Abstract
Strategies to improve the early diagnosis of prostate cancer will provide opportunities for earlier intervention. The blood-based prostate-specific antigen (PSA) assay is widely used for prostate cancer diagnosis but specificity of the assay is not satisfactory. An algorithm based on serum levels of PSA combined with other serum biomarkers may significantly improve prostate cancer diagnosis. Plasma glycan-binding IgG/IgM studies suggested that glycan patterns differ between normal and tumor cells. We hypothesize that in prostate cancer glycoproteins or glycolipids are secreted from tumor tissues into the blood and induce auto-immunoglobulin (Ig) production. A 24-glycan microarray and a 5-glycan subarray were developed using plasma samples obtained from 35 prostate cancer patients and 54 healthy subjects in order to identify glycan-binding auto-IgGs. Neu5Acα2–8Neu5Acα2–8Neu5Acα (G81)-binding auto-IgG was higher in prostate cancer samples and, when levels of G81-binding auto-IgG and growth differentiation factor-15 (GDF-15 or NAG-1) were combined with levels of PSA, the prediction rate of prostate cancer increased from 78.2% to 86.2% than with PSA levels alone. The G81 glycan-binding auto-IgG fraction was isolated from plasma samples using G81 glycan-affinity chromatography and identified by N-terminal sequencing of the 50 kDa heavy chain variable region of the IgG. G81 glycan-binding 25 kDa fibroblast growth factor-1 (FGF1) fragment was also identified by N-terminal sequencing. Our results demonstrated that a multiplex diagnostic combining G81 glycan-binding auto-IgG, GDF-15/NAG-1 and PSA (≥2.1 ng PSA/ml for cancer) increased the specificity of prostate cancer diagnosis by 8%. The multiplex assessment could improve the early diagnosis of prostate cancer thereby allowing the prompt delivery of prostate cancer treatment.
Keywords: Prostate cancer, Glycan-binding auto-IgG, Biomarker, PSA, GDF-15/NAG-1
INTRODUCTION
Prostate cancer is one of the most common malignant tumors among the male population, causing cancer-related deaths [1]. Early diagnosis strategies could improve prostate cancer outcomes by providing care at the earliest possible stage. Serum biomarkers are widely used for cancer diagnosis and the monitoring of disease progression [2, 3]. However, the biomarkers sometimes cannot distinguish cancer from benign or inflammatory diseases.
Prostate-specific antigen (PSA) is a protein produced by normal, as well as malignant tissues and is often used for prostate cancer diagnosis. Approximately 20 out of 250 plasma samples from prostate cancer-free individuals had PSA levels ≥4 ng/ml, probably due to benign prostatic hyperplasia (BPH) or prostatitis [2–4]. Previously at our laboratory, a combined score of PSA and growth differentiation factor-15 [GDF-15, also designated as non-steroidal anti-inflammatory drug-activated gene-1 (NAG-1) and macrophage inhibitory cytokine-1 (MIC-1)] was shown to improve specificity of prostate cancer detection, suggesting that a signature panel could improve cancer diagnosis. The majority of cancer patients suffer from NAG-1-mediated cachexia at advanced cancer stages [5]. Thus, this PSA and NAG-1 combinatory diagnosis presented a weakness at early prostate cancer diagnosis with high incidence of false-negative reports.
Over the past several years, researchers have identified a number of new biochemical components of cancer cell surfaces [6–12]. These include surface antigens, lipids, proteoglycans and glycolipids, all of which alter cell-cell and cell-extracellular matrix communications [9, 7]. Glycans can be found attached to proteins or lipids as in glycoproteins and glycolipids, respectively, and, in general, are located on the exterior surface of cells [13]. Studies analyzing the presence of IgG/IgM in plasma suggested different glycan patterns between normal and malignant cells [6–8, 14].
