pcDNA3-Dlx5DC, pcDNA3-Dlx5DN, pcDNA3-Dlx6, pcDNA3-Dlx6DC, pcDNA3-Dlx6DN have been previously described [12]. 2.2.2. Reporter plasmids Approximately 1 kb of murine Gbx2 genomic sequence was amplified with primers 50 -ACACCTCGAGAGAGGATGACAGCGAGCTTCG-30 and 50 -GTGTAAGCTTGAGCAAACATTCCAGTTTTAATGC-30 and cloned as a XhoI-HindIII fragment into a pGL3-Basic vector that had been modified to contain the minimal promoter sequence present in the pGL4.2x series of vectors (Promega). Approximately 1 kb of murine Bglap1 (Osteocalcin) genomic sequence was amplified with primers 50 -GCAGCTCGAGGCGCTAGGTTACTTT-30 and 50 -TGTGAAGCTTGTCTGTTCTGCACCC-30 and cloned as a XhoI-HindIII fragment into pGL3-Basic. Approximately 2.5 kb of murine Ibsp (Bone sialoprotein) genomic sequence was amplified with primers 50 -CTGCGGTACCGTGTCTAGAAAGCACTGT-30 and 50 -CTGGCTCGAGTGAGTGGCACGGATT-30 and cloned as a KpnI-XhoI fragment into pGL3-Basic. pSNM-luc, containing approximately 0.5 kb of the human c-MYC proximal promoters, is described in Ref. [13]. 2.3. In situ hybridization An 867 bp sequence corresponding to the CDS of chicken Dlx5 was cloned into the BamHI and HindIII sites of pBlueIISK. Antisense riboprobe was synthesised with T7 RNA polymerase from a template linearized with BamHI. An 804 bp sequence corresponding to the CDS of chicken Dlx6 was cloned into the BamHI and HindIII sites of pGEM11Zf(). Antisense riboprobe was synthesised with SP6 RNA polymerase from a template linearized with BamHI. Whole embryos or adjacent cryosections were hybridized with digoxygenin-labelled probes, in parallel, as described in Ref. [14]. 2.4. Transcription assays Subconfluent HEK293T cells (ATCC CRL-11268) were maintained in high glucose DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 mg/ml streptomycin, 2 mM L-glutamine at 37 C and 5% CO2. Unless noted otherwise in the figure legend, cells were transfected in 12-well dishes with 16 ng pRL-SV40-Renilla (Promega), 400 ng of firefly luciferase reporter plasmid, and varying amounts of pcDNA3 plasmids encoding Dlx polypeptides. Total pcDNA3 backbone levels were kept constant with empty vector; for
experiments involving large ranges of expression plasmid, basal activity of reporter plasmids was measured at two or three different levels of empty pcDNA3 for comparison with Dlx-encoding plasmids. For experiments involving GAL4 fusions, reporter activities were compared to basal levels measured in the presence of equal amounts of plasmid encoding the GAL4 DNA binding domain alone. Twenty-four hours post transfection (or 48 h for primary cells), cells were scraped into chilled PBS and centrifuged for 1 min at 3000 rpm. Cells were resuspended in passive lysis buffer (Promega) and incubated on ice for 30 min. Luciferase levels were detected using the Dual-Luciferase Reporter Assay System (Promega) in a Turner TD 20e luminometer. 2.5. Immunoblotting Cells were transfected with 4 mg plasmid in 35 mm dishes, collected 48 h post-transfection in chilled PBS and lysed with sonication in high salt lysis buffer (50 mM Tris, pH 8; 500 mM NaCl; 1% Triton X-100), supplemented with Complete Mini Protease Inhibitor (Roche). Protein concentration was quantified using a BCA protein assay kit (Pierce) and 80 mg total protein was separated by SDS-PAGE, and transferred to PVDF membrane. Antibodies: anti-cMYC (9E10 monoclonal hybridoma supernatant), anti-b-actin (Sigma, A5441), HRP-conjugated goat-a-mouse (Pierce, 31,430). Immunoreactive proteins were visualized with Western Lightning Chemiluminescence Reagent Plus (PerkinElmer Life Sciences, NEL105001EA) and imaged using a Bio-Rad molecular Imager Chemi-Doc XRSþ and ImageLab software. 