Structure and epitope distribution of heparan sulfate is disrupted in experimental lung hypoplasia: a glycobiological epigenetic cause for malformation?

Heparan sulfate (HS) is present on the surface of virtually all mammalian cells and is a major component of the extracellular matrix (ECM), where it plays a pivotal role in cell-cell and cell-matrix cross-talk through its large interactome. Disruption of HS biosynthesis in mice results in neonatal death as a consequence of malformed lungs, indicating that HS is crucial for airway morphogenesis. Neonatal mortality (~50%) in newborns with congenital diaphragmatic hernia (CDH) is principally associated with lung hypoplasia and pulmonary hypertension. Given the importance of HS for lung morphogenesis, we investigated developmental changes in HS structure in normal and hypoplastic lungs using the nitrofen rat model of CDH and semi-synthetic bacteriophage ('phage) display antibodies, which identify distinct HS structures. The pulmonary pattern of elaborated HS structures is developmentally regulated. For example, the HS4E4V epitope is highly expressed in sub-epithelial mesenchyme of E15.5 - E17.5 lungs and at a lower level in more distal mesenchyme. However, by E19.5, this epitope is expressed similarly throughout the lung mesenchyme. We also reveal abnormalities in HS fine structure and spatiotemporal distribution of HS epitopes in hypoplastic CDH lungs. These changes involve structures recognised by key growth factors, FGF2 and FGF9. For example, the EV3C3V epitope, which was abnormally distributed in the mesenchyme of hypoplastic lungs, is recognised by FGF2. The observed spatiotemporal changes in HS structure during normal lung development will likely reflect altered activities of many HS-binding proteins regulating lung morphogenesis. Abnormalities in HS structure and distribution in hypoplastic lungs can be expected to perturb HS:protein interactions, ECM microenvironments and crucial epithelial-mesenchyme communication, which may contribute to lung dysmorphogenesis. Indeed, a number of epitopes correlate with structures recognised by FGFs, suggesting a functional consequence of the observed changes in HS in these lungs. These results identify a novel, significant molecular defect in hypoplastic lungs and reveals HS as a potential contributor to hypoplastic lung development in CDH. Finally, these results afford the prospect that HS-mimetic therapeutics could repair defective signalling in hypoplastic lungs, improve lung growth, and reduce CDH mortality.


Abstract Background
Heparan sulfate (HS) is present on the surface of virtually all mammalian cells and is a major component of the extracellular matrix (ECM), where it plays a pivotal role in cell-cell and cell-matrix cross-talk through its large interactome. Disruption of HS biosynthesis in mice results in neonatal death as a consequence of malformed lungs, indicating that HS is crucial for airway morphogenesis. Neonatal mortality (~50%) in newborns with congenital diaphragmatic hernia (CDH) is principally associated with lung hypoplasia and pulmonary hypertension. Given the importance of HS for lung morphogenesis, we investigated developmental changes in HS structure in normal and hypoplastic lungs using the nitrofen rat model of CDH and semi-synthetic bacteriophage ('phage) display antibodies, which identify distinct HS structures.

Results
The pulmonary pattern of elaborated HS structures is developmentally regulated. For example, the HS4E4V epitope is highly expressed in sub-epithelial mesenchyme of E15.5 -E17.5 lungs and at a lower level in more distal mesenchyme. However, by E19.5, this epitope is expressed similarly throughout the lung mesenchyme.
We also reveal abnormalities in HS fine structure and spatiotemporal distribution of HS epitopes in hypoplastic CDH lungs. These changes involve structures recognised by key growth factors, FGF2 and FGF9. For example, the EV3C3V epitope, which was abnormally distributed in the mesenchyme of hypoplastic lungs, is recognised by FGF2.

-3 -
The observed spatiotemporal changes in HS structure during normal lung development will likely reflect altered activities of many HS-binding proteins regulating lung morphogenesis. Abnormalities in HS structure and distribution in hypoplastic lungs can be expected to perturb HS:protein interactions, ECM microenvironments and crucial epithelial-mesenchyme communication, which may contribute to lung dysmorphogenesis. Indeed, a number of epitopes correlate with structures recognised by FGFs, suggesting a functional consequence of the observed changes in HS in these lungs. These results identify a novel, significant molecular defect in hypoplastic lungs and reveals HS as a potential contributor to hypoplastic lung development in CDH. Finally, these results afford the prospect that HS-mimetic therapeutics could repair defective signalling in hypoplastic lungs, improve lung growth, and reduce CDH mortality.