Microarray analysis, a high-throughput screening (HTS) method suitable to elucidate interactions of various glycan structures with auto-antibodies, identified a breast cancer-specific glycan, Globo H, and antibody biomarkers [12]. Therefore, we hypothesize that in prostate cancer, glycoproteins, exosomes or entire cancer cells are secreted from tumor tissues into blood and induce auto-immunoglobulin (Ig) production providing a novel biomarker for prostate cancer diagnoses. The aims of this study are to verify whether the measurement of the auto-IgG or IgM in blood could improve prostate cancer diagnosis, to improve prostate cancer diagnosis by combining glycan-binding IgG biomarker(s) with protein cancer biomarkers, PSA and NAG-1, and, finally, to analyze the glycan prostate cancer biomarker-binding proteins and auto-antibodies.
METHODS
Subjects:
Blood samples were obtained from 35 prostate cancer patients and 54 age-matched healthy male subjects from a previously conducted case-control study of prostate cancer at Henry Ford Health System (Detroit, MI – IRB number 1018). Plasma was isolated and the coded sample was sent to Detroit R&D, Inc. The subjects selected in this study were subjected to a tissue biopsy during consultation at Henry Ford Hospital for prostate cancer diagnosis. Controls were age, sex and race frequency matched men randomly sampled from the Henry Ford Hospital patient database who did not have a history of prostate cancer. (Table 1).
Table 1.
Patient characteristics
| Variable | Controls (n=54) | Cases (n=35) |
|---|---|---|
| Age | 61.8 ± 8.3 | 60.1 ± 7.1 |
| Race | ||
| White | 33 (61.1%) | 22 (62.9%) |
| African American | 21 (38.9%) | 13 (37.1%) |
| BPH | 12 (22.2%) | 5 (14.7%) |
| PSA | 3.5 ± 4.5% | 38.9 ± 136.6 |
| Gleason Grade | ||
| 6 | 17 (48.6%) | |
| 7 (3+4) | 6 (17.1%) | |
| 7 (4+3) | 5 (14.2%) | |
| 8 | 4 (11.4%) | |
| Focal* | 3 (8.6%) | |
| Stage | ||
| 1C | 23 (65.7%) | |
| 2A | 6 (17.1%) | |
| 2B | 1 (2.9%) | |
| 2C | 5 (14.3%) |
Could not assign a Gleason grade
PSA and NAG-1 (GDF-15) assays:
Levels of PSA were from routine clinical information obtained at Henry Ford Health System and levels of NAG-1 were assessed using an ELISA kit from Detroit R&D, Inc. (Detroit, MI).
Glycan microarray production:
Glycan and protein microarray analyses were developed using Poly-L-Lysine slides (Thermo Scientific, Pittsburgh, PA) with disuccinimidyl suberate (DSS, Pierce, Rockford, IL) activation. For the microarrays, 24 glycans and a control polyacrylamide polymer (PAA) without sugar (G0, negative control) (GlycoTech, Gaithersburg, MD) (Table 2) and 25 antibodies (R&D Systems, Minneapolis, MN), respectively, were spotted (0.5 mg/ml, 0.4 nl/spot) using a piezoelectric Nanoplotter (GeSiM Co.). Each slide contained 16 chips and each chip had 4 identical blocks (Fig. 1A) including 3–4 Alexa647-conjugated streptavidin (Invitrogen) spots (positive control, 12–16 data points), 1–2 PBS spots (negative control), 2 no sugar-containing PAA carrier (G0, negative control) and 24 glycan spots or 25 antibody spots, respectively. A 5-glycan subarray was developed to confirm the results of the 24-glycan microarray (Fig. 2A and 2B). Each subarray slide contained 16 chips and each chip had 4 identical blocks (Fig. 1A) with 2 Alexa647-conjugated streptavidin spots (positive control, 8 data points), 2 PBS spots (negative control), 2 no sugar-containing PAA carrier (G0, negative control) and 5 glycan spots. The glycans were G05 (α-D-Man-PAA), G08 (L-Rhamnose-PAA), G09 (GlcNAc-PAA), G24 (Fucα1–3GlcNAcβ-PAA) and G81 (Neu5Acα2–8Neu5Acα2–8Neu5Acα-sp-PAA).
Table 2.
