Endogenous immune response to gangliosides in patients with confined prostate cancer



This article is corrected by:

  1. Errata: Erratum Volume 136, Issue 4, E204–E206, Article first published online: 10 December 2014


Our study investigated whether endogenous IgM antibodies to gangliosides occur in patients with early stages of prostate cancer (CaP) patients, after defining ganglioside profiles of CaP cell lines. Immune and resorcinol staining detected the presence of gangliosides GM3, GM2, GD3, GD2 and GD1a but not GM1a, GD1b or GT1b in the extracts of normal prostatic epithelial cells (PrEC) and neoplastic androgen-insensitive (PC-3, DU145) and -sensitive (LNCaP-FGC and LNCaP-FGC-10) CaP cells. Using a sensitive ELISA, developed and validated in our laboratory, the titers of IgM against 8 gangliosides from sera of patients with benign prostatic hyperplasia (BPH) (n = 11), organ-confined (T1/T2, n = 36) and unconfined (T3/T4, n = 27) CaP and age-matched healthy men (n = 11) were determined double-blinded. Using ANOVA and Fisher's least significant difference (LSD) methods, the log-titers among different groups were compared. CaP patients differed from healthy and BPH patients in increased titers against GD1a and decreased titers against GD3. Titers of antibodies to other gangliosides exhibited no difference between CaP patients and others. The specific augmentation of anti-GD1a IgM in patients with organ-confined CaP (stage T1/T2) but not in patients with unconfined CaP (stage T3/T4) or BPH or in healthy controls is striking. This finding together with identification of GD1a as a major ganglioside in CaP cell lines and with the accruing studies on the immunosuppressive nature of GD1a indicates that augmentation of anti-GD1a IgM in confined CaP may signify an early endogenous immune response to eliminate a “danger signal” from tumor microenvironment and circulation. © 2005 Wiley-Liss, Inc.

Prostate cancer (CaP) progresses imperceptibly in its early stages and is potentially curable when organ confined. A better understanding of the tumor biology that distinguishes healthy prostatic cells from cancer cells and organ-confined cancer from unconfined cancer would enable development of biomarkers for early detection and target antigens for novel therapies. Gangliosides, sialyl-oligosaccharides with ceramides, are overexpressed on tumor cell surfaces1 and differ in their nature and distribution between normal and cancer cells and among different types of cancers.1, 2, 3, 4, 5, 6, 7 Although gangliosides of melanoma, colon cancer and pancreatic cancer have been characterized, there is little information on the ganglioside profile of CaP.8, 9, 10

Tumor-associated gangliosides, when shed into the circulation,11, 12, 13, 14, 15 may elicit an endogenous IgM antibody response.16 In patients with advanced colon cancer, cryoablation (freezing and thawing) of hepatic metastases causes tumor necrosis and releases tumor-associated gangliosides into the circulation.16 Gangliosides released because of tumor necrosis augment IgM antibodies without the need for exogenous adjuvants.16 Antiganglioside IgM antibodies occur naturally in healthy individuals,17 but in cancer patients the level of antiganglioside antibodies varies according to the stage of disease and tumor progression.15, 18

We hypothesize that the endogenous IgM response to gangliosides might be an early immunologic event of tumorigenesis. Our pilot study examines whether specific antiganglioside IgM antibodies occur in the sera of patients with early-stage CaP, after comparing the profile of tumor-associated gangliosides in androgen receptor-positive (LNCaP-FGC and LNCaP-FGC-10) and -negative (PC-3 and DU 145) ATCC CaP cell lines. In our study, we used chemical staining and immunostaining to profile the expression of gangliosides on normal prostate epithelial cells and prostate cancer cells. We then examined the profile of antiganglioside antibodies in the sera of patients with confined (stage T1/T2) and unconfined (stage T3/T4) CaP, benign prostate disease or no prostate disease.


ANOVA, analyses of variance; ATCC, American Type Culture Collection; BPH, benign prostatic hyperplasia; CaP, prostate cancer; ELISA, enzyme-linked immunosorbent assay; HPTLC, high-performance thin-layer chromatography; HSA, human serum albumin; PBS, phosphate-buffered saline; PrEC, prostate epithelial cells; TSG, total serum gangliosides.

Material and methods

Cell lines

Normal prostate epithelial cell (PrEC) line was obtained from Cambrex BioScience (Walkersville, MD) and cultured in growth media PrEGM (CC-4177, Cambrex). Two androgen receptor-negative CaP cell lines (PC-3 [ATCC CRL-1435] and DU 145 [ATCC HTB-81]) and 2 androgen receptor-positive CaP cell lines (LNCaP Clone FGC [ATCC CRL-1740] and LNCaP Clone FGC-10 [ATCC CRL-10995]) were obtained from the American Type Culture Collection repository. All 4 CaP cell lines were established from metastases. Cell culture conditions, detachment, harvesting and cryopreservation of the cell lines are described elsewhere.19

Extraction and isolation of gangliosides

Glycolipids were extracted from CaP cells as described earlier.19 Gangliosides were isolated on columns (ENVI-Chrom P; Supelco, Bellefonte, PA) containing a resin made of small, nonionic, highly cross-linked styrene-divinylbenzene beads.19

High-performance thin-layer chromatography (HPTLC)

Ganglioside signatures of CaP cells were analyzed by HPTLC as described earlier.19 The following gangliosides were used as reference standards after they had been screened for purity and homogeneity: GM3 (Sigma [St. Louis, MO]: G 5642), GM2 (Sigma: G 8397), N-glycol-GM3 (gift from Dr. Adriana Carr, Center for Molecular Immunology, Havana, Cuba), GM1a (Sigma: G 7641), GD3 (Calbiochem [La Jolla, CA]: 345752), GD1a (Sigma: G 2392), GD2 (Advanced Immunochemical [Long Beach, CA]: IG6), GD1b (Sigma: G 8146) and GT1b (Sigma: G 3767). The gangliosides were spotted onto the plates using Linomat (CAMAG Scientific, Wilmington, NC). The plates were prerun in chloroform to eliminate neutral lipid and other contaminants that may interfere with the mobility of gangliosides. Gangliosides were visualized by heating at 100°C after spraying resorcinol-HCl reagent (10 ml of 2% resorcinol in water, 40 ml concentrated HCl, 0.125 ml 0.1M copper sulphate). Each chromatogram represented gangliosides extracted from 25 × 106 cells.