Background
The majority of the extracellular proteins involved in regulating embryonic development interact with heparin/heparan sulfate (HS), and, moreover, require HS for their cellular activities [1]. These include proteins required for lung morphogenesis [2][3]. For example, not only are fibroblast growth factors (FGFs) essential for lung development [4][5][6][7][8][9][10], but they require HS for FGF receptor activation and subsequent signalling [11][12][13]. Due to its vast interactome and location at the cell surface and within the extracellular matrix (ECM), HS is ideally positioned to integrate biochemical regulators of lung development with mechanical stimuli required for normal lung growth [14][15].
HS is a linear polysaccharide consisting of N-acetyl glucosamine-glucuronic acid disaccharide repeats. Chains are variably modified by N-deacetylation/N-sulfation of N-acetyl glucosamines, O-sulfation at various positions and conversion of glucuronic acid to its C-5 epimer, iduronic acid. These modifications do not occur at every potential site within a chain, resulting in a diverse range of HS chain structures displayed by a cell [1]. Moreover, HS is post-synthetically remodelled by 6-Oendosulfatase enzymes, which selectively remove sulphate groups [16][17][18]. HS chains are usually attached to core proteins to form HS proteoglycans (HSPGs), which are expressed by most mammalian cells and represent a major component of the cell surface and ECM. Individual cells of a tissue display a variety of HS chains, which, in addition to being structurally complex and diverse, are dynamic, altering over time and with cellular physiology [3]. Since interactions between HS and proteins are mediated by specific HS structures, changes in HS structure in vivo are likely to alter HS:protein binding events and related signalling. Characterising HS fine structure in vivo is, therefore, important as it equates to a view of HS function.
Obtaining structural information on native HS is challenging due to the non-template nature of HS biosynthesis (unlike proteins or nucleic acids). Tissue HS is typically analysed by extraction and purification. However, the inherent averaging of this approach limits the information to an overall assessment of the mixed population of HS structures present, and all spatial information is lost. In addition, due to the relative low immunogenicity of HS, only a limited number of HS specific monoclonal antibodies are available [19][20][21]. However, single chain variable fragment (scFv) antibodies generated by bacteriophage ('phage) display methodology [22-23] allow specific classes of structures in HS to be probed in situ. We have demonstrated recently that these antibodies display distinct specificities for different HS structures in vitro and, therefore, the individual HS epitopes they recognise, though structurally complex, are unique [24]. Limited analysis of one antibody shows that these probes are suitable to identify the diversity of HS in different cellular compartments of fetal HS plays a fundamental role in airway morphogenesis. In Drosophila and mice, disrupted HS biosynthesis in vivo results in defective airway branching, which in the mouse, results in lethal neonatal respiratory insufficiency [26][27][28][29]. In addition, digestion of endogenous HS in cultured lung explants using heparitinases or inhibition of HS sulfation with sodium chlorate, disrupts branching [30][31]. Further compelling evidence for a critical role of HS in lung disease is the association in humans and mice of mutations in the HSPG, glypican-3, with lung hypoplasia and congenital diaphragmatic hernia (CDH), known as Simpson-Golabi-Behmel syndrome [32][33][34][35]. It is, therefore, clear that HS is crucial for lung development and identifies HS as a potential contributor to pulmonary pathologies such as pulmonary hypoplasia in CDH [3]. CDH is characterised by a diaphragmatic defect, herniation of abdominal contents into the thoracic cavity and pulmonary hypoplasia. The high neonatal morbidity and mortality is largely attributed to severe respiratory insufficiency resulting from hypoplastic lung growth and pulmonary hypertension. Hitherto, most work exploring the pathogenesis of human birth defects has focussed on the identification of changes in gene expression and/or in protein levels. However, very few truly genetic causes of these defects have been identified. Since strong evidence supports a central role for HS as a master regulator of extracellular proteins controlling embryogenesis, we have investigated HS structure during development of the lung, and in the pathogenesis of CDH and pulmonary hypoplasia using HS specific 'phage display antibodies. We utilised the teratogen-induced rodent model of CDH, which uses nitrofen (2,4dichlorophenyl-p-nitrophenyl ether) to induce congenital malformations in the offspring of treated pregnant dams with striking similarity to human CDH [36][37][38].
Here, we demonstrate that HS undergoes structural alterations during normal lung development and that there are pronounced abnormalities of HS structure and epitope distribution in hypoplastic lungs. Aberrations in epithelial basement membrane structure and composition were also identified in hypoplastic lungs, reflecting a specific abnormality in the ECM. The functional importance of these changes to HS structure during normal development and in hypoplastic lung is illustrated by our finding that a number of the HS epitopes are analogous to structures recognised by the critical morphogenetic growth factors, FGF2 and FGF9. Hence, these novel glycobiological defects may contribute to defective lung morphogenesis via altered interactions between HS and key signalling molecules such as FGFs. In addition, altered contacts between the ECM and cell surface are likely to interrupt mechanotransduction across a tissue, which is crucial for morphogenesis. HS may, therefore, play significant biochemical and biomechanical roles in the pathogenesis of pulmonary hypoplasia in CDH.

