Bladder cancer is a common urologic malignancy. Transitional cell carcinoma of the bladder accounts for 90% of bladder tumors. Squamous cell carcinoma and adenocarcinoma make up the remaining 10%. Pathology parameters such as tumor grade and stage indicate the ability of a bladder tumor to invade/metastasize and the depth of invasion, respectively. A Key characteristic of bladder tumors is their heterogeneity in the ability to invade, recur and metastasize. Low-grade (G1) bladder tumors are mostly confined to the mucosa (Stage Ta) and rarely invade lamina propria and beyond (Stage ≥T1). Given sufficient time, most high-grade bladder tumors will invade muscle (Stage ≥T2), perivesicle fat (Stage T3) and beyond (Stage T4).1 Bladder tumor recurrence, which usually is the occurrence of a new tumor in the bladder, is quite common; 50–80% of patients with bladder cancer will have a new tumor in the bladder within 3-years of resection of the initial tumor.2 In addition, patients with low-grade tumors may subsequently have a high-grade bladder tumor in their bladder. Therefore, patients with bladder cancer are monitored every 3–6 months with cystoscopy (an endoscopic procedure) and urine cytology.3
Several noninvasive urine tests to detect bladder cancer have been reported in the literature including, BTA-Stat/TRAK, NMP22, UroVysion, UBC, U-Cyt™, telomerase, BLCA-4 and HA-HAase.4, 5 Among these the HA-HAase test measures urinary hyaluronic acid (HA) and hyaluronidase (HAase) levels to detect cancer and to evaluate its grade.6 The HA-HAase test has shown ∼90% sensitivity and ∼85% specificity to detect bladder cancer occurrence and to monitor its recurrence. Furthermore, the test may be useful for early detection of bladder cancer.7, 8
HA is a glycosaminoglycan made up of repeating disaccharide units D-glucuronic acid and N-acetyl-D-glucosamine.9, 10 HA concentration is elevated in several carcinomas, including lung, breast, bladder, colon and prostate.11, 12, 13, 14, 15, 16 In tumor tissues, HA promotes tumor cell adhesion, proliferation and migration. It also offers protection to tumor cells from contact inhibition of growth and possibly immune surveillance.17, 18, 19 Recently, HA has been shown to constitutively regulate ErbB2 phosphorylation and signaling complex formation in HCT116 colon carcinoma and MCF-7 cells.20 Small HA fragments are angiogenic and are generated by hyaluronidases (e.g., tumor cell-derived HAase, HYAL1). However, HA oligosaccharides may also inhibit anchorage-independent growth of tumor cells at concentrations ≥100 μg/ml.21, 22, 23 In cells, HA is synthesized at the plasma membrane by any one of the 3 HA-synthases (HAS). HAS1, HAS2 and HAS3 are transmembrane proteins, which synthesize different sizes of HA at different kinetic rates.24 For example, HAS1 and HAS2 synthesize relatively large molecular mass HA (2 × 105–2 × 106 Da), whereas HAS3 synthesizes HA polymers between 1 × 105 and 1× 106 Da.25 Expression of HAS3 has been shown to increase tumor growth, invasion and anchorage independent growth in colon and prostate carcinomas.26, 27, 28 HAS1 and HAS2 expression increases following malignant transformation and promotes tumor growth.29 Among the 3 HAS genes, HAS1 expression correlates with invasion, disease progression and poor survival in colon, ovarian and endometrial carcinomas.30, 31, 32 When compared with HAS2 and HAS3, HAS1 expression is elevated in multiple myeloma patients.33, 34 Recently, the expression of HAS1 splice variant (HAS1-vb), generated by intronic splicing of HAS1 transcript, was shown to correlate with survival in multiple myeloma patients.35 However, it is unknown whether these splice variants are functionally active, in terms of HA synthesis.
In this study, we examined the expression of wild type HAS1 in bladder tumor and normal bladder tissues and cells by reverse transcription (RT) real time RT-PCR, northern blot analysis, immunohistochemistry and immunoblotting. We also correlated HAS1 expression with HA expression, and HA urine test results.
CIS, carcinoma in situ; HA, hyaluronic acid; HAase, hyaluronidase; HAS, HA-synthase; HAS1-S, HAS1-sense; HAS1-AS, HAS1-antisense; ROC, Receiver Operating Characteristic curve.
Material and methods
Bladder specimens: Fresh specimens
Fresh bladder tissues were obtained from patients with bladder cancer, who were undergoing cystectomy for muscle invasive disease (Stage ≥T2) or transurethral resection of bladder tumor for superficial disease. These specimens included 6 low-grade (i.e., G1) tumors and 27 high grade tumors (i.e., G2 or G3). The stage distribution for the tumors was 7 superficial (i.e., Ta + T1) and 26 muscle invasive (i.e., Stage ≥T2). Among normal bladder tissues (n = 15), 12 were from age matched organ donors (40–65 years) and 3 normal bladder urothelial specimens from patients with bladder cancer. All tumor tissue specimens were brought to the laboratory within 30 min after obtaining the specimens following surgery and frozen at −70°C. These specimens were used for HAS1 transcript analysis and HA level measurement.
