Purpose: Angiogenesis and vascular endothelial growth factor (VEGF) expression are associated with a poor outcome in bladder cancer. To understand more about the mechanisms, we studied the role of delta-like 4 (DLL4), an endothelial-specific ligand of the Notch signaling pathway, in bladder cancer angiogenesis.

Experimental Design: The expression of DLL4, CD34, and VEGF were studied in a cohort of 60 bladder tumors and 10 normal samples using quantitative PCR. In situ hybridization was used to study the pattern of DLL4 expression in 22 tumor and 9 normal samples. Serial sections were also stained for CD34 and α-smooth muscle actin (α-SMA) using conventional immunohistochemistry.

Results: The expression of DLL4 was significantly up-regulated in superficial (P < 0.01) and invasive (P < 0.05) bladder cancers. DLL4 expression significantly correlated with CD34 (P < 0.001) and VEGF (P < 0.001) expression. The in situ hybridization studies showed that DLL4 was highly expressed within bladder tumor vasculature. Additionally, DLL4 expression significantly correlated with vessel maturation as judged by periendothelial cell expression of α-SMA, 98.7% of DLL4-positive tumor vessels coexpressed α-SMA, compared with 64.5% of DLL4-negative tumor vessels (P < 0.001). High DLL4 expression may have prognostic value in superficial and invasive bladder.

Conclusion: DLL4 expression is associated with vascular differentiation in bladder cancer; thus, targeting DLL4 may be a novel antiangiogenic therapy.

Bladder cancer results in over 13,000 deaths per year in the United States alone. Despite advances in the management of bladder cancer, the need for new treatment modalities remains. As with most solid tumors, bladder cancer growth and metastatic progression is dependent on the acquisition of an adequate blood supply through the process of angiogenesis (1). The role of angiogenesis in bladder cancer has been extensively studied ever since Chodak et al. (2) showed that urine from patients with bladder cancer contained proangiogenic growth factors (3).

The Notch signaling pathway has recently been shown to play an important role in angiogenesis (46). It is an evolutionarily conserved signaling pathway involved in cell fate determination, cellular differentiation, proliferation, survival, and apoptosis (79). In mammalian cells, it comprises five transmembrane Notch ligands (Jagged1, Jagged2, DLL1, DLL3, and DLL4) and four Notch receptors (Notch1-4). Ligand receptor binding leads to the cleavage and subsequent translocation from the cell membrane to the nucleus of the Notch intracellular domain (10, 11). In the nucleus, the Notch intracellular domain interacts with the transcription factor CSL to regulate the transcription of the basic helix-loop-helix proteins hairy/enhancer of split (HES) and HES-related protein (HEY; refs. 1214).

Delta-like 4 (DLL4) is an endothelial-specific ligand expressed at sites of vascular development and angiogenesis (15, 16). DLL4 expression has previously been shown to be up-regulated within the vasculature of breast and renal tumors. In utero DLL4 knockout mice die of severe vascular defects, and, interestingly, haploinsufficiency of DLL4 also results in embryonic lethality from severe vascular defects (15, 17, 18). Similar haploinsufficiency phenotypes for angiogenic pathways have only been previously described for vascular endothelial growth factor (VEGF) knockout mice (19, 20). In human endothelial cells, we have recently shown that DLL4 plays an important role in regulating endothelial cell proliferation, migration, survival, and network formation (21).

In this study, we have investigated the expression of DLL4 in transitional cell carcinoma of the bladder, studied the role of DLL4 in vessel maturation, and carried out an initial evaluation of the role of DLL4 in prognosis.

Tissue samples. Human bladder tissue was obtained postoperatively from the Department of Urology, Churchill Hospital, Oxford, United Kingdom. All patients gave signed, informed consent for their tissues to be used for scientific research. Ethical approval for the study was obtained from the Clinical Ethics Committee, Oxford Radcliffe Hospitals NHS Trust, Oxford, United Kingdom.

Tumor samples were obtained following either transurethral resection (TURBT) or radical cystectomy. Bladder cancers were staged using the tumor-node-metastasis classification (22). Tumor samples were divided into three groups depending on their stage: Ta tumors, T1 superficial tumors (Ta and T1 tumors were collectively designated as superficial tumors), and T2-4 invasive tumors. For the quantitative PCR (qPCR) analysis, a cohort of 10 normal bladder samples and 60 bladder tumors were studied. Areas of normal bladder and tumor were identified at cytoscopy, excised, and snap frozen in liquid nitrogen. For the in situ hybridization/immunohistochemistry vessel counting study, the series comprised six Ta (TURBT), six T1 (TURBT), six T2-4 (TURBT), and four T2-4 (cystectomy) samples. Truly normal full thickness bladder sections were difficult to source, so microscopically normal areas within radical cystectomy specimens (four samples) and normal ureters (five samples) obtained following organ donation were used to represent “normal tissue.”

RNA extraction and reverse transcription. Total RNA from frozen tissue samples was extracted after morselizing in liquid nitrogen and homogenizing in TRI-Reagent (Sigma-Aldrich, St. Louis, MO). The quality of RNA extracted was assessed using RNA 6000 Nano Chips and the Agilent 2100 Bio-analyzer (Agilent Technologies, Paulo Alto, CA). cDNA was synthesized by reverse transcribing total RNA using the High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA).

Real-time qPCR. Real-time qPCR and analysis was done using the method previously described by Patel et al. (21). Human β-actin was used as a reference gene to normalize for differences in the amount of total RNA in each sample. Primer/probe kits were purchased as Assays-on-Demand from Applied Biosystems: DLL4, VEGF, and CD34. Relative quantitation of gene expression was done using the method described by Pfaffl (23). The normal bladder sample with the median expression of β-actin was designated the comparator.