Immunostaining of thin-layer chromatograms

The following murine monoclonal antibodies against gangliosides were obtained from Seikagaku America (Falmouth, MA): GMB16 (anti-GM1a IgM; Cat. No.: 370685), GGR12 (anti-GD1b; Cat. No.: 370660), GD1a-1 (anti-GD1a IgG1; Cat. No.: 370706-1), GD1a-2 (anti-GD1a IgG2a; Cat. No.: 370706-2), GMR17 (anti-GD1a/GT1b IgM; Cat. No.: 370705) and GMR5 (anti-GT1b IgM; Cat. No.: 370675). Other monoclonal antibodies used in our study included KM696 (anti-GM2 IgM; Kyowa Hakko Kogyo, Tokyo, Japan), 14.G2a (anti-GD2 IgG2a; BD Biosciences (San Diego, CA); Cat. No.: 554272), MB3.6 (anti-GD3 IgG3; BD Biosciences; Cat. No.: 554274), and 14F7 (anti-GM3 [NeuGc] IgG1; gift from Dr. Adriana Carr). We recently reported the specificity of each monoclonal antibody when tested with a panel of 8 different gangliosides and a standardized ELISA protocol.19, 20

Gangliosides were separated on aluminum-backed silica gel plates, and the plates were dipped in 0.2% polyisobutyl-methacrylate in hexane for 1 min. After drying, the plates were blocked with PBS-1% HSA for 30 min, washed with PBS and dried. Each plate was overlaid with a monoclonal antibody diluted in PBS-1% HSA for 2 hr at 37°C. Primary antibodies were diluted as follows: GMB16 (1:400); KM696 (1:1,000); 14.G2a (1:500); MB 3.6 (1:250); GGR 12 (1:200); GMR 5 (1:500); 14F7 (1:1,000); GMR 17 (1:800); and GD1a-1 (1:500). The plates were then washed 3 times with PBS (for 3 min), dried and incubated with biotinylated rabbit anti-mouse secondary antibody (Jackson Immunoresearch, Pittsburgh, PA; rabbit anti-mouse IgG, Cat. No.: 315-065-008; rabbit anti-mouse IgM, Cat. No.: 315-065-049) diluted at 1:500 for 1 hr at 37°C. The plates were washed thrice in PBS, dried and incubated for 45 min in peroxidase-conjugated streptavidin (Jackson Immunoresearch, Cat. No.: 016-030-084) diluted in PBS at 1:2,500. The plates were developed with 4-chloro-1-naphthol solution (10 ml) and hydrogen peroxide (5 μl). A blue-grey color indicated specific binding of the antibody.

Patient population

Serum specimens were collected from 11 patients with benign prostatic hyperplasia (BPH), 36 patients with stage T1/T2 CaP (organ-confined) and 27 patients with stage T3/T4 (unconfined) CaP. Tumor-related variables included clinical stage, biopsy Gleason score and serum PSA; for our study, total serum ganglioside level was also considered a tumor-related variable. Informed consent was obtained from each patient in accordance with the Saint John's Health Center and John Wayne Cancer Institute Institutional Review Board (IRB) policy. Joint IRB committees of John Wayne Cancer Institute and Saint John's Health Center approved the study protocol.

Fifteen healthy male volunteers who had donated blood at Saint John's Health Center between January 1998 and November 2003 and had given consent for their blood samples to be used for research purposes were initially screened for serum IgM autoantibody titers against different gangliosides. Eleven of the 15 were selected as control subjects by computer matching of their ages with the ages of BPH and CaP patients (Table I). No control subject had active medical problems or was taking any medications.

Table I. Age, Disease Status and Treatment Status of 4 Groups in Clinical Study of Endogenous Response to Tumor-Associated Gangliosides
Age (years)Stage/gradePrior treatmentPSATotal gangliosides
  1. PSA, prostate-specific antigen; CaP, prostate cancer; AA, androgen ablation; CRx, chemotherapy; XRt, radiation; RP, radical prostectomy; BPH, benign prostatic hyperplasia