Paraffin embedded bladder tissues (n = 68) were obtained from patients with bladder cancer (mean age: 67.9 years; range 44–82 years). Fifty-six tissues were positive for bladder cancer and 12 specimens had no evidence of the disease (i.e., normal urothelium). The grade distribution was 8, G1, 21, G2 and 27, G3 specimens. The stage distribution for the specimens was 28, Ta, 13, T1 and 12 ≥ T2 (i.e., 6, T2, 5, T3 and 1, T4) and 3 carcinoma in situ (CIS) specimens.
Out of the 68 bladder cancer cases, on whom there were archival bladder specimens available (as described above), urine specimens were collected from a total of 47 patients; Thirty five patients had active bladder cancer and 12 had no evidence of tumor at the time of urine collection.
Bladder cancer cells
Human bladder cancer lines, HT1376, HT1197, RT4, UMUC-3, TCC-Sup, T24 and J82 were obtained from ATCC. 253-J Parent and 253-J Lung lines were kindly provided by Dr. Colin Dinney, M.D. Anderson Cancer Center. HAS1 transfectants (HAS1-S 1, HAS1-AS 1 and vector) were generated as follows: HAS1 cDNA containing the entire coding region was amplified by RT-PCR from HT1376 total RNA, using HAS1-L (24–43; accession no. NM_001523): 5′AGACCCACTGCGATGAGACA3′ and HAS1-R (1779–1760): 5′GCTGGACTCA CACTGGAC3′ primers. HAS1 cDNA was cloned into a pEF6-v5-His bidirectional eukaryotic expression vector plasmid (InVitrogen, Carlsbad, CA) to generate HAS1-sense (HAS1-S) and HAS1-antisense (HAS1-AS) cDNA constructs, which were then subcloned into a pTracer-EF/V5-His A vector (InVitrogen). HT1376 cells were transfected with pTracer-EF/v5/His A vector, HAS1-S or HAS1-AS cDNA constructs, using the Effectene™ reagent (Qiagen, Valencia, CA) and the transfectants were selected in 3.5 μg/ml blasticidin.
Total RNA from was isolated from frozen bladder tissues (∼100–200 mg), and RNA quality was determined by agarose gel analysis and spectrophotometry. RNA was reverse transcribed using Superscript II kit (InVitrogen, CA). cDNAs were subjected to real time PCR in a Bio-Rad iCycler iQ real time PCR system (BioRad, Hercules, CA), using HAS1 primers and a detection probe. The sequences for HAS1 (accession no. NM_001523) primers and probe were as follows: HAS1-L 825–842: 5′-GGTGGGGACGTGCGGATC-3′; HAS1-R 926–949: 5′-ATGCAGGATACACA GTGGAAGTAG-3′; 5′ Fam flurophore/3′ Tamra quencher HAS1 probe 890–915: 5′-CCCGCTC CACATTGAAGGCTACCCAG-3′. As a control, each cDNA sample was simultaneously subjected to β-actin (accession no. NM_001101) PCR, using the following primers and detection probe: β-actin-L 305–321 5′-CAACTGGGACGACATGGA-3′; β-actin-R 401–418: 5′-GTTGGCCTTGGGGTTCAG-3′; 5′ Fam flurophore/3′ Tamra quencher β-actin probe 325–351: 5′-AATCTGGCACCA CACCTTCTACAA TGA-3′. PCR conditions were as follows: 1 cycle at 95°C for 3 min, 50 cycles at 95°C for 15 sec and 60°C for 60 sec, followed by 1 cycle at 95°C for 60 sec, and finally 1 cycle at 55°C for 60 sec. The threshold cycle (Ct) of each sample was determined and the relative level of HAS1 expression (2ΔCt) was calculated by obtaining a ΔCt value (HAS1 Ct - β-actin Ct) and then expressing it as arbitrary units (1/2ΔCt × 100). HAS 1 levels were normalized to β-actin, and therefore, the amount of tissue used for analysis, delay in tissue processing, etc did not affect the results.
cDNAs obtained from normal and bladder tumor tissues by RT reaction were subjected to semiquantitative PCR to amplify HAS1-va, HAS1-vb and HAS1-vc variants. Primers: HAS1-va (AY916554), forward 21–40: 5′CCTTCAAGGCGCT CGGAGAT3′ and reverse 310–291: 5′TGGAGGTGTACCTAGAGAAC3′. HAS1-vb (AY916555), forward 31–50: 5′GCTCGGAGATTCGGTGGACT3′ and reverse 334–315: 5′GGCGAGGAATGAGGGCA TCA3′. HAS1-vc (AY916556), forward 31–50: 5′GCTCGGAGATTCGGTGGACT3′ and reverse 437–418: 5′ACCTGGTCCCCTCAGCTTAC3′. PCR conditions: (i) 95°C/10 min (ii) 10 cycles of 94°C/30 sec, (70–60°C/30 sec), 72°C/1 min (iii) 44 cycles of 94°C/30 sec, 60°C/30 sec and 72°C/1 min. PCR products were separated on a 1.2% agarose gel, and the ethidium bromide stained bands were scanned for intensity using Kodak 1-D gel analysis software. The intensity of the bands was normalized to β-actin expression and then expressed as an HAS1/β-actin ratio.
