|Home | About | Journals | Submit | Contact Us | Français|
Users may view, print, copy, download and text and data- mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use: http://www.nature.com/authors/editorial_policies/license.html#terms
Heparan sulfate proteoglycans (HSPGs) bind to multiple growth factors/morphogens and regulate their signaling. 6-O-sulfation (6S) of glucosamine within HS-chains is critical for many of these ligand interactions. Sulf-1 and Sulf-2, which are extracellular neutral-pH sulfatases, provide a novel post-synthetic mechanism for regulation of HSPG function by removing 6S from intact HS-chains. The Sulfs can thereby modulate several signaling pathways, including the promotion of Wnt signaling. We found induction of SULF2 transcripts and Sulf-2 protein in human lung adenocarcinoma and squamous cell carcinoma, the two major classes of non-small cell lung cancers (NSCLC). We confirmed widespread Sulf-2 protein expression in tumor cells of 10/10 surgical specimens of human lung squamous carcinomas. We studied five Sulf-2+ NSCLC cell lines, including two which were derived by cigarette-smoke transformation of bronchial epithelial cells. shRNA-mediated Sulf-2 knockdown in these lines caused an increase in 6S on their cell surface and in parallel reversed their transformed phenotype in vitro, eliminated autocrine Wnt signaling, and strongly blunted xenograft tumor formation in nude mice. Conversely, forced Sulf-2 expression in non-malignant bronchial epithelial cells produced a partially transformed phenotype. Our findings support an essential role for Sulf-2 in lung cancer, the leading cancer killer.
Lung cancer is the leading world-wide cause of cancer deaths with the majority of cases resulting from the exposure of airways to tobacco-smoke carcinogens (Sato et al., 2007). Further understanding of the molecular basis of lung cancer is needed, since current treatment options are frequently inadequate (Osada and Takahashi, 2002; Sato et al., 2007). Although heparan sulfate proteoglycans (HSPGs) are ubiquitous elements of the cell surface/ECM and perform critical signaling functions in development and normal physiology (Bernfield et al., 1999; Bishop et al., 2007; Habuchi et al., 2004), they have received scant attention in lung cancer. HSPGs consist of heparan sulfate (HS) chains containing repeating uronic acid and glucosamine disaccharide units with the chains covalently linked to a restricted set of core proteins. The functions of HSPGs are mediated through their binding to a myriad number of ligands, including growth factors and morphogens. The pattern of four sulfation modifications on the sugars largely dictates binding specificity (Esko and Lindahl, 2001; Gallagher, 2001). The 6-O-sulfation (6S) of glucosamine is known to be critical in the binding of many protein ligands (Gallagher, 2001; Habuchi et al., 2004). One mechanism for the spatiotemporal regulation of 6S in the embryo and adult is through expression of the heparan sulfate 6-O-sulfotransferases, which add sulfate esters to nascent HS chains (Brickman et al., 1998; Habuchi et al., 2004; Habuchi et al., 2007).
A novel post-synthetic mechanism for regulating 6S on HSPGs emerged several years ago with the cloning and characterization of two novel extracellular sulfatases (Sulf-1 and Sulf-2) in human (Morimoto-Tomita et al., 2002), following the original identification of Sulf-1 in the quail embryo (Dhoot et al., 2001). These neutral-pH endosulfatases remove 6S in highly sulfated subdomains of intact heparin/HSPGs (Ai et al., 2003; Lamanna et al., 2008; Morimoto-Tomita et al., 2005; Viviano et al., 2004). Desulfation of HSPGs/heparin by the Sulfs reduces interactions with protein ligands, including Wnts, FGFs, and GDNF (Ai et al., 2003; Ai et al., 2007; Uchimura et al., 2006), and modulates their signaling. The importance of these enzymes in development is established by the abnormal phenotypes observed in single and double sulf knockouts in mice (Ai et al., 2007; Holst et al., 2007; Lum et al., 2006; Ratzka et al., 2008). Consistent with many precedents for onco-developmental genes, Sulf-2 has been suggested to be relevant in several cancers (Lai et al., 2008; Morimoto-Tomita et al., 2005; Nawroth et al., 2007). Here, we present evidence that Sulf-2 promotes human lung carcinogenesis.