36 patients with organ-confined CaP (stages T1 and T2)
85T1a / 3/3Untreated2.6 
87T1a / 5/2AA/CRx1.607.5
71T1c / 3/3Untreated2 
66T1c / 3/3Untreated8.7513
74T1c / 3/3Untreated 10.6
72T1c / 3/3Untreated2.4315.8
72T1c / 3/3Untreated  
68T1c / 3/3Untreated 18.1
80T1c / 3/3Untreated3.7013.4
80T1c / 3/3Untreated4.0012.1
80T1c / 3/3Untreated4.5016.0
80T1c / 3/3Untreated5.0 
69T1c / 3/3Untreated5.7013.1
77T1c / 3/3Untreated19.115.0
49T1c / 3/3Untreated3.216.5
80T1c / 3/3Untreated11.817.1
55T1c / 3/3Untreated5.824.3
60T1c / 3/3Post RP6.7021.6
74T1c / 3/3RP0.0015.8
64T1c / 3/4RP AA1.47 
74T1c / 3/4RP  
86T1c / 3/4AA2.91 
73T1c / 3/4XRt0.09 
84T1c / 4/4Untreated3.2018.4
70T1c / 4/4XRt 12.3
68T2a / 3/2RP0.80 
93T2a / 3/3XRt/AA0.0990.6
80T2b / 3/3RP 3/912.10 
76T2a / 3/3Untreated3.9517.4
76T2a / 3/3Untreated 18.9
60T2a / 3/3Untreated6.5818.9
77T2a / 4/3Untreated5.515.8
68T2a / 3/4Untreated2.118.1
67T2b / 3/3Untreated  
64T2b / 3/4Untreated 16.5
61T2c / 3/4Untreated6.720.2
27 patients with unconfined CaP (stages T3 and T4)
91T3 / 3/3AA  
80T3 / 3/4Untreated 16.3
82T3 / 3/4XRt/AA21.8010
85T3 / 3/4XRt5.3118.3
81T3 / 3/4RP AA1.47 
70T3 / 3/4Untreated4.73 
69T3 / 3/4Untreated4.7315
70T3 / 3/4Untreated4.7317.3
85T3 / 3/4XRt5.3112.3
77T3 / 3/4RP4.30 
93T3 / 4/4Untreated1.3019.5
85T3 / 4/4Untreated2.9118.1
67T3 / 4/4Untreated4.929.7
67T3 / 4/4Untreated  
67T3 / 4/4Untreated  
82T4 / 3/4RP/XRt/CRx20017.8
84T4 / 3/4RP/AA7.5018.6
93T4 / 3/4AA7.0018.1
80T4 / 3/4AA28.0017.9
70T4 / 3/4AA30.00 
80T4 / 3/5Untreated25.4015.5
70T4 / 3/5Untreated25.4017
70T4 / 3/5Post CRx  
71T4 / 4/3RP/XRt67.0020
44T4 / 4/3Untreated17.016.00
62T4 / 4/5RP/AA0.10 
73T4 / 5/5XRt/CRx0.7020.6
11 patients with BPH
46BPH 0.3612.8
53BPH 0.6813.4
76BPH  10.8
75BPH 4.3012.6
73BPH 1.5010.4
72BPH 5.198.7
50BPH  8.7
75BPH 1.40 
82BPH 2.40 
64BPH 4.5715.4
86BPH 2.70 
11 healthy males

In all cases, serum was recovered within 6 hr after blood sampling, aliquoted, coded and stored at −70°C in JWCI's serum bank. For antibody analyses, sera were thawed once, vortexed extensively for 45 sec and aliquoted for further dilution. All analyses were carried out in a double-blinded fashion.

Measurement of total serum gangliosides (TSG)

TSG levels in 100 μl of serum were measured by an assay for lipid-bound sialic acid.5 Briefly, the concentration of sialic acid in serum extract was determined in 100 μl of chloroform/methanol (v/v: 2/1) by the resorcinol-HCl method21 after transferring the supernatant to a clean tube. The resorcinol/HCl reagent was mixed with supernatant and then heated at 100°C for 15 min. After cooling at room temperature, a butylacetate:n-butanol (v/v: 85/15) mixture was added and the organic layer containing the chromogen was read at an absorbancy of 580 nm. By using a standard curve obtained with commercial NeuAc (Sigma), the TSG was expressed as mg/dL. No phosphotungstic acid was used for the measurement of lipid-bound sialic acids. For routine analyses, varying concentrations of HSA served as a negative control. HSA spiked with a known amount of purified gangliosides served as a positive control. TSG levels were measured with an automated system at Dianon Systems (Stratford, CT).

Approximately 30% of the sialic acids extracted with this protocol may be derived from sialoproteins or proteolipids with sialic acids. However, some of the sialic acid-containing proteins may include ganglioside-binding transport proteins or immunoglobulins complexed with gangliosides. The reliability and reproducibility of this assay was confirmed by a double-blind analysis. The coefficient of variation was always <15%.

Quantitation of cell-surface antigens

Quantitation of cell-surface antigens was performed using methods described previously.22, 23 Trypsin was strictly avoided for harvesting the cells. All cells were washed once with RPMI-4% HSA and then suspended in cold RPMI-4% HSA. Cells were counted for viability using 0.2% trypan blue. Suspensions of 0.5 × 106 cells in microcentrifuge tubes containing 60 μl of solution were treated with secondary antibody (background), with primary and secondary antibody (experimental), or with class-matched isotypes of the primary antibody (negative control). The final dilution of the primary (120 μl) antibodies (determined by box titration) varied with different monoclonals, as follows: 14F.7 (anti-GM3-N-glycolyl IgG1, 1:48,000), KM696 (anti-GM2 IgM, 1:10,000), 14.G2a (anti-GD2 IgG2a, 1:1,500), MB3.6 (anti-GD3 IgG3, 1:2,000), GGR12 (anti-GD1b IgG3, 1:200), GMR17 (anti-GD1a/GT1b IgM, 1:8,000), GMR5 (anti-GT1b IgM, 1:10,000) and GMB16 (anti-GM1a IgM, 1:2,000). When the monospecificity of these antibodies was assessed by cell-suspension ELISA, all monoclonals except GMR17 exhibited remarkable specificity for their respective gangliosides; GMR17 stained both GD1a and GT1b.

ELISA for serum antiganglioside antibodies

GM1, GM2, GM3, GD1a, GD1b, GD2, GD3 and GT1b were obtained from 1 or more of 5 commercial manufacturers: Sigma-Aldrich (Dallas, TX); Calbiochem-Novabiochem (Pasadena, CA); Alexis Biochemicals (San Diego, CA); Advanced Immunochemical (Long Beach, CA); and Accurate Chemical and Science (Westbury, NY). Purity and homogeneity of these commercially obtained gangliosides were compared by chemical staining with resorcinol-HCl and by immunostaining with monospecific monoclonal antibodies19 (Fig. 1). Purified GD1a contained a minor contaminant identified as GD2 by its mobility and by its staining with anti-GD2 antibody 14.G2a (Fig. 1b). The level of this contaminant varied with different commercial preparations but was lowest in GD1a obtained from Sigma-Aldrich. GD3 obtained from Calbiochem was free of contamination by resorcinol and by immunostaining (Fig. 1c). Immunostaining revealed no contaminants in GM2, GD3 and GD1a (Fig. 1d). GD2 contained an alkali-susceptible fraction identified as O-acetylated GD2.19 Both GD2 and O-acetylated GD2 were stained by anti-GD2 MAb 14.G2a.

Figure 1.