Whole cell lysates of bladder cancer cells and HT1376 transfectants were subjected to immunoblot analysis using a rabbit polyclonal anti-HAS1 IgG (1:3000 dilution). The anti-HAS1 IgG was prepared against a 17-amino acid carboxyl terminal sequence (563GVRRLCRRRTGGYRVQV579) in HAS1 protein.
Metabolic labeling and immunoprecipitation
Pulse-labeling and immunoprecipitation was conducted as described before.36 Subconfluent cultures of HT1376 cells (5 × 105 cells/6 cm dish) were incubated at 37°C for 4 hr in L-methionine free DMEM, followed by pulse labeling with Tran35S label (200 μCi/ml) for 10 min. The cells were washed in DMEM + 10 mM unlabeled methionine and then incubated in the same medium for various time periods as indicated. The radioactively labeled cells were first washed in phosphate buffered saline and solubilized in RIPA buffer. The solubilized extracts were immunoprecipitated using either anti-HAS1 IgG or normal rabbit IgG and goat anti-rabbit IgG agarose beads. The immunoprecipitates were analyzed by 8% SDS-polyacrylamide gel electrophoresis followed by fluorography.
Northern blot analysis
Total RNA isolated from bladder cancer cells and tissues was subjected to northern blot analysis using a 376-bp [32P] α-dCTP labeled HAS1 cDNA probe (bp 1232–1608; HAS1 GenBank accession no. NM001523.1). As controls, northern blot membranes were reprobed with a [32P] α-dCTP labeled GAPDH cDNA probe (BD Biosciences, San Jose, CA).
HA-test (HA-ELISA like assay)
Bladder tissue extracts, serum-free culture conditioned media and urine specimens were assayed for HA levels using an HA ELISA like assay.6 The HA levels (μg/ml, or ng/ml) were normalized to total protein (mg/ml). On the HA urine test, HA levels ≥500 ng/ml constitute a positive test.6
HAS1 and HA immunohistochemistry
Three-micrometer sections of bladder tissues placed on positively charged slides were deparaffinized, rehydrated and subjected to antigen retrieval.13 For HAS1 staining, specimens were incubated with the anti-HAS1 IgG (2.0 μg/ml) at room temperature for 2 hr. The slides were developed using the Dako LASB kit (DakoCytomation, Carpentaria, CA) and 3,3′-diaminobenzidine staining. Three individuals graded slides for intensity as 0 (no staining), 1+, 2+ and 3+. In some tissues, there was heterogeneity in staining, some of which was due to the “edge effect” often seen in immunohistochemistry. To account for heterogeneity in staining (if any), the overall staining grade for each slide was assigned based on the staining intensity of the majority of the tumor tissue in the specimen. However, if ∼50% of the tumor cells in the section were assigned 1+ staining, and the other 50% as 3+, the overall staining grade was 2+. If the staining distribution was that ∼50% of the tumor cells stained 2+ and the remaining stained 3+, the overall staining inference was assigned as 3+. The tissue sections were then grouped as low- (0–1+) or high- (2+ – 3+) grade staining. For HA staining, specimens were stained using a biotinylated HA binding protein (2 μg/ml) and the slides were graded as described previously.13
Bladder tumor and normal bladder tissue extracts (2-mg protein) were separated on a Sepharose S-300 column (1.5 × 110 cm) equilibrated with phosphate buffered saline + 0.05% Tween-20. The column fractions (1.2 ml/fraction) were assayed by both the HA ELISA-like assay and the modified carbazole assay, as described previously.37, 38 The column was calibrated using HA polymers of various sizes (2.0 × 106 D–8,000 D) obtained from Genzyme Corp (Cambridge, MA).
Sensitivity, specificity and accuracy parameters were calculated from the 2 × 2 contingency table, as described previously.4 Since the distribution of HAS1 and HA expression in tumor tissues and exfoliated cells was nonparametric, we used the Mann-Whitney U test to compare 2 groups (e.g., normal and tumor) and the Kruskal-Wallis test followed by Dunn's multiple comparison test to compare multiple groups when determining the statistical significance of differences in HAS1 transcript levels, tissue HA levels and HAS1/HA staining intensity in various groups. The McNemar test was used to evaluate the correlation between HAS1 levels (transcript or protein) and HA levels.