In situ hybridization. Radioactive in situ hybridization was done on paraffin-embedded tissue sections by the in situ hybridization service, Cancer Research UK London Research Institute, following a method described by Poulsom et al. (24). The probe used to detect DLL4 was a 741 bp fragment located from position 1,775 to 2,516 bp (16). After 14 days, sections were visualized using a Nikon Eclipse ME600 dark field microscope, an Optronics digital camera, and Magnafire Version 2.0.

Immunohistochemistry. Paraffin-embedded tissue blocks were serially sectioned 4 μm in thickness, dewaxed, and rehydrated in serial alcohol washes. Endogenous peroxidase activity was blocked with 0.03% hydrogen peroxide (20 minutes). Double immunostaining for CD34 and α-smooth muscle actin (α-SMA) was done by sequentially adding CD34 primary antibody 1:200 for 30 minutes (Novocastra, Newcastle-upon-Tyne, United Kingdom, QBEnd/10), DAKO (Carpinteria, CA) Checkmate EnVision Detection kit (peroxidase/3,3′-diaminobenzidine-rabbit/mouse, K5007) for 30 minutes, 3,3′-diaminobenzidine substrate for 2 minutes, α-SMA primary antibody 1:1,000 for 30 minutes (DAKO-1A4), Envision AP Polymer 30 minutes, and Vector Blue substrate for 6 minutes. Sections were mounted using Aquamount.

Analysis of DLL4, CD34, and α-SMA expression in bladder cancer.DLL4 (in situ hybridization) and CD34/α-SMA (immunohistochemistry) expression in serial sections was quantified. First, the distribution of DLL4 expression was assessed by two observers (N.P. and M.D.). Twenty random fields within the DLL4 in situ sections were scanned using a light microscope at a magnification of ×100 and assessed for the presence of any DLL4-positive vessels. Second, tissues were further analyzed by identifying the two to three areas of highest DLL4 expression. These areas were photographed at ×100 magnification using a Nikon Eclipse ME600 dark field microscope, an Optronics digital camera, and Magnafire version 2.0. The equivalent areas were identified upon the serial sections and captured accordingly. The digital images were then used to quantify the vasculature for CD34, α-SMA, and DLL4.

Statistics. Data are presented as mean ± SE. Statistical analysis was done using GraphPad Prism. Tests used included t test, one-way ANOVA, Mann-Whitney and χ2 tests. Statistical significance was concluded when P < 0.05 and is denoted in the figures with an asterisk.

Quantitative analysis of DLL4 expression in bladder cancer using qPCR. Quantitative PCR was used to study gene expression in a cohort of bladder tumors and normal bladder samples (Table 1). Relative gene expression was assessed using the method of Pfaffl (23), a modified method of comparative quantitation. In this technique, the expression of DLL4 was calculated relative to the reference gene β-actin. The expression of β-actin was equivalent between the normals and tumors (β-actin expression normals: median Ct = 23.0, range 22.1-23.8, n = 10; β-actin expression tumors: median Ct = 22.8, range 22.0-23.8, n = 60, P = 0.3).

Table 1.

Patient demographics for the qPCR bladder cancer samples

NormalsTaT1T2T3-4SuperficialsInvasivesTumors
Number 10 11 32 11 43 17 60 
Mean age (y) 76.4 71.0 71.2 69.5 68.3 71.2 69.1 70.57 
Age range (y) 68-87 62-82 42-89 56-83 46-91 42-89 46-91 42-91 
Sex         
    M 10 25 10 34 14 48 
    F 12 
Operation Cystoscopy and biopsy TURBT TURBT TURBT T3 (5) = TURBT; T4 (1) = CYST TURBT TURBT = 16; CYST = 1 TURBT = 59; CYST = 1 
Adjuvant intravesical therapy  Nil = 4; MMC = 2; BCG = 2; both = 1; N/A = 2 Nil = 21; MMC = 2; BCG = 6; both = 1; N/A = 2   Nil = 25; MMC = 4; BCG = 8; both = 2; N/A = 4   
Invasive cancer treatment    DXT = 7; CYST = 2; N/A = 1 DXT = 1; CYST = 2; Nil = 3  DXT = 8; CYST = 4; Nil = 1; N/A = 1  
NormalsTaT1T2T3-4SuperficialsInvasivesTumors
Number 10 11 32 11 43 17 60 
Mean age (y) 76.4 71.0 71.2 69.5 68.3 71.2 69.1 70.57 
Age range (y) 68-87 62-82 42-89 56-83 46-91 42-89 46-91 42-91 
Sex         
    M 10 25 10 34 14 48 
    F 12 
Operation Cystoscopy and biopsy TURBT TURBT TURBT T3 (5) = TURBT; T4 (1) = CYST TURBT TURBT = 16; CYST = 1 TURBT = 59; CYST = 1 
Adjuvant intravesical therapy  Nil = 4; MMC = 2; BCG = 2; both = 1; N/A = 2 Nil = 21; MMC = 2; BCG = 6; both = 1; N/A = 2   Nil = 25; MMC = 4; BCG = 8; both = 2; N/A = 4   
Invasive cancer treatment    DXT = 7; CYST = 2; N/A = 1 DXT = 1; CYST = 2; Nil = 3  DXT = 8; CYST = 4; Nil = 1; N/A = 1  

Abbreviations: CYST, cystectomy; DXT, radiotherapy; BCG, bacille Calmette Guérin; MMC, mitomycin C; Nil, no further treatment; N/A, not applicable.