Results
Seven HS 'phage display antibodies were chosen on the basis that they have been raised against HS/heparin from a variety of tissue sources and display distinct binding specificities, which we have characterised in depth [24]. The data on antibody binding specificities and epitope structures are summarised in Table 1. (dp) degree of polymerisation, i.e., a disaccharide is a dp2, tetrasaccharide a dp4, etc.

HS structure changes during normal rat lung development
Different staining patterns were observed in developing rat lung with the various HS antibodies, indicating recognition of distinct HS epitope structures ( Figure 1 and Table 2). One particular HS epitope recognised by HS4C3V was not identified in developing rat lungs of any age. The use of adult rat kidney as a positive control demonstrates that this structure is indeed absent from lung ( Figure 1). The remaining six antibodies show distinct patterns of staining in fetal rat lungs, and the various lung compartments display different HS structures (additional files 1, 2, 3, 4, 5 and 6, summarised in Table 2). For example, in airway epithelium, HS structures recognised by EV3C3V, AO4B08V and HS3A8V are displayed, whereas the other three antibody epitopes are not. At E13.5, only the EV3C3V HS epitope is displayed by the epithelium, and at E15.5, the AO4B08V and HS3A8V HS epitopes are also present ( Figure 2). The spatial localisation of these HS epitopes was shown to change during lung morphogenesis, indicating alterations in the structure of native HS chains during mammalian lung development (Figures 1 and 3 and additional files 1, 2, 3, 4, 5 and 6). For example, the HS epitope recognised by HS3A8V is displayed exclusively by epithelial basement membranes at E13.5 at a high level. At the early pseudoglandular period (E15.5), this structure is additionally displayed at a high level throughout the mesenchyme and by the airway epithelium. At the late pseudoglandular period (E17.5), mesenchymal expression of this epitope remains high; however, distribution of the structure becomes particularly concentrated around smaller developing airways and epithelial staining is no longer present. At canalicular (E19.5) and saccular (E21.5) fetal stages, HS3A8V epitope levels in the mesenchyme decrease and epitope distribution becomes more widespread ( Figure 3 and Table 2). In contrast, the HS epitope identified by EW4G1V is not present in E13.5 rat lungs; however, at E15.5 it is detected in epithelial basement membranes and in the surrounding mesenchyme at a low level ( Figure 3 and Table 2). At E17.5, levels of this HS epitope drastically increase, particularly in epithelial basement membranes, while epitope levels in the mesenchyme remain low and its distribution becomes concentrated around smaller distal airways. From E19.5 -E21.5, distribution of the EW4G1V epitope becomes more widespread throughout the mesenchyme ( Figure 3 and Table 2).
In the pulmonary vasculature, HS3B7V shows a unique staining profile, specifically highlighting regions in the outer tunica media of arterial walls ( Figure 4). This HS structure is absent from the inner medial layer, in contrast to the other five antibody epitopes, which were present and showed comparable vascular staining, highlighting the medial layer of both arteries and veins, predominantly the basal lamina surrounding smooth muscle cell layers ( Figure 4).