3. Results & discussion In knockout mice, where differences were apparent, loss of two Dlx5 alleles had a greater effect on morphology and downstream gene expression compared to loss of two Dlx6 alleles [9], suggesting that Dlx5 makes a greater quantitative contribution to the shared function(s). Additionally, where published studies permit a direct comparison, Dlx5 was expressed at higher levels than Dlx6 [9,15]. Thus, expression levels could account for the observation that Dlx5 makes a quantitatively larger contribution to pharyngeal arch development than Dlx6. We asked whether this was a conserved feature in avian embryos. Similarly sized antisense riboprobes to GDlx5 and GDlx6 were used in parallel on whole embryos and tissue sections to measure relative transcript abundance in a variety of developing tissues (Fig. 1). At the earliest developmental stage examined (HH8), Dlx5 and Dlx6 expression appeared to be very similar in the anterior neural folds, where cranial neural crest are undergoing delamination from the neural epithelium (Fig. 1 A/A0 ). At later times, and in most other tissues examined, stronger signals were consistently detected with Dlx5 riboprobes. Differences were particularly obvious in the ventral telencephalon (Fig. 1 B/B0 ), pharyngeal arches (Fig. 1 B/B0 and C/C0 ), and limbs (Fig. 1D/D0 ). Given the similarity between Dlx5 and Dlx6 expression in the cranial neural folds, we interpreted the disparities in signal accumulation in other tissues and at later stages as being representative of differences in transcript abundance, and not due to variation in hybridization or stability of the two riboprobes. Thus, in tissues where they are co-expressed, Dlx5 was typically expressed at higher levels than Dlx6. This is consistent with studies in mouse embryos where Dlx5 was seen to be expressed at higher levels in cranial tissues [9,15]. Such differential expression of Dlx genes may result from orientation-dependent activity of intergenic enhancers. Alternatively, cis-regulatory elements like those found upstream of Dlx1 [16] may account for the higher levels of transcription activity of one gene of the bigene cluster. The quantitative inputs to gene regulatory networks depend on
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
373
Fig. 1. Comparative expression of Dlx5 and Dlx6 in chick embryos. Whole-mount or adjacent section in situ hybridization of chicken embryos at various stages with antisense riboprobes against Dlx5 or Dlx6 as shown. (A, A0 ) Dorsal view of the anterior end of Hamburger and Hamilton stage (HH) 8 embryos; anterior is up. (B, B0 ) Adjacent coronal sections through the head and pharyngeal region of a HH27 embryo. (C, C0 ) Lateral view of the head and pharyngeal arches of HH15 embryos. (D, D0 ) Dorsal view of dissected limbs from the left side of HH26 embryos; anterior is up, distal is to the left. Each pair of embryos or adjacent sections come from the same experiment and were treated equivalently. Abbreviations: anp, anterior neuropore; cnc, cranial neural crest; ey, eye; fb, forebrain; fl, forelimb; fnp, frontonasal process; hl, hindlimb; md, mandibular branch of the first pharyngeal arch; nf, neural fold; ov, otic vesicle; pa1, first pharyngeal arch etc.; vt, ventral thalamus. Scale bars: 0.25 mm (A0 ), 0.5 mm (B0 , C0 ), 1.0 mm (D0 ). Panel B is reproduced from Ref. [14] with permission from the International Journal of Developmental Biology.