The relative expression of DLL4 was significantly elevated in bladder tumors compared with normal bladders (mean DLL4 expression in tumors = 4.5 ± 0.4, n = 60; mean expression of DLL4 in normal samples = 2.4 ± 0.2, n = 10, P = 0.01). In comparison to normal tissue, DLL4 expression was noted to be highest in superficial tumors (mean expression = 4.7 ± 0.6, n = 43, P < 0.05) followed by invasive tumors (mean expression = 3.9 ± 0.6, n = 17, P > 0.05; Fig. 1A).

Fig. 1.

Analysis of DLL4, CD34, and VEGF expression using qPCR. A, DLL4 expression was significantly up-regulated in superficial bladder cancer compared with normal bladder tissue. B, the expression of CD34 was significantly higher in the normal tissues than in the bladder tumors. C, the DLL4/CD34 ratio was significantly higher in superficial and invasive bladder cancer compared with normal tissue. D, DLL4 expression was significantly correlated with CD34 expression in bladder tumors (Spearman rank correlation coefficient = 0.6, P < 0.001, n = 60). E, VEGF expression was significantly elevated in both superficial and invasive bladder cancer. F, the expression of DLL4 significantly correlated with VEGF expression (Spearman rank correlation coefficient = 0.4, P = 0.001, n = 60. *, P < 0.05; **, P < 0.01; ***, P < 0.001). Bars, SE.

Fig. 1.

Analysis of DLL4, CD34, and VEGF expression using qPCR. A, DLL4 expression was significantly up-regulated in superficial bladder cancer compared with normal bladder tissue. B, the expression of CD34 was significantly higher in the normal tissues than in the bladder tumors. C, the DLL4/CD34 ratio was significantly higher in superficial and invasive bladder cancer compared with normal tissue. D, DLL4 expression was significantly correlated with CD34 expression in bladder tumors (Spearman rank correlation coefficient = 0.6, P < 0.001, n = 60). E, VEGF expression was significantly elevated in both superficial and invasive bladder cancer. F, the expression of DLL4 significantly correlated with VEGF expression (Spearman rank correlation coefficient = 0.4, P = 0.001, n = 60. *, P < 0.05; **, P < 0.01; ***, P < 0.001). Bars, SE.

Close modal

Quantitative PCR was done for the endothelial marker CD34 to determine if the differences in DLL4 expression were simply due to differences in vascular density between normal and tumor samples. The relative expression of CD34 was highest in the normal samples (mean expression = 1.2 ± 0.2, n = 10) when compared with superficial (mean expression 0.6 ± 0.1, n = 43) and invasive tumors (mean expression = 0.5 ± 0.1, n = 17; Fig. 1B).

A ratio of DLL4 expression to CD34 expression (DLL4/CD34 ratio) was calculated by dividing the relative expression of DLL4 by that of CD34. The mean DLL4/CD34 ratio for normal samples was 3.2 ± 1.2, for superficial tumors was 10.6 ± 1.0 (P < 0.001), and for invasive tumors was 9.1 ± 4.0 (P < 0.001; Fig. 1C). The expression of DLL4 correlated significantly with the expression of CD34 in tumors (Spearman rank correlation coefficients: all tumors = 0.6, n = 60, P < 0.001; superficial tumors = 0.6, n = 47, P < 0.001; invasive tumors = 0.6, n = 13, P < 0.05), but not in normal bladder tissue controls (normal samples = 0.2, P > 0.05; Fig. 1D).

The expression of VEGF was also assessed using qPCR. The mean expression of VEGF in both superficial and invasive tumors was significantly higher than in normal bladder tissue (44.6 ± 5.3 and 45.9 ± 13.6 versus 4.0 ± 1.0; P < 0.001; Fig. 1E). The Spearman rank correlation coefficient of VEGF and DLL4 expression was significant only for superficial tumors (Spearman rank correlation coefficients all tumors = 0.4, P = 0.001, n = 60; superficial tumors = 0.4, P < 0.01, n = 43; invasive tumors = 0.3, P = 0.3, n = 17; normals = 0.4, P > 0.3, n = 10; Fig. 1F).

DLL4 mRNA expression pattern.In situ hybridization was used to assess the pattern of DLL4 mRNA expression in a series of 22 bladder cancers, 4 normal bladders, and 5 normal ureters. The expression of DLL4 was compared with that of the pan-endothelial marker CD34. Like CD34, DLL4 expression was confined to the vascular endothelium, unlike CD34; however, DLL4 expression was only noted within a subset of vessels (Fig. 2A).

Fig. 2.

DLL4, CD34, and α-SMA expression in bladder cancer. A, in situ hybridization was used to study the pattern of DLL4 expression in bladder cancer. Serial sections were also stained for CD34 and α-SMA using conventional double immunohistochemistry. Vessels in the suburothelial layer of normal ureter were predominantly microvessels staining only for CD34 (black arrows). Vessels within superficial and invasive bladder tumors stained positive for CD34 (black arrow) and for α-SMA (blue arrows). DLL4 expression was confined to the vascular endothelium. Endothelial expression of DLL4 is shown in the light (red arrows) and dark field (green arrows) in situ hybridization. Quantitation studies of the expression of DLL4, CD34, and α-SMA were done at ×100 magnification. B, representative fields of view. DLL4 light fields, black autoradiographic silver grains; DLL4 dark field, white reflections; CD34, brown staining; α-SMA, blue staining.

Fig. 2.