HAS1 mRNA expression is elevated in bladder tumor tissues
We employed real-time RT-PCR to measure the HAS1 transcript in bladder tissues. As shown in Figure 1a, the expression of HAS1 transcript is elevated from 5.1- to 9.8-fold in low- and high-grade bladder tumor tissues, when compared with normal bladder tissues. Among the 15 normal bladder tissues, 3 were from pathologically normal specimens adjacent to bladder tumor. In these specimens, the HAS1 transcript level (4.7 ± 2.2) was still 5- to 10-fold lower than that in the tumor specimens, suggesting that HAS1 expression is probably elevated only in neoplastic tissues. The increase in the level of HAS1 transcript in both low- and high-grade bladder tumors was statistically significant when compared with normal tissues (p < 0.001; Dunn's multiple comparison test). Although high-grade bladder tumors had 2-fold higher levels of HAS1 transcript, this difference was not statistically significant (p > 0.05). Among normal and bladder cancer specimens, some specimens showed much higher HAS1 transcript levels when compared with the mean. Elimination of the 2 outliers from normal specimens (HAS1 transcript levels 10.3–10.1) decreased the mean HAS1 transcript levels to 3.3 ± 0.66). These normal bladder specimens with high HAS1 values were obtained from organ donors. Removal of the 1 outlier from the low-grade category reduced the mean HAS1 transcript levels to 14.2 ± 5.7; however, the difference in the HAS1 levels between G1 and normal bladder specimens was statistically significant (p < 0.05). This patient with such high HAS1 transcript level had multiple recurrences before and also after the resection of the tumor, which was analyzed in this study. Removal of the 2 outliers in the G2 + G3 category decreased the mean HAS1 levels to 18.2 ± 2.3; however, the difference in HAS1 levels among normal bladder specimens and G2 + G3 bladder tumor specimens was statistically significant (p < 0.001). One of these patients with high HAS1 value had a G3–T3 tumor but no prior recurrence; however, the patient was lost to followup. Another patient with high HAS1 transcript levels had a G3 T2b tumor with CIS (node negative), had multiple recurrences and was treated with both intravesical chemotherapy and immunotherapy. No progression was observed in this patient 18 months after cystectomy. These results show that HAS1 transcript levels might be highly elevated among patients with high stage disease and/or multiple recurrences.
Figure 1b shows the HAS1 transcript levels in superficial (Stage Ta or T1) and invasive (Stage ≥T2) bladder tumors. As observed for tumor grade, HAS1 transcript levels are elevated 4.5- to 10-fold in superficial and invasive bladder tumors when compared with normal bladder tissues (p > 0.05). Although the sample size was small, we compared HAS1 transcript levels between Ta and T1 tumors to examine whether there were any differences in HAS1 levels between tumors that were confined to the mucosa and those that had already expressed the ability to penetrate into the lamina propria. The mean HAS1 levels in Ta tumors (16.4 ± 6.8; n = 4) were 1.8-fold less than the levels in T1 tumors (29.1 ± 22.2; n = 3), but this difference was not statistically significant (p = 0.47; t test), possibly because of low number of specimens in each category. Removal of the 2 outliers from normal bladder, 1 outlier from the Ta+T1 and 2 outliers from the T2 categories did not alter the statistical significance of the differences between normal and bladder cancer specimens, i.e., the difference between normal and Ta + T1 tumors (p < 0.05) and between normal and ≥ Stage T2 tumors (p < 0.001) was statistically significant.
We next examined whether HAS1 transcript levels in tumor specimens might be influenced by the number of recurrences, presence of multiple tumors, and/or treatment with intravesical chemotherapy or immunotherapy prior to surgery. Out of the 33 patients, information on prior recurrences, as well as, recurrences in the bladder following tumor resection was available on 28 patients. HAS1 levels among those patients with no-prior recurrence (15. 8 ± 5.3; n = 17) and those who at least had 1 recurrence (13.5 ± 3.0; n = 11) were not statistically different (p >0.05). However, among the patients with no recurrences, 2 had very high HAS1 levels. Elimination of these patients from the sample, resulted in a decrease in HAS1 levels in the nonrecurred group (8.2 ± 1.4), and the difference in HAS1 levels between patients who recurred and those who did not, reached statistical significance (p = 0.028; t test). Clinical information on disease progression was available on 21 patients; 15 did not progress and 6 had pelvic, lung and/or liver metastasis. The difference in HAS1 levels among those with (12.7 ± 3.2; n = 6) or without (9.6 ± 2.5; n = 15) disease progression is not statistically significant (p = 0.089; t test). Removal of the 2 outliers with very high HAS1 levels in the nonprogressed group decreased HAS1 levels (7.5 ± 1.5) and this difference almost reached statistical significance (p = 0.058; t test). The differences in HAS1 transcript levels among those who did not receive intravesical chemotherapy/immunotherapy (7.8 ± 1.3; n = 14), those who received intravesical treatment (27.9 ± 10.2; n = 6) was statistically significant (p = 0.0083; t test). In the study cohort, 2 patients received neoadjuvant chemotherapy but the HAS1 transcript levels in these 2 patients were not different from those who did not receive any treatment (7.2 ± 4.5; n = 2; p > 0.05). Information on multiple tumors in the bladder was not available on >95% of the patients in this cohort, and therefore, we could not evaluate the influence of multiple tumors on HAS1 transcript levels. These results show that HAS1 transcript levels may correlate with recurrence and treatment. For disease progression, HAS1 transcript levels were higher in progressed vs. nonprogressed patients and showed a trend toward statistical significance.