Source and assessment of purity of gangliosides used in ELISA by TLC-resorcinol staining and immunostaining with specific murine monoclonal antibodies (MAbs) against all gangliosides listed in Cabot.24(a) Different commercial preparations (Sigma, Alexis, Advance Immunochemical and Calbiochem) of 3 nmol of the ganglioside GD1b, purified from human and bovine brain, were analyzed by TLC and stained by resorcinol-HCl. The positions of the contaminating glycolipids are indicated by the thin line on the right. No contamination could be detected in the GD1b obtained from Alexis and hence used for ELISA. (b) Different commercial preparations (sources: Calbiochem, Alexis, Accurate and Sigma) of 3 nmol of GD1a were resolved on TLC and stained both by resorcinol and monoclonal antibodies for all gangliosides including GD1a (GMR17) and for GD2 (14.G2a). The figure shows staining done with MAbs GMR17 and 14.G2a. Resorcinol staining exhibited a distinct GM1-like contaminant in the preparation obtained from Accurate. Such a contaminant was feeble in the preparation obtained from Calbiochem and Alexis. No such contaminant was detectable in the preparation obtained from Sigma. Interestingly, there is a contaminant at the position of GD2 stained by anti-GD2 MAb 14.G2a in the preparations obtained from Calbiochem, Alexis and Sigma. The contaminant level is very low in Sigma. Because Sigma preparation did not contain GM1-like contaminant and traces of GD2 contaminant, we have used GD1a obtained from Sigma. (c) The commercial preparations (Alexis, Calbiochem and Sigma) were free of contaminants stainable by resorcinol or MAbs. We have selected GD3 from Calbiochem. Sigma stopped the supply of GD3. (d) We have also tested purity by immunostaining after mixing all gangliosides (GM3, GM2, GM1, GD1a, GD1b, GD2, GD3, GD1b and GT1b). TLC immunostaining of each ganglioside was done with specific MAb. We have used Seikagaku MAb 52-566 for GD3 (Calbiochem), KM696 for GM2 (Sigma), GMR17 for GD1a (Sigma) and 14.G2a for GD2 (Advanced Immunochemical). 14.G2a also identified O-acetylated GD2 in the preparation. The resorcinol profile of the mixture is shown on the left.

After screening for purity and homogeneity, gangliosides were coated onto microtiter plates, in accordance with an ELISA protocol described earlier.20 This ELISA considers the following factors: (i) microtiter plate type; (ii) purity of the commercial ganglioside; (iii) antigen-coating strategy; (iv) nature of protein in the blocking buffer; (v) concentration and nature of the detergent in the wash buffer; (vi) background; and (vii) positive and negative controls.23 Briefly, the microtiter plates (Falcon Probind 3915) were coated with an ethanolic suspension of gangliosides (3 nmol/100 μL/well) (Sigma) and dried in a vacuum desiccator for 2 days. The plates were blocked with PBS-4% HSA, pH 7.2, for 90 min. Sera were diluted to 1:100 and incubated in a water bath for 30 min at 37°C. Sera were further serially diluted to 1:6,400, overlaid on the plates, and incubated at 37°C for 2 hr. The plates were washed 5 times with washing buffer (PBS-0.1% HSA-0.1% Tween-20). Anti-human IgM coupled to peroxidase (Jackson Immunoresearch) was used as the second antibody at a dilution of 1:5,000, and the plates were incubated with second antibody (100 μl/well) for 1 hr at 37°C. The plates were washed as before. The substrate, o-phenylenediamine hydrochloride (OPDE) (20 mg) dissolved in citrate-phosphate buffer (pH 5.0) and hydrogen peroxide (10.5 μL/25 ml of buffer), was added to the plates (100 μL/well) and incubated for 45 min in the dark at 25 μC. The enzymatic oxidation was arrested with 120 μL of 6N H2SO4. The absorbancy difference at 490 nm and 650 nm was measured in a UV max kinetic microplate reader (Molecular Devices, Sunnyvale, CA). The values were corrected for background (wells without antigen) and expressed as titers. The reliability and reproducibility of this assay were previously validated.23 Because the assay detects IgM that can recognize and bind to gangliosides coated onto the plates, it measures the titer of free IgM but not IgM-ganglioside complexes.

Statistical analysis

Analyses of variance (ANOVA) were performed to compare antiganglioside IgM antibody titer among healthy volunteers, BPH patients and patients with stage T1/T2 or stage T3/T4 CaP. Data were log-transformed for compatibility with the assumption of normal distribution and common variance. Fisher's least significant difference (LSD) method was used for pairwise comparisons of values significant at the 0.05 level.


Ganglioside profiles of PrEC and CaP cells

Three different methodologies were used to characterize ganglioside profiles from the 5 cell lines. In the first method, 1- and 2-dimensional chromatograms of glycolipid extracts were stained with resorcinol-HCl; bovine brain gangliosides were used as reference standards. In the second method, 1- and 2-dimensional chromatograms were stained with specific murine monoclonal antibodies. In the third method, tumor cell-surface gangliosides were measured directly by cs-ELISA with murine monoclonal antibodies to gangliosides.

Resorcinol characterization of gangliosides

Resorcinol staining provides approximate identification of the gangliosides in CaP cells compared to the mobility of the standard bovine brain gangliosides. Figure 2a shows the ganglioside profiles identified by resorcinol-HCl staining of 1-dimensional chromatograms for normal and CaP cell lines. In general, gangliosides stained much less intensely in PrEC cells than in CaP cell lines. LNCaP-FGC and LNCaP-FGC-10 CaP cell lines stained less intensely than PC-3 and DU-145 CaP lines. Based on the staining intensity of 1-dimensional chromatograms, the relative distribution of gangliosides in normal PrEC was as follows: GM3 > GD1a = GT1b > GM2 = GD3 = GD1b > GM1a (commonly known as GM1). The profile in PC-3 and DU 145 was as follows: GD1a > GM1 > GM2 > GD3 > GM3. GT1b and GD1b could not be identified. In LNCaP-FGC and LNCaP-FGC-10 cell lines, resorcinol staining revealed the presence of GM1 > GM2 > GM3. Other gangliosides were not detectable in 1-dimensional chromatograms.