HS structure is abnormal in hypoplastic lungs from nitrofen-induced CDH
Identification of HSPGs with the 3G10 antibody, which recognises the neo-epitope generated on all HSPGs following heparitinase digestion of the HS chains, indicated that there is no gross disruption in the overall spatial localisation of HSPGs in hypoplastic lungs. However, levels of HSPGs are reduced, particularly at E15.5 -E17.5 and in epithelial basement membranes ( Figure 5A). Analysis of specific HS epitopes with 'phage display antibodies indicated an abnormality in the fine structure of HS in hypoplastic lungs, which was also more marked in lungs of earlier gestation (Figures 5, 6 and 7, summarised in Table 2). Levels of a number of HS epitopes are reduced (AO4B08V) or lost (HS3A8V, EV3C3V and EW4G1V) from the airway epithelium at E15.5 and E17.5 compared to normal lungs ( Figure 5B and Table 2).
Moreover, all of the HS structures analysed are displayed at a lower level by epithelial basement membranes. Although a number of HS structures are also shown to be reduced in hypoplastic lung mesenchyme (HS4E4V, HS3A8V and AO4B08V at E15.5 -E17.5) ( Figure 5C and additional files 2, 3 and 4), HS structural alterations are more complex than a simple reduction or loss of epitopes. Levels of the HS epitope identified by EV3C3V are increased in hypoplastic lung mesenchyme, and in addition, the spatial distribution of this HS structure is abnormal ( Figure 6 and Table   2). In E15.5 -E17.5 control lung mesenchyme, a gradient of EV3C3V epitope distribution is observed, with sub-epithelial areas adjacent to smaller, distal airways displaying high epitope levels and more proximal regions of the lung displaying a lower level of the structure. However, in nitrofen treated hypoplastic lungs, this gradient is not present and epitope distribution is more widespread throughout the entire lung mesenchyme ( Figure 6 and Table 2). At E19.5 -E21.5, EV3C3V epitope levels and distribution are comparable to that observed in control lungs.
Pulmonary arteries of nitrofen-treated hypoplastic lungs have thickened vessel walls with increased smooth muscle content, contributing to the persistent pulmonary hypertension associated with CDH [39][40][41]. Although we confirmed marked thickening in arterial walls of nitrofen-treated lungs, no difference in HS or HSPG staining was observed. All HS antibodies highlighted the tunica media, with five predominantly staining the basal lamina surrounding the layers of smooth muscle and HS3B7V specifically highlighting an outer region of the arterial walls.

HS/HSPG staining identifies abnormalities in epithelial basement membranes
Abnormalities in epithelial basement membrane HSPG expression and HS structure were identified in hypoplastic lungs. Basement membranes appear thinner, and levels of both HSPGs ( Figure 7A) and specific HS epitopes ( Figure 7B and C) are reduced.
In addition, staining with HS antibodies revealed discontinuities in basement membrane HS distribution, which are not observed with 3G10 immunohistochemistry.
To examine the general structure of epithelial basement membranes in hypoplastic lungs and evaluate whether abnormalities are HS/HSPG specific, hypoplastic rat lungs were probed with an antibody to laminin ( Figure 7D), an integral component of basement membranes. Immunohistochemical detection of laminin indicated that epithelial basement membranes are indeed thinner; however, no discontinuities in laminin staining were observed ( Figure 7D). The abnormally fine laminin staining of epithelial basement membranes was more pronounced in lungs at the pseudoglandular stages of development (E15.5 and E17.5), reminiscent of HS and HSPG basement membrane staining.

Functional analysis of HS epitopes using ELISA
To further our understanding of the functional consequences of abnormal HS structure identified in hypoplastic lungs, we analysed the specificities of the HS antibodies in competition ELISAs with FGF2 and FGF9, which are known to be involved in lung morphogenesis [4,[42][43][44]. The relative binding affinities of FGFs for HS epitopes were evaluated by determining IC 50 values, defined as the concentration of FGF that inhibits antibody binding to HS by 50 % (Figure 8). This allowed us to investigate possible overlap between antibody epitope structures and FGF binding sites in HS, to identify potential biological functions of the antibody epitopes.
FGF2 competed for all six antibody epitopes to varying extents ( Figure 8A and B), whereas FGF9 only competed for two antibody epitopes, recognised by HS3B7V and HS3A8V ( Figure 8C and D values of 670 nM ± 18 nM and 1.8 µM ± 0.81 µm, respectively ( Figure 8C and D).