Next, we determined the potential of HAS1 real-time PCR to detect bladder cancer. Using the receiver operating characteristic (ROC) curve, we determined that a cut-off limit of 6 arbitrary units yields the best trade-off between sensitivity and specificity among normal and bladder cancer patients. This cut-off limit was not set to distinguish between “superficial or invasive tumors” As shown in Table I, high HAS1 transcript expression has reasonably high sensitivity and specificity to detect bladder tumors regardless of tumor grade or stage. If the cut-off limit was raised to 10 arbitrary units, the specificity with respect to normal bladder specimens would be 100%. At this cut-off limit, 3/7 superficial tumors (i.e., stage Ta + T1) and 16/26 muscle invasive tumors (i.e., Stage ≥T2) had HAS1 transcript levels greater than 10 arbitrary units, i.e., a sensitivity of 42.3–61.5%, respectively. Although when compared with superficial tumors more number of muscle invasive tumor specimens had HAS1 transcript levels higher than 10 arbitrary units, this difference was not statistically significant (p = 0.42), perhaps due to the small number of superficial tumors.
Table I. Sensitivity, Specificity and Accuracy of HAS1 mRNA Expression to Detect Bladder Cancer1
Arbitrary unit (as described in “Materials and Methods” section) 6 was used as the cut-off limit to determine sensitivity and specificity parameters.
G2 + G3
Stage Ta + T1
Stage ≥ T2
To confirm real-time PCR results, we examined HAS1 transcript expression in bladder tissues and cells by Northern blot analysis. As shown in Figure 1c, all of the 7 bladder cancer cell lines that were tested express HAS1 transcript. Furthermore, the expression of HAS1 mRNA is elevated in bladder tumor tissues when compared with normal bladder tissues. Image analysis showed that HAS1 mRNA expression is 2- to 4-fold elevated in bladder tumor tissues when compared with normal bladder tissues. These results show that bladder cancer cells express HAS1 transcript and HAS1 mRNA expression is elevated in tumor tissues.
Detection of HAS1 variants in bladder tumor tissues
Recently, 3 HAS1 splice variants that are generated by intronic splicing events have been identified.35 Since the length of the unique sequences in these 3 splice variants is < 50 bases, when compared with the sequence of the wild type HAS1 transcript, it is not possible to design primers and probes that specifically detect only a particular HAS1 variant by real-time PCR. Therefore, we employed semiquantitative RT-PCR to detect the relative expression of the 3 HAS1 variants in normal and bladder tumor tissues. As shown in Figure 2a, a 289-bp HAS1-va PCR product is detected in both bladder tumor and normal bladder tissues. As shown in Figure 2b, the expression of HAS1-va is 2.3-fold higher in bladder tumor tissues when compared with normal bladder tissues and this increase is statistically significant (p = 0.0193; Mann-Whitney U test). The difference in the mean HAS1va levels among normal bladder (275 ± 222.1) and TaG1 (492.7 ± 32.1; p < 0.05) was statistically significant. The difference in the mean HAS1va levels among normal bladder and Stage ≥ T2 G2 + G3 (691.2 ± 74.3; p < 0.001) was also statistically significant (Tukey's multiple comparison test). However, the differences in HAS1va levels among G1 and G2 + G3 bladder cancer specimens were not statistically significant. It is noteworthy that the HAS1-va product was detected only after 44 PCR cycles, in addition to 10 cycles of a touch down PCR protocol. In contrast, by semiquantitative RT-PCR, a PCR product was amplified specifically from the HAS1 wild type transcript after 30 cycles, in the same bladder tissues (data not shown). This suggests that wild type transcript is the major HAS1 transcript that is expressed in bladder tissues. It is noteworthy that none of the bladder cancer cell lines expressed HAS1-va transcript. Furthermore, RT-PCR analysis also showed that HAS1-vb and HAS1-vc transcripts were not expressed in bladder tissues (neoplastic and nonneoplastic) and bladder cancer cell lines.
Measurement of tissue HA levels and size determination
Since HAS1 directs HA synthesis, we measured HA levels in bladder cancer tissues (G1, n = 3; G2 + G3, n = 12) and normal bladder tissues (n = 9) in which HAS1 transcript levels were measured by real-time PCR. For HA level determination, we needed about 300–500 mg of tissue and for many specimens so much material was not available; therefore, we could measure HA levels in 15 bladder cancer and 9 normal bladder tissues. As shown in Figure 3a, HA levels in both low- and high-grade bladder tumor tissues are elevated (∼2.5-fold) when compared with normal bladder tissues (p < 0.01; Dunn's multiple comparison test). Furthermore, there is also a correlation between increased tissue HA levels (>2.4 μg/mg) and increased HAS1 transcript levels (>6 arbitrary units; McNemar test, p < 0.001).