Figure 2.

Ganglioside profiles in normal PrEC and CaP cell lines stained by resorcinol-HCl. (a) Unidimensional chromatograms show the ganglioside profile of normal PrEC (lane 1), PC-3 (lane 2), DU 145 (lane 3), LNCaP-FGC-10 (lane 4) and LNCaP-FGC (lane 5). The standards were bovine brain GM2 and GD1a (lane 6) and bovine brain gangliosides (lane 7). Each lane contains ganglioside extract obtained from 25 million cells. The concentration of each ganglioside in the bovine brain standard is 3 nm. The upper horizontal line demarcates GM1 and the lower horizontal line refers to the position of GD1a. The bottom line represents the point of application. The solvent system is chloroform/methanol/0.2% CaCl2 (v/v/v: 55/45/10). (b) Two-dimensional chromatograms compare all cell lines except LNCaP-FGC with the standard bovine brain gangliosides run simultaneously. Each lane contains ganglioside extract obtained from 25 million cells. The concentration of each ganglioside in the bovine brain standard is 3 nmoles. The vertical line represents the direction of flow of the first solvent system, chloroform/methanol/0.2% CaCl2 (55/45/10, v/v/v). The horizontal line represents the direction of flow of the second solvent system, chloroform/methanol/2.5M NH4OH in 0.25% KCl 50/40/10, v/v/v). GD1a and GM2 are the most prevalent gangliosides in all cell lines.

Resorcinol staining of 2-dimensional chromatograms identified GD1a as a common ganglioside in all CaP lines, including LNCaP-FGC (Fig. 2b). GM1, GM2 and GM3 were also abundant, whereas disialo-gangliosides GD3 and GD2 were present in small amounts. A spot identical to GD1b was observed in PC-3 but not in DU 145 and LNCaP-FGC. A spot identical to GT1b was detected in all cell lines except LNCaP-FGC. Several additional spots could not be identified with standards, suggesting they may be unique to prostate cancer cells.

Immunochemical characterization of gangliosides

For specific identification of the gangliosides, 1- and 2-dimensional chromatograms were immunostained with murine monoclonal antibodies specific for each ganglioside except GM3, for which no antibody exhibited specificity. All of the lots of the monoclonal antibodies were used after testing for their specificity by assessing their relative reactivity to all of the gangliosides on ELISA.24

KM696, the monoclonal specific for GM2, stained a single fraction in PrEC cells (Fig. 3a) and 2 distinct spots in the 2-dimensional chromatograms of all 4 CaP lines (Fig. 3b, arrows). The weaker of the 2 spots could be a GM2 with an altered fatty acid chain (change in length, double bonds or hydroxyl groups) or altered sialic acid. Presence of GM2 in CaP cells is confirmed by both resorcinol and immunostaining. Using KM696 in immunohistochemistry, Zhang et al.10 have observed no difference in the intensity of GM2-staining between normal and neoplastic prostatic epithelial cells.

Figure 3.

Immunochemical characterization of ganglioside extracts from CaP cell lines and normal PrEC. (a) Unidimensional chromatogram shows that PrEC extracts reacted with KM696 (murine monoclonal antibody specific for GM2) but not with GMB 16 (specific for GM1) or GMR 5 (specific for GT1b). Bovine brain standards, GM2, GM1 and GT1b, reacted with all 3 antibodies. (b) Two-dimensional chromatogram shows that KM696 recognized only GM2 in all cell lines. The 2 fractions of GM2 are indicated by a larger band (right arrow) and a weak band (left arrow). The solvent systems used for first and second runs are reported in Material and Methods. (c) Two-dimensional chromatogram identifies GD2 and 14.G2a-reactive gangliosides. Standard bovine gangliosides (3 nmol each) stained in resorcinol-HCl. Two fractions in PC-3 and LNCaP-FGC and 1 fraction in DU 145 were stained with 14.G2a. (d) Two-dimensional chromatogram identifies GD1a. Standard bovine gangliosides (3 nmol each) and DU 145 stained in resorcinol-HCl. The 3 monoclonal antibodies (Seikagau America: clones GD1a-1, GD1a-2a and GD1a-2b) to GD1a stained GD1a distinctly in DU 145. The low reactivity of clone IgG2a suggests that the concentration of antibody may be low. (e) Two-dimensional chromatogram identifies GD1a with clone GD1a-1 (IgG1) in both PC-3 and LNCaP cell lines.

14.G2a (IgG2a), the monoclonal specific for GD2, stained 2-D chromatograms of all cell lines (Fig. 3c). Whereas 14.G2a stained only 1 spot in DU 145, it stained 2 distinct spots in PC-3 (1 spot above GD2). The nature of the upper spot is not known. Presence of GD2 in CaP cells is confirmed by both resorcinol and immunostaining.

GD1a was identified by 3 monoclonal clones: GD1a-1, GD1a-2a and GD1a-2b. All 3 antibodies stained 1 spot in 2-D chromatograms of DU 145 (Fig. 3d). Only 1 of the clones was used to stain 2-D chromatograms of PC-3 and LNCaP-FGC. Results of resorcinol and immunostaining indicated that GD1a is a major ganglioside of all CaP cell lines (Fig. 3e).

MB3.6, the monoclonal specific for GD3, was applied to 1-dimensional chromatograms of all cell lines. Only DU 145 exhibited a faint staining. It appears that GD3 is not a major component of CaP cells.