DLL4, CD34, and α-SMA expression in bladder cancer. A, in situ hybridization was used to study the pattern of DLL4 expression in bladder cancer. Serial sections were also stained for CD34 and α-SMA using conventional double immunohistochemistry. Vessels in the suburothelial layer of normal ureter were predominantly microvessels staining only for CD34 (black arrows). Vessels within superficial and invasive bladder tumors stained positive for CD34 (black arrow) and for α-SMA (blue arrows). DLL4 expression was confined to the vascular endothelium. Endothelial expression of DLL4 is shown in the light (red arrows) and dark field (green arrows) in situ hybridization. Quantitation studies of the expression of DLL4, CD34, and α-SMA were done at ×100 magnification. B, representative fields of view. DLL4 light fields, black autoradiographic silver grains; DLL4 dark field, white reflections; CD34, brown staining; α-SMA, blue staining.

Close modal

Twenty random fields within each of the 9 normal (180 fields), 12 superficial bladder tumor (240 fields), and 10 invasive bladder tumor (200 fields) sections were scored for the presence or absence of DLL4-positive vessels (Fig. 2B). The presence of a single DLL4-positive vessel was sufficient for a field to be regarded as DLL4 positive (maximum score: 20 of 20 fields, minimum score: 0 of 20). Compared with normal tissue (positive random fields 1.4 ± 0.7), DLL4-positive vessels were most abundant in superficial tumors (positive random fields 14.3 ±1.2, P < 0.001) followed by invasive tumors (positive random fields 8.0 ±1.4, P < 0.05; Fig. 3A). The differences are likely to be greater as eight of the nine normal cases had less than two DLL4-positive random fields, with the remaining normal case having seven DLL4-positive fields. The number of positive cases and their distributions for each tissue group are shown in Fig. 3B.

Fig. 3.

Quantitation of DLL4 in situ hybridization studies. The expression of DLL4 was studied in 20 random fields per case (10 normal, 12 superficial bladder tumors, and 10 invasive bladder tumors) using in situ hybridization. A, a significantly higher number of random fields had DLL4-positive vessels in superficial (P < 0.001) and invasive tumors (P < 0.05) compared with normal tissue. B, the expression of DLL4 was uncommon within the normal vasculature. Eight of nine normal cases had less than two DLL4-positive random fields; this compares to 0 of 12 superficial tumors and 1 of 10 invasive tumors. C, the percentage of DLL4-positive vessels within DLL4 hotspot areas was significantly higher in superficial tumors and invasive tumors compared with both normal bladder and ureteric tissue. Expression of DLL4 was noted to be higher in invasive TURBT specimens than in invasive cystectomy specimens. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Bars, SE.

Fig. 3.

Quantitation of DLL4 in situ hybridization studies. The expression of DLL4 was studied in 20 random fields per case (10 normal, 12 superficial bladder tumors, and 10 invasive bladder tumors) using in situ hybridization. A, a significantly higher number of random fields had DLL4-positive vessels in superficial (P < 0.001) and invasive tumors (P < 0.05) compared with normal tissue. B, the expression of DLL4 was uncommon within the normal vasculature. Eight of nine normal cases had less than two DLL4-positive random fields; this compares to 0 of 12 superficial tumors and 1 of 10 invasive tumors. C, the percentage of DLL4-positive vessels within DLL4 hotspot areas was significantly higher in superficial tumors and invasive tumors compared with both normal bladder and ureteric tissue. Expression of DLL4 was noted to be higher in invasive TURBT specimens than in invasive cystectomy specimens. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Bars, SE.

Close modal

The expression of DLL4 was further studied by identifying DLL4 “hotspots.” For each tumor, the two areas with the highest concentration of DLL4-positive vessels were identified and photographed using a light and dark field microscope. In the case of normal tissue, the three most positive DLL4 areas were photographed; in the absence of DLL4 expression, representative photographs were taken to include both suburothelial microvessels and mature stromal vessels.

The proportion of vessels positive for DLL4 within the hotspots was higher in superficial and invasive tumor samples compared with both normal bladder tissue and normal ureter (mean percentage of positive hotspot vessels: normal bladder = 0.6 ± 0.4, n = 12; normal ureter = 10.5 ± 5.3, n = 15; Ta TURBT = 67.5 ± 5.2, n = 12, P < 0.001, compared with normal ureter and bladder; T1 TURBT = 70.6 ± 8.0, n = 12, P < 0.001, compared with normal ureter and bladder; T2-4 cystectomy = 24.7 ± 6.7, n = 8, compared with normal ureter and bladder; T2-4 TURBT = 70.0 ± 6.8, n = 12, P < 0.001, compared with normal ureter and bladder; Fig. 3C). Within the invasive tumor group, DLL4 expression was noted to be higher in the TURBT specimens than in the cystectomy specimens although this was not statistically significant.

Coexpression of DLL4 and α-SMA. Serial sections of each case were also stained for CD34 and α-SMA using conventional double immunohistochemistry (Fig. 2B). The DLL4 hotspot areas were identified within these serial sections and corresponding matched photographs were taken.

DLL4 hotspots were studied within normal (27 fields, 9 cases) and tumor tissue (44 fields, 22 cases). Similar numbers of vessels were present within the different tissue types (normal bladder = 41.7 ± 8.9, normal ureter 31.5 ± 7.2, Ta TURBT = 52.3 ± 8.0, T1 TURBT = 47.2 ± 9.5, T2-4 cystectomy = 41.8 ± 6.1, T2-4 TURBT = 37.6 ± 4.9).

