Heparan sulfate proteoglycans (HSPGs) are a family of complex molecules that are critically important for normal embryonic development. Their basic molecular structure consists of a variable number of heparan sulfate (HS) glycosaminoglycan chains covalently attached to a core protein (Casu and Lindahl, 2001). The heterogeneity of HSPGs relates to changes in the core protein, as well as to the type and size of the HS chains. HSPGs are constituents of the cell membrane and extracellular matrix (ECM) that regulate the structural integrity and permeability of the ECM. Furthermore, they bind to and control the biological function of signaling molecules, such as growth factors and morphogens (Allen and Rapraeger, 2003; Johnson et al., 2004; Kreuger et al., 2006). These interactions with signaling molecules underlie in part the importance of HSPGs in development (Lin et al., 1999; Galli et al., 2003; Bornemann et al., 2004).
The structure of HS defines its interactions with other molecules, and is largely determined by its biosynthesis (Binari et al., 1997; Bellaiche et al., 1998; Bornemann et al., 2004) and posterior enzymatic degradation by heparanase. While disruption of either process appears to disturb normal development, heparanase manipulations have mainly involved exogenous treatment of the developing embryo with bacterial heparatinase (Yost, 1992; Brickman and Gerhart, 1994; Itoh and Sokol, 1994; Walz et al., 1997). As such, the role of the endogenous heparanase in embryonic development is not well understood. Overexpression of heparanase in mouse and Xenopus produces no overt early embryonic phenotype (Zcharia et al., 2004, 2005; Bertolesi et al., 2008). Yet, loss-of-function studies argue for an important role for heparanase in developmental events. Knockdown of heparanase in Xenopus results in a failure of gastrulation (Bertolesi et al., 2008), and while development occurs normally in heparanase mouse mutants, an up-regulation of a different class of proteolytic enzymes, the metalloproteinases, may by some unknown means compensate for heparanase loss (Zcharia et al., 2009).
Three distinct forms of heparanase are present in the Xenopus embryo (Bertolesi et al., 2008). Xenopus HpaL is a latent preproheparanase of approximately 67 kDa (XHpaL; 65 kDa in humans), which is posttranslationally processed to generate amino (8 kDa) and carboxy (50 kDa) termini products, which associate noncovalently to generate the active enzyme (HpaL active; Levy-Adam et al., 2003; McKenzie et al., 2003). A third heparanase, Hpa short (HpaS), is produced in Xenopus and humans by alternative splicing of the heparanase gene in which exon 5 is skipped to produce a protein missing 58 residues (Nasser et al., 2007; Bertolesi et al., 2008). While the enzymatically active long form (Hpa Active) will degrade and remodel HSPGs, HpaS is not enzymatically active and promotes cell adhesion and migration (Goldshmidt et al., 2003; Zetser et al., 2003; Gingis-Velitski et al., 2004; Sotnikov et al., 2004; Bertolesi et al., 2008).
These distinct roles suggest that the function of heparanase in development is likely more complex than expected. Moreover, recent studies indicate that heparanase and HSPGs also play important roles inside the cell. For instance, nuclear colocalization of the HSPG syndecan 1 with endogenous heparanase may regulate gene transcription (Schubert et al., 2004; Chen and Sanderson, 2009; Zong et al., 2009). A critical step in understanding the role of heparanase during development is analysis of the dynamic and spatial aspects of its expression. To this end, we studied heparanase expression in the developing Xenopus embryo by generating a Xenopus heparanase antibody, and identified and analyzed the tissue-specific activity of the promoter region upstream of the XHpa gene.
The Long, Short, and Active Forms of Heparanase Exhibit Distinct Expression Profiles Through Embryonic Development
We first asked whether XHpa expression is dynamically and spatially regulated through the major period of embryonic development, from the fertilized egg through to the four day old larvae (stage 44). These data could then be used to better understand protein regulation at the transcriptional level. Previously, we investigated the differential expression of XhpaL and XhpaS in whole embryos at the mRNA level by reverse transcriptase-polymerase chain reaction (RT-PCR), and XHpa at the protein level with a commercially available monoclonal antibody raised against the human Hpa (clone HP3/17; Bertolesi et al., 2008). Because XHpaS was barely detectable with the human Hpa antibody by Western blot analysis, this antibody could not be used to address the tissue specific expression of the XHpa protein.
Thus, we generated an antibody specific to Xenopus laevis heparanase. The αXHpa antibody was raised against a peptide localized to the carboxy end of proheparanase (319SVPGKRIWLG ETSS332). Because this peptide also resides within the 50 kDa mature heparanase subunit, the antibody was expected to recognize the XHpaL and the XHpaS isoforms, as well as the mature active form (XHpa active). Indeed, three bands of the expected molecular weights are detected by Western blot in protein lysates obtained from Xenopus embryos at different developmental stages (Fig. 1A). The upper (67 kDa) and middle (57 kDa) bands correspond to bands of the same molecular weight in cell extracts from transfected COS-7 cells overexpressing XHpaL and XHpaS, respectively. The low affinity human Hpa (clone HP3/17) antibody recognizes similar molecular weight bands (Fig. 1A). A more intense band of approximately 50 kDa, corresponding to the molecular weight of the active form of XHpaL, is detected in extracts from X. laevis embryos.