The 24 glycan and 1 control (no sugar) polyacrylamide (PAA) polymers used for a glycan microarray analysis
| ID | Number of Saccharide | Glycan-Polyacrylamide (PAA) | Control | PC |
|---|---|---|---|---|
| G0 | Control (No Sugar) | HOCH2(HOCH)4CH2NH-PAA | N/A | |
| G01 | Mono | α-D-Glucose-PAA | 3.5± 0.28 | 3.4± 0.41 |
| G02 | Mono | β-D-Glucose-PAA | 10.8± 0.15 | 11± 0.04 |
| G03 | Mono | α-D-Glalactose-PAA | 7.8± 0.21 | 7.6± 0.27 |
| G04 | Mono | β-D-Galactose-PAA | 7.2± 0.22 | 7.9± 0.23* |
| G05 | Mono | α-D-Man-PAA | 2.1± 0.18a | 3.3± 0.34a |
| G07 | Mono | α-L-Fucose-PAA | 3.6± 0.29 | 3.6± 0.20 |
| G08 | Mono | α-L-Rhamnose-PAA | 7.7± 0.28 | 8.7± 0.18* |
| G09 | Mono | β-GlcNAc-PAA | 6.8± 0.24 | 6.2± 0.25* |
| G10 | Mono | α-GalNAc-PAA | 6.6± 0.22 | 6.3± 0.25 |
| G12 | Mono | α-Neu5Ac-PAA | 3.1± 0.21 | 2.9± 0.22 |
| G17 | Di | Blood Type A: GalNAcα1–3Galβ-PAA | 9.6± 0.19 | 9.5± 0.13 |
| G18 | Di | Blood Type B: Galα1–3Galβ-PAA | 8.4± 0.23 | 8.4± 0.21 |
| G19 | Di | Blood Type H: Fucα1–2Galβ-PAA | 4.8± 0.32 | 4.6± 0.27 |
| G20 | Di | Galβ1–3GlcNAcβ-PAA (Lec) | 7.0± 0.27 | 7.1± 0.24 |
| G21 | Di | Galβ1–4Glcβ-PAA | 6.1± 0.22 | 6.0± 0.24 |
| G22 | Di | Galβ1–4GlcNAcβ-PAA | 3.6± 0.29 | 3.2± 0.34 |
| G23 | Di | Galβ1–3Glcβ-PAA | 4.1± 0.29 | 4.3± 0.25 |
| G24 | Di | Fucα1–3GlcNAcβ-PAA | 6.9± 0.27 | 6.6± 0.29* |
| G25 | Di | Fucα1–4GlcNAcβ-PAA | 5.7± 0.22 | 5.6± 0.34 |
| G26 | Di | GalNAcα1–3GalNAcβ-PAA | 10.2± 0.10 | 10.2± 0.08 |
| G37 | Tri | Fuc1–2Gal1–3GlcNAc- PAA [Led(H type 1)] | 3.5± 0.22 | 3.7± 0.33 |
| G71 | Di | GlcNAcβ1–3GalNAcα-PAA | 6.9± 0.23 | 6.3± 0.26 |
| G81 | Tri | Neu5Acα2–8Neu5Acα2–8Neu5Acα-sp-PAA | 4.7± 0.25b | 4.6± 0.29b |
| G98 | Tetra | Galβ1–4GlcNAcβ1–3Galβ1–4Glcβ-PAA | 4.5± 0.19 | 4.2± 0.25 |
Levels of glycan in plasma sample of Control and Prostate Cancer (PC). Data points of the auto-IgG were log-transformed (log2) and quantile normalized. Data are mean ± SEM.
p˂0.05 compared to control using an independent T-Test.
Correlated to PSA.
Correlated to NAG-1 using Pearson correlation.
Fig. 1. Microarray analyses.
(A) Protein and glycan-binding auto-IgG microarray platforms. Sixteen chips in a slide and each chip contains ~20–30 proteins or glycans in quadruplicate. As an example, coreprotein arrangement is shown. (B) Schematic diagrams of glycan-binding protein and auto-IgG microarrays
Fig. 2. Scatter plots of Control (horizontal) vs. Prostate cancer (vertical) for comparison of glycan-binding auto-IgG levels in plasma samples obtained from 35 prostate cancer patients and 54 healthy subjects.
