|Home | About | Journals | Submit | Contact Us | Français|
Specific interactions of growth factors with heparan sulfate are thought to regulate stages of branching morphogenesis in developing mammalian organs, but the evidence derives mostly from studies of explanted tissues or cell culture. We recently provided in vivo evidence that inactivation of Ndst1, the predominant N-deacetylase/N-sulfotransferase gene essential for the formation of mature heparan sulfate, results in a highly specific defect in murine lobuloalveolar development. Here, we demonstrate a highly penetrant dramatic defect in primary branching by mammary epithelial-specific inactivation of Ext1, a subunit of the copolymerase complex that catalyzes the formation of the heparan sulfate chain. In contrast to Ext1 deletion, inactivation of Hs2st (which encodes an enzyme required for 2-O-sulfation of uronic acids in heparan sulfate) did not inhibit ductal formation but displayed markedly decreased secondary and ductal side-branches as well as fewer bifurcated terminal end buds. Targeted conditional deletion of c-Met, the receptor for HGF, in mammary epithelial cells showed similar defects in secondary and ductal side-branching, but did not result in any apparent defect in bifurcation of terminal end buds. Although there is published evidence indicating a role for 2-O sulfation in HGF binding, primary epithelial cells isolated from Hs2st conditional deletions were able to activate Erk in the presence of HGF and there appeared to be only a slight reduction in HGF-mediated c-Met phosphorylation in these cells compared to control. Thus, both c-Met and Hs2st play important, but potentially independent, roles in secondary and ductal side-branching. When considered together with previous studies of Ndst1-deficient glands, the data presented here raise the possibility of partially-independent regulation by heparan sulfate-dependent pathways of primary ductal branching, terminal end bud bifurcation, secondary branching, ductal side-branching and lobuloalveolar formation.
Epithelial branching morphogenesis is a key feature of a number of organs, including salivary glands, kidneys, lungs and mammary glands. Unlike other branched organs, much of branching morphogenesis of the mammary gland occurs after birth. At birth, the mammary gland consists of a rudimentary epithelial ductal system contained within the stroma of the fat pad, which will undergo slow growth. At puberty, estrogen and progesterone initiate a cycle of reciprocal inductive signaling between the ductal epithelia and cells of the surrounding stroma, resulting in iterations of bifurcation and elongation of the terminal end buds of the ductal epithelium. This process of branching morphogenesis leads to the formation of an extensive network of branched epithelial ducts that fills the fat pad. In addition, ductal side-branching also occurs in which new ducts form by budding from the existing ducts. During pregnancy, secretory alveolar structures develop and differentiate to produce milk during lactation. Following weaning, involution takes place, which returns the mammary gland to a virgin-like state [reviewed in (Sternlicht, 2006; Sternlicht et al., 2006)]. Thus, mammary gland development can be viewed as occurring in distinct stages that are characterized by sexual development and reproduction (Imagawa et al., 1990; Sunil et al., 2000). These stages of branching morphogenesis have proven more difficult to separate in other organs, such as the kidney, where the role of the mesenchymal component is more complicated (Shah et al., 2010; Shah et al., 2009).
In addition to hormonal requirements (i.e., estrogen, progesterone, growth hormone, and prolactin), mammary gland development has also been shown to depend on the action of numerous local growth factors (Hovey et al., 2002), including fibroblast growth factors (FGFs) (Mailleux et al., 2002), Wnt proteins (Brisken et al., 2000; Humphreys et al., 1997), parathyroid hormone-related protein (PTHrP) (Dunbar et al., 2001), Hedgehog proteins (Hh and Ihh) (Lewis et al., 2001; Lewis et al., 1999), transforming growth factor beta (TGFβ) (Forrester et al., 2005; Joseph et al., 1999) and inhibin-βb (Robinson and Hennighausen, 1997), insulin-like growth factors (IGF) 1 and 2 (Brisken et al., 2002; Ruan and Kleinberg, 1999) and IGF-binding protein-5 (IGFBP-5) (Chapman et al., 1999; Grimm et al., 2002), hepatocyte growth factor (HGF) (Yang et al., 1995), amphiregulin (Luetteke et al., 1999), EGF (Luetteke et al., 1999; Wiesen et al., 1999), heregulin (HRG) (Li et al., 2002), as well as ligands of EGF receptors (Xie et al., 1997) and ERBB4 receptors (Long et al., 2003; Tidcombe et al., 2003). Many of these growth factors bind to heparan sulfate proteoglycans (HSPGs) which are expressed in mammary epithelial cells (Delehedde et al., 2001) and whose expression has been demonstrated to increase in mammary gland during pregnancy and lactation (Warburton et al., 1982). Interaction with HSPGs is hypothesized to: 1) protect growth factors against proteolytic degradation; 2) create a storage depot for later release; 3) facilitate assembly of signaling complexes (co-receptor activity); 4) enable clearance by endocytosis; and 5) regulate their diffusion through the tissue (Bernfield et al., 1999; Delehedde et al., 2001; Lander et al., 2002).