both the steady state levels of individual regulatory factors in the nucleus and the strength of their transcriptional activities. To the extent that quantitative differences between paralogous genes are considered, expression levels are typically emphasized while the contribution made to the collective functional protein pool by the activity of individual paralogs is rarely considered and even more rarely tested. We therefore sought to directly compare the transcription activities of the paralogous Dlx5 and Dlx6 proteins. This is especially relevant for these two factors, whose amino- and carboxy-terminal domains (NTD and CTD respectively) have diverged to the point where it is difficult to appreciate their familial relationship and shared gene ancestry from protein sequence alignments [12]. To make a meaningful comparison we chose to measure Tmax for each protein under a variety of conditions. We define Tmax as the maximum response of a given reporter in a given cell type. To find Tmax, we titrated each expression vector against a fixed amount of reporter to identify the maximal effectordependent reporter activity before further levels of exogenous effector started sequestering limiting endogenous basal transcription factors or co-factors and suppressing the reporter response (squelching). In this way, we could directly compare the maximal activity of the Dlx5 and Dlx6 proteins in a way that does not rely on equal transfection efficiencies or precise measurements of protein abundance or stability nor without identifying the underlying mechanisms responsible for that activity, like differential modification of the Dlx5 and Dlx6 proteins in a particular cell type or their ability to occupy binding sites on the reporter. We first asked whether the amino acid sequence divergence between the paralogous Dlx5 and Dlx6 proteins had resulted in different intrinsic transcriptional potencies. As a first step, we fused
the individual amino- or carboxy-terminal domains to the heterologous DNA binding domain (DBD) from the S. cerevisiae transcription factor GAL4, maintaining the native arrangement of each Dlx domain relative to the DBD. We chose the heterologous cell line HEK293T for these experiments to represent a suitable generic mammalian cell in which to measure intrinsic Dlx transcriptional activities independently of associations with co-evolved protein partners. Expression plasmids were co-transfected with a luciferase reporter plasmid containing five tandem copies of the GAL4 upstream activation sequence and a Renilla luciferase-encoding plasmid as a normalizer for variations in cell number, transfection efficiency, and/or protein extraction. To calibrate the sensitivity of this reporter, we first cotransfected the effector plasmid pcDNA3-GAL4:VP16. We observed a robust dose-dependent activation over a 30-fold range of effector plasmid, up to a maximum stimulation of 800-fold over the basal reporter activity in HEK293T cells (Fig. S1). Such robust stimulation was likely a result of the tandem array of five UAS binding sites present in the reporter. Further increases in expression plasmid resulted in a loss of activation (not shown), likely due to the squelching of general transcription factors from the episomal reporter [17]. In contrast, titration of a Dlx5(N):GAL4 expression plasmid showed a dose-dependent stimulation of the reporter by the Dlx5 NTD, to a maximum of 7-fold activation; thereafter, increasing the amount of co-transfected plasmids, even over two orders of magnitude, did not greatly perturb reporter activity (Fig. 2A). The Dlx6 NTD activity peaked at 3-fold activation, and behaved similarly when overexpressed at higher levels (Fig. 2A). Thus, the NTD’s of Dlx5 and Dlx6 showed modest transcriptional activation, with the Dlx5 NTD being twice as active in this assay
374
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
Fig. 2. The amino-terminal and carboxy-terminal domains of Dlx5 and Dlx6 have weak intrinsic transcriptional activation activity. (A, C) Transcription assays with a 5x GAL4 UAS-luciferase reporter co-transfected with the indicated amounts of pcDNA3 plasmids encoding GAL4 DNA binding domain fusions to Dlx5 or Dlx6 amino-terminal domains (N) or Dlx5 or Dlx6 carboxy-terminal domains (C) respectively. Graphs represent the average fold activation of the reporter for a minimum of three independent transfections, each done in triplicate, ± SEM. *, P < 0.05; ns, P > 0.05 (Mann-Whitney U test). (B, D) Immunoblots of expressed proteins in HEK293T cells probed with monoclonal antibodies against the cMyc epitope tag (a-Myc) or b-actin (a-Actin). Note that the Dlx polypeptides were re-detected with the anti-mouse secondary antibody to the a-Actin in panel B. Migration of selected protein standards is indicated at right.