For the 24 glycan microarray (A) and the 5-glycan subarray (B) analyses, glycan-spotted slides were incubated with 50 μl plasma (50-fold diluted in 5% BSA), detected with biotinylated anti-human IgG, followed by visualization of the biotinylated IgG using Alexa647-conjugated streptavidin (4 data points/sample, in total, 140 data points for prostate cancer and 188–216 data points for control). Green lines, no change or 1.5-fold increase/decrease. (C) Glycan auto-IgG microarray analysis images are shown for plasma samples of representative prostate cancer (right) and control (left).
Glycan and protein microarray detection:
Slides containing glycans were blocked with 5% BSA in PBS for 1 hr and incubated with 40 μl control or prostate cancer plasma samples diluted 1:50 (5% BSA in PBS) overnight at room temperature. Several dilutions (1:1, 1:25, 1:50, 1:100, 1:1,000) were tested before the final ratio, 1:50, was established. Slides were washed and either incubated with biotinylated anti-human IgG or IgM secondary antibody (Sigma-Aldrich) diluted 1:200 in blocking solution (Detroit R&D, Inc.) for 2 hours (Fig. 1B). Slides containing capture antibodies were blocked and incubated with plasma samples and then incubated with pooled biotinylated detection antibodies (R&D Systems) for 2 hours. All slides were washed 3 times and incubated with Alexa647-conjugated streptavidin (Molecular Probe, Eugene, OR) for 2 hr, washed, visualized and quantified using a Genepix 4100A (Molecular Devices). Protein microarray was performed as previously carried out by Seurynck-Servoss, et al. [15].
Characterization of the glycan-binding auto-IgG (denatured) in prostate cancer and control plasma samples:
Glycan-specific Ig fractions were isolated using the glycan-affinity columns produced with biotinylated G81. Plasma samples were selected from a control (#3) and a prostate cancer patient (#74) with G81-binding auto-IgG levels similar to the average of the 54 control or 35 prostate cancer (PCa) plasma samples. Plasma samples (400 μl) were incubated with 400 μg (32 μl) of biotinylated G81 (Neu5Acα2–8Neu5Acα2–8Neu5Acα, GlycoTech) for 2 hr at room temperature. Then 500 μl of PBS pre-washed streptavidin agarose resin (Thermo Scientific) was added to the plasma sample mix and the solution was incubated at 4°C for 72 hours. The resin was washed 5 times with PBS and 60 μl sample dilution buffer (SDB) was added to the G81/streptavidin-resin. The auto-IgG in SDB solution was heated at 100°C for 5 min and 35 μl of sample dissolved in SDB was obtained and proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) (8–16% gradient gel, BioRad) for Western blot, proteomic and N-terminal amino acid sequencing analyses (1 μl, 2.5 μl and 4 μl, respectively). For proteomic analysis, the gel was stained with Coomassie Blue stain (1% Coomassie Blue, 50% methanol and 10% acetic acid) and destained with 50% methanol and 10% acetic acid.
Proteomic analysis:
Streptavidin biotinylated G81-binding proteins were separated by SDS-PAGE, stained with Coomassie Blue stain and excised. Tryptic digestion and NanoLC-Orbitrap/MS analysis were performed by the Proteomics Core Facility at Wayne State University. The Mascot program (Matrix Sciences) was used for database search.
Western blot analysis:
G81-binding proteins without or with papain digestion were transferred to nitrocellulose, blocked with 5% BSA in PBS, incubated with secondary antibodies for human IgG or IgM (Sigma-Aldrich) and rabbit anti-fibroblast growth factor-1 (FGF1) IgG (BosterBio, Pleasanton, CA) followed by incubation with a secondary antibody for rabbit IgG, respectively, and developed by GloBrite ECL reagent (Detroit R&D). The proteins separated by SDS-PAGE were also transferred to PVDF (Waters/Merck Millipore) membrane for N-terminal amino acid sequencing.