Initially, we looked for dysregulated expression of SULF genes in human lung cancer by mining public microarray data. Relative to normal lung, SULF1 was markedly increased by 3–6 fold in both adenocarcinoma (p = 0.006) and squamous cell carcinoma (p = 0.004) (Supplementary Fig. 1a), the two major classes of non-small cell lung carcinoma. Data from the Consortium for Functional Glycomics (http://www.functionalglycomics.org/glycomics/publicdata/microarray.jsp) provided information about both SULFs in lung cancer (Fig. 1a). Paired samples of lung squamous carcinoma and nonmalignant neighboring tissue were obtained from 10 patients undergoing surgical resection. SULF1 increased in 10/10 pairs with a mean increase of 18 ± 2.4-fold (p=0.0005). SULF2 increased in 8/10 pairs with a mean increase of 3 ± 0.3-fold (p=0.003). qPCR analysis of SULF1 and SULF2 in archived cases of lung carcinoma verified these findings (Fig. 1b,c). (SULF1 and SULF2 increased 12 ± 1.5 fold (p=0.008) and 4 ± 0.3 fold (p=0.05), respectively in squamous carcinomas and 3 ± 0.4 fold for both (p=0.003 and p=0.002, respectively) in adenocarcinomas).
We next determined SULF expression in a series of NSCLC cell lines, all of which form tumors in immunocompromized mice (Supplementary Table 1). In addition, we evaluated two novel cell lines, B-ST and P-ST, which were obtained by exposing BEAS2B cells (a non-malignant human bronchial epithelial cell line, denoted “B–C” for “BEAS2B-control”) and primary human bronchial epithelial cells, (denoted “P-C” for “primary-control”) to an aqueous extract of cigarette smoke (Lemjabbar-Alaoui et al., 2006). B-ST and P-ST (“ST” denotes “smoke-transformed’) are immortalized, exhibit a transformed phenotype in culture, and form tumors when injected into nude mice (Lemjabbar-Alaoui et al., 2006). The tumors have features of squamous cell carcinoma (Supplementary Fig. 2), and thus B-ST and P-ST cells can be classified as NSCLC lines. By RT-PCR (Fig. 1d) and qPCR (Supplementary Fig. 1b), B-ST and P-ST cells strongly expressed SULF2, whereas the parental cells (B–C and P-C) had minimal levels. In addition, 3/16 of the conventional NSCLC cell lines were positive for SULF1 and 5/16 were positive for SULF2. No line showed simultaneous expression of both SULFs. Among the SULF+ cell lines, four could be classified as squamous cell carcinomas, five as adenocarcinomas, and one as a large cell carcinoma (Supplementary Table 1). There was no statistically-significant correlation between SULF expression and whether or not the cell lines harbored oncogenic mutations in KRAS or EGFR (Supplementary Table 1).
We concentrated on Sulf-2 for mechanistic studies because of its more frequent expression in the lines (7/18). We confirmed that all seven SULF2+ lines (including B-ST and P-ST) were strongly positive for Sulf-2 protein in both conditioned media (75 kDa) and detergent lysates (125 kDa) (Fig. 1e). The 125 and 75 kDa components correspond respectively to the unprocessed pro-protein and the amino-terminal subunit of Sulf-2 (Morimoto-Tomita et al., 2002; Tang and Rosen, 2009). B–C and P-C cells showed no detectable signal. Thus, smoke-induced transformation is accompanied by a dramatic induction of Sulf-2.
We performed immunocytochemistry for Sulf-2 in NSCLC tumors (ten cases each of squamous cell carcinoma and adenocarcinoma) selected from archived surgical specimens (Supplementary Table 2). Normal lung tissue from the same specimens served as controls. Paraffin sections were stained with a newly developed Sulf-2 mAb (Supplementary Fig. 3). All ten cases of squamous cell carcinoma showed some degree of staining for Sulf-2 in the tumor cells. The staining was patchy and accentuated in the cells at the periphery of nests (Fig. 2). The percentage of positive cells was variable and the intensity of staining ranged from faint to intense (Supplementary Table 2). No correlation was found between tumor differentiation and the percentage of positive cells or intensity of staining. In striking contrast, there was no staining of adenocarcinoma cells in any of the cases. This was surprising since three of the NSCLC cell lines with SULF2 expression (Calu-3, Calu-6, and A549) could be classified as adenocarcinomas (Supplementary Table 1). However, both adenocarcinomas and squamous cell carcinomas showed conspicuous staining of tumor stroma, including spindle-shaped cells and endothelial cells of blood vessels. In control tissue distant from the tumors, endothelial cell staining was also seen (Fig. 2b), but normal airway epithelium, except for rare basal cells, was negative for Sulf-2.