Evidence supporting a role for HSPGs in mammary development derives from studies of genetically altered mouse strains. Mice deficient in syndecan-1, a plasma membrane HSPG, show a mild reduction in secondary and tertiary branching (Alexander et al., 2000; Liu et al., 2003), whereas mice lacking the membrane proteoglycan CD44v3 display defective lobuloalveolar expansion (Yu et al., 2002). In contrast, transgenic mice expressing human heparanase, a degradative enzyme secreted by cells that cleaves the heparan sulfate chains, induces hyper-branching of the mammary epithelia (Zcharia et al., 2004), suggesting that the defects in the syndecan-1 and CD44v3 mutants may reflect activities associated with the core proteins of the proteoglycans rather than the sulfated glycosaminoglycan chains. Interestingly, transgenic mice lacking heparanase showed abundant side branches and alveolar structures compared to wild type animals (Zcharia et al., 2009).
The glycosaminoglycan chains of the HSPGs assemble by the alternating addition of N-acetylglucosamine and glucuronic acid while attached to core proteins of proteoglycans by a copolymerase complex composed of the proteins, Ext1 and Ext2. The disaccharides undergo further processing by numerous enzymes, including members of the N-deacetylase-N-sulfotransferase (Ndst) family, the uronyl C5-epimerase (Hsglce) and 2-O-sulfotransferase (Hs2st), the glucosaminyl 6-O-sulfotransferases (Hs6st) and 3-O-sulfotransferases (Hs3st) (reviewed in (Esko and Lindahl, 2001)). Clustered subsets of residues undergo these processing reactions, resulting in modified domains of variable length separated by unmodified or sparsely modified domains. The modifications catalyzed by these enzymes generally occur sequentially (i.e., Ndst→ HsGlce →Hs2st→Hs6st→Hs3st), but some reactions do not go to completion, creating unique patterns of modified residues in sections of the chains. Mutants systemically lacking many of these enzymes succumb embryonically, obviating further studies of the mammary gland, or have no reported defects in mammary gland development or function (Bullock et al., 1998; Grobe et al., 2005; Habuchi et al., 2007; HajMohammadi et al., 2003; Hasegawa and Wang, 2008; Izvolsky et al., 2008; Li et al., 2003; Ringvall et al., 2000).
Recently we showed that inactivation of the heparan sulfate biosynthetic enzyme, Ndst1, in mammary epithelial cells results in a highly specific defect in lobuloalveolar development in mice (Crawford et al., 2010). However, primary and secondary branching morphogenesis appeared normal, suggesting the possibility that early events in ductal extension and branching might not depend on heparan sulfate. In this study, we have investigated this possibility and show that ductal branching in the mammary gland does depend upon heparan sulfate based on conditional deletion of Ext1. Moreover, we show that secondary branching of the ductal epithelium and bifurcation of terminal end buds depends on sulfation of heparan sulfate catalyzed by Hs2st. The branching defect of Hs2st-deficient mice is similar to that observed in mice lacking c-Met, the signal transduction receptor for HGF. However, these pathways appear to be largely independent based on activiation of c-Met signaling in Hs2st-deficient epithelia.
All animals were handled in accordance with protocols for the humane treatment of animals approved by the Institutional Animal Care and Use Committee and Animal Subjects Committee at the University of California San Diego. Mice bearing a loxP-flanked allele of Ext1 (Ext1f/f) were described previously (Inatani et al., 2003). Mice bearing a loxP-flanked allele of c-Met (c-Metf/f) were obtained from Dr. S.S. Thorgeirsson (NIH, Maryland) (Huh et al., 2004). Mice bearing a loxP-flanked allele of 2-O-sulfotransferase (Hs2stf/f) were described previously (Stanford et al., 2010). The MMTV Cre line “A” mice in the 129 background was obtained from Dr. T. Wynshaw-Boris (University of California, San Diego) (Wagner et al., 2001; Wagner et al., 1997). Cross breeding between the genotypes was initiated to obtain flox/floxMMTV cre- and flox/floxMMTV cre+ littermates. Only male mice carrying the MMTVCre allele were used for breeding to avoid deletion of the conditional allele by Cre expression in oocytes. All the experiments were done with mice on a mixed background with littermate controls. Quantitative and qualitative aspects of phenotypes did not change with further backcrossing.