N-terminal amino acid sequencing analysis:
The PVDF membrane was stained with Coomassie Blue Stain (1% Coomassie Blue, 50% methanol and 10% acetic acid) and destained with 50% methanol and 10% acetic acid. The 50 kDa G81-binding auto-IgG band in the prostate cancer plasma sample was excised and N-terminal sequencing analysis was performed at the Protein Facility of Iowa State University using a 494 Procise® Protein Sequencer/140C Analyzer (Applied Biosystems).
Papain digestion of the G81-binding plasma protein:
A fraction of the eluates of G81-binding proteins obtained from prostate cancer (#74) and control (#3) plasma samples (125 μl) were digested with papain using the Pierce Fab micro preparation kit (Thermo Scientific). Briefly, a Zeba desalting spin column was washed 3 times with digestion buffer, pH 7, and 125 μl of neutralized eluate fraction of G81-binding protein sample was added to the column for buffer exchange. The papain column was prepared by adding 65 μl papain resin to a 0.8 ml column and washing the resin twice with 130 μl digestion buffer. The flow-through fraction from the desalting column was added to the pre-washed papain column and incubated for 6 hr and the column was centrifuged to separate the digested protein from the papain resin. Gradient SDS-PAGE with 8–16% gel (Bio-Rad) was carried out with the digest (20 μl) and the proteins were transferred to an Immobilon PVDF membrane, stained with Coomassie Blue Stain and destained with 50% methanol and 10% acetic acid. The 25 kDa band of the digested G81-binding proteins in the prostate cancer plasma sample was excised and sent to the Protein Facility at Iowa State University for N-terminal sequencing.
Statistical Analyses:
Decision tree (recursive partitioning) analysis to partition the two groups (prostate cancer patients vs. healthy subjects) of samples for PSA, NAG-1 (GDF-15) and G81 auto-IgG was performed using JMP statistical software. The decision tree algorithm selected a point to provide the optimal separation between the two groups. Thus, each parameter (predictor) cut-off is automatically selected for optimal predictive accuracy. Levels of PSA (ng/ml), NAG-1 (GDF-15) (pg/ml) and glycan-binding auto-IgG (arbitrary unit) by glycan microarray analysis of plasma samples were obtained from the 35 prostate cancer patients and 54 healthy subjects. The glycan-binding auto-IgG data for each sample were log-transformed (log2) and quantile normalized using the GeneSpring v13 to ensure the distribution of normalized signal values equivalent across all samples (Fig. 2A and 2B). The glycans applied to the glycan subarray analysis were G5, G08, G09, G24 and G81, and independent T-test was used to find a significant difference between control and prostate cancer.
RESULTS
Sample characteristics:
Our donor pool was composed of 89 male subjects, 54 of whom were identified as controls and 35 as prostate cancer patients following biopsies. Age and race were similar among control subjects and prostate cancer patients (Table 1). The majority of prostate cancer patients were in Stage 1C (65.7% of the prostate cancer patients), whereas the remaining prostate cancer patients were in prostate cancer Stage 2 (Table 1).
Microarray and subarray results:
The glycan-binding auto-IgG profile constructed from the 24-glycan auto-IgG microarray analysis (Table 2) showed that all of the 24 glycan auto-IgG signals were distributed between G0 (no sugar-containing polymer, HOCH2(HOCH)4CH2NH-PAA) and the positive control A647 (Alexa647-conjugated streptavidin) (Fig. 2A). No glycan signal was detected when anti-human IgM (secondary antibody) was used suggesting that most of the glycans were bound to auto-IgG. The normalized signal values revealed that the majority of auto-IgG bound to glycans in prostate cancer were similar to those found in control plasma samples including Fuc1–2Gal1–3GlcNAc-PAA [Led (H type 1, Globo H)] (G37) (Table 2, Fig. 2C) which has been identified to be a breast cancer biomarker [12].