Discussion and Conclusions
HS is a master regulator of morphogenesis and is essential for lung development. In the present study, we have demonstrated spatiotemporal alterations in HS structure and distribution during normal and hypoplastic lung development using HS specific 'phage display antibodies. Moreover, we show that a number of antibody epitopes are also structures recognised by FGFs, suggesting that abnormal distribution of these epitopes in hypoplastic CDH lungs may alter FGF binding, with functional consequences for morphogenesis.

HS structure in normal fetal rat lungs
A number of HS scFv antibodies have been used previously to analyse HS structures in adult human lungs [45]. Only one of these antibodies used in the previous work is used in the present study (EV3C3V). In adult lungs, the EV3C3V HS epitope is identified in airway epithelial cells and their underlying basement membranes and in basement membranes surrounding smooth muscle cells of blood vessels [45]. We also identified the EV3C3V epitope in epithelial and smooth muscle cell basement membranes at all developmental stages (Figures 2, 4, 6 and additional file 5) and in airway epithelial cells at E13.5 -E17.5 ( Figure 2). This demonstrates a degree of conservation in HS structure between species.
We demonstrate that distinct cellular compartments of the lung display a variety of HS chains of different structure, which are modified during lung development, with temporal variation in both expression level and spatial localisation of HS epitopes identified by HS antibodies.
The epitope for one antibody, HS4C3V, was not identified in fetal lungs, however, it was present in kidney tissue ( Figure 1) and has previously been detected in immunoblots of solubilised fetal lung extracts [24]. This raises the possibility of cryptic binding sites for the antibodies in situ, and is an important consideration when interpreting immunohistochemical data, i.e., the absence of an epitope in situ does not necessarily mean it is not present in a tissue, rather, the epitope may be masked, for example, by endogenous proteins bound to HS. The antibodies, therefore, specifically highlight free binding structures in HS. Indeed, occupied, cryptic binding sites in HS in situ have been identified previously in growing mammary glands probed with FGF2 [46].
The antibody binding specificities (Table 1)  binding for HS4E4V [24]. In contrast to HS4E4V, HS3A8V binds to more highly sulfated HS, particularly N-sulfated sequences (Table 1), and displays a wider epitope distribution throughout fetal lung mesenchyme (Figures 1 and 3). Hence, more highly sulfated HS in proximal mesenchyme may allow binding of HS3A8V, but not HS4E4V, while a more heterogeneous population of HS structures in distal mesenchyme, allows binding of both.
The localisation of specific HS structures within the developing lung and their importance for the coordination of branching morphogenesis has been hinted at in previous studies, e.g., expression of the epitope recognised by the HS specific monoclonal antibody, 10E4, was shown to alter rapidly with lung growth in vitro [31].
In addition, blanket addition of heparin to lung cultures lacking endogenous sulfated GAGs after sodium chlorate treatment results in generalised epithelial expansion, indicating a global growth response, rather than defined branching [31]. These data, together with ours, suggest that distinct HS structures are specifically displayed by the various lung cell types to direct localised signalling and spatiotemporally restricted morphogenetic cues, e.g., epithelial budding. In addition, modification of HS fine structure during development, together with alterations in HS microenvironment via dynamic HS:protein interactions, will influence crucial cell-cell and cell-matrix communication governing lung morphogenesis. HS structural dynamics are, therefore, likely to be an important regulator of fetal lung morphogenesis.