In the course of the hotspot study, a total of 3,008 vessels were identified as staining positive for CD34. Of these vessels, 2,025 were within tumor sections (22 cases) and 983 in normal tissue sections (9 cases). Taken as a whole, 2,403 (79.9%) hotspot vessels showed periendothelial expression of α-SMA, whereas 1,226 (40.8%) hotspot vessels expressed DLL4 and 589 (19.6%) vessels expressed neither (n = 31 cases). Further evaluation showed that 1,210 of 1,226 (98.7%) of DLL4 positive vessels were also positive for α-SMA; in comparison, only 1,193 of 1,782 (66.9%) of DLL4-negative vessels coexpressed α-SMA (n = 31 cases; P < 0.001, χ2 test; Fig. 4A).

Fig. 4.

DLL4 and vessel maturation. DLL4 and α-SMA expression was analyzed by creating χ2 contingency tables. DLL4 expression was significantly associated with α-SMA expression (P < 0.001).

Fig. 4.

DLL4 and vessel maturation. DLL4 and α-SMA expression was analyzed by creating χ2 contingency tables. DLL4 expression was significantly associated with α-SMA expression (P < 0.001).

Close modal

These data were then analyzed after subclassifying into normal and tumor vasculature (9 normal, 22 tumor cases). Within the normal tissue, 687 of 983 (69.9%) vessels were positive for α-SMA. Of the normal hotspot vessels, 26 of 983 (2.6%) were DLL4 positive, 100% of which coexpressed α-SMA. Nine hundred fifty-seven of 983 vessels were DLL4 negative of which 661 (69.1%) also expressed perivascular α-SMA (P = 0.001, χ2 test; Fig. 4B).

Analysis of the tumor vasculature showed 1,716 of 2,025 (84.7%) vessels were α-SMA positive and 1,200 of 2,025 (59.3%) vessels were DLL4 positive. Of the 1,200 DLL4-positive vessels, 1,184 (98.7%) also expressed α-SMA, whereas only 532 of 825 (64.5%) DLL4-negative tumor vessels stained positive for α-SMA (P < 0.001, χ2 test; Fig. 4C). These results indicate that DLL4 is significantly up-regulated and colocalized together with α-SMA in the tumor vasculature.

Relation of DLL4 expression to outcome. Follow-up was available for 39 of the 43 superficial tumors (91%) and 16 of the 17 invasive tumors (94%). The mean duration of follow-up for the superficial cases was 12 months (range 1-36 months, median = 6) and for the invasive cases was 33 months (range 5-84 months, median = 23.5).

In the case of superficial cancers, the disease-free period was measured by determining the time to first recurrence (Fig. 5A). Kaplan-Meier curves were used to show that the median time to recurrence for patients in whom the expression of DLL4 was above the median was 6 months, whereas that for patients with low levels of DLL4 was 18 months, but this was not statistically significant.

Fig. 5.

DLL4 as a prognostic markers. High DLL4 expression was associated with an earlier median time to recurrence in superficial bladder cancer (A; P > 0.05) and earlier median time to cancer-specific death in invasive cancer (B; P > 0.05).

Fig. 5.

DLL4 as a prognostic markers. High DLL4 expression was associated with an earlier median time to recurrence in superficial bladder cancer (A; P > 0.05) and earlier median time to cancer-specific death in invasive cancer (B; P > 0.05).

Close modal

Superficial bladder cancer is often treated with adjuvant intravesical chemotharpy. The tumor samples used in this study were collected when the administration of mitomycin C immediately after TURBT was not routine practice. Of the 39 patients with superficial tumors for whom follow-up was available, 14 subsequently received adjuvant courses of intravesical therapy; 4 received mitomycin C, 8 bacille Calmette Guérin, and 2 both. Of these 14 patients, 10 were noted to have high levels of DLL4 expression.

Kaplan-Meier curves were used to assess the time to cancer-specific death in the invasive cancers (Fig. 5B). Of the 16 patients with available follow-up, 8 received radical radiotherapy (high DLL4 = 4, low DLL4 = 4), 4 underwent radical cystectomies (high DLL4 = 3, low DLL4 = 1), and 4 were treated palliatively (high DLL4 =1, low DLL4 = 3). Patients with high DLL4 levels had a median survival of 7.5 months compared with 44.5 months for those with low DLL4 levels (hazard ratio = 2.40). Because of the small numbers in this study, no statistical significance was shown upon comparison of the curves.

The aim of this study was to evaluate DLL4 as a potential novel antiangiogenic target in bladder cancer. Using quantitative PCR, we showed that DLL4 expression is up-regulated in transitional cell carcinoma of the bladder. Additionally, quantitation of CD34 expression was done using qPCR as a means of assessing sample vascular density. Interestingly, the expression of CD34 was higher in the normal samples than in the tumor samples. These normal bladder samples were obtained as cold cup biopsies at cytoscopy, and so contained a high number of suburothelial microvessels. The presence of these vessels within the normal tissues is the most likely explanation for the high CD34 expression levels seen in the normal samples.

In keeping with previous work (25), we showed that VEGF mRNA expression was up-regulated in bladder cancer. In our study, the expression of DLL4 also significantly correlated with the expression of CD34 and VEGF in bladder cancers. These results matched our previous findings in clear-cell renal cell carcinoma that DLL4 expression correlated with VEGF expression (21) and also supported in vitro studies that have shown the regulation of DLL4 in endothelial cells by VEGF (21, 26).