Primary mammary epithelia were isolated and cultured following an established protocol (Pullan et al., 1996). Number 3 and 4 glands were excised and chopped with a razor blade, and digested with 0.2% trypsin and 0.2% collagenase A (Roche, Nutley, NJ). Cells were enriched by differential centrifugation. Tissue culture plates were precoated with 100ml/cm2 of Ham's F12 (Invitrogen, Carlsbad, CA) medium containing 20% heat-inactivated fetal bovine serum (FBS) (Atlanta Biologicals, Lawrenceville, GA) and 1 mg/ml fetuin (Sigma, St. Louis, Mo). Cells were cultured in Ham's F12 medium containing 10% heat-inactivated FBS, 5mg/ml insulin, 1mg/ml hydrocortisone, 5 ng/ml epidermal growth factor, 50 μg/ml gentamycin, 100 U/ml penicillin, and 100 μg/ml streptomycin (all from Sigma). The medium was changed after the second day of culture and subsequently on every other day; cells were cultured for a total of 5-7 days.
The binding of fibroblast growth factor 2 (FGF2) was determined using flow cytometry. Isolated mammary epithelial cells were incubated with biotinylated FGF2 in Hams F12 medium (Invitrogen, Carlsbad, CA) with 0.5 % BSA (Sigma, St. Louis, MO) for 1 hour with shaking at 4°C. Cells were washed twice in PBS and incubated in PBS containing streptavidin-APC for 20 minutes with shaking at 4°C. Samples where FGF2 was omitted were included as negative controls. To show that FGF2 binding was related to heparan sulfate expression, controls were also performed after pre-treatment of mammary epithelial cells with 5 μU/ml of a mixture of heparin lyases I, II, and III (Ibex, Montreal, Quebec) for 4 hours at 37°C. An adenovirus containing Cre recombinase (AdCre) was also used to inactivate heparan sulfate biosynthetic enzyme genes in vitro as described (Li et al., 2000). AdCre and adenovirus containing green fluorescent protein (AdGFP) were obtained from the Vector Core Development Lab at the University of California, San Diego. Cells were treated for 90 min twice over four days with 108 pfu/ml, washed with PBS and cultured in normal growth medium. Flow cytometry using biotinylated FGF and strepavidin phycoerythrin-Cy5 showed >10-fold decrease in fluorescence of ~99% of the cells (Bai and Esko, 1996).
Histological analyses were performed by the Cancer Center Histology Core at the University of California, San Diego. Whole mounts were stained with hematoxylin (Ip and Asch, 2000). Hematoxylin and eosin staining of sections was performed by standard procedures.
Isolated primary mammary epithelial cells [Hs2stf/fMMTVCre- (control), c-Metf/fMMTVCre- (control), Hs2stf/fMMTVCre+ (mutant) or c-Metf/fMMTVcre+ (mutant)] were grown in 6-well tissue culture dishes as described above for 5-7 days. Cells were then subjected to 4-5 hrs of serum deprivation followed by the addition of various concentrations of HGF for 20 minutes. Cells were lysed in 1%NP-40 buffer and protein content was determined with the Bradford assay (BioRads, Hercules, CA) using BSA as a standard. Ten micrograms of protein from cultured cells was electrophoresed on BioRad precast Ready gels and transferred to nitrocellulose with a semidry blotting apparatus. The following antibodies were used: phospho-Erk1/2 (p44/42, Thr202/Tyr204), Erk, Met and phospho-Met (Tyr1234/1235) (Cell Signaling Technology, Beverly, MA); phosphoFAK (pY397), FAK (Invitrogen, Carlsbad, CA); α-Actin (Santa Cruz Biotechnology, Santa Cruz, CA). HRP-conjugated anti-rabbit antisera was obtained from Biorad (Hercules, CA). HRP-conjugated anti-mouse IgG were obtained from Amersham Biosciences (Piscataway, NJ). HRP was detected using the SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL).