HS structure and epitope distribution are abnormal in hypoplastic CDH lungs
Using the nitrofen rat model of CDH, we investigated the potential role of HS in hypoplastic lung development, since HS has been shown to be crucial for normal lung morphogenesis. In addition, it has been shown previously that nitrofen-exposed rat lung explants respond abnormally to exogenous FGF1 and FGF2 and also heparin, suggesting a possible defect in the FGF:FGFR:HS signalling system [47][48]. Notably, in humans, a mutation in the gene encoding the HSPG, glypican-3, features multiple congenital anomalies, including CDH [32][33][34][35].
Following analysis of HS fine structure and its developmental regulation during normal rat lung morphogenesis, HS structure and distribution was analysed in hypoplastic lungs from rats with nitrofen-induced CDH. In hypoplastic lungs, HSPG expression is reduced, and in addition, specific abnormalities in HS structure and spatial distribution were observed, which cannot simply be a consequence of an overall reduction in the level of HSPGs, since the level of some epitopes is increased, e.g., EV3C3V ( Figure 6). As the overall distribution of HSPGs appears the same in control and hypoplastic lungs, abnormalities in the spatial distribution of specific epitopes is likely to reflect alterations in HS fine structure and irregular localisation of discrete HS structures displayed by the various lung cell types. In addition, the occupancy or availability of antibody binding sites may differ in hypoplastic lungs due to differences in protein binding events. Changes in HS structure in the lung is likely to modify HS:protein interactions, since these rely on specific HS structures [1]. This in turn, will affect various signalling systems, as HS:protein interactions have been shown to be functionally significant, regulating transport and effector functions of the protein ligand. For example, HS plays a key role in FGF signalling, which is fundamental for lung morphogenesis.
HS facilitates interactions between FGFs and FGFRs and is also required for sustained FGFR activation and subsequent cellular signalling [11][12][13]. Cells expressing FGFRs but lacking HSPGs are unresponsive to FGF unless heparin/HS is added [11][12] and treatment of cells with sodium chlorate or heparinase blocks the biological activity of FGFs, an effect which can be restored by the addition of exogenous heparin [13]. In an ex vivo model system of epithelial branching morphogenesis using mouse salivary gland, modification of FGF:HS binding affinities was shown to impact upon the morphogenetic effect of FGFs [55]. FGF10 with a reduced affinity for HS (via single amino acid mutations in the heparin binding site), formed abnormal gradients due to altered transport properties, resulting in an induction of epithelial branching rather than elongation observed with wild type FGF10 [55]. In contrast, reduced affinity of FGF10 for its receptor, FGFR2b, affected only the extent of the response, without altering the nature of the response.
Abnormal HS structure and distribution observed here in hypoplastic lungs can, therefore, be expected to contribute to defective lung morphogenesis via aberrant epithelial-mesenchymal signalling as a result of altered HS:protein interactions.

Competitive selectivity of FGFs and antibodies provides insight into structure:function relationships of epitopes and a functional consequence of abnormal HS in hypoplastic lungs
Competitive binding assays with FGFs and HS antibodies allowed us to demonstrate that antibodies recognise specific HS structures that are also recognised by key FGF morphogens, thereby revealing biological relevance of epitopes. Alterations in FGF bindings sites in the lung has previously been shown to have functional consequences for morphogenesis [30].
FGF2 is expressed in the developing lung and is important for lung morphogenesis [43][44]56]. The HS binding specificity of FGF2 is well characterised, requiring N-sulfated and 2-O-sulfated HS structures [57][58] and at least a tetrasaccharide [59] for binding. In competitive binding assays, FGF2 competed with all six HS antibodies to varying extents. Of particular note is the effectiveness of FGF2 to compete with EV3C3V, which was significantly higher compared to competition with the other antibodies ( Figure 8A and B Functional analysis of epitope structures enables us to suggest potential biological consequences of abnormal epitope distribution in developing lungs. In CDH hypoplastic lungs, the EV3C3V epitope was identified at a higher level compared to normal lungs, and in addition, was shown to be abnormally distributed in hypoplastic lung mesenchyme ( Figure 6). This may indicate an increase in the number of EV3C3V epitopes and, therefore, structures recognised by FGF2, in the HS synthesised by the lung cells. Alternatively, the availability of these structures may be increased in the lung mesenchyme due to altered expression of proteins that bind this class of structures. Moreover, these lungs respond abnormally to FGF2 [47]. Addition of FGF2 to nitrofen-treated lung explants results in increased lung area and formation of dilated, cystic airways, whereas FGF2 was shown to have minimal effect on the growth of control lung explants [47]. Our data describing an increased number of available EV3C3V/FGF2 binding structures in the mesenchyme of hypoplastic lungs provides a possible explanation for this abnormal response to exogenous FGF2.
Epitopes analogous to structures recognised by FGF9, i.e., the HS3A8V and HS3B7V epitopes, were also abnormally expressed in nitrofen-treated lungs. Both epitopes were expressed at a reduced level in epithelial basement membranes and in addition, showed abnormal mesenchymal expression (Table 2 and additional files 1 and 3). The HS3B7V epitope was displayed at a low level in sub-epithelial mesenchymal compartments of E19.5 hypoplastic lungs, and this is a compartment which was not stained in normal lungs (Table 2 and additional file 1). The HS3A8V epitope was identified at a high level in sub-epithelial mesenchyme at E15.5 and E17.5 in normal lungs, however, in hypoplastic lungs, mesenchymal expression was reduced and the partitioning of HS3A8V staining in sub-epithelial and sub-mesothelial mesenchyme was not evident ( We have shown here that hypoplastic lungs exhibit abnormal epithelial basement membranes. Immunodetection of laminin, HSPGs and HS structures demonstrated an abnormally thinned basement membrane. Additionally, HS epitopes, but not HSPGs or laminin, were displayed discontinuously in hypoplastic lungs, suggesting abnormal localisation of HS epitopes and/or availability of binding sites in HS.