In situ hybridization was used for the first time to show that DLL4 expression is up-regulated in bladder cancer and has an endothelial cell–specific pattern of expression. Quantitation of DLL4 expression was done by assessing expression within both random fields and DLL4 hotspots. DLL4 expression within random fields was more widespread within bladder cancer samples than normal tissues. Further subclassification of the tumors showed that DLL4 expression was most widespread within the superficial cases. O'Brien et al. (27) reported that tumor vessel formation in superficial and invasive bladder cancers was driven by differing angiogenic pathways. Angiogenesis in papillary-type superficial bladder tumors is associated more strongly with VEGF and results in characteristic mature, well-defined, functional fibrovascular cores that do not penetrate the underlying muscle (28). In invasive cancers, new vessel formation can also arise as a result of interactions between tumor cells and native vessels within the muscularis layer. The production of metalloproteases disrupts the extracellular matrix and liberates growth factors like basic fibroblast growth factor (29). Vessels formed under the influence of these varying angiogenic pathways may show differences in their characteristics.

It is interesting to note that DLL4 expression was highest in the TURBT samples whether superficial or invasive. The majority of tumor tissue obtained at TURBT is from exophytic tumors. The vessels within these areas of tumor are neoangiogenic tumor vessels, and most likely arise from the suburothelial microvessels. The vasculature within the cystectomy specimens, however, is composed of a combination of neoangiogenic vessels and preexisting bladder vessels. The difference in DLL4 expression between the cystectomy and TURBT specimens suggests that DLL4 expression is enhanced in neoangiogenic vessels, of which there may be more in the TURBT samples.

Angiogenic vessels mature upon the recruitment of pericytes (30). To understand if DLL4 played a role in vessel maturation in human cancer, serial sections were immunostained for CD34 and α-SMA, and an assessment was made of vessel maturation within the DLL4 hotspot areas. This is the first time that vessel maturity has been assessed in bladder cancer. α-SMA expression was found on nearly 70% of normal vessels and 85% of tumor vessels studied. The vessels that lacked α-SMA expression were most often microvascular in nature.

Vessel maturation has previously been studied in a number of different tumor types. Eberhard et al. (31) showed that the degree of vessel maturation was highly variable between different tumors, in breast cancer nearly 70% of vessels were α-SMA positive, whereas only 10% to 20% of glioblastoma and renal cancer vessels were α-SMA positive. More recently, confocal microscopy was used by Morikawa et al. (32) to show that >97% of pancreatic tumor vessels expressed α-SMA compared with 22% of normal pancreatic vessels. They also found that α-SMA expression was absent from a significant proportion of capillaries and microvessels within the normal vasculature; these vessels did, however, stain for desmin, another marker of periendothelial cells.

A highly significant association was noted between DLL4 expression and α-SMA expression in both normal and tumor vessels. The expression of α-SMA in normal specimens was lower in DLL4-negative vessels (normal 69%, tumor 65%) compared with DLL4-positive vessels (normal 100%, tumor 99%). This suggests that DLL4 may play a role in regulating vessel maturation by interacting with Notch receptors upon pericytes.

The mechanism by which endothelial expressed DLL4 regulates vessel maturation needs further investigation. DLL4 has been shown to signal through both Notch1 and Notch4 between endothelial cells (3335), but the effect of endothelial DLL4 on pericytes is not reported. However, evidence that Notch plays a role in vessel maturation comes from a number of sources. Vessels within mice deficient for DLL4 have an absence of smooth muscle/pericyte coverings (15). This phenotype is similar to that of platelet-derived growth factor-B or platelet-derived growth factor receptor-B–deficient mice (36, 37).

The human syndrome CADASIL is a vascular disease characterized by mutations of the Notch3 receptor leading to abnormalities of vascular smooth muscle (4). A recent in vitro study by Sweeney et al. (38) showed that vascular smooth muscle cells express Notch 1, Notch 3, HES1, HES5, HEY1, HEY2, and HEYL. They showed that inhibition of RBP-Jκ activity led to decreased proliferation, increased apoptosis, and increased migration in vascular smooth muscle cells. Clearly, our clinical study provides justification for more extensive in vitro work.

Tumor vessels typically show disorganized vascular patterning, which in conjunction with their abnormal ultrastructures results in chaotic blood flow, high interstitial pressures, and increased vessel leakiness (30). Tumor vasculature consists of a mix of immature vessels devoid of pericyte coverings and more mature vessels with pericyte/smooth muscle cell coats. Endothelial cell proliferation and sprouting have been to shown to be most prevalent within the immature vessels, suggesting that it is these vessels that are angiogenic (32, 39, 40). These vessels are also the ones most susceptible to antiangiogenic therapies. Benjamin et al. (41) studied vessel maturation in prostate cancer. They showed that androgen blockade decreased production of VEGF, which, in turn, led to the regression only of immature vessels lacking pericyte coverings in prostate cancer. They showed that almost 40% of tumor vessels were positive for α-SMA; however, after androgen blockade, the proportion of vessels coated with α-SMA rose to 79%.

Gee et al. (39) also found that vessels covered with α-SMA were protected from the antiangiogenic therapy, which in their case was interleukin-12. These findings have important implications for the development of future antiangiogenic therapies. Tumors that have a high proportion mature vessels are less likely to respond to for example anti-VEGF therapies, as the pericytes seem to offer protection against growth factor deprivation. We have recently shown in human endothelial cells that overexpression of DLL4 inhibits proliferation by down-regulating VEGF receptor 2 and NRP1 expression (42), providing an additional mechanism for resistance to such therapies. The association shown between DLL4 expression and α-SMA expression suggests that targeting DLL4 in addition to VEGF may, by targeting both mature and immature vessels, prove to be of greater therapeutic benefit and we are currently investigating this in xenograft models.