We recently analyzed mice bearing a mammary gland-specific inactivation of Ndst1 (Crawford et al., 2010), a member of a family of enzymes that catalyzes the initial sulfation of heparan sulfate chains (Esko and Selleck, 2002). Although the animals displayed no readily detectable defects in branching morphogenesis of the ductal epithelium, reduced sulfation of heparan sulfate resulting from the deletion of Ndst1 inhibited lobuloalveolar development but not lactational differentiation of the mammary epithelial cells (Crawford et al., 2010). Since the inactivation of Ndst1 only partially reduced sulfation of the heparan sulfate chains, we tested if a more severe reduction in heparan sulfate might affect branching by inactivation of Ext1. Deletion of Ext1 was selectively done in mammary epithelial cells by cross breeding mice with a conditional allele of Ext1 (Ext1f/f) to MMTVCre mice (Inatani et al., 2003; Wagner et al., 1997), which gave rise to littermate pairs of Ext1f/fMMTVCre+ (mutant) mice and Ext1f/fMMTVCre- (wildtype) mice. Mice of both genotypes were seen at the expected Mendelian frequency. Dissection did not demonstrate any overt change in the size of the mammary fat pads (data not shown).
Initial examination of the effect of the mutation on ductal branching was performed at 10 to 12 weeks postpartum, a point at which ductal branching morphogenesis should be complete. Whole-mounts of the fourth inguinal mammary gland revealed perturbed growth of the ductal network in the mutant. A scoring system was devised to determine the extent of inhibition of growth (Fig. 1A). Glands that exhibited absence of ductal growth received a score of 1. Ductal trees that grew from the nipple but stopped prior to the lymph node were given a score of 2. A tree that progressed past the lymph node but halted before the end of the fat pad distal to the nipple received a score of 3. Growth that completely filled the fat pad received a score of 4. Using this system, all wildtype (Ext1f/fMMTVCre-, n = 10) mice and heterozygotes (Ext1f/wtMMTVCre+, n = 36) had a score of 4 (i.e. the fat pad was completely filled with ductal epithelia; Fig. 1B and 1I). In contrast, 31% of mutant mice (Ext1f/fMMTVCre+) were devoid of any ductal growth (n = 54; Fig. 1C and 1I). About 2-4% of mutant mice had scores of 2 or 3, but 63% of the glands had scores of 4 (Fig 1D). A comparison of tissue sections prepared from a wildtype mouse with a score of 4 and a mutant with a score of 1 confirmed that the mutant lacked ductal structures (Fig 1E and 1F). We also examined mutant versus wildtype glands at 5 weeks postpartum, a period of rapid ductal expansion. Ext1f/fMMTVCre- glands all exhibited a wildtype phenotype at 5 weeks (Fig. 1G), while Ext1f/fMMTVCre+ glands showed a lack of ductal structures at a frequency similar to the 10-12 week analysis (Fig. 1H).
This data suggested that Ext1 is required for mammary ductal branching, but the variation in the phenotype was such that only about 37% of animals displayed a growth phenotype. To determine if incomplete penetrance of the Ext1 deletion might reflect inefficient Cre recombination, mammary epithelial cells were isolated from glands without prior knowledge of their score. The presence or absence of heparan sulfate on the cells was then determined by incubation of the cells with biotinylated FGF2/strepavidin-phycoerythrin-Cy5, which binds to cells in a heparan sulfate-dependent manner (Bai and Esko, 1996; Crawford et al., 2010). As expected, Ext1f/fMMTVCre- epithelia (Ext1 is still expressed) bound FGF2 (Fig. 2A-C grey line). Treatment of isolated Ext1f/fMMTVCre- mammary epithelial cells with heparin lyase (also known as heparinase), an enzyme that cleaves heparan sulfate chains, showed a two-log reduction in FGF2 binding (Fig. 2A, black line), while infection of Ext1f/fMMTVCre- primary epithelial cells with exogenous Adenovirus-Cre (which has previously been shown to achieve 95% inactivation of floxed loci (Rohlmann et al., 1996)) resulted in a comparable reduction in FGF2 binding (Fig 2B, black line). These results clearly indicate the requirement for heparan sulfate in the binding of FGF2 to the isolated mammary epithelial cells. To address the question of the efficiency of Cre recombination and the penetrance of Ext1 deletion, FGF2 binding to Ext1f/fMMTVCre+ primary mammary epithelial cells isolated from glands (without knowledge of the extent of ductal growth or branching within the glands) revealed a FGF2 binding profile remarkably similar to wildtype (Fig. 2C). Since 31% of the glands from the Ext1f/fMMTVCre+ animals were completely lacking in ductal growth, any isolated primary mammary epithelial cells from these glands would have to have come from those in which some ductal growth occurred. Moreover, since Ext1 is critical for heparan sulfate chain biosynthesis, its deletion should completely eliminate heparan sulfate expression on mammary epithelial cells. Therefore the observed FGF2 binding profile of the primary mammary epithelial cells isolated from the Ext1f/fMMTVCre+ glands indicates that these cultured cells were wildtype arising from ductal cells that had not undergone Cre-mediated recombination.