compared to the Dlx6 NTD. In contrast, the Dlx5 and Dlx6 CTD’s were indistinguishable in their behaviour on this reporter; both peaked at 3- to 4-fold activation, but without a clear dose-response of the reporter and large excesses of the effectors did not result in squelching (Fig. 2C). Given the clustered DNA binding sites present in this reporter and the stable expression of all polypeptides in HEK293T cells (Fig. 2B, D) the measured activities of individual Dlx5 and Dlx6 domains in this heterologous context were low; only the Dlx5 NTD showed a clear dose-dependent activation of the reporter. The absence of a strong squelching response, even in the presence of a large excess of fusion protein, also points to a weak interaction with core transcription factors. Thus, in a heterologous DNA binding context, Dlx domains were unable to interact effectively with the transcriptional machinery. The Dlx5 NTD has previously been reported to be a strong activator of a 5x UASeluc reporter [18] but, in that case, the NTD was fused carboxy-terminal to the GAL4 DBD and so is not directly comparable to our experiments or to the wild type Dlx5 protein. We next asked whether the Dlx5 and Dlx6 domains would function differently in the context of their native homeodomain. We selected a 1 kb enhancer sequence from the murine Gbx2 locus for this test since Gbx2 appears to require a threshold of Dlx function for mandibular arch expression, and Gbx2 expression was minimally affected in Dlx5/ and Dlx6/ embryos but was lost or considerably reduced in compound mutants [9]. A Dlx-responsive enhancer has been identified at the Gbx2 locus [9], and this region contains four TAATT core binding sites for Dlx proteins (Fig. 3A). When co-transfected with a Gbx2enh-luc reporter plasmid, both Dlx5(N þ H) and Dlx6(N þ H) polypeptides stimulated in a doseresponsive manner, having a maximal activity of 10- to 16-fold
over basal reporter activity in HEK293T cells (Fig. 3B). The Dlx5 and Dlx6 amino-terminal domains were not distinguishable statistically. Conversely, the Dlx6 CTD was almost three times more active on the Gbx2 enhancer compared to the Dlx5 CTD, stimulating reporter transcription an average of 44-fold, compared to 16-fold for the Dlx5 CTD (Fig. 3C). All polypeptides were stably expressed in HEK293T cells (Fig. 3D). Given that Dlx6 domains were more active than the corresponding domains of Dlx5, we asked whether this would be reflected in the activities of the full-length proteins. We therefore co-transfected pcDNA3 constructs encoding wildtype Dlx5 and Dlx6 (Fig. 3F) with the same Gbx2-dependent reporter in HEK293T cells. Dlx5 and Dlx6 were indistinguishable on this reporter, having a dose-responsive stimulation to a maximum of 18- to 19-fold (Fig. 3E). To compare this level of Dlx5 and Dlx6 activity with that of VP16 on this reporter, we substituted the amino-terminal domain of Dlx5 with the VP16 activation domain and co-transfected the resulting pcDNA3 expression plasmid with the Gbx2-dependent reporter. The VP16:Dlx5 chimeric protein showed a dose-responsive stimulation to a maximum (average) of 50-fold (Fig. S2A). While VP16:Dlx5(eH þ C) showed higher activity on the Gbx2enh-luc reporter (approximately 2-fold more active compared to Dlx5), the differential in activities was much smaller than in the heterologous context of the GAL4 DBD (VP16 > 100-fold more active compared to Dlx5 NTD). We also confirmed that the trans-activation activity of Dlx5 was dependent on binding to the reporter by comparing the activity of a mutated form of Dlx5 in which three amino acids in the amino-terminal arm of the homeodomain were substituted with alanine. This Dlx5HDm protein showed weak stimulation of the Gbx2 reporter at the highest level of co-transfected plasmid tested (Fig. S2B). Thus, a Dlx5 protein