The biological function of DLL4 is not confined to vessel maturation. Knockout studies have shown that haploinsufficiency of DLL4 results in embryonic lethality from severe vascular defects (15, 17, 18). The demonstration of vascular defects in heterozygous mice is uncommon and highlights the importance of DLL4 in vascular development. Similar phenotypes have only been described for VEGF knockout mice (19, 20). We have recently shown that down-regulating DLL4 expression in vitro using RNA interference significantly inhibits endothelial cell proliferation, migration, and network formation (21). Similarly, we have shown up-regulation of Delta 4 expression disrupts endothelial function (42), a phenotype previously found for Notch4, where in vivo up-regulation or down-regulation disrupts angiogenesis (43, 44). This implies that a narrow range of notch signaling is needed for optimum angiogenesis. Thus, targeting DLL4 may inhibit tumor angiogenesis by blocking multiple endothelial cell functions.

Endothelial expression of DLL4 may represent an intermediate step toward final differentiation by promoting maturation and functionality; thus, it may be an important target for tumors resistant to anti-VEGF therapy.

Exploratory analysis of the value of DLL4 as a prognostic marker was assessed by relating DLL4 expression with clinical follow-up data. In superficial and invasive disease, high levels of DLL4 conferred an increased risk of tumor recurrence or cancer-specific death, respectively. We additionally noted that 10 of the 14 patients that received adjuvant intravesical therapy had high levels of DLL4 expression. Adjuvant mitomycin C or bacille Calmette Guérin is commonly used in superficial bladder cancers thought to be at high risk of recurrence and progression. The risk stratification of superficial bladder tumors is based on the number of tumors at first presentation, number and frequency of recurrences, and the tumor histology. In this study, we observed for the first time that conventionally classified high-risk tumors seemed to express high levels of DLL4. The intensive nature of the microscopic studies meant that we could not examine a larger cohort; however, we are currently trying to develop antibodies to investigate this in much larger numbers.

Here, for the first time, we have identified DLL4 as a novel angiogenic target in bladder cancer. We have shown that DLL4 (a) is preferentially expressed in tumor vasculature, (b) is associated with vessel maturation, and (c) may have prognostic value in superficial and invasive cancer. Taken with our recent findings that an optimal level of DLL4 is necessary for endothelial cell function (21, 42), targeting DLL4 either by inhibition or overexpression may be an approach worthy of further investigation in the management of superficial and invasive bladder cancer.

Grant support: Cancer Research UK Molecular Oncology Laboratories, The 6th Framework Programme of the European Union (Angiotargeting), The Royal College of Surgeons of England, The British Urological Foundation, and the National Translational Cancer Research Network.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: N.S. Patel and M.S. Dobbie contributed equally to this work.

We thank Toby Hunt, Rosemary Jeffrey, Leticia Campo, Stephen Fox, and Professor Roy Bicknell for their contributions to this work.