The data presented above indicates that Ext1 is critical for ductal growth and branching morphogenesis, implying a requirement for heparan sulfate. Previously we showed that inactivation of Ndst1 in mammary epithelia, which reduces the overall level of sulfation of the chains, resulted in a selective effect on lobuloalveolar development (a late stage of mammary gland development) (Crawford et al., 2010). To examine whether the pattern of sulfation might affect different stages of mammary gland development in vivo, we targeted the uronyl 2-O-sulfotransferase gene (Hs2st).1 Hs2st transfers sulfate to iduronic acid residues, and more rarely to glucuronic acids, located to the reducing side of N-sulfated glucosamine residues generated by Ndst1. This modification prevents the back conversion of iduronic acid to glucuronic acid and positions the 2-O-sulfate groups in an orientation important for binding of many growth factors involved in mammary gland development. Hs2st was selectively inactivated in mammary epithelial cells by cross breeding Hs2stf/f mice to MMTVCre mice (Stanford et al., 2010), which resulted in littermate pairs of Hs2stf/fMMTVCre+ (mutant) and Hs2stf/fMMTVCre- (wildtype) mice at the expected Mendelian frequency. Hs2stf/fMMTVCre- glands at 5 and 12 weeks appeared to have a level of branching comparable to wildtype (Fig. 3A and 3C). Hs2stf/fMMTVCre+ glands, on the other hand, showed a reduction in overall branching (Fig. 3B and 3D), with long stretches of epithelial tube without any apparent branch points (Fig. 3B and 3D). The total number of branch points per high-powered field was counted and compared between the wildtype and mutant glands (Fig. 3E and 3F), revealing a two-fold reduction in branching in the mutant (Fig. 3G). Ductal branching is divided into two different mechanisms: primary branching (facilitated by bifurcation and proliferation of the terminal end buds (TEBs)) and secondary branching from existing ductal structures. At 5 weeks of age (during ductal branching) the number of TEBs was reduced in the mutant and a complete lack of secondary ductal branching was observed compared to the wildtype (Fig. 3A and 3B).
To determine the extent of alteration in cell surface heparan sulfate in Hs2stf/fMMTVCre+ glands, we isolated mammary epithelia and again measured the binding of FGF2, which depends on 2-O-sulfation of heparan sulfate for binding (Bai and Esko, 1996). Primary mammary epithelia isolated from wildtype glands bound FGF2 (Fig. 3H, grey line), but cells isolated from Hs2stf/fMMTVCre+ glands exhibited greatly reduced FGF2 binding (Fig. 3H, black line). These results suggest that the deficiency in branching seen in Hs2st-deficient glands resulted from the change in the cell surface heparan sulfate of the epithelial cells. Furthermore, mammary epithelial cell proliferation apparently does not depend the loss of 2-O-sulfation, whereas complete loss of heparan sulfate caused by Ext1 deletion is incompatible with proliferation.
Among the various growth factors that require interaction with heparan sulfate, hepatocyte growth factor (HGF) has been suggested to play a key role in mammary gland development. For example, expression of HGF and its receptor, c-Met, have been found to correlate with stages of ductal branching morphogenesis (Rosario and Birchmeier, 2003; Yang et al., 1995), and the genes are expressed during virgin branching morphogenesis (6 weeks) and during the process of ductal hyperbranching in the early stages of pregnancy (Niranjan et al., 1995). Perturbation of HGF signaling in whole mammary gland cultures inhibits branching morphogenesis (Yang et al., 1995), and high levels of HGF in mammary glands may induce branching (Yant et al., 1998).