We found previously that Xenopus hpaL transcripts are initially high before gastrulation and up-regulated selectively during late gastrulation and early neurulation, while transcripts for both isoforms are abundant in tailbud through tadpole stages (Bertolesi et al., 2008). We wished to verify these data at the protein level by Western blot analysis with the anti XHpa antibody. This confirmation had not been possible previously for XHpaS, because of the particularly low affinity shown by the anti human Hpa antibody for this isoform.
In general, analyses by Western blot using the anti XHpa antibody are consistent with the reported gene expression (Bertolesi et al., 2008), and further show expression of all three forms of heparanase (long, short, and active) in unfertilized eggs and stage 2 embryos (Fig. 1B). XHpaL and XHpaS levels are initially high before and over the period of gastrulation (stages 2–11), but decrease in the neurulating embryo (stages 15 to 17). After neurulation, both isoforms are up-regulated. Interestingly, in contrast, XHpaL active shows little developmental control through these early periods of embryonic development, but is dramatically up-regulated in the stage 40 tadpole (Fig. 1B). In fact, the active heparanase is the main form detected in adults, with high levels of expression in muscle, heart, brain, stomach, liver, lung, kidney, and spleen (Fig. 1C). While XHpaS is expressed embryonically, its levels start to decrease at stage 32 and continue to abate over the next day or two of development (stage 40 and 44; Fig. 1B). XHpaS is undetectable in stage 52 tadpoles, just before metamorphosis (data not shown), and the adult only shows low level expression in ovary (two of three experiments; Fig. 1C).
The long splice variant heparanase from X. tropicalis was recently identified (gene bank number BC160516), and the amino acid sequence is 87% identical to that of X. laevis. Two amino acids critical for antibody recognition, however, differ within the peptide sequence used to generate the XHpa antibody (324HV325 instead of 324RI325): Western blot analysis shows that the heparanase from X. tropicalis is essentially undetected by the XHpa antibody. Of note, the human anti-Hpa antibody recognizes both X. tropicalis and X. laevis heparanase in an overexpression system (Fig. 1A).
Maternally Expressed Heparanase Localizes to the Jelly Coat, Plasma Membrane, and Nucleus
Next we used the XHpa antibody to examine exactly where and when in the developing embryo the protein is present. Of note, an affinity-purified antibody raised against a peptide sequence encoded by the spliced exon 5 (174NVLLRNDNNEWNTS187) of heparanase, with the aim of differentiating between XHpaL and XHpaS, failed to react to XHpa (data not shown).
We first asked where the maternal heparanase is localized by immunohistochemistry of transverse cryostat sections of eggs before, or immediately after, fertilization. The specificity of the antiserum to recognize heparanase in tissue sections was confirmed by lack of staining in embryos treated with the preimmune serum (Supp. Fig. S1; which is available online). Distinct from mammalian eggs, amphibian eggs are surrounded by a complex ECM that consists of both a vitelline envelope proximal to the egg, and a three layered thick jelly coat, consisting of complex and partially characterized glycoproteins that are deposited around the egg as it passes through the oviduct (Hedrick and Nishihara, 1991; Guerardel et al., 2000; Zong et al., 2009). The jelly coat provides a protective environment for the developing embryo, and, by modulating sperm binding to the vitelline membrane, blocks poly-spermy. This coat can be removed by treating eggs with L-cysteine. Heparanase immunoreactivity is detected in the more external layer of the jelly coat of untreated embryos, but not in those embryos devoid of a jelly coat (Fig. 2A). Two major immunopositive domains are observed in both oocytes and embryos immediately after fertilization: The cell membrane, and an intense intracellular aggregate found in close proximity to nuclear DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) staining (Fig. 2A,B). This pattern is present through stages 2–10 (Fig. 2B,C; Supp. Fig. S1), although the membrane staining diminishes by stage 6, only to become more evident again at early gastrula stages (stage 9/10; Fig. 3). Of interest, the anti XHpa antibody labels uniformly the plasma membrane of cells in the animal pole of stage 10 embryos, but exhibits a more punctate staining of the plasma membrane of cells in the vegetal pole (Fig. 3A,B). Moreover, XHpa immunoreactivity is present in the nucleus of ectodermal cells of the animal pole of gastrulating embryos (stage 10; Fig. 3C).
Heparanase Expression During Neurulation, Organogenesis, and Tailbud Stage
The ectodermal layer of the gastrulating embryo shows heparanase expression and a particularly strong signal is detected in the neurogenic region located dorsal to the blastocoel, from which the nervous system and associated organs originate. The mesodermal layer immediately below the ectoderm also expresses heparanase (Fig. 3C). Not surprisingly, given the early domains of heparanase expression, heparanase immunoreactivity is present at all subsequent stages of development both in mesodermally derived tissues and in the central nervous system (CNS; Fig. 4). In the CNS, there is strong and highly localized expression of Hpa in the forebrain, midbrain and hindbrain (Fig. 4D,G). In the head, Hpa is expressed in neuronally based organs, such as the eye, the olfactory placodes, the pineal gland and the otic vesicle (Fig. 4C–E,G). Several other organs such as heart, liver and kidney, which arise from the mesoderm, also show heparanase expression at tadpole stages (Fig. 4B,D–F). Heparanase expression is particularly strong in the pronephros, including the pronephric tubule and duct (Fig. 4B,D–F), and is present as early as stage 24, when the cells in these tissues start to differentiate (data not shown).
Together, these results indicate that expression in the mesoderm and ectoderm in the gastrulating embryo continues through tadpole stages, with heparanase present in the nervous system and in mesodermally derived organs. Of note, the Western blot data argue that the immunoreactivity in the tadpole and larvae likely corresponds to XHpa active.