1
Folkman J. Tumor angiogenesis: therapeutic implications.
N Engl J Med
1971
;
285
:
1182
–6.
2
Chodak GW, Scheiner CJ, Zetter BR. Urine from patients with transitional-cell carcinoma stimulates migration of capillary endothelial cells.
N Engl J Med
1981
;
305
:
869
–74.
3
Streeter EH, Harris AL. Angiogenesis in bladder cancer-prognostic marker and target for future therapy.
Surg Oncol
2002
;
11
:
85
–100.
4
Shawber CJ, Kitajewski J. Notch function in the vasculature: insights from zebrafish, mouse and man.
BioEssays
2004
;
26
:
225
–34.
5
Iso T, Hamamori Y, Kedes L. Notch signaling in vascular development.
Arterioscler Thromb Vasc Biol
2003
;
23
:
543
–53.
6
Li JL, Harris AL. Notch signaling from tumor cells: a new mechanism of angiogenesis.
Cancer Cell
2005
;
8
:
1
–3.
7
Artavanis-Tsakonas S, Matsuno K, Fortini ME. Notch signaling.
Science
1995
;
268
:
225
–32.
8
Weinmaster G. The ins and outs of notch signaling.
Mol Cell Neurosci
1997
;
9
:
91
–102.
9
Radtke F, Raj K. The role of Notch in tumorigenesis: oncogene or tumour suppressor?
Nat Rev Cancer
2003
;
3
:
756
–67.
10
Mumm JS, Kopan R. Notch signaling: from the outside in.
Dev Biol
2000
;
228
:
151
–65.
11
Weinmaster G. Notch signaling: direct or what?
Curr Opin Genet Dev
1998
;
8
:
436
–42.
12
Nakagawa O, McFadden DG, Nakagawa M, et al. Members of the HRT family of basic helix-loop-helix proteins act as transcriptional repressors downstream of Notch signaling.
Proc Natl Acad Sci U S A
2000
;
97
:
13655
–60.
13
Davis RL, Turner DL. Vertebrate hairy and Enhancer of split related proteins: transcriptional repressors regulating cellular differentiation and embryonic patterning.
Oncogene
2001
;
20
:
8342
–57.
14
Iso T, Kedes L, Hamamori Y. HES and HERP families: multiple effectors of the Notch signaling pathway.
J Cell Physiol
2003
;
194
:
237
–55.
15
Gale NW, Dominguez MG, Noguera I, et al. Haploinsufficiency of δ-like 4 ligand results in embryonic lethality due to major defects in arterial and vascular development.
Proc Natl Acad Sci U S A
2004
;
101
:
15949
–54.
16
Mailhos C, Modlich U, Lewis J, et al. Delta4, an endothelial specific notch ligand expressed at sites of physiological and tumor angiogenesis.
Differentiation
2001
;
69
:
135
–44.
17
Duarte A, Hirashima M, Benedito R, et al. Dosage-sensitive requirement for mouse Dll4 in artery development.
Genes Dev
2004
;
18
:
2474
–8.
18
Krebs LT, Shutter JR, Tanigaki K, et al. Haploinsufficient lethality and formation of arteriovenous malformations in Notch pathway mutants.
Genes Dev
2004
;
18
:
2469
–73.
19
Ferrara N, Carver-Moore K, Chen H, et al. Heterozygous embryonic lethality induced by targeted inactivation of the VEGF gene.
Nature
1996
;
380
:
439
–42.
20
Carmeliet P, Ferreira V, Breier G, et al. Abnormal blood vessel development and lethality in embryos lacking a single VEGF allele.
Nature
1996
;
380
:
435
–9.
21
Patel NS, Li JL, Generali D, et al. Up-regulation of δ-like 4 ligand in human tumor vasculature and the role of basal expression in endothelial cell function.
Cancer Res
2005
;
65
:
8690
–7.
22
Sobin LH. TNM classification of malignant tumours, 6th ed. New York: Wiley; 2002.
23
Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR.
Nucleic Acids Res
2001
;
29
:
e45
.
24
Poulsom R, Longcroft JM, Jeffery RE, Rogers LA, Steel JH. A robust method for isotopic riboprobe in situ hybridisation to localise mRNAs in routine pathology specimens.
Eur J Histochem
1998
;
42
:
121
–32.
25
Crew JP, O'Brien T, Bradburn M, et al. Vascular endothelial growth factor is a predictor of relapse and stage progression in superficial bladder cancer.
Cancer Res
1997
;
57
:
5281
–5.
26
Liu ZJ, Shirakawa T, Li Y, et al. Regulation of Notch1 and Dll4 by vascular endothelial growth factor in arterial endothelial cells: implications for modulating arteriogenesis and angiogenesis.
Mol Cell Biol
2003
;
23
:
14
–25.
27
O'Brien T, Cranston D, Fuggle S, Bicknell R, Harris AL. Different angiogenic pathways characterize superficial and invasive bladder cancer.
Cancer Res
1995
;
55
:
510
–3.
28
Crew JP. Vascular endothelial growth factor: an important angiogenic mediator in bladder cancer.
Eur Urol
1999
;
35
:
2
–8.
29
O'Brien TS, Fox SB, Dickinson AJ, et al. Expression of the angiogenic factor thymidine phosphorylase/platelet-derived endothelial cell growth factor in primary bladder cancers.
Cancer Res
1996
;
56
:
4799
–804.
30
Jain RK. Molecular regulation of vessel maturation.
Nat Med
2003
;
9
:
685
–93.
31
Eberhard A, Kahlert S, Goede V, et al. Heterogeneity of angiogenesis and blood vessel maturation in human tumors: implications for antiangiogenic tumor therapies.
Cancer Res
2000
;
60
:
1388
–93.
32
Morikawa S, Baluk P, Kaidoh T, et al. Abnormalities in pericytes on blood vessels and endothelial sprouts in tumors.
Am J Pathol
2002
;
160
:
985
–1000.
33
Shawber CJ, Das I, Francisco E, Kitajewski J. Notch signaling in primary endothelial cells.
Ann N Y Acad Sci
2003
;
995
:
162
–70.
34
Rao PK, Dorsch M, Chickering T, et al. Isolation and characterization of the notch ligand δ4.
Exp Cell Res
2000
;
260
:
379
–86.
35
Shutter JR, Scully S, Fan W, et al. Dll4, a novel Notch ligand expressed in arterial endothelium.
Genes Dev
2000
;
14
:
1313
–8.
36
Hellstrom M, Kalen M, Lindahl P, Abramsson A, Betsholtz C. Role of PDGF-B and PDGFR-β in recruitment of vascular smooth muscle cells and pericytes during embryonic blood vessel formation in the mouse.
Development
1999
;
126
:
3047
–55.
37
Lindahl P, Johansson BR, Leveen P, Betsholtz C. Pericyte loss and microaneurysm formation in PDGF-B-deficient mice.
Science
1997
;
277
:
242
–5.
38
Sweeney C, Morrow D, Birney YA, et al. Notch 1 and 3 receptors modulate vascular smooth muscle cell growth, apoptosis and migration via a CBF-1/RBP-Jk dependent pathway.
FASEB J
2004
;
18
:
1421
–3.
39
Gee MS, Procopio WN, Makonnen S, et al. Tumor vessel development and maturation impose limits on the effectiveness of anti-vascular therapy.
Am J Pathol
2003
;
162
:
183
–93.
40
Abramsson A, Berlin O, Papayan H, et al. Analysis of mural cell recruitment to tumor vessels.
Circulation
2002
;
105
:
112
–7.
41
Benjamin LE, Golijanin D, Itin A, Pode D, Keshet E. Selective ablation of immature blood vessels in established human tumors follows vascular endothelial growth factor withdrawal.
J Clin Invest
1999
;
103
:
159
–65.
42
Williams CK, Li JL, Murga M, Harris AL, Tosato G. Upregulation of the Notch ligand Delta-like 4 inhibits VEGF-induced endothelial cell function.
Blood
2006
;
107
:
931
–9.
43
Uyttendaele H, Ho J, Rossant J, Kitajewski J. Vascular patterning defects associated with expression of activated Notch4 in embryonic endothelium.
Proc Natl Acad Sci U S A
2001
;
98
:
5643
–8.
44
Krebs LT, Xue Y, Norton CR, et al. Notch signaling is essential for vascular morphogenesis in mice.
Genes Dev
2000
;
14
:
1343
–52.