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
375
Fig. 3. Context-dependent differences in Dlx5 and Dlx6 transcription activities. (A) Schematic of the distal Gbx2 enhancer used in the reporter construct. TAATT-containing motifs that represent likely Dlx binding sites are indicated with asterisks above or below the line, depending on orientation. Relative binding site spacing within the enhancer is to scale, distance from the Gbx2 transcription start site (arrow) is indicated in base pairs. (B, C, E) Transcription assays in HEK293T cells with a Gbx2enh-luciferase reporter cotransfected with the indicated amounts of pcDNA3 plasmids encoding Dlx5DC or Dlx6DC (B), Dlx5DN or Dlx6DN (C), or Dlx5 or Dlx6 (E). (D, F) Immunoblots of expressed proteins in HEK293T cells probed with monoclonal antibodies against the c-Myc epitope tag (a-Myc) or b-actin (a-Actin). Migration of selected protein standards is indicated at right. (G) Transcription assays in primary HH stage 22 pharyngeal arch cells with the Gbx2enh-luciferase reporter co-transfected with the indicated amounts of pcDNA3 plasmids encoding Dlx5 or Dlx6. Graphs represent the average fold activation of the reporter for a minimum of three independent transfections, each done in triplicate, ± SEM. *, P < 0.05; ns, P > 0.05 (Mann-Whitney U test).
376
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377
whose DNA binding is compromised showed much reduced transactivation activity. We next sought to compare the transcriptional properties of Dlx5 and Dlx6 on other target sequences to ask whether their quantitatively indistinguishable behaviour was a general feature. Based on published studies showing Dlx-responsiveness, we selected two cell type-specific Dlx-regulated targets, Bglap1 (Osteocalcin) [19] and Ibsp (Bone sialoprotein) [20], and a cell cycle regulator, c-Myc [21]. When co-transfected with a luciferase reporter containing the proximal 1008 bp of the Bglap1 gene (Fig. S3A), both Dlx5 and Dlx6 stimulated with the same activities, to a maximum of approximately 5-fold (Fig. S3B). Similarly, when co-transfected with a luciferase reporter containing the proximal 2472 bp of the Ibsp gene (Fig. S3C), both Dlx5 and Dlx6 stimulated with the same activities, to a maximum of approximately 3-fold (Fig. S3D). Finally, when co-transfected with a luciferase reporter containing the proximal 454 bp of the c-Myc gene, and incorporating both transcription start sites (Fig. S3E), both Dlx5 and Dlx6 repressed with the same activities, to a maximum of approximately 3-fold (Fig. S3F). Taken together, our results show that, in a heterologous cellular context, Dlx5 and Dlx6 behave in a quantitatively indistinguishable manner on a variety of natural cis-regulatory sequences containing dispersed DNA binding sites. Finally, we asked whether these two regulatory proteins were differentially adapted to the cell types in which they are normally expressed. To this end we re-examined the transcriptional activities of Dlx5 and Dx6 on the Gbx2enh-luc reporter in primary pharyngeal arch cells; a cell population that normally expresses Dlx5, Dlx6, and Gbx2. Both proteins stimulated the transcription of the Gbx2dependent reporter to higher levels in primary pharyngeal arch cells, compared to HEK293T (compare Fig. 3E and G). We also saw, for the first time, a clear