Thus, we investigated whether deletion of HGF signaling, might affect mammary gland branching by conditionally deleting the c-Met gene, which encodes the HGF tyrosine kinase receptor, c-Met. Whole-mount glands stained with hematoxylin from 10 week old c-Metf/fMMTVCre- (control) (Fig. 4A and 4C) and c-Metf/fMMTVcre+ (mutant) animals (Fig. 4B and 4D) were compared and analyzed. Initial examination revealed that at 10 weeks all fat pads were full of branched epithelial ductal structures (Fig. 4A and 4B), suggesting that primary branching morphogenesis of the ductal epithelium was unaffected by deletion of c-Met. In support of this notion, wildtype (Fig. 4E and 4G) and mutant glands (Fig. 4F and 4H) were found to exhibit comparable growth and branching of ductal structures at 6 weeks, a period of rapid primary branching of the ductal epithelium. Higher magnification examination of these glands revealed comparable number of TEBs and primary branches (Fig. 4G vs. 4H). Nevertheless, there appeared to be a clear reduction in the extent of secondary and ductal side-branching in the c-Metf/fMMTVCre+ mammary gland at 10 weeks (compare Fig. 4A and 4C with Fig. 4B and 4D). This was supported by quantitation of branching in the 10 week mammary glands, which revealed ~35% reduction in overall branching in c-Metf/fMMTVCre+ compared to the c-Metf/fMMTVcre- mice (Fig. 4I).
To ensure the c-Metf/fMMTVcre+ glands were deficient in c-Met, mammary epithelia from mutant and wildtype mice were isolated and the expression of c-Met protein was analyzed (Fig. 5A). c-Metf/fMMTVcre+ glands showed a reduction in Met protein as compared to the wildtype epithelia (Fig. 5A). Previous work found that breast carcinoma cells phosphorylated focal adhesion kinase (FAK) in response to HGF treatment (Beviglia and Kramer, 1999). When the resting state of phosphorylated FAK (pFAK) was analyzed, a reduction in pFAK was observed, with no decrease in total FAK levels (Fig. 5A). MAP kinase, a known downstream Met effector, was activated as evidenced by the accumulation of phorphorylated Erk (pErk) after stimulation with HGF (Fig. 5B). Comparison of pErk levels in HGF-treated primary mammary epithelial cells isolated from wildtype and mutant demonstrated the loss of HGF-mediated signaling in the absence of c-Met (Fig. 5B). Interestingly, treatment of wildtype primary mammary epithelial cells with HGF in the presence of heparan lyase, which cleaves the heparan sulfate, revealed a similar reduction in HGF-mediated activation of pErk (Fig. 5C). Taken together, these findings suggest that the disruption in branching seen in the conditional deletions of heparan sulfate biosynthetic enzymes might be due to attenuation of HGF signaling.
In an attempt to address this question, primary mammary epithelial cells from Hs2stf/fMMTVCre- and Hs2stf/fMMTVCre+ mice were isolated and the levels of pErk were determined in the presence and absence of HGF (Fig. 6). Western blots revealed a slightly elevated basal level of pErk in the untreated mammary epithelial cells isolated from the Hs2stf/fMMTVCre+ mice (Fig. 6), as observed previously in embryonic cells from animals bearing a systemic deletions in Hs2st (Deakin et al., 2009; Merry et al., 2001). In contrast to the inhibition of signaling observed after heparinase treatment, deletion of Hs2st in the cells did not attenuate HGF signaling (Fig. 6). The addition of FGF2 to the Hs2st-deficient cells also resulted in Erk activation (Fig. 6). Phosphorylation of c-Met following treatment with HGF was also examined (Fig. 6). While there appears to be a slight decrease in the level of c-Met phosphorylation in the Hs2stf/fMMTVCre+ cells compared to control, HGF is still able to bind to c-Met in the absence of 2-O sulfated heparan sulfate and trigger receptor phosphorylation (Fig. 6). Since this phosphorylation of c-Met is blocked following treatment with heparan lyase (Fig. 6), it is possible that heparan sulfate sulfated at other positions could be involved in HGF binding.
To examine if heparan sulfate present in growing mammary ducts is necessary for pubertal branching, we targeted mammary epithelial heparan sulfate production by inactivating the biosynthetic enzymes responsible for heparan sulfate polymerization (Ext1) and modification (Hs2st) using Cre recombinase under the control of the MMTV promoter. We also examined if HGF signaling was required by the conditional deletion of its receptor, c-Met. We found that altering heparan sulfate production had a striking effect on branching morphogenesis and that the targeting of 2-O-sulfaton of heparan sulfate caused a selective effect on side branching and terminal end bud bifurcation. Similarly, c-Met deletion led to a similar side branching phenotype, with a lesser effect on terminal end buds. Together with the previously data published on the role of Ndst1 in lobuloalveolar development (Crawford et al., 2010), the new findings support the idea that heparan sulfate and specific heparan sulfate modifications regulate primary branching, terminal end bud bifurcation, secondary branching, ductal side-branching and lobuloalveolar development. Remarkably, these effects are partly separable in vivo.