We employed a previously developed lentiviral shRNA strategy (Nawroth et al., 2007) to achieve knockdown of Sulf-2 in NSCLC lines. We transduced five of the Sulf-2+ lines (H460, H292, Calu-6, P-ST and B-ST cells) with either a specific shRNA (PLV-1413 or PLV-1143) (Nawroth et al., 2007) or a control shRNA (PLV-Ctrl) containing an irrelevant sequence. Sulf-2 protein expression was strongly reduced (>90%) in all of the lines after transduction with PLV-1413 or PLV-1143, as compared to the control cells (Fig. 3a). To determine the effects of Sulf-2 knockdown on the sulfation status of HSPGs, we used a phage-display antibody (RB4CD12), whose binding to HSPGs depends on 6S (Dennissen et al., 2002). We have verified that this antibody can be used as a cell surface-staining reagent to report the presence of the Sulfs within cells (M. Hossain, T. Hosono, R. Tang, T. H. van Kuppevelt, G. J. Jenniskens, S. D. Rosen, and K. Uchimura, manuscript submitted). Transduction of Sulf-2+ lines with PLV-1413 increased the RB4CD12 epitope on the cell surface relative to that of control-treated cells by 30–37% (Fig. 3b). The same treatment had no effect on the epitope in a Sulf-2− line (H1975).
Whereas no difference in the growth was observed between the PLV-Ctrl transduced cells and the parental cells, both SULF2-specific shRNAs induced a marked reduction in the 5 Sulf-2+ lines (46–50 % reduction) (Fig. 3c). PLV-1413 transduction had no effect on H1975 cells compared to controls. Mutation of two cysteines in the catalytic domain of HSulf-2 (S2ΔCC) totally abrogates its sulfatase activity (Morimoto-Tomita et al., 2002) and results in a dominant negative form (Nawroth et al., 2007). Transfection of S2ΔCC into the five Sulf-2+ lines decreased cell growth but had no effect on H1975 cells (Supplementary Fig. 4).
Transduction of cells with either of the two Sulf-2 specific shRNAs inhibited cell proliferation, assayed by measuring BrdU incorporation, by 44 to 50% in all five Sulf2+ lines, when compared to controls (Fig. 3d). Sulf-2 knockdown increased apoptosis (annexin V staining) by 2–3 fold in the Sulf-2+ lines (Fig. 3e). In both cases, there were no effects on H1975 cells. These results indicate that Sulf-2 promotes cell growth in lung cancer cell lines by increasing both proliferation and cell survival.
Sulf-2 knockdown also induced 40–60% increased cell-substratum adhesion to fibronectin or collagen (Supplementary Fig. 5a), and 15–45% decreased serum-induced cell migration (Supplementary Fig. 5b). Correlated with this increased adhesion, the mobility of PLV-1413 transduced cells (H292 and P-ST), measured after “wounding” cell monolayers, was reduced by 37–50% (Fig. 3f). There was no difference in wound closure between PLV-1413-transduced and control H1975 cells (not shown).
Anchorage-independent cell growth is one of the hallmarks of transformed cells and is exhibited by the NSCLC lines and the smoke-transformed lines studied above. We therefore investigated the effect of Sulf-2 knockdown on the ability of the cells to grow in soft agar. Sulf-2 knockdown in the Sulf2+ lines reduced the number of colonies formed in soft agar by 70–82% (Fig. 3g), as compared to controls. In contrast, transduction of PLV-1413 had no effect on colony formation by H1975 cells.
These loss-of-function findings suggested that Sulf-2 may have a transforming capability in lung epithelial cells. To examine the functional consequences of Sulf-2 expression in lung cells, we overexpressed either active or inactive Sulf-2 (S2ΔCC) in two non-malignant human bronchial epithelial cell lines (BEAS2B and 16HBE14o-), both of which are negative for the Sulfs. Transduction resulted in Sulf-2 expression levels comparable to those found in high-expressing lung cancer cell lines (Fig. 4a).