quantitative difference in the activities of these two proteins. Dlx6 activated reporter transcription to approximately 120-fold over basal levels, while Dlx5 peaked at approximately 45-fold (Fig. 3G). Thus, in a nuclear environment for which Dlx5 and Dlx6 proteins are naturally adapted, quantitative differences in transcription activities become apparent, likely reflecting differential recruitment or interaction with co-evolved cofactors. In evolutionary terms then, downstream gene regulatory networks likely responded to the collective input from the Dlx5/6 bigene, influenced by the expression level of each paralog and the potency of each protein as a transcriptional regulator. Dlx5 and Dlx6 proteins appear to be highly adapted to discrete cellular and regulatory contexts and function sub-optimally in other cell types. Consistent with the notion that Dlx5 and Dlx6 proteins are not sufficient for threshold levels of target gene activation in heterologous cellular contexts, Dlx5 was unable to induce ectopic mineralization in Col2-Dlx5 transgenic mice [22]. Additionally, and in contrast to Sox9 [23], neither Dlx5 nor Dlx6 could overcome the cell-crowding requirement for chondrogenesis in vitro in otherwise competent mesenchymal limb bud precursors [12]. Finally, widespread misexpression of Dlx5 or Dlx2 in the early chick head produced stereotypical phenotypes of ectopic bone and cartilage [24], suggesting that not all cranial mesenchyme is competent to respond to misexpressed Dlx proteins. Taken together, our observations suggest that Dlx proteins are not “master” regulatory factors of skeletogenesis, but rather that they exert their pro-differentiation effects in cell types that are permissive to respond. More generally, our data emphasize the need to measure transcriptional activities in an appropriate cellular context when comparing the effects of any two proteins on a specific regulatory sequence. The kind of quantitative approach taken here will also likely have general relevance when it comes to understanding axial
patterning via combinatorial transcription factor action. Acknowledgements We thank Scott Baynton, Robin Hallett, and Hui Zhu for technical help, and Mark Mindon (University of Toronto) for pSNM-luc. This work was supported by the Natural Sciences and Engineering Research Council of Canada (Discovery Grant 249773-2013 to AJB and a Canadian Graduate Scholarship-MSc to RKM). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.bbrc.2016.07.044. Transparency document Transparency document related to this article can be found online at http://dx.doi.org/10.1016/j.bbrc.2016.07.044. References [1] A.J. Bendall, C. Abate-Shen, Roles for Msx and Dlx homeoproteins in vertebrate development, Gene 247 (2000) 17e31. [2] P. Kraus, T. Lufkin, Dlx homeobox gene control of mammalian limb and craniofacial development, Am. J. Med. Genet. A 140 (2006) 1366e1374. [3] G.R. Merlo, B. Zerega, L. Paleari, S. Trombino, S. Mantero, G. Levi, Multiple functions of Dlx genes, Int. J. Dev. Biol. 44 (2000) 619e626. [4] G. Panganiban, J.L. Rubenstein, Developmental functions of the Distal-less/Dlx homeobox genes, Development 129 (2002) 4371e4386. [5] N. Ghanem, O. Jarinova, A. Amores, Q. Long, G. Hatch, B.K. Park, J.L. Rubenstein, M. Ekker, Regulatory roles of conserved intergenic domains in vertebrate Dlx bigene clusters, Genome Res. 13 (2003) 533e543. [6] M.J. Depew, C.A. Simpson, M. Morasso, J.L. Rubenstein, Reassessing the Dlx code: the genetic regulation of branchial arch skeletal pattern and development, J. Anat. 207 (2005) 501e561. [7] A.J. Bendall, Direct