Expression of Heparanase in X. laevis Is Driven by Two Promoters
Our expression data indicate that the expression of XHpa is both temporally and spatially regulated during embryonic development. To establish whether transcriptional control is a potential underlying mechanism that explains developmental changes in XHpa protein expression, we identified and amplified by PCR a DNA segment corresponding to the 2.1-kb region upstream of the XHpa gene from Xenopus tropicalis. The cloned DNA contains the putative promoter region, and the first exon and part of the second exon of XHpa, stopping 55 nucleotides upstream of the translation start codon (numbered as +1 hereafter; Fig. 5A). To confirm the presence of a functional promoter in this region of DNA, the 2.1-kb DNA fragment (−2199: −55) was subcloned into a luciferase reporter system. The engineered construct was used to transiently transfect two Xenopus cell lines, A6 and XTC: The A6 cell line derives from adult Xenopus kidney, and XTC is a fibroblast cell line obtained from X. laevis tadpoles (Rafferty and Sherwin, 1969; Pudney et al., 1973). Both cell lines express endogenous Xenopus HpaL, as determined by RT-PCR (Fig. 5B), and XHpa active protein (Western blot, data not shown), but do not express the short isoform. After transfection, cells were cultured for 24–28 hr, and then harvested and analyzed for luciferase activity. Activity increases 10 and 13 times over the luciferase construct without a promoter (pGL3-basic) for A6 and XTC cells, respectively (Fig. 5A). The Xenopus heparanase promoter is also functional in mammalian cells, because COS-7 cells, derived from monkey kidney and PC12 cells, obtained from rat pheochromocytoma, show 12 and 4 times increases over the basic promoter activity, respectively (Fig. 5A).
Potential transcription binding sites were identified by analysis of the genomic sequence using MatInspector (Genomatrix) and Tfsi-tescan software. Two TATA binding sites, several sites for octamer binding protein complexes, and AP1 sites are present (Fig. 6). The 2.1-kb fragment containing the Xenopus Hpa promoter was aligned with the corresponding region of the human and mouse Hpa promoters to determine similarities between the DNA and to identify promoter elements conserved between the three species. Interestingly, while human and mouse sequences show a high degree of identity (more than 80%) in the first 600 bp upstream of the translation start site (Jiang et al., 2002; de Mestre et al., 2003; de Mestre et al., 2007), Xenopus DNA is highly divergent (38% identity with both human and mouse). Moreover, Sp1 and Ets-relevant elements (ERE) sites, critical for human heparanase promoter activity (Jiang et al., 2002) are not present in Xenopus. Furthermore, the octamer binding and TATA box sites present in the Xenopus sequence are not conserved in human. These data argue that different regulatory mechanisms have evolved between these species.
Blasting the cloned sequence against Xenopus expressed sequence tag (EST) let us identify several heparanase EST clones that were isolated by using 5′RACE-PCR (Sanger Institute Xenopus blast serve; http://www.sanger.ac.uk/Projects/X_tropicalis/). Clones with four different transcription start sites, corresponding to nucleotides −356, −348 and −327 on exon 1, and nucleotide −120 on exon 2 (sites 1, 2, 3, and 4, respectively) are depicted in Figure 6. Three other sites of transcription initiation, at nucleotides −330, −101, and −98 (sites A and B/C in Fig. 6), have also been described for the human gene (Hulett et al., 1999; Vlodavsky et al., 1999; Dong et al., 2000; Jiang et al., 2002). Alternative transcriptional start sites do not alter the putative translation initiation start codon, which remains unaffected and localized inside exon 2.
To identify possible regulatory sequences, and to determine the minimal promoter region, we conducted a functional analysis of the −2199: −55 XHpa promoter construct by generating several deletions of the 5′ and 3′ ends. Luciferase reporter system analyses of all constructs were performed in transfected A6 and XTC cells lines. In addition, constructs were injected into both blastomeres of 2−cell stage Xenopus laevis embryos, and embryos processed at stage 10 for luciferase activity.
Deletion of the XHpa promoter within 224 bp upstream of the ATG translation initiation site actually increases luciferase activity by two-fold in A6 and XTC cells (compare −2199: −55 vs. −2199: −224; Fig. 7A). In contrast, constructs deleted upstream of nucleotide −473 (−2119: −473 and −2199: −621) lose promoter activity such that levels of luciferase are similar to those seen for the reporter without promoter. These results suggest that the sequences for cis-regulatory elements required for basal promoter function are present in the region between −473 and −224.
Surprisingly, progressive 5′ deletions of the −2199: −55 construct generate a construct, −224: −55, which lacks the first exon, and the basal promoter region determined above, but that can drive luciferase activity (Fig. 7A). These results suggest the presence of a second basal promoter in the region between −224 and −55. To confirm and narrow down the limits of the two basal promoter regions, additional reporter constructs were made (−473: −224; −473: −331; −621: −331; −224: −55; −224: −117; −331: −224 and −117: −55). Luciferase assays using these constructs expressed in all three systems clearly show that two nonoverlapping regions, one upstream of exon 1 (−473 to −331; named P1), and another in the intron 1 (−224 to −117; named P2), are able to independently drive reporter gene expression (Fig. 7A).