Sulf-2 expression in both BEAS2B and 16HBE14o- cell lines resulted in decreased staining by RB4CD12, indicating that the enzyme had the expected activity on cell surface HSPGs (Fig. 4b). However, S2ΔCC expression in these cells had no effect on this epitope, confirming the inactivity of this Sulf-2 mutant. Cells expressing S2ΔCC retained the flattened shape of the parental cells, whereas cells with active Sulf-2 (Supplementary Fig. 6a) were more rounded. In keeping with these morphological changes, the cells with active Sulf-2 exhibited decreased cell-substrate adhesion by 40–60% (Supplementary Fig. 6b). Furthermore, in both the wound repair assay (Fig. 4c) and chemotaxis assay (Supplementary Fig. 6c), the migration of active Sulf-2 expressing cells was markedly enhanced relative to that of cells with S2ΔCC or parental cells. With respect to cell growth, active Sulf-2, but not S2ΔCC, induced a modest (25–28%) but statistically significant increase (Fig. 4d). As for anchorage-independent growth, BEAS2B and 16HBE14o- cells, expressing Sulf-2, formed colonies in soft agar after 18 days, whereas neither parental cells nor S2ΔCC-transduced cells formed any colonies (Fig. 4e). We established 9 clones of Sulf-2-expressing BEAS2B cells from colonies in soft agar, all of which showed increased cell proliferation (not shown), cell survival (Fig. 4f), and anchorage-independent growth upon re-testing (not shown). Thus, over-expression of Sulf-2 in lung epithelial cells induced phenotypic characteristics of transformed cells.
To examine the contribution of Sulf-2 to the tumorigenicity of the NSCLC cell lines, we injected control-transduced or Sulf-2 knockdown cells subcutaneously into nude mice and monitored tumor growth. Tumors derived from H460, H292, Calu-6, P-ST and B-ST cells with knockdown grew at markedly slower rates than those from the control cells (60–85% reduction) (Fig. 5). Furthermore, the harvested tumors were greatly reduced in weight (Supplementary Fig. 7), consistent with an observed decrease in cell proliferation and an increase in apoptosis within the tumors (Supplementary Figs. 8a and 8b). Transduction of H1975 cells with SULF2 shRNA had no effect on tumorigenicity (Fig. 5).
To examine the effect of Sulf-2 overexpression on non-malignant bronchial epithelial cells, we injected BEAS2B or 16HBE14o- cells transduced with Sulf-2 into nude mice. Even after 2 months, no tumors formed from either cell type with or without Sulf-2 (not shown). Thus, although Sulf-2 overexpression in bronchial epithelial cells induced many features of malignant transformation in culture, the introduction of Sulf-2 alone did not result in full transformation to a tumorigenic state.
Previous work has implicated the Sulfs as positive regulators of Wnt signaling in development (Dhoot et al., 2001; Freeman et al., 2008) and in pancreatic cancer cell lines (Nawroth et al., 2007). These enhancing effects were demonstrated for canonical Wnt signaling, which involves the translocation of β-catenin to the nucleus and its participation in a transcriptional complex (Clevers, 2006; Logan and Nusse, 2004). The analysis of Emerson and colleagues (Ai et al., 2003) strongly supports a model whereby HSPGs sequester Wnt ligands and the Sulf acts to reduce the affinity of HSPG for Wnt, freeing or altering Wnt association with HSPGs, and thus allowing Wnt access to its signal-transduction receptors on the cell surface. To determine whether the five Sulf-2+ lines exhibited canonical Wnt signaling, we used the TOP/FOP flash assay (Clevers, 2006), which quantifies β-catenin-dependent transcriptional activity. All five cell lines exhibited TOP/FOP flash activity, with a range from 4 to 13 (Supplementary Fig. 9a). To check whether the signaling was autocrine in nature (i.e., triggered by extracellular Wnts), we employed sFRP and WIF-1, which are soluble extracellular inhibitors of Wnt signaling (Kawano and Kypta, 2003). When the five cell lines were transfected with sFRP or WIF cDNA, TOP/FOP flash activity was inhibited by 50–60% (Supplementary Fig. 9b). Consistent with these results, the addition of either sFRP1 or WIF protein to the culture medium of the cells reduced TOP/FOP flash activity to a comparable extent (Fig. 6a). Furthermore, the inhibitors markedly reduced cell growth by 30–50% (Fig. 6b). We next asked whether Wnt signaling was regulated by Sulf-2. Knockdown of Sulf-2 in the five cell lines resulted in a 45–50% inhibition of TOP/FOP flash activity (Fig. 6c). In addition, Sulf-2 knockdown cells demonstrated a favored plasma membrane localization for β-catenin, whereas a nuclear localization was favored in the control cells (Fig. 6d, Supplementary Fig. 9c). Also, Sulf-2 knockdown (H292 cells) led to a 3–4 fold reduction in the expression of several Wnt target genes (Cyclin D1, Cyclin D2, Cyclin D3, JUN, and MYC), which are involved in cell growth (http://www.stanford.edu/~rnusse/wntwindow.html) (Supplementary Table 3). Importantly, the addition sFRP1 or WIF-1 to the culture medium of knockdown cells produced no further inhibition of Wnt signaling (Supplementary Fig. 9d) or cell growth (Fig. 6e), consistent with the model in which Sulf-2 directly regulates the mobilization of Wnt ligands from HSPG sequestration (Ai et al., 2003). It is conceivable that Sulf-2 knockdown could also have indirect effects on Wnt signaling as there was a ≈2-fold reduction in the expression of two WNT genes among the six WNTs expressed in H292 cells (Supplementary Table 3).