evidence of allele equivalency at the Dlx5/6 locus, Genesis 54 (2016) 272e276. [8] A. Beverdam, G.R. Merlo, L. Paleari, S. Mantero, F. Genova, O. Barbieri, P. Janvier, G. Levi, Jaw transformation with gain of symmetry after Dlx5/Dlx6 inactivation: mirror of the past? Genesis 34 (2002) 221e227. [9] J. Jeong, X. Li, R.J. McEvilly, M.G. Rosenfeld, T. Lufkin, J.L. Rubenstein, Dlx genes pattern mammalian jaw primordium by regulating both lower jaw-specific and upper jaw-specific genetic programs, Development 135 (2008) 2905e2916. [10] V. Hamburger, H.L. Hamilton, A series of normal stages in the development of the chick embryo, J. Morphol. 88 (1951) 49e92. [11] I. Sadowski, B. Bell, P. Broad, M. Hollis, GAL4 fusion vectors for expression in yeast or mammalian cells, Gene 118 (1992) 137e141. [12] S.-h.C. Hsu, B. Noamani, D.E. Abernethy, H. Zhu, G. Levi, A.J. Bendall, Dlx5- and Dlx6-mediated chondrogenesis: differential domain requirements for a conserved function, Mech. Dev. 123 (2006) 819e830. [13] L.M. Facchini, S. Chen, W.W. Marhin, J.N. Lear, L.Z. Penn, The Myc negative autoregulation mechanism requires Myc-Max association and involves the cmyc P2 minimal promoter, Mol. Cell. Biol. 17 (1997) 100e114. [14] H. Zhu, A.J. Bendall, Dlx3 is expressed in the ventral forebrain of chicken embryos: implications for the evolution of the Dlx gene family, Int. J. Dev. Biol. 50 (2006) 71e75. [15] G.R. Merlo, S. Mantero, A.A. Zaghetto, P. Peretto, S. Paina, M. Gozzo, The role of Dlx homeogenes in early development of the olfactory pathway, J. Mol. Histol. 38 (2007) 347e358. [16] N. Ghanem, M. Yu, J. Long, G. Hatch, J.L. Rubenstein, M. Ekker, Distinct cisregulatory elements from the Dlx1/Dlx2 locus mark different progenitor cell populations in the ganglionic eminences and different subtypes of adult cortical interneurons, J. Neurosci. 27 (2007) 5012e5022. [17] S. Natesan, V.M. Rivera, E. Molinari, M. Gilman, Transcriptional squelching reexamined, Nature 390 (1997) 349e350. [18] Y. Masuda, A. Sasaki, H. Shibuya, N. Ueno, K. Ikeda, K. Watanabe, Dlxin-1, a novel protein that binds Dlx5 and regulates its transcriptional function, J. Biol. Chem. 276 (2001) 5331e5338. [19] M.Q. Hassan, A. Javed, M.I. Morasso, J. Karlin, M. Montecino, A.J. van Wijnen, G.S. Stein, J.L. Stein, J.B. Lian, Dlx3 transcriptional regulation of osteoblast differentiation: temporal recruitment of Msx2, Dlx3, and Dlx5 homeodomain proteins to chromatin of the osteocalcin gene, Mol. Cell. Biol. 24 (2004) 9248e9261. [20] M.D. Benson, J.L. Bargeon, G. Xiao, P.E. Thomas, A. Kim, Y. Cui, R.T. Franceschi,
A. Quach et al. / Biochemical and Biophysical Research Communications 478 (2016) 371e377 Identification of a homeodomain binding element in the bone sialoprotein gene promoter that is required for its osteoblast-selective expression, J. Biol. Chem. 275 (2000) 13907e13917. [21] J. Xu, J.R. Testa, DLX5 (Distal-less Homeobox 5) promotes tumor cell proliferation by transcriptionally regulating MYC, J. Biol. Chem. 284 (2009) 20593e20601. [22] H. Zhu, A.J. Bendall, Dlx5 is a cell autonomous regulator of chondrocyte
377
hypertrophy in mice and functionally substitutes for Dlx6 during endochondral ossification, PLoS One 4 (2009) e8097. [23] C. Healy, D. Uwanogho, P.T. Sharpe, Regulation and role of Sox9 in cartilage formation, Dev. Dyn. 215 (1999) 69e78. [24] C.T. Gordon, I.M. Brinas, F.A. Rodda, A.J. Bendall, P.G. Farlie, Role of Dlx genes in craniofacial morphogenesis: Dlx2 influences skeletal patterning by inducing ectomesenchymal aggregation in ovo, Evol. Dev. 12 (2010) 459e473.