Staining with murine monoclonal antibodies to GD1b (GGR12), GM1a (GMB16) and GT1b (GMR 5) failed to stain 1-dimensional chromatograms of all cell lines (data not shown). Failure of GMB16 to stain GM1a is interesting because resorcinol-HCl stained GM1 intensely and the precursor of GD1a is GM1a. It is possible that GM1a may not occur in detectable quantity in CaP cells; instead there may be a variant of GM1 or neolacto- or globo- series of a glycolipid that migrates similar to GM1. Recently, we were able identify the fraction as GM1b.19

Cell-surface expression of gangliosides

Gangliosides not identified in chromatograms of cell extracts may be detected by direct monitoring of the cell surface with ganglioside-specific monoclonal antibodies.22, 23 Figure 4 shows the expression of gangliosides on the surface of the 4 CaP cell lines. PrEC cells expressed low levels of gangliosides on the cell surface; the only detectable ganglioside in PrEC cells was GM1a, which was also found in low levels on CaP cell lines. Cell-surface expression of GM2 varied as follows: LNCaP-FGC > DU 145 > PC-3 > LNCaP-FGC-10. GD1a was found prominently on the cell surface of DU 145, less on PC-3 and not on other CaP cells. GD1b and GT1b, which were not detected by HPTLC immunostaining, were located on the surface of all CaP cell lines. GD2 was expressed on the cell surface of CaP cell lines. GD1a was the most abundant ganglioside in CaP cells, followed in decreasing order by GM2, GM1b, GD2 and GM3.

Figure 4.

Cell-surface expression of gangliosides on normal PrEC and CaP cell lines. The cell-surface density of gangliosides (density of expression in Cs-ELISA) was comparable to total fluorescence intensity (TFI) (mean FI × number of events).25 For each cell line, analyses were done in triplicate and negative controls included treatment of cells with nonspecific isotype of each monoclonal antibody (IgM, IgG1, IgG2a, IgG3). Reactivity of the isotypes with cell surface is negligible. GM3 was not tested because no specific antibody is commercially available. Normal epithelial cells failed to express gangliosides other than GM1a, but GM1a expression was negligible in all cell lines. All CaP lines expressed GM2 on the cell surface; GM2 expression was as follows: LNCaP-FGC > DU 145 > PC-3 > LNCaP-FGC-10. N-glycolyl GM3 and GD3 were negligible. GD1a was the most prominent ganglioside in DU 145 > PC-3. Cell-surface expression of GD1a was low in other cell lines, particularly in LNCaP-FGC and LNCaP-FGC10. All CaP cells expressed low levels of GD1b and GT1b on the cell surface.

Endogenous IgM response to gangliosides

To determine whether CaP-associated gangliosides elicit an endogenous immune response, we measured titers of antiganglioside IgM antibodies in the sera of CaP patients, patients with BPH and age-matched healthy male volunteers (Table I). Titers derived from ELISA analyses were converted to natural log titers; mean and median values were calculated (Table II). Because these values were compatible with the assumptions of the normal distribution and common variances, ANOVA was done. Two-tailed p-values revealed significant intergroup differences in titers of anti-GD1a and anti-GD3 but not anti-GM1, anti-GM2, anti-GM3, anti-GD2, anti-GD1b or anti-GT1b (Table II). The LSD method used for pair-wise comparison showed that the log titers of anti-GD3 IgM were significantly lower in BPH and CaP groups than in controls. Log titers of anti-GD1a IgM were similar between control and BPH groups. Log titers of anti-GD1a were significantly higher in patients with confined CaP (T1/T2) than in healthy controls (p = 0.02), patients with BPH (p < 0.008) or patients with unconfined CaP (p = 0.002) (Table III).

Table II. ANOVA Assessment of p-Values Showing Significant Differences in Log Titers of Anti-GD3 and Anti-GD1a IgM Antibodies in Sera from Healthy Controls and Patients with Benign Prostatic Hyperplasia (BPH) or Prostate Cancer (CaP)
Log titerHealthy (n = 11)BPH (n = 11)T1/2 CaP (n = 36)T3/4 CaP (n = 27)ANOVA
Mean ± SDMedMean ± SDMedMean ± SDMedMean ± SDMedp-value
GM14.88 ± 0.614.614.73 ± 0.424.615.33 ± 1.014.615.09 ± 0.854.610.154
GM25.74 ± 0.845.995.48 ± 1.434.615.40 ± 1.054.615.41 ± 1.084.610.820
GM35.27 ± 0.984.614.89 ± 0.764.615.20 ± 1.034.614.93 ± 0.734.610.493
GD35.90 ± ± 0.004.614.93 ± 0.794.614.61 ± 0.004.61<0.0001
GD24.82 ± 0.514.615.42 ± 1.314.615.67 ± 1.135.705.50 ± 1.084.610.152
GD1a5.04 ± 0.604.614.93 ± 0.614.615.77 ± 1.115.705.04 ± 0.754.610.0033
GD1b6.00 ± 1.026.384.89 ± 0.654.615.35 ± 1.114.615.50 ± 1.314.610.146
GT1b5.93 ± 0.966.405.95 ± 1.006.556.22 ± 1.406.316.34 ± 1.036.550.699
Table III. Pairwise Comparison by Fisher's Least Significant Difference Method Reveals that Anti-GD1a IgM Log Titers Significantly Distinguish Confined CaP (T1/T2) From Healthy, Benign Prostatic Hyperplasia (BPH) and Unconfined Prostate Cancer (CaP) (T3/T4)
Pairwise comparisonsAnti-GD3Anti-GD1a
  • *

    Lower than healthy volunteers.

Healthy vs. BPH<0.0001*0.776
Healthy vs. T1/2 CaP<0.0001*0.0201
Healthy vs. T3/4 CaP<0.0001*0.996
BPH vs. T1/2 CaP0.1380.0079
BPH vs. T3/4 CaP1.0000.730
T1/2 CaP vs. T3/4 CaP0.0460.002