Complementing the knockdown approach, we found that overexpression of the dominant negative S2ΔCC in the five cell lines inhibited Wnt activity by 40–60%, while overexpression of wild-type Sulf-2 produced no augmentation of Wnt signaling (Fig. 6f). In contrast, forced expression of Sulf-2 in cells that lacked it (BEAS2B and 16HBE14o-) increased TOP/FOP flash activity 4–5 fold from a baseline value of 1 (Fig. 6g).
The early view in the literature was that both Sulfs were tumor suppressors (Dai et al., 2005; Lai et al., 2003; Lai et al., 2004; Li et al., 2005). This belief originated from experiments in which forced expression of a Sulf in several tumor lines caused reduced growth-factor signaling by HB-EGF, FGF-2 or HGF, and diminished tumorigenicity. The negative effects of Sulfs on FGF-2 signaling are consistent with the requirement for 6S on HSPGs in the FGF-2 signaling complex (Gallagher, 2001). The ability of Sulf-1 to inhibit certain growth factor responses may explain why SULF1 expression is reduced in subsets of tumors (ovarian, hepatocellular) (Lai et al., 2003; Lai et al., 2004) and in certain developmental contexts (Freeman et al., 2008; Langsdorf et al., 2007; Wang et al., 2004). However, consistent with a potential oncogenic role, one or both SULF genes are over-expressed in subsets of multiple tumors (breast, pancreatic, hepatocellular carcinoma, head and neck, and lung) (Castro et al., 2008; Grigoriadis et al., 2006; Kudo et al., 2006; Lai et al., 2004; Lai et al., 2008; Morimoto-Tomita et al., 2005; Nawroth et al., 2007). A strong case for a positive contribution has emerged for Sulf-2 in particular. Not only is SULF2 overexpressed in hepatocellular carcinoma, but high expression is associated with a worse prognosis (Lai et al., 2008). Unbiased screening studies have identified SULF2/sulf2 as a candidate cancer-causing gene in human breast cancer and mouse brain cancer (Johansson et al., 2004; Sjoblom et al., 2006). Finally, knockdown of Sulf-2 in pancreatic adenocarcinoma cell lines (Nawroth et al., 2007) and hepatocellular cell lines (Lai et al., 2008) reverts their transformed phenotypes in culture and impedes their tumorigenicity in nude mice. The present study is the first to investigate Sulf-2 in NSCLC. The expression of Sulf-2 in NSCLC tumors, taken together with the loss-of-function experiments in NSCLC cell lines and gain-of-function experiments in non-malignant bronchial epithelial cells, constitute the strongest evidence to date for a positive contribution in malignancy. Notably, two of the NSCLC lines studied represent novel models of tobacco smoke-induced carcinogenesis, which is responsible for 85% of lung cancer (Sato et al., 2007). Our findings provide impetus for the further investigation of Sulf-2 in NSCLC and other cancers in which it is over-expressed.
By monitoring the RB4CD12 epitope, we confirmed that the transforming activity of Sulf-2 depends on its activity as an HSPG endosulfatase. Previous studies have shown that HSPGs facilitate the progression of tumors by promoting their growth through direct effects on tumor cells (Aikawa et al., 2008; Alexander et al., 2000; Capurro et al., 2005) or indirect effects on the tumor microenvironment (Aikawa et al., 2008; (Fuster and Esko, 2005). The upregulated Sulf-2 expression in the stroma of NSCLC tumors (in particular adenocarcinomas) may be involved in carcinogenesis mechanisms of the latter type, for example in promoting angiogenesis or metastasis (Aikawa et al., 2008; Fuster and Esko, 2005). The extensive heparanase literature further highlights the importance of HSPGs in cancer (Parish et al., 2001; Vlodavsky et al., 2007). This endoglycosidase cleaves HS chains at a few sites and produces large fragments that retain growth factor-binding activity. Heparanase is upregulated in many tumors and promotes tumor growth and metastasis (Vlodavsky et al., 2007). Like heparanase, the Sulfs provide a postsynthetic mechanism to edit HSPGs, but do so without cleaving its chains.