Mammary epithelia lacking Ext1 do not polymerize heparan sulfate and fail to undergo primary ductal branching morphogenesis after birth and in response to pubertal hormones released at the onset of the estrous cycle. This phenotype is dramatic and highly penetrant and has not been observed in proteoglycan core protein gene knockouts (Alexander et al., 2000; Liu et al., 2003; Yu et al., 2002). Some variability in the penetrance was observed, but this appears to be due to inefficient Cre-recombinase activity (Fig. 2) and selective growth of cells that retain the capacity to synthesize heparan sulfate (Turlo et al., 2010). The failure of cells completely deficient in heparan sulfate to divide probably reflects the interaction of heparan sulfate with multiple heparin-binding growth factors, as well as ECM formation and tissue organization. A major effort needs to be devoted to defining the specific growth factors and matrix molecules affected in this system. Nevertheless, this is the first clear demonstration of a role for heparan sulfate in ductal branching morphogenesis.
Heparan sulfate does not occur as free chains in tissues, but rather as covalent complexes with proteoglycan core proteins. Over 17 different heparan sulfate proteoglycans are known (Bishop et al., 2007) and although mutants have not been made in all of these genes, inactivation of syndecans-1, -3, -4, glypican-3, and collagen 18 have not been reported to exhibit major defects in mammary gland branching morphogenesis (Alexander et al., 2000; Cano-Gauci et al., 1999; Echtermeyer et al., 2001; Ishiguro et al., 2001; Liu et al., 2003; Zcharia et al., 2004; Zhou et al., 2004). Many growth factors relevant to mammary gland development are known to bind to heparan sulfate, including FGF family members (Makarenkova et al., 2009; Rapraeger et al., 1991), EGF family members (Aviezer and Yayon, 1994) and HGF (Zioncheck et al., 1995). Single growth factor knockouts in the mammary gland rarely abrogate ductal growth, suggesting that compensation by different sets of growth factors could account for the mild phenotypes. Because many of these growth factors bind to heparan sulfate, altering the formation of heparan sulfate is equivalent to a combined mutation of multiple growth factors, which may explain the more penetrant phenotypes observed when heparan sulfate is genetically modified.
Based on these data it was surprising that inactivation of HGF signaling following the deletion of its receptor, c-Met, resulted in such a clear branching phenotype. HGF has long been implicated in in vitro branching of epithelial cells (Cantley et al., 1994; Montesano et al., 1991; Santos et al., 1994; Santos et al., 1993) and organ culture (Santos et al., 1993), but to our knowledge, there is no knockout of c-Met that exhibits a branching phenotype, although a combined EGFR and c-Met knockout exhibits decreased kidney branching (Ishibe et al., 2009), as predicted by in vitro data (Barros et al., 1995; Sakurai et al., 1997). The loss of secondary branches and ductal side-branches in c-Metf/fMMTVcre+ mice demonstrates a key role for HGF in these processes. Since growth and branching from the terminal end buds appeared to be largely unaffected in the c-Metf/fMMTVcre+ mutant glands, the data suggest the mechanism(s) of secondary branch outgrowth from the trailing primary ducts is quite different from that involved in primary ductal branching (Sternlicht, 2006; Sternlicht et al., 2006).
Our findings also suggest that HGF-mediated secondary branching is heparan sulfate dependent, as treatment of primary ductal epithelial cells with heparin lyases abrogated HGF-mediated activation of MAP kinase as evidenced by a loss of phospho-Erk and c-Met phosphorylation (Fig. 6). Interestingly, the phenotype is similar to that seen in the Hs2st-conditional knockout, which not only had a reduced number of terminal end buds (and thus displayed reduced primary branching) but also decreased secondary and ductal side-branching. Because Ndst1-deficient glands exhibit normal secondary and side-ductal branching (Crawford et al., 2010), the data here raised the possibility that HGF-c-Met signaling might be partly dependent upon 2-O-sulfated heparan sulfate for modulation of its action. Although in vitro binding data supports the idea that HGF binding and/or signaling depends on 2-O-sulfation (Ashikari-Hada et al., 2004), signaling assays performed here using primary mammary epithelial cells indicate that c-Met phosphorylation and activation of Erk is not dependent in vitro upon 2-O-sulfated heparan sulfate (Fig. 6). As stated above, this is consistent with other in vitro data which indicates that the systemic deletion of Hs2st does not interfere with HGF-mediated activation of Erk in isolated primary cells (Deakin et al., 2009; Merry et al., 2001). In addition, an apparently normal signaling response is also seen in these mutant cells in the presence of FGF2, which is also consistent with a recent showing that lacrimal gland branching is more sensitive to 6-O-sulfation than 2-O-sulfation (Merry et al., 2001; Qu et al., 2011). The reasons for this remain unclear, although the increased basal activity of Erk seen in our assays (Fig. 5D) raises the possibility that, at least in these in vitro assays, this signal transduction pathway is overly sensitive or hyperactivated. While this may indicate a role for other growth factors in branching of the mammary ductal epithelium, it has proven difficult to link any growth factors and their receptors to Hs2st (Merry and Wilson, 2002).