Since HAS1 directs synthesis of HA that is of higher molecular mass than that synthesized by HAS3, we performed Sepharose S-300 gel filtration chromatography to determine the size of the HA species present in bladder tumor tissues. As shown in Figure 3b, the size of the large HA polymer present in a G3 bladder tumor tissue extract is ∼2 × 106 Da, and this size is consistent with the HA polymer synthesized by HAS1.25 As we previously reported, in addition to the HA species of intermediate size (∼105 D), small HA fragments of molecular mass 8–15 kDa (i.e., 13–25 disaccharide units) are present in the G3 bladder tumor tissue extract. Figure 3b also shows that only a single large molecular mass HA species is present in both low-grade bladder tumor and normal bladder tissue extracts. These results suggest that HAS1 is most likely contributing to HA synthesis in both normal and neoplastic bladder tissues, and the presence of HA fragments in high grade bladder tumor tissue extracts is most likely due to the degradation of the large HA polymer by HAase.
Detection of HAS1 protein in bladder cancer cells
Since the wild type HAS1 transcript appears to be the major transcript in bladder tissues, we generated an anti-HAS1 peptide IgG that detects only the wild type HAS1 protein. We performed immunoblot analysis on 9 bladder cancer cell lines and stable HAS1-S and HAS1-AS HT1376 transfectants, using the anti-HAS1 IgG. As shown in Figure 4a, anti-HAS1 IgG detects 2 proteins of molecular mass 60 and 40 kDa in RT4, HT1376 and HT1197 cells and only the 40 kDa protein in the remaining 6 bladder cancer cell lines. Both the 60–40 kDa proteins are related to HAS1, since the expression of both of these proteins is ∼90% downregulated in a HAS1-AS transfectant when compared with HT1376 vector and HAS1-S transfectants. The 40 kDa HAS-1 related protein is not one of the known HAS1-variant proteins (i.e., HAS1-va, HAS1-vb or HAS1-vc), since the anti-HAS1 c-terminal peptide IgG used for immunoblot analysis does not recognize these variant proteins. In addition, as described above, the bladder cancer cell lines do not express any of the 3 HAS1 variants.
To determine the relationship between the 40 and 60 kDa proteins, we conducted a pulse chase experiment following metabolic labeling of HT1376 cells with 35S methionine. 35S-labeled HAS1 protein was immunoprecipitated using the anti-HAS1 IgG. As shown in Figure 4b, at 0 min chase, the anti-HAS1 IgG immunoprecipitates a ∼ 40 kDa protein. After 30–60 min chase, both the ∼40–60 kDa proteins are visible. This suggests that the 40 kDa protein is a precursor of the 60 kDa HAS-1 protein. At later times, only the 60 kDa protein is visible, suggesting that 40 kDa protein is posttranslationally modified to yield mature HAS1 protein. After 60 min of chase only the 60 kDa protein is visible. The ½ life of 60 kDa HAS1-protein was 2 hr. No HAS-related smaller products that represent degraded forms of the 60 kDa HAS1 protein were detected at various chase periods up to 24 hr. Therefore, it is likely that the 40 kDa protein observed in all bladder cancer cell lines, by immunoblotting, is the precursor of the mature HAS1 protein.
We also measured HA levels present in the conditioned media of various bladder cancer cells. As shown in Figure 4c, all bladder cancer cells secrete HA. Among these, RT4 cells secrete the highest amount of HA, followed by HT1376. In the HAS1-AS transfectant, HAS1 expression is ∼ 90% downregulated and these cells secrete 0.4 ± 0.2 μg HA/mg protein in their conditioned media when compared with 3.1 ± 0.5 μg HA/mg protein, secreted by HT1376 wild type cells. On the contrary, HAS1-S transfectant overexpress HAS1 and secrete 1.7-fold more HA in their conditioned media (5.4 ± 0.5 μg/mg protein). This suggests that HAS1 contributes significantly to HA production in bladder cancer cells.
HAS1 and HA localization in bladder tissues
We compared HAS1 and HA staining in 68 archival bladder tissues by immunohistochemistry. As shown in Figure 5 (panels a, b), normal bladder tissues do not stain for HAS1 or HA suggesting low to no expression of these molecules. Both low- and high-grade bladder tumor tissues stain with high intensity for HAS1 and the staining is in tumor cells (Fig. 5, panels c, e). Patchy but intense HA staining is observed in a low-grade bladder tumor tissue, whereas a high-grade bladder tumor specimen stains intensely and uniformly for HA (Fig. 5, panels d, f).