Our investigation examined the endogenous immune response to gangliosides in patients with early stages of CaP and its correlation with CaP-associated gangliosides. If the findings are further validated, it may help to develop and establish a reliable immunologic marker for early detection of confined CaP. Our study is based on 4 tenets. First, tumorigenesis may involve the biosynthesis of ceramides,24 which mediate apoptosis. Tumor cells can avoid ceramide-mediated apoptosis by glycosylating ceramide24, 25, 26, 27 and storing the glycosylated products as sialyllactosylceramides or gangliosides in the cytoplasm or on the outer layer of the bilayered cell membrane. Second, the qualitative and quantitative profiles of gangliosides differ between normal and neoplastic cells and among cancers.1 For example, GM3 is the dominant ganglioside in melanocytes, whereas GD3 is the dominant ganglioside in melanoma cells that have metastasized.1, 2, 3, 4 Pancreatic adenocarcinoma and hepatic metastases of colorectal carcinomas express GD3 poorly but overexpress GM2, GD1b and GT1b.5, 6, 16 Third, during tumor cell proliferation and necrosis, the tumor-associated gangliosides may be released into the tumor microenvironment and enter into the circulation. Finally, circulating tumor-gangliosides suppress a variety of immunofunctions.28 In view of their immunosuppressive potential to alter immune functions, we perceive that they are recognized as “Danger Signals” by the host's immune system,29, 30 as evidenced by the endogenous antiganglioside immune response. The endogenous IgM against gangliosides may facilitate elimination of these danger signals to restore immune competence of the host. The present investigation provides supporting evidence for the above tenets.

Ganglioside profiles of normal and neoplastic prostateepithelial cells

This is not the first study to investigate gangliosides of normal and neoplastic prostatic epithelial cells. Shiraishi et al.8 used chemical, enzymatic and immunostaining procedures to analyze ganglioside profiles of benign human prostate tissue. They showed that the monosialoganglioside fraction contained GM3 and GM1 plus multiple species of monosialylated lactosamine; the disialoganglioside fraction contained GD3, GD1a, GD2 and GD1b and the GT1b was the major trisialoganglioside. Satoh et al.9 characterized prostate cancer tissue by its reduced expression of lactosyl and globoside series glycolipids, reduced expression of long-chain glycolipids and unchanged expression of GM3 and GD3. They concluded that prostate carcinogenesis was associated with inhibition of sugar chain elongation but not sialylation. The primary observations of the above studies are based on the resorcinol-HCl staining of the gangliosides resolved on 1-dimensional chromatography, which cannot specifically identify several long-chain gangliosides, including GD1a.

In the present investigation, we have characterized gangliosides, after examining their profiles in 2-dimensional chromatography and staining with both resorcinol-HCl and epitope-specific monoclonal antibodies. The specificities of the monoclonal antibodies used in our investigation were reported earlier.19 Our findings establish the absence of GM1a, GD1b and GT1b in the cell lines analyzed and confirm the presence of GD1a and GD2. Although resorcinol staining showed that all CaP cell lines contained abundant GM3, GM2 and GM1 (monosialogangliosides), immunostaining with GMB16 revealed that the intense resorcinol-positive GM1 fraction was not GM1a. We have recently identified this ganglioside as GM1b.19

High expression of GM2 in cancer cell lines is not surprising; Zhang et al.10 used immunohistochemical staining with KM696 to identify abundant GM2 in normal prostatic epithelial cells and in primary and metastatic prostate cancer cells. Their study of biopsy specimens eliminated concerns that elevated GM2 levels might be an in vitro artifact of increased N-acetyl GalNAc-transferase (GM2-synthase) activity under tissue culture conditions.31

Immunosuppression by gangliosides released from tumor cells

Several studies support the contention that the gangliosides in the cytoplasm and on the cell surface are released into the tumor microenvironment and circulation.11, 12, 13, 14, 15 In pancreatic cancer, melanoma and sarcoma, the level of serum gangliosides correlates with tumor burden.5, 15, 18 Circulating tumor-derived gangliosides can suppress a variety of cellular immune functions,28 possibly by interfering with cytokine function. For example, melanoma-associated GD1b binds to IL-2, an antitumor cytokine.33

There is considerable evidence to implicate GD1a as an immunosuppressive ganglioside. GD1a significantly inhibits myeloid colony formation,34 and it induces protein secretion and mRNA expression of interleukin-10 (the cytokine-inhibiting cytokine) in T cells by regulating the protein-tyrosine kinase signaling pathway.35 Exposure of dendritic cells to GD1a reportedly ablates the LPS-induced upregulation of co-stimulatory molecules CD80 and CD86, reduces expression of CD40 and CD83 and inhibits nuclear binding of NF-kappa B and release of IL-12 and TNF-α.36 Another study reported that GD1a failed to induce a normal T-cell proliferative response to tetanus toxoid.37 Co-culture of T cells with highly purified GD1a can inhibit production of IFN-γ mRNA and protein.38 Finally, GD1a can promote tumor angiogenesis by enhancing VEGF response of endothelial cells in the tumor microenvironment.39

We believe that increased levels of specific anti-GD1a IgM may represent the host's attempt to arrest the immunosuppressive effects of GD1a. Such an endogenous immune response is not surprising in the context of Matzinger's hypothesis,29, 30 which proposes that self-antigens may act as danger or distress signals that elicit an immune response in the absence of exogenous adjuvants.

Endogenous immune response to gangliosides

Gangliosides do not require T-cell help to produce antibodies.32 Human antiganglioside antibodies are invariably IgM and occur at low levels in healthy individuals.5, 17 We previously observed an endogenous antiganglioside antibody response in patients with early sarcoma18 and during postoperative monitoring of patients who had undergone cryosurgical ablation of liver metastases from colon carcinomas.16 In the latter group, necrosis induced by cryosurgical ablation of tumors increased the serum level of total gangliosides. Postoperatively, this increase elicited an increase in IgM titers against GM2 and/or GD1b and/or GT1b (gangliosides found on colorectal cancer cells that metastasize to the liver) but not against GM3 and GM1 (gangliosides found on normal liver cells).