Heparan sulfate could also play a role in other stages of mammary gland development, such as embryonic stages of differentiation or formation of the initial placode. Additional studies will be needed using mice that express the Cre recombinase at earlier stages of development to address this question, since the Cre mice employed in this study requires hormonal activation of the MMTV promoter during the estrous cycle (Wagner et al., 2001; Wagner et al., 1997). Future experiments should focus on the role of other heparan sulfate modifying enzymes, including the Hs6sts, Hs3sts, and HsGlce (also known as C5 epimerase). Systemically, mutations in Hs6st and Hsglce show perinatal embryonic lethality (Habuchi et al., 2007; Izvolsky et al., 2008; Li et al., 2003), but whether mammary buds exist in these animals is not known.
We emphasize here that it is not obvious that deletion of enzymes involved in heparan sulfate polymerization, or those involved in heparan sulfate modification, should affect separate stages of mammary gland branching morphogenesis. Systemic deletion of C5 epimerase (Hsglce), which is responsible for conversion of glucuronic acid to iduronic acid, results in kidney agenesis and lung defects, but other organs are reported to develop normally (Li et al., 2003). However, since the deletion of Hsglce is neonatally lethal (with null animals dying immediately after birth), the role of epimerization in mammary gland development was not examined (Li et al., 2003). Hs2st systemic deletion gives a somewhat similar kidney phenotype, apparently sparing other organs (Bullock et al., 1998; Stanford et al., 2010). Moreover, it has recently been shown that this is not due to a primary defect in epithelial branching morphogenesis (Shah et al., 2010). Furthermore, to our knowledge, no isolated deletion of c-Met in another organ has been shown to result in a branching defect. Although it is clear that there is some conservation of branching programs, our results point to the distinct nature of morphogenetic pathways in different branching organs. This could be due to organ-specific branching mechanisms, organ-specific morphogenetic molecules (e.g., growth factors, matrix molecules, integrin receptors, downstream signaling effectors and transcription factors) and/or different levels of “redundancy” of such molecules, as well as differences in the expression and composition of heparan sulfate. For example, recent studies of the lacrimal gland branching indicates a requirement for specific modification of heparan sulfates at the tip of the lacrimal gland bud (Pan et al., 2008; Qu et al., 2011), which is in contrast to the data for mammary gland development presented in an earlier study, where the terminal end bud appears to be insensitive to Ndst1 deletion, at least during early branching events (Crawford et al., 2010). Finally, differences in growth factor requirements in postnatal and hormonally dependent branching organs (i.e., breast) compared with those in which branching occurs largely or entirely prenatally (e.g., kidney, pancreas, lung, salivary gland, and lacrimal gland) could also play a role. Taken together, the data suggests that mammary branching may be somewhat unique in its absolute dependence upon heparan sulfate biosynthetic enzymes and/or heparan sulfate-dependent growth factor signaling in primary branching, terminal end bud bifurcation, secondary and ductal side-branching, as well as lobuloalveolar development.
The authors would like to thank Dr. Thorgeirsson for the generous gift of the c-Met null mice. This work was supported by grant GM33063 and HL57345 (to J.D.E.) and grants DK57286, DK65831 and DK79784 (to S.K.N). O.G. was supported by an NRSA grant AI058916.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1While the other sulfotransferases could also play a role in ductal branching morphogenesis, we decided to focus on Hs2st due to the fact that (i) knockouts of glucosaminyl 3-O sulfotransferases (Hs3st1 and Hs3st2) are apparently healthy (HajMohammadi et al., 2003), (ii) three isoforms of glucosaminyl 6-O sulfotransferases (Hs6st1-3) exist, (iii) only one isoform of Hs2st is found in the mammary gland, and (iv) knockouts of Hs2st have been shown to affect epithelial branching morphogenesis in the kidney (Bullock et al., 1998).
University at Buffalo, South Campus
Buffalo, NY 14215