Clinical implications
Current treatments for CDH primarily focus on postnatal management to address the consequences of pulmonary hypoplasia and hypertension, including strategies aimed at providing adequate tissue oxygenation via inhaled nitric oxide, high frequency oscillatory ventilation and extracorporeal membrane oxygenation (ECMO). However, CDH mortality has not been greatly improved. Results from the present work suggest that heparin/HS based therapeutics may be beneficial in ameliorating hypoplastic lung growth in CDH. Indeed, a number of glycotherapeutics are emerging for the repair of damaged tissue, e.g., skin and bone [92][93][94][95][96] and for the treatment of some cancers [97][98][99]. The morphogenetic effect of chemically modified heparins has previously been investigated in salivary gland branching morphogenesis [100]. Using a similar rationale, investigating the effect of various engineered heparins on lung growth may help develop a class of HS structures able to ameliorate lung hypoplasia, reduce smooth muscle cell proliferation and increase vascular branching. With chemically modified heparins available, which possess low or zero anticoagulant activity [101], this is an exciting potential future therapeutic avenue.

FGF2 and FGF9 synthesis and purification
Full-length human recombinant FGF2 with an N-terminal hexahistidine tag was produced in E. coli exactly as described [102]. FGF9 (Uniprot Accession: P31371; residues: 1-208) with a 6×Histidine tag and a TEV cleavage site (26 amino acids, MKHHHHHHPMSDYDIPTTENLYFQGA) at the N-terminus was expressed in C41 E.coli cells using a modified pET-24b vector (pETM-11, kind gift from Dr Paul Elliott, University of Liverpool), which provides the sequences of the 6×Histidine tag and the TEV cleavage site. Protein was produced in bacteria using an auto induction system [103]. Cells were grown at 37°C, for 7 h in Terrific Broth, and FGF-9 production was induced at 22°C for 16 h. Cell pellets were lysed by sonication in   HS3B7V shows a unique pattern of vascular staining in fetal lungs, specifically highlighting the outer tunica media of arterial walls (arrow) and leaving the inner tunica media unlabelled. In addition, this antibody does not stain pulmonary veins.
(Some weak, nuclear staining was observed with HS3B7V on occasion, including in veins. However, this was not sensitive to heparinase digestion and is therefore nonspecific staining). The remaining antibodies highlight the entire tunica media layer in the walls of both arteries and veins (only AO4B08V, HS4E4V and EV3C3V are shown, however, HS3A8V and HS4E4V display comparable blood vessel staining).
Fetal rat lungs were probed with HS antibodies followed by rabbit VSV-G tag antibody and FITC conjugated goat anti-rabbit IgG. Scale bar represents 10 µm and all images are the same magnification. (A) artery, (V) vein, (aw) airway, (m) media HSPG levels, identified by 3G10, are reduced in hypoplastic rat lungs, particularly at E15.5 and E17.5 and in epithelial basement membranes (A). Analysis of specific HS epitopes with 'phage display antibodies revealed an abnormality in HS fine structure.
A number of epitopes are reduced or lost from the epithelium e.g., AO4B08V and HS3A8V, respectively (B). In addition, a number of epitopes, e.g., HS4E4V, are reduced in the lung mesenchyme (C) and all epitopes are reduced in epithelial basement membranes (B, C).
Hypoplastic lungs from rats with nitrofen-induced left sided CDH and control lungs from rats fed olive oil alone were probed with 3G10 after initial digestion of lung HS with heparitinase to reveal the 3G10 neo-epitope on all HSPGs. Bound antibody was then detected with FITC conjugated goat anti-mouse IgG. As a negative control, sections were incubated with heparitinase buffer alone without enzyme, leaving the 3G10 neo-epitope concealed. Incubation of lung sections with HS 'phage display antibodies was followed by rabbit VSV-G tag antibody and FITC conjugated goat anti-rabbit IgG. Scale bars represent 10 µm. (ep) epithelium, (bm) basement membrane, (me) mesenchyme In normal development, a gradient of EV3C3V epitope distribution is observed in the mesenchyme of E15.5 -E17.5 lungs, with high epitope levels in sub-epithelial mesenchyme around distal airways (arrowhead) and lower levels around proximal airways. This organised gradient of EV3C3V epitope distribution is lost in lungs of nitrofen treated rats, which display the structure at a high level throughout the mesenchyme at E15.5 -E17.5. At E19.5 -E21.5, EV3C3V epitope levels and distribution are comparable to control lungs.