The Sulfs positively modulate several signaling pathways including Wnt, BMP, Hedgehog, and GDNF (Ai et al., 2003; Ai et al., 2007; Danesin et al., 2006; Dhoot et al., 2001; Nawroth et al., 2007; Viviano et al., 2004). Of these, promotion of canonical Wnt signaling is the best validated (Dhoot et al., 2001; Freeman et al., 2008; Nawroth et al., 2007; Tang and Rosen, 2009). Aberrant activation of Wnt signaling is an important mechanism underlying the pathogenesis of many tumors (Clevers, 2006; Klaus and Birchmeier, 2008; Logan and Nusse, 2004; Miller et al., 1999). In some cases (e.g., colorectal cancers), the pathway is activated by mutations in intracellular signaling intermediates such as APC, axin or β-catenin (Clevers, 2006). Wnt signaling can also be activated by altered receptor-ligand interactions at the level of the cell surface, involving changes in expression of Wnts, soluble extracellular Wnt antagonists, or Wnt receptors/co-receptors (Hoang et al., 2004; Huguet et al., 1994; Janssens et al., 2004; Suzuki et al., 2004). Tumors and cell lines driven by this type of mechanism (“autocrine Wnt signaling”) are subject to inhibition by extracellular administration of Wnt antagonists (Bafico et al., 2004; DeAlmeida et al., 2007; Derksen et al., 2004; Groen et al., 2008; Kim et al., 2007; Mikami et al., 2005; Schlange et al., 2007; You et al., 2004).
In NSCLC, there is emerging evidence for aberrant Wnt signaling, which is based on altered receptor/ligand interactions or membrane proximal events rather than mutations in downstream components (Furuuchi et al., 2000; (Ohgaki et al., 2004; Sunaga et al., 2001; Winn et al., 2006). Thus, promoters of autocrine Wnt signaling (Wnt1, Wnt2, disheveled) are upregulated (Huang et al., 2008; Nakashima et al., 2008; Uematsu et al., 2003; You et al., 2004) and conversely, WIF-1 is down-regulated in NSCLC tumors (Mazieres et al., 2004). Studies with NSCLC cell lines corroborate the importance of these dysregulated Wnt elements in promoting cell growth (Kim et al., 2007). Strikingly, all five Sulf-2+ NSCLC lines studied exhibited autocrine Wnt signaling, which is compatible with the involvement of Sulf-2 in regulating the bioavailability of extracellular Wnt ligands through its endosulfatase action on HSPGs. Consistent with this role for Sulf-2, knockdown of Sulf-2 or the introduction of a dominant negative form of Sulf-2 in these lines caused pronounced inhibition of canonical Wnt signaling, while conversely the forced expression of Sulf-2 in normal bronchial epithelial lines induced Wnt signaling. Furthermore, inhibition of autocrine Wnt signaling or knockdown of Sulf-2 in the NSCLC lines produced equivalent effects on cell growth, which were not augmented by combining both treatments. These findings not only corroborate the importance of autocrine Wnt signaling in the growth of NSCLC lines (Kim et al., 2007) but also establish that Sulf-2 is required for this autocrine pathway.
Lung cancer is a heterogeneous disease involving dysregulation of a number of signaling pathways, most frequently involving EGFR and KRAS (Osada and Takahashi, 2002; Sato et al., 2007). The relationship of Sulf-2 to these pathways remains to be determined. Intriguingly, Hynes and colleagues have demonstrated that autocrine Wnt signaling can transactivate EGFR in breast cancer cell lines (Schlange et al., 2007). Although further validation is required, the results presented above raise the possibility that Sulf-2 could be a therapeutic target in NSCLC. Sulf-2 is an extracellular enzyme and thus is potentially amenable to inhibition by either antibody-based or small molecule drugs. Since Sulf-2 is secreted and may enter body fluids, it also merits consideration as a potential biomarker of lung cancer.