Table II shows the distribution of HAS1 and HA staining in bladder tissues. Ten of the 12 normal bladder tissues (83.3%) either did not stain or showed low intensity of staining for HAS1, whereas 12/12 tissues showed low-grade staining for HA. It is noteworthy that these normal bladder specimens were obtained from patients with bladder cancer, suggesting that HAS1 and HA expression does not change as a result of field changes in the urothelium, often seen in patients with bladder cancer. HAS1 staining was high in 44 of the 56 bladder tissues (78.6%), but the staining did not correlate with tumor grade or stage (p > 0.05). The increase in HAS1 staining in bladder tumor tissues, when compared with normal tissues is statistically significant (χ2: 17.3; p < 0.0001; relative risk = 4.7). Similar to HAS1 staining, HA staining is also high in bladder tumor tissues when compared with normal tissues (χ2: 26.4; p < 0.0001; relative risk = 5.3). Contrary to HAS1 staining, HA staining intensity increases with tumor grade; 50% of low-grade vs. 96.3% of high-grade bladder tumor tissues stain with high intensity (Table II). However, there was no substantial difference in the staining patterns of HAS1 and HA in Ta and T1 tumors (Table II). In this cohort, there were 3 patients with CIS, all of these had primary CIS and these patients were previously treated with intravesical immunotherapy (i.e., Bacille Calmette-Guerin). As shown in Table II, tumor specimens from 2 of these 3 patients showed high HAS1 staining and all 3 specimens overexpressed (i.e., high staining) HA.
Table II. Comparison of Has1 and HA Immunohistochemical Inferences1
High-grade intensity of staining for HAS1 or HA in bladder tumor tissues constituted a true positive inference. Low-grade staining intensity in normal bladder tissues constituted a true negative inference. Conversely, low-grade staining intensity in bladder tumor tissues constituted false-negative and high-grade staining intensity in normal bladder tissues constituted false-positive inferences, respectively. Sensitivity and specificity were calculated from these inferences. Specificity of HAS1 staining was 83.3% (10/12) and for HA 100% (12/12).
McNemar test showed a significant relationship between HAS1 staining and HA staining in G2 + G3 tumor tissues (p = 0.004; χ2 = 4.0; relative risk = 4.0). However, since only 50% of G1 tissues stain with high intensity for HA vs. 75% of the tissues staining high for HAS1, the correlation between HAS1 and HA staining was not significant in low-grade tissues (McNemar test; p > 0.05). There was also no statistically significant correlation between HA staining and tumor stage (p > 0.05).
In this study, the majority of specimens were obtained from patients who were operated in a county hospital, and therefore, follow-up information regarding disease progression, recurrence and/or treatment was available only on 18 patients. Among these 18 patients, the differences in HAS1 or HA staining inferences among those who recurred or nonrecurred was statistically significant (p = 0.036; Fisher's test). Out of the 18 patients, 2 had disease progression and HAS1 and HA inferences in the tumor specimens of these patients were high. The difference in HAS1 and HA staining inferences among patients who received prior intravesical treatment (chemotherapy or immunotherapy) and those who did not was statistically significant (p = 0.025; Fisher's test); comprehensive information regarding intravesical treatment was not available on several patients with superficial tumors, and therefore, this group of patients could not be analyzed as a separate category.
Correlation between HAS1 expression and HA urine test
Since elevated urinary HA levels are a sensitive and specific marker for detecting bladder cancer, regardless of tumor grade, we determined whether increased HAS1 expression in bladder tumor tissues constitutes a positive HA test. As shown in TableIII, HAS1 and HA staining inferences in bladder tissues and the HA test were positive in 82.9, 88.7 and 88.7% of individuals with bladder cancer, respectively. Statistically, there was a significant association between HAS1 staining and HA staining, as well as, between HAS1 staining and the HA urine test (p < 0.001). In this cohort, the HA-HAase test, which measures both urinary HA and HAase levels, had 91.4% sensitivity and 83.3% specificity.
Table III. Comparison of HAS1 and HA Immunohistochemical Localization with HA Urine Test Findings
HA urine test
In this study, we demonstrate that HAS1 type HA-synthase is expressed in bladder cancer cells and the expression is upregulated, both at the transcriptional and translational levels. Furthermore, increased HAS1 expression correlates with increased HA levels, in bladder tumor tissues, and also with a positive HA test. With the exception of lung tissues, HAS1 expression is not upregulated in most normal tissues, and it may have some role in angiogenesis.24, 35 However, recent studies on multiple myeloma, endometrial and ovarian carcinomas show that HAS1 expression is increased in tumor tissues and malignant cells.29, 30, 31, 32, 33, 34 Consistent with these observations, our results show that HAS1 transcript levels are up-regulated 5- to 10-fold in bladder tumor tissues regardless of tumor grade and stage.