To determine if the two promoters are present and functional in the human Hpa gene, we used a previously characterized human Hpa promoter to generate similar constructs to those described above (Jiang et al., 2002). Because the human Hpa promoter was not functional in any of the Xenopus cell systems (data not shown), we used COS-7 cells instead. The human Hpa promoter contains two regions with Sp1 binding sites (Sp1-C in the first region, and Sp1-B and Sp1-A in the second region), alternated by three regions with ERE sites (very distal: ERE-F and ERE-E; distal: ERE-D and ERE-C; and proximal: ERE-B and ERE-A; Figs. 6, 7B; Jiang et al., 2002). In agreement with previous results, the maximum promoter activity occurs with constructs that contain the distal and proximal Sp1 and ERE sites (constructs −472: −57 and −331: −57). Deletion of the most distal ERE sites (F and E) increases luciferase activity two-fold (−622: −57 vs. −472: −57), suggesting that these sites are not necessary for promoter activity.
The construct −220: −57 produces close to the maximum luciferase response detected in COS-7 cells, suggesting that Sp1-B/A plus ERE-C/D are sufficient for promoter activity. However, the construct −472: −220, which lacks the proximal ERE tandem (ERE-B/A) and Sp1-B/A, reveals a strong functional promoter (25 times higher than the basal luciferase construct). The luciferase activity generated by this construct is reduced five-fold by deletion of 150 bp from the 5′ end (construct −331: −220), although luciferase activity remains significantly higher than basal levels (Fig. 7B). These data suggest that other cis-responsive elements in the region −472: −220 are able to drive reporter gene expression. Both human Hpa promoter constructs deleted downstream of −331 (−472: −331 and −622: −331) fail to show promoter activity, while the equivalent Xenopus constructs are functional. These data further support the idea that different regulatory mechanisms have evolved between these two species. However, the first intron of both species retain promoter activity, because constructs containing only intron 1 sequence (Xenopus, −224: −117; human, −220: −120) drive luciferase expression in COS-7 cells.
Oct-25 and Oct-60 Induce Expression of Heparanase in Early Embryos
Our promoter analysis indicates that there are putative transcription factor binding sites in the promoter of Xenopus that are not there in human, and vice versa. These data indicate that different transcription factors control the expression of human and Xenopus heparanase. To test this idea, we investigated whether Xenopus heparanase expression is regulated by an octamer transcription factor (Oct), for which the Xenopus, but not the human, promoter has multiple consensus binding sites (numbers 1 to 5 in Fig. 6). The octamer transcription factors are a large group of POU family transcription factors (Class V) that recognize a specific octameric DNA consensus sequence (Hinkley et al., 1992; Schonemann et al., 1998). We were particularly interested in Oct regulation in that Oct binding sites 4 and 1 are present in the promoter regions P1 and P2, respectively. To determine if these binding sites are necessary for heparanase expression, site-directed mutations were made in the oct-4 site (P1 mutated), oct-1 site (P2 mutated), or both (P1+P2 mutated). Luciferase activity of the respective promoter constructs were first analyzed in microinjected blastomeres. A significant decrease in luciferase activity was detected for all mutated constructs with respect to the wild type promoter (Fig. 8A). Of interest, mutation of either the P1 or P2 oct site produced a comparable decrease in luciferase activity, and no additive effect was detected in the double mutated expression construct, suggesting that the oct-4 and oct-1 sites are both used to modulate hpa transcription. Of note, P1 and P2 oct site mutants showed 6−7 times more activity than the basal pGL3 construct, which suggests additional regulatory sites are sufficient to drive heparanase expression.
Oct-25, Oct-60, and Oct-91 are members of the Oct family that are expressed in Xenopus oocytes and developing embryos at gastrulation and early neurulation (Hinkley et al., 1992; Cao et al., 2006). To determine if Oct transcription factors can modulate heparanase expression the hpa promoter (−2119: −55) luciferase construct was co-injected with a plasmid encoding Oct-25, Oct-60, or Oct-91 into both blastomeres of two-cell stage embryos. Embryos were then processed at stage 12 for luciferase activity. Heparanase promoter activity was significantly increased by 2–3 times with Oct-25 and Oct-60 overexpression. In contrast, Oct-91 overexpression did not modify luciferase expression (Fig. 8B). These results argue that Oct proteins are capable of regulating heparanase promoter activity. Of interest, co-expression of Oct-25 and Oct-60 with the P1, P2, and P1+P2 mutated promoter constructs still induced 2 times more luciferase activity compared with vector alone co-transfected blastomeres (data not shown). Most likely Oct-25 and Oct-60 use additional oct-binding sites to induce heparanase promoter activity, although we cannot rule out that overexpressed Oct proteins can overcome and function through the mutated P1 and P2 oct-binding sites. Together, these results indicate that at least two Oct-binding sites present in P1 and P2 may regulate heparanase promoter in blastomeres, and that Oct sites in other regions of the promoter may mediate Oct-25 and Oct-60 regulation.
Differential Activity of Xenopus Promoters P1 and P2 in Distinct Tissues
Next, we compared the activity of the two Xenopus promoters in four different expression systems (Xenopus blastulas, and A6, XTC, and COS-7 cell lines). Our data indicate that the two promoters exhibit differential activity in distinct tissues and cell types. For instance, in A6 and XTC cell lines promoter activity for constructs containing P1 alone is approximately 6–8 times higher than for those constructs with only P2. Further, results with the A6 and XTC cell lines argue that while P2 has functional promoter activity by itself (construct −224: −55 and −224: −117), it competes with, or acts as a repressor of, P1 when both are present: The constructs that contain only P1 show more activity than constructs including both P1 and P2 (compare −2119: −55 vs. −2119: −244; −473: −55 vs. −473: −224; −621: −55 vs. −621: −331 vs. −331: −55). Of interest, the apparent “repressor” property of P2 over P1 is not observed when the promoter constructs are introduced into two-cell stage Xenopus embryos or COS-7 cells. Here, promoter activity of the P1 only construct is similar to that of constructs with P2 alone, or both P1 and P2 (Fig. 7A,B). These results suggest that P1 and P2 have differential functional activity in distinct tissues.