Descriptions are provided in Supplementary Materials
For analyzing SULF expression in published gene microarray studies, we used the oncomine database (www.oncomine.org) and the Consortium for Functional Glycomics data of Dr. Rittenhouse-Olson (http://www.functionalglycomics.org/glycomics/publicdata/microarray.jsp, #887). For analyzing SULF expression by PCR and RT-PCR, methodology is provided in Supplementary Materials.
The peptide-based polyclonal antibody against HSulf-2 was produced as described (Morimoto-Tomita et al., 2005). Monoclonal antibodies were produced by ProSci Inc. A sulf2 null mouse (Lum et al., 2006) (provided by Drs. Joanna Phillips and Zena Werb) was immunized with recombinant HSulf-2 protein (15 micrograms over three injections). 2B4, one of the resulting mAbs, was used. Its characterization is presented in Supplementary Fig. 3.
Under UCSF CHR approval (H1060-27616-01), archived paraffin-embedded tissue blocks were obtained for ten cases each of squamous lung carcinoma and adenocarcinoma. 4 μm sections were subjected to heat-induced epitope retrieval with Trilogy (Cell Marque) and stained with 2 μg/ml of 2B4 for 2 hours. A mAb directed to Herpes Simplex virus I/II served as a control. Antibody binding was visualized by the Dako Envision (+) system (cat #K4011).
Methodology is provided in Supplementary Materials
H460, H292, A549, Calu-6, Calu-3, and BEAS2B cells were from American Type Culture Collection. B-ST and P-ST are described in Results 16HBE14o- cells were from Dr. Keith Mostov (UCSF). All cells were grown in RPMI-1640 with 10% FBS and penicillin/streptomycin. HEK 293 cells were maintained in DMEM with 10% FBS and penicillin/streptomycin.
To determine the effects of Sulf-2 knockdown on HSPGs, we used a phage-display antibody (RB4CD12) (Dennissen et al., 2002) with an irrelevant antibody as a control. A Cy3-conjugated mouse anti-VSV secondary antibody (Sigma) was used and cells were analyzed by flow cytometry.
Methodology is provided in Supplementary Materials.
Methodology is provided in Supplementary Materials.
Cells were grown to confluency on 6-well plates. Scratches were made with a pipette tip. Cell migration was quantified by measuring the width of the wounds at 0, 24, 48 and 72 hours.
Methodology is provided in Supplementary Materials.
Methodology is provided in Supplementary Materials.
5×106 cells, transduced with either control or Sulf-2 shRNA were injected into the flanks of 5-week-old athymic male BALB/c nu/nu mice (5 mice/group). Tumor growth was monitored every 2–3 days. Groups of animals were sacrificed when the largest tumor nodules attained 0.8 cm (after 25 – 40 days). Tumors were harvested, weighed and processed for histology. All procedures had UCSF IACUC approval. Some of the mice were injected intraperitoneally with BrdU 2 hrs prior to sacrifice. Tissue sections were immunostained with a BrdU antibody (BD Biosciences) or with the active caspase 3 antibody. Tissue sections were incubated with the corresponding peroxidase-conjugated secondary antibody with visualization with Nova Red and counterstained with hematoxylin. The numbers of BrdU+ or active caspase3+ cells were counted in four high power field (HPF) for each tissue section with 3 sections per tumor.
All numerical data were calculated as means ± standard deviations or ± SEMs. Differences between groups were compared with a Student’s t–test. Paired Tumor/Control samples were compared with a paired two-tailed t-test. A p value ≥ 0.05 was considered significant.
RB4CD12 was a kind gift of Drs. Guido Jenniskens and Toin van Kuppevelt of Nijmegen Medical Center. We thank Dr. Roman Nawroth for preliminary work with lung cancer cell lines. We thank Drs. Zena Werb, Joanna Phillips, and Inna Maltseva for their comments on the manuscript. The work was performed with support from the Tobacco-Related Disease Research Program of the University of California Grant 17RT-0216, NIH P01 AI053194, and NIH R21 CA122025, Susan Komen Breast Cancer Foundation PDF0402844 to SDR; NIH RO1 HL075602 to Dr. Zena Werb; The Larry Hall and Zygielbaum Memorial Trust, and The Kazan, McClain, Edises, Abrams, Fernandez, Lyons & Farrise Foundation to DMJ; and NIH R01CA125030 to BH.
Supplementary information is available at the journals website.