This is the first study in which both HAS1 and HA expression has been examined in the same set of tumor tissues. The results presented in this study, as well as, a previous report from our laboratory13 demonstrate that increased HA concentration in bladder tumor tissues is contributed by both tumor associated stroma and tumor cells. Furthermore, a correlation exists between increased HA levels in bladder tumor tissues and a positive HA urine test.13 Results presented in this study show that a similar correlation also exists between increased HAS1 expression in bladder tumor tissues and a positive HA urine test. It is possible that the expression of other HAS proteins, i.e., HAS2 and HAS3, is also upregulated in bladder tumor tissues and contributes to increased HA levels, nonetheless, a positive correlation between HAS1 expression and increased tissue/urinary HA levels suggests that HAS1 expression in tumor cells is contributing, at least partly, to the increased HA levels in bladder tumor tissues.
The notion that HAS1 expression contributes to increased HA levels in bladder tumor tissues is further supported by our observation that the large HA polymer that is present in tumor tissue extracts has a molecular mass of 2 × 106 Da. This molecular mass of bladder tumor-associated HA is consistent with the length of the HA polymer that is synthesized by HAS1.25 Since HAS1 expression is elevated in malignant cells,24, 30, 31, 32, 33, 34 it may explain the high accuracy of the HA test in detecting patients with bladder cancer. Detection of HA fragments only in a high-grade bladder tumor tissue extract is also consistent with our earlier observations that in high-grade bladder cancer, the levels of HYAL1 type HAase are elevated. Thus, tumor-derived HAase plausibly degrades tumor-associated HA to generate angiogenic HA fragments.
Recently, Adamia et al have shown the presence of 3 HAS1 variants in CD19 + B cells from multiple myeloma patients.35 Our data on RT-PCR in 9 bladder cancer lines show that none of these cell lines express any of these 3 HAS1 variants. Among these 3 variants, the expression of only HAS1-va variant is observed in bladder tissues. It is noteworthy that in bladder tissues (i.e., normal and tumor), the HAS1 wild type PCR product is detected at PCR cycle 30, when compared with the HAS1-va PCR product, which is detected at cycle 44. This means that the expression of HAS1 wild type transcript in bladder tissues is significantly higher (i.e., ≥ 214 times) than the expression of HAS1-va transcript. Furthermore, the wild type HAS1 transcript is contributing to elevated HA levels in bladder tumor tissues and that the functional significance of HAS1-va remains uncertain in bladder cancer.
Tumor-associated HA is increased in several carcinomas and is known to promote tumor metastasis. HAS expression also appears to be associated with tumor metastasis. For example, HAS3 expression in tumor cells increases tumor growth and promotes anchorage-independent growth of colon cancer cells.39, 40 Both HAS1 and HAS2 levels are upregulated following oncogenic transformation of fibroblasts, which results in increased cell motility and tumor growth.29 Our recent results on HYAL1 HAase corroborate these findings and show that the HA-HAase system is a molecular determinant of bladder tumor growth, invasion and angiogenesis.41 Since both HAS1 and HYAL1 levels are elevated in bladder cancer and the HA-HAase system functions in bladder tumor growth and progression, it may explain why the HA-HAase test has high accuracy in detecting bladder cancer.
Our data show that the expression of HAS1 and HA in bladder tumor tissues is consistent with HA urine test results. For example, as with urinary HA levels, which are elevated in patients with bladder cancer, regardless of the tumor grade,6 HAS1 and HA expression in bladder tumor tissues is also similarly elevated in both low- and high-grade bladder tumors. Furthermore, determination of HAS1 expression in exfoliated cells by RT-real time PCR shows that HAS1 transcript levels are elevated in both low- and high-grade bladder tumor cells. This suggests that in bladder tumor tissues HA synthesized by HAS1 is released in urine, which then results in a positive HA urine test. We have previously shown that HYAL1 levels are elevated in G2/G3 bladder tumor tissues,13 which is consistent with the observation that urinary HAase levels are elevated in G2/G3 bladder cancer patients.6 It is noteworthy that the combined HA-HAase test detects bladder cancer regardless of the tumor grade with ∼90% accuracy.13
HAS1 expression has been shown to correlate with invasion, disease progression and poor survival in colon, ovarian and endometrial carcinomas.30, 31, 32 Although the number of bladder cancer patients with sufficient followup and preoperative clinical information related to recurrence, treatment and progression was small, our data indicate that higher HAS1 expression may be related to frequent recurrence, treatment response and disease progression. We are currently using tissue microarrays to examine the prognostic potential of HA, HAS1 and HYAL1 in bladder cancer.
In summary, HAS1 expression is elevated in bladder cancer where it also correlates with increased HA levels in bladder tumor tissues and a positive HA urine test. Furthermore, HAS1 expression may have prognostic potential related to tumor recurrence and progression.
The authors thank the invaluable input of Drs. Naoko Iida and Marvin Young, and Mrs. Marie Selzer, University of Miami Miller school of Medicine, during the course of this work. They also thank Soum D. Lokeshwar for technical assistance. The authors are grateful to the assistance of Ms. Teresa Rodriguez in specimen collection. They also thank the kind assistance of Dr. Martha Reyes and Mrs. Cynthia Soloway in providing clinical information related to tumor recurrence, treatment and progression.