To test this idea further, the luciferase activity of constructs containing P1 or P2 alone, or that included both P1 and P2, were studied in vivo in distinct regions of the nervous system of Xenopus embryos, which we found express heparanase (see above). cDNA plasmids encoding the different luciferase promoter variants were introduced into either the brain or eye of Xenopus embryos at stage 27 by electroporation. To normalize for transfection efficiency a renilla construct was co-electroporated along with the luciferase promoter constructs, whereas a pCS2-GFP construct was co-electroporated to verify targeting specifically to either eye or brain: GFP expression in eye structures including the lens and neural retina was referred to as “eye” expression, while “brain” expression included embryos with GFP in the forebrain, midbrain and/or hindbrain (Fig. 9A). Embryos were allowed to develop until stage 37/38, at which time luciferase and renilla activity in each individual embryo were measured. A minimum of five embryos with correct gene targeting were measured for each construct analyzed.
In the brain, the normalized luciferase activity of the Xenopus Hpa promoter construct containing P1 alone (−473: −331) is approximately 4 times higher than that of the construct which included only the P2 promoter (−224: −55). Similar to what is observed in A6 and XTC cells, the construct with both P1 and P2 (−473: −55) exhibits significantly less activity than P1 alone (Fig. 9B). In contrast, no statistical difference is observed between any of the three constructs when expressed in eye (Fig. 9B). These results suggest that heparanase expression in different tissues is differentially controlled by P1 and P2, and that depending on the context, P2 may “repress” the functional activity of P1.
While the blastomere data suggest that Oct proteins work through Oct binding sites in P1 and P2 to control heparanse expression in the early embryo, it is less clear they do so in the nervous system. Certainly, Oct-25 and Oct-60 are not expressed in Xenopus embryos after stage 14 (Hinkley et al., 1992). Thus, to ask whether the Oct binding sites in P1 and P2 are still capable of regulating heparanase promoter activity in the developing nervous system, in the absence of an identified Oct transcriptional regulator, we electroporated the oct binding site 1 and 4 mutated constructs into brain and eye and asked whether promoter activity was similarly affected as seen in the blastomere assay. Promoter constructs were co-electroporated with oct binding site 1 and 4 mutated cDNA plasmids at stage 27 into either the eye or the brain of Xenopus embryos. In contrast to what was seen for promoter activity of these constructs in the early embryo, in both eyes and brain the luciferase activity of the promoter constructs with oct binding sites 1 and 4 mutated is similar to that of the wild type promoter (Fig. 9C). These data provide further support for the notion that Oct transcription factors use the oct 1 and 4 binding sites to regulate heparanase promoter activity in the early embryo. Oct-25 and Oct-60 overexpression in both eye or brain is still able to induce heparanase promoter activity two- to three-fold (Fig. 9C), arguing that an Oct-dependent regulatory mechanism would still be capable of functioning in the tadpole nervous system to control heparanase expression, although possibly through Oct family members other than Oct-25 and Oct-60 (Wolf et al., 2009).
In this study we determined that XHpa protein expression is spatially and dynamically regulated, and that transcriptional mechanisms could play a role. Our results indicate that: (1) all three heparanase isoforms are maternally expressed, with protein detected in prefertilized zygotes; (2) XHpaS is a developmental variant, with expression largely restricted to embryos (up to stage 44), with the exception of ovary in the adult frog; (3) in the early embryo (up to stage 10), heparanase is localized to both the plasma membrane and the nucleus; (4) levels of XHpaL and XHpaS decrease postgastrulation and early neurulation, and then rise again, whereas levels of the mature XHpaL stay constant over this time period; (5) several organs and tissues express heparanase during development, but expression is mainly detected in structures of the developing nervous system; (6) two promoters with distinct activities in different tissues drive the expression of heparanase in Xenopus, and (7) Oct25 and Oct60 transcription factors may modulate heparanase expression in the early embryo, because mutation of specific Oct binding sites within the heparanase promoter and overexpression of Oct-25 and Oct-60 altered the activity of the heparanase promoter reporter gene. These data argue that heparanase is widely expressed during development, but localization and levels are finely regulated.
Our work, and those of others, indicates that XHpaL is enzymatically active and processes HSPGs, whereas enzymatically inactive splice variants in various species bind HSPGs and promote cell adhesion and migration (Goldshmidt et al., 2003; Zetser et al., 2003; Sotnikov et al., 2004; Bertolesi et al., 2008). Our expression data argue that the different forms of heparanase are regulated tightly both spatially and temporally in the egg, embryo and the adult. Expression of heparanase has been reported as early as the morula mouse embryo (Revel et al., 2005). We found that XHpaL, XHpaS, and XHpa active are maternally expressed, supporting the idea of a potential role for heparanase in early embryonic development. Because of the large size of Xenopus embryos, we were able to show that a fraction of the maternal heparanase is found in proximity to the DNA. Potentially, the heparanase close to the DNA corresponds to XHpa active, because it is also detected with the human anti-Hpa antibody that mainly recognizes the active form (data not shown). In support, in myeloma and fibroblast cell lines a fraction of the active form of heparanase colocalizes with syndecan-1 in the nucleus, where it appears to help silence gene expression (Chen and Sanderson, 2009; Zong et al., 2009). Of note, in Xenopus, gene transcription starts at the midblastula transition (stage 8) and increases approximately 20 fold over the next hour (Newport and Kirschner, 1982), just before the last time point at which heparanase is detected in the nucleus (stage 10).