In the present investigation of the antiganglioside response to human prostate cancer, CaP patients had lower levels of anti-GD3 IgM than did healthy controls. This difference was significant (p < 0.0001) as assessed by Fisher's least significant difference method, which considers differences in sample size. Although our analysis did not consider differences in mean age of the CaP, BPH and healthy control populations (75 ± 11, 70 ± 13 and 64 ± 12 years, respectively), we previously reported that the levels of antiganglioside antibodies in healthy individuals tend to decline with age and remain low after the age of 50.17 Decreased levels of anti-GD3 IgM in patients with CaP might also reflect either decreased shedding of the ganglioside or lack of immunogenicity for tumor-associated GD3, as claimed earlier.40 Although GM3 and GM2 are the major gangliosides of CaP cells, they failed to induce significant IgM responses in patients with CaP. These 2 gangliosides are prevalent in both normal and neoplastic CaP cells.10 Although GM2 is considered immunogenic in vivo,41, 42, 43 absence of a significant endogenous immune response to GM2 in CaP patients could be due to overexpression of GM2 both in normal cells and in CaP cells.10 Possibly, GM2 may not be an ideal tumor antigen to be targeted by active specific or passive immunotherapy.

Both resorcinol-HCl staining and immunostaining of 1- and 2-dimensional chromatograms with 14.G2a identified GD2 in CaP cells but not in normal epithelial cells. GD2 is considered to be immunogenic in human cancers41, 44 and considered as a potential target for both passive44, 45, 46 and active specific immunotherapies47 of melanoma. Although we could not observe any significant difference in the anti-GD2 response among our study groups, we did not test the sera against O-acetylated GD2 only.

The most interesting and salient finding of our investigation is the increased production of endogenous IgM antibody against CaP-associated GD1a in the sera from patients with organ-confined CaP but not in sera from patients with unconfined CaP. GD1a is not CaP specific; it occurs in neural48, 49, 50, 51 and extraneural3, 52, 53, 54, 55, 56 tissue and sera57 of patients with cancer. It is also found in normal tissue.4, 58, 59, 60, 61, 62, 63 Despite this lack of specificity, GD1a appears to be immunogenic. Kanda and Watanabe64 showed that GD1a greatly enhanced spontaneous IgM production by human peripheral blood mononuclear cells (PBMC) in vitro, without affecting proliferation or viability of PBMC; however, exposure of isolated B cells alone to GD1a did not alter IgM production. The GD1a-induced production of IgM by PBMC was blocked by anti-IL-6 or anti-IL-10 antibodies. GD1a also increased the production of IL-6 and IL-10 by monocytes but not by T cells or B cells. Production of both cytokines was blocked by calmodulin (CAM)-dependent phosphodiester (PDE) inhibitors but not by other signal-transducing enzyme inhibitors. The authors suggest that GD1a may indirectly enhance B-cell production of IgM by increasing monocyte production of IL-6 and IL-10 via CAM-dependent PDE activity. Possibly, a similar mechanism may be operating in the patients with confined CaP, when GD1a is newly released into the tumor microenvironment and into the circulation. We are investigating the mechanism-specific augmentation of anti-GD1a IgM in cancer patients.

We propose that the spontaneous induction of IgM against CaP-associated GD1a begins with the abnormal proliferation of neoplastic prostatic epithelial cells. Abnormal proliferation is accompanied by high levels of cell necrosis; necrotic cells may release gangliosides into the tumor microenvironment. Among these tumor-gangliosides, GD1a may induce infiltrating or regional lymph node or peripheral blood monocytes to produce anti-inflammatory cytokines (IL-10 and/or IL-6), which may stimulate B cells. These B cells may recognize GD1a as a danger signal and produce IgM against GD1a, to eliminate GD1a from tumor microenvironment and to restore immunocompetence and homeostasis. As levels of circulating GD1a decrease, we speculate that anti-GD1a IgM may target GD1a-rich tumor cells, which may contribute to the reduction in the endogenous IgM response against GD1a.

Expression of GD1a by CaP cells may be altered at subsequent stages of disease progression. Satoh et al.9 observed that the staining of frozen CaP tissue with APG1, a murine monoclonal antibody that binds to sialoglycosphingolipids, decreased in intensity with increasing grade of cancer. This suggests a change in the pattern of gangliosides during early phases of neoplastic differentiation and disease progression. Although the exact nature of the ganglioside that decreased with increasing grade is not known, our observed difference in the anti-GD1a IgM response between confined and unconfined CaP suggests that GD1a may be the most prevalent ganglioside of confined CaP and expression of this ganglioside might decrease with tumor progression.

Although the double-blinded design of our study lends support to our findings, a larger study of patients with untreated organ-confined and unconfined CaP is necessary to validate a pivotal role for GD1a in the endogenous immune response to CaP. One of the major limitations of our study is the availability of cell lines from organ-confined CaP. Although we have relied on ganglioside profiles of ATCC cell lines, there is a need for a detailed immunohistochemical characterization of gangliosides in CaP cells from tumor biopsies to validate our observations. Comparing the profiles of gangliosides from tumor biopsies and antiganglioside IgM antibodies between primary and metastatic CaP may provide a better understanding of the host response to tumorigenesis.

Our preliminary study identified the humoral immune response to a tumor ganglioside as a potentially valuable tool to understand the biologic changes associated with progression of human CaP. Because it is not possible to detect the quantity of specific tumor gangliosides shed into the circulation, we used a sensitive ELISA protocol for antiganglioside IgM; this assay has been validated for screening sera of melanoma patients receiving a cancer vaccine in a Phase III multicenter clinical trial based at the John Wayne Cancer Institute.47 In our study, this ELISA revealed that serum levels of anti-GD1a IgM antibody were significantly higher in patients with organ-confined CaP than in healthy controls (p = 0.02), patients with BPH (p < 0.008) or patients with unconfined CaP (p = 0.002). The uniquely elevated endogenous immune response to GD1a in our study merits further investigation as an immunologic marker of neoplastic transformation of prostate epithelial cells.


We thank Dr. J. Portoukalian, INSERM, Hospital Edouard, Lyon, France, for advice and guidance regarding purification and characterization of tumor gangliosides; T.S. Saravanan, Ph.D., and M. Verma, M.B., B.S., for technical and database support; Ms. K. Hirsch, Mr. A. Blackstone (Figs. 2 and 3) and Ms. M. Rice (Figs. 1 and 4) for the preparation of figures; and Ms. G. Berry for meticulous editing of the manuscript.