Hypoplastic lungs from rats with nitrofen-induced left sided CDH and control lungs from rats fed olive oil alone were probed with EV3C3V followed by rabbit VSV-G tag antibody and FITC conjugated goat anti-rabbit IgG. Scale bar represents 10 µm and all images are the same magnification. (aw) airway, (me) mesenchyme FGF2 and FGF9 competed with a number of antibodies for HS binding, indicating that epitope structures are analogous to structures recognised by these growth factors.
FGF2 competed for all six epitopes to variable extents, but most significantly with EV3C3V. FGF9, in contrast, showed more competitive selectivity and was only able to compete for two epitope structures, recognised by HS3B7V and HS3A8V.
PMHS was biotinylated and immobilised on streptavidin coated microtitre plates.
Equilibrium binding of HS antibodies in the presence of various concentrations of
In normal lungs, the HS epitope recognised by HS3A8V is restricted to epithelial basement membranes at E13.5. From E15.5, distribution of the epitope is more widespread and is present in epithelial basement membranes and throughout the mesenchyme, particularly in sub-epithelial mesenchyme. Epithelial cells also display this HS structure transiently at E15.5 and (more weakly) at E17.5. In hypoplastic lungs, mesenchymal expression of the HS3A8V epitope is reduced, particularly at E15.5 and E17.5, and epithelial staining observed in normal lungs is lost.
Additionally, irregularities in epithelial basement membrane staining are observed. As a negative control, endogenous HS was digested with heparitinase prior to antibody incubation.
Expression of the AO4B08V HS epitope increases during the course of normal lung development. At E13.5, it is only weakly expressed by epithelial basement membranes, and at E15.5, is additionally displayed at a low level in the mesenchyme and airway epithelium. From E17.5 -E21.5, levels of this epitope increases in basement membranes and throughout the mesenchyme. In hypoplastic lungs, however, expression of the AO4B08V epitope is reduced in the epithelium and underlying basement membranes, and in addition, basement membranes appear discontinuous. In lung mesenchyme, however, the AO4B08V epitope structure is displayed at a higher level compared to normal lungs. As a negative control, endogenous HS was digested with heparitinase prior to antibody incubation.
In normal lungs, the EV3C3V epitope is displayed by the epithelium at E13.5 -E17.5 and in the underlying basement membranes at E13.5 -E21.5. A gradient of epitope expression is observed in the mesenchyme, with highest levels in sub-epithelial mesenchyme around smaller, distal airways and lower levels in sub-mesothelial mesenchyme. However, in hypoplastic lungs, this gradient of mesenchymal expression is lost, and the EV3C3V epitope is more extensively and evenly distributed throughout the entire mesenchyme. In addition, epithelial staining is lost from hypoplastic lungs and basement membrane staining is irregular. As a negative control, endogenous HS was digested with heparitinase prior to antibody incubation.
- 46 -In normal developing lungs, the HS structure identified by EW4G1V is absent at E13.5. From E15.5 onwards, however, it is present in all epithelial basement membranes and also at a low level in the mesenchyme, with increased levels at E21.5.
This epitope is transiently expressed by the epithelium at E15.5. In hypoplastic lungs, levels of this epitope appear to be raised somewhat in the mesenchyme compared to normal lungs and simultaneously reduced in epithelial basement membranes. As a negative control, endogenous HS was digested with heparitinase prior to antibody incubation.