The role(s) of heparanase in development needs to be elucidated. Experiments with exogenous introduction of bacterial heparanase into embryos at different developmental stages (Yost, 1992; Brickman and Gerhart, 1994; Itoh and Sokol, 1994; Walz et al., 1997) say more about a requirement of HSPGs in different biological processes than the biological role of the endogenous heparanase. Defining the role of heparanase will require an understanding of when and where XHpaL is active. In agreement with this idea, the overexpression of preproheparanase did not affect early embryonic events in Xenopus and mouse (Bertolesi et al., 2008; Zcharia et al., 2009). Of interest, our Western data argue that processing of heparanase to become enzymatically active is temporally regulated. Likely, tight spatial control is also required.
The unprocessed HpaL and enzymatically inactive HpaS isoforms, however, are also present throughout development. Both are maternally expressed, with a dip in levels around gastrulation before zygotic expression begins at the end of neurulation. The regulation of the expression of these two isoforms begins to differ dramatically in the tadpole. While HpaL is expressed through larval development and into the adult, HpaS appears to be an embryonic variant, with expression restricted to the ovary in the adult. In support, XHpaS expression is not detected in A6 and XTC cell lines, which derive from adult frog kidneys and mature metamorphic tadpoles, respectively. The low level expression detected in adult ovaries may explain the presence of XHpaS in eggs and the early embryo. Heparanase is also expressed in human ovary (Haimov-Kochman et al., 2005), but which variant is unknown.
Our data argue that while heparanase expression in certain tissues is conserved, for instance in both Xenopus and chick heparanase is expressed by migrating epiblast cells and the developing nervous system (Goldshmidt et al., 2001), some degree of species specific regulation is present. For instance, human heparanase is expressed in fetal but not adult liver, whereas Xenopus adult liver continues to express the protein. The Hpa promoter studies also support the idea of species-specific differences in heparanase expression between human and Xenopus. Our analyses of the human and Xenopus Hpa promoter regions suggest that all cis elements responsible for transcription in the human Hpa gene localize 331 bp upstream of the translational start site, while in Xenopus additional elements localize further upstream of this position. In Xenopus, two physically independent promoters are present in the 5′ upstream region tested: P1 (−473 to −331) and P2 (−224 to −117). In contrast, in humans two diffuse domains exhibit promoter activity: A proximal region containing the Sp1-B/A and ERE-B/A sites, and a more distal domain with Sp1-C and ERE-D/C sites. Furthermore, these Sp1 and ERE sites are important in Hpa human expression (Jiang et al., 2002), but are not present in the Xenopus promoter, which instead contain sites for the Octamer binding proteins.
Our results indicate that at least in the early embryo these sites may function to up-regulate heparanase expression. Mutations of the two oct binding sites present in promoter region P1 and P2 (oct sites number 4 and 1 on Fig. 6, respectively) reduce heparanase promoter activity in the gastrulating embryo, but not in the nervous system of tadpoles. Furthermore, overexpression of Oct-25 and Oct-60 stimulates luciferase activity in the in vivo heparanase promoter assay. These two transcription factors are enriched in the animal pole of the developing embryo during gastrulation and early neurulation, and Oct-25 overexpression promotes neural induction while suppressing epidermal differentiation (Cao et al., 2006; Takebayashi-Suzuki et al., 2007). The mechanism by which Oct-25 and Oct-60 induce heparanase promoter activity, and if it is mediated through oct binding sites outside of the P1 and P2 promoter regions will require additional studies. Oct protein regulation of transcription is complex, and the presence of an octamer motif per se, and binding of Oct protein, does not necessarily mean activation and/or repression of gene expression. For instance, the regulation by Oct-25 of Gsc and Mix2 expression depends on its interacting partners (Cao et al., 2008). The requirement for specific partners may explain why Oct-91 failed to alter heparanase promoter activity in blastomeres, but stimulated the response in eye and brain.
Our results indicate differences, in addition to those in transcription factor consensus binding sites, between the heparanase promoter in human and Xenopus. Not surprisingly, given the different transcription sites in the promoters of the two species, the human promoter was not functional in any of the Xenopus systems studied. Yet, certain aspects of the exon–intron distribution appear to be conserved between the two species, including: (i) 14 coding exons; (ii) the presence of a spliced exon 1 constitutive of a 5′untranslated region (UTR); (iii) alternative transcription initiation sites with the start translation codon localized in exon 2 (Hulett et al., 1999; Vlodavsky et al., 1999; Dong et al., 2000; Jiang et al., 2002; Bertolesi et al., 2008); and (iv) the presence of cis-elements in the first intron able to drive heparanase expression in both species, as we showed in this study.
It is likely that each of the Xenopus hpa promoters drives transcription at different initiation sites, generating two splice forms that differ in their 5′ UTRs: The P1 transcript would contain exon 1 as part of the 5′ UTR and the P2 transcript would start with exon 2. Whether the inclusion of exon 1 in the 5′UTR is biologically relevant will need to be determined. In this regard, it is interesting that changes in the 3′UTR of heparanase transcripts in humans regulates heparanase stability and gene expression at the posttranscriptional level (Arvatz et al., 2010). Although we were unable to distinguish two discretely spaced promoters in the human heparanase gene, studies by Dong and coworkers, suggest that two different splicing forms result from differential transcription initiation (Dong et al., 2000).
Our data argue that heparanase is widely expressed during development, but protein localization and levels are finely regulated. Furthermore, the luciferase reporter experiments suggest that transcriptional mechanisms may in part control spatial and temporal aspects of heparanase protein expression, in that the two hpa promoters, P1 and P2, present different activities in a tissue-dependent manner. For instance, P1 is twice as effective at inducing luciferase expression in the brain as in the eye. Moreover, in certain tissues, such as brain but not eye, P2 acts as a repressor of P1, decreasing the activity of the P1 promoter. Finally, oct binding sites within the promoters may contribute to this tissue specific regulation, being important in gastrulating embryos, but not the differentiating nervous system. The biological significance of having two promoters with differential activity, and their roles in regulating heparanase expression, are two main issues to pursue in the future.
Antibodies and Reagents
Antibodies against the peptide 319SVPGKRIWLGETSS332 of Xenopus heparanase (XHpa) were generated in rabbits by GenScript Corporation (Piscataway, NJ). A cysteine residue was added following the final S to enable an efficient coupling of the peptide to Kehole Limpet Hemocyanin (KLH). The KLH-conjugated peptide was injected into rabbits and the specificity of affinity purified antibodies was evaluated by enzyme-linked immunosorbent assay and Western blot analysis.
Embryos, Blastomere Injections and Brain and Eye Electroporation
X. laevis embryos were obtained from chorionic gonadotrophin (Intervet Canada Ltd.) induced egg production and in vitro fertilization according to standard procedures (Gimlich and Gerhart, 1984). Embryos were staged according to Nieuwkoop and Faber (1994). Both blastomeres of two-cell stage embryos were injected by using a glass micropipette with 10 ηl of solution containing different DNA constructs (see below; 40 ηg/μl). For luciferase reporter experiments, embryos were lysed at stage 10–12 for renilla/luciferase measurements. Brain and eyes were electroporated separately as described previously (Atkinson-Leadbeater et al., 2010; Hocking et al., 2010). Briefly, DNA (0.37 μg/μl) was injected into the central brain ventricle, or into the space between the eye primordium and the brain, of stage 27 embryos. Two custom made platinum-wire electrodes, spaced 4 mm apart, were placed on either side of the head of the embryo, and a Grass Technologies S44 stimulator was used to apply 10 square, 50 ms, 50 V pulses, spaced 1 s apart. Procedures involving frogs and embryos were approved by the Animal Care and Use Committee, University of Calgary.
Transient Transfection of Cell Lines
A6 and XTC cell lines were maintained in 55% L15 medium plus 0.02% glutamine and streptomycin antibiotics supplemented with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA). Cells grew at room temperature (22–25°C) in regular atmosphere. COS-7 and PC12 cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) and F12 medium (Sigma, St. Louis, MO), respectively. Media were supplemented with 0.02% glutamine, and 10% FBS without antibiotics. PC12 cultures were also supplemented with 15% horse serum. Both cell lines were grown in a humidified 5% CO2: 95% air atmosphere at 37°C. Transient transfection of the four cell lines was carried out on 70–80% confluent monolayers in 24-well dishes with Lipofectamine 2000 (2 μl/well; Invitrogen). Cells were keep in their optimum growth medium without FBS for 6–8 hr, followed by additional recovery and expression time of 24–28 hr in medium with 10% Fetal Calf Serum.
Protein extracted from cell lines or Xenopus embryos was measured using a BCA protein assay kit (Thermo Scientific, IL). Proteins were separated on 10% polyacrylamide gels, transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad, Hercules, CA), and immunoblotted with anti XHpa (1/1,000 dilution), anti human heparanase (HHpa monoclonal clone HP3/17, 1/500 dilution; Cedarlane Lab, Canada) or goat anti-actin (C-11, 1/1,000 dilution; Santa Cruz Biotech Inc.) antibodies. Specific peroxidase-conjugated secondary antibodies were used to detect protein expression by enhanced chemiluminescence (Perkin-Elmer Corp.; UK).
Section and Whole-Mount Immunohistochemistry
Transverse cryostat sections (12 μm) of 4% paraformaldehyde-fixed Xenopus embryos were immunostained with a polyclonal rabbit anti-XHpa antibody. The whole-mount immunohistochemistry protocol was a modification of one described previously (Davis et al., 1991). Briefly, embryos were fixed in methanol:DMSO (4:1) overnight at 4°C, bleached in fix including 7% H2O2 overnight at room temperature and then stored in methanol at −20°C. The embryos were then hydrated through a methanol series to phosphate buffered saline (PBS), washed three times in PBS containing 0.5% Triton X-100 (PBS-T), and blocked with PBS-T containing 5% goat serum (PBS-GT). The embryos were incubated overnight at 4°C with anti xHpa antibodies diluted 1:50 in PBS-GT, washed with PBS-GT and incubated overnight at 4°C with goat anti-rabbit Alexa red (Invitrogen). Embryos were then extensively washed in PBS-GT and digital images were taken with an AxioCam HRc (Zeiss) on the Stemi SVII stereomicroscope (Zeiss). Images were processed for brightness and contrast with Adobe Photoshop 7.0.
Total RNA was obtained from A6 and XTC cells lines using TRIzol (Invitrogen, CA) according to the manufacturer's protocol. PCR conditions and primer sequences have been published previously (Bertolesi et al., 2008).
Cloning of the 5′ Flanking Region, Exon 1, Exon 2, and Intron 1 of XHpa Gene
DNA isolated from liver of X. tropicalis adult male (a gift from Dr Marko Horb, Institut de Recherches Cliniques de Montreal, Canada) was used as a template for PCR. The primers (Forward: GGAGAGTTGCAAAGGCC TGAT and Reverse: TGCCACGGCA GGAATCTATTA) were designated against the scaffold 214 containing the partial sequence of X. tropicalis DNA. This region contains the 5′ flanking sequence, exon1, intron 1, and a fragment of exon 2 of the X. tropicalis hpa gene. PCR reactions were performed in a total volume of 25 μl and contained: 200 ng of DNA template, 0.3 mM dNTP, 1 mM MgSO4, 0.3 μM of each forward and reverse primer, 1 unit of Pfx DNA polymerase in a 1× enhancer solution and 1× buffer of the platinum Pfx amplification set (Invitrogen). The cycling conditions consisted of denaturation at 94°C for 30 sec, annealing at 55° for 30 sec and extension at 68°C for 4 min for 35 cycles. The amplification product was gel purified by using the QIAquick Gel Extraction kit (Quiagen, Mississauga, ON, Canada), and cloned into a pJET2.1 vector (Fermentas; pJET-Prom XHpa −2119: −55). Here and elsewhere sequence integrity was verified by automated sequencing at the DNA Services Facility, University of Calgary.
Luciferase Reporter Assays
A construct containing the Xenopus Hpa 5′ flanking sequence, between nucleotides −2119 to −55 from the beginning of XHpa translation, was subcloned from the pJET2.1 vector into pGL3 basic by using StuI/SmaI and BglII restriction enzymes. Several additional Xenopus Hpa promoter fragments were generated. NheI at the 5′ end and Hind III at the 3′ end, in combination with an internal enzyme (for, e.g., MscI; PstI; NdeI), were used to delete portions of the promoter. For those constructs that lacked a specific enzyme recognition site at the required location, point mutations were introduced to generate the recognition site with the help of Quik-Change Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). PCR amplification with specific primers and subcloning into pGL3 plasmid was the main strategy used for generating several constructs. The same strategy was followed to generate different fragments from the human Hpa promoter (a gift from Dr. Xiulong Xu, St Luke's Medical Center, Chicago, IL).
The OctBP-4 and OctBP-1 site mutants of Hpa promoter −2119: −55 were made by using the Quik- Change Site-Directed Mutagenesis Kit (Stratagene). PCR used the appropriate plasmid as template and the primers OctBP-4 mut (5′-ATTTATACTTATGA GTACTGCAGCACAACCACACCCC-3′) or OctBP-1 mut (5′-CTATTTTGACAT CAGCCTGTACATATGTGCAACTCTT GCA-3′). The underlined nucleotides of the primers indicate the mutation of OctBP-4 (AGTAATGCAT to AGTA CTGCAG) and OctBP-1 (ATGTAAT to CTGTACA) sites. pCS2 expression vectors with the full length sequence for Oct-25, Oct-60, and Oct-91 were a gift from Dr Walter Knöchel lab (Cao et al., 2006, 2008). A6, XTC, PC12 and COS-7 cells were seeded in 24-well plates. After 24 to 48 hr, cells were transfected using Lipofectamine 2000 (Invitrogen) with the various luciferase vectors (0.4 μg), and the pRL-TK renilla vector (0.2 μg). Approximately 24 hr after transfection, cells were washed in PBS and then harvested in a Dual-Glo Luciferase Reagent. Luciferase activities were measured from triplicate wells by using a TD 20/20 luminometer (Turner Designs) with the help of a commercial Dual-Glo luciferase assay system (Promega). Data were normalized to levels of renilla activity. For the electroporation studies, a plasmid expressing GFP (pCS2-GFP) was co-transfected along with the luciferase and renilla constructs, and a minimum of 5 embryos showing high GFP expression in the desired target area (either brain or eye) were analyzed individually. All promoter data analysis are presented as mean ± SEM unless otherwise stated. Data are expressed as times over control, which refers to the pGL3-basic/pTK-renilla group normalized to a value of one. The statistical methods used were repeated measures analysis of variance, and when appropriate two-tailed Student's t-test for unpaired data. P < 0.01 was considered statistically significant.
The authors thank Dr. Vanina Zaremberg for helpful comments on the manuscript, and Dr. Xiulong Xu, Dr. Marko Horb, and Dr. Walter Knöchel for their kind gifts of the human Hpa promoter construct, the X. tropicalis DNA and the Oct protein expression constructs, respectively. This work was supported by an operating grant from the Natural Sciences and Engineering Research Council of Canada to S.M. B.S. and G.M. were supported by summer studentships from the Alberta Heritage Foundation for Medical Research (AHFMR), S.D. by a summer studentship from the Canadian Institutes of Health Research Training Grant in Genetics, Child Health and Development, and S.M. is a Tier II Canada Research Chair in Developmental Neurobiology and an AHFMR Scientist.