Auto-IgGs bound to 5 glycans were identified to develop 5-glycan auto-IgG subarray analysis by comparison of glycan levels in prostate cancer with levels in controls (increased or decreased) after the 24 glycan auto-IgG microarray analysis or by correlation of the levels to PSA and GDF-15/NAG-1 levels, i.e., D-Man G05, was positively correlated with PSA (0.32) whereas Neu5Acα2–8Neu5Acα2–8Neu5Acα-sp G81 was negatively correlated with NAG-1 (−0.31) (Table 2). The 5 glycans are G05 (D-Man-PAA), G08 (L-Rhamnose-PAA), G09 (GlcNAc-PAA), G24 (Fucα1–3GlcNAcβ-PAA) and G81 (Neu5Acα2–8Neu5Acα2–8Neu5Acα-sp-PAA).
As detected in the first 24-glycan auto-IgG microarray analysis, levels of the auto-IgGs bound to 5 glycans were distributed between G0 (negative control, linker-PAA polymer) and A647 (positive control, Alexa-conjugated streptavidin) (Fig. 2B). Both G0 and A647 were located near or on the diagonal line as expected, indicating no difference between groups. As previously detected by the 24 glycan auto-IgG microarray analysis, levels of auto-IgG bound to G09 (GlcNAc) and G24 (Fucα1–3GlcNAcβ) were lower in prostate cancer plasma samples compared to controls (Table 2 and Fig. 2A, 2B and 3A). The G81 (Neu5Acα2–8Neu5Acα2–8Neu5Acα) auto-IgG levels were higher in prostate cancer patients (Fig. 2B, 2C and 3A). No difference between the groups was found in respect to auto-IgG bound to G08 (L-Rhamnose-PAA) and low signal was detected with G05 (α-D-Man-PAA) auto-IgG (Fig. 2B and 3A).
Fig. 3. Partition (decision tree) analysis for prostate cancer and control prediction using plasma samples obtained from 35 prostate cancer patients and 54 healthy subjects.
(A) Subarray results. Results are presented as a mean ± SEM. *p˂0.05 compared to control using an independent T-Test. (B) A recursive partitioning analysis of PSA, NAG-1 (GDF-15) and G81 auto-IgG levels performed using JMP statistical software. The decision tree algorithm selected points of 2.1 ng/ml for PSA, 320 pg/ml for NAG-1 (GDF-15) and 5.66 (arbitrary unit) for G81 glycan-binding auto-IgG to provide the optimum separation between the two groups. The area under the receiver operating characteristic curve (ROC) is shown for each biomarker combination. (C) Comparison of the optimum separation analysis with 2.1 ng PSA/ml or 4.0 ng PSA/ml.
Decision tree analysis for prostate cancer and control prediction using PSA, NAG-1 (GDF-15) and G81-binding:
Decision tree analysis revealed that PSA, NAG-1, and G81-binding auto-IgG (Neu5Acα2–8Neu5Acα2–8Neu5Acα) provided optimal accuracy as a combinatorial predictor in classifying patients into the two groups. The combination of NAG-1, G81-binding auto-IgG and PSA values (2.1 ng/ml) improved the diagnosis of prostate cancer (increased specificity and sensitivity) compared to PSA alone from 0.78 to 0.86 (8% increase), as measured by the area under receiver operating characteristic (ROC) curves (Fig. 3B). To investigate the reproducibility of this improvement when using random subsets of our data, we performed a 10-fold cross validation by randomly selecting ~10% of the samples for independent testing, while training the decision tree with the remaining samples. The trained decision tree was then used to make category predictions for the test set. This procedure was repeated 10 times. The average area under the curve (AUC) for the random test sets improved by an average of 9.5% when using the combined panel of PSA, NAG-1, and G81 (0.81) compared to PSA alone (0.74), consistent with the improvement observed above with the overall sample set.
For the overall dataset, the ability to correctly diagnose prostate cancer decreased to 0.82 when the point of separation of the 2 groups for PSA is set at 4.0 ng/ml as used at clinics (Fig. 3C).
In a protein microarray analysis carried out at our laboratory, prostate cancer prediction increased by 4% after combining EGFR, ICAM-1 and RANTES levels to PSA levels as compared to PSA alone (78%).
Proteomic analyses:
Western blot analysis of the G81-binding proteins obtained from 11 μl control and 11 μl prostate cancer plasma samples confirmed that the level of auto-IgG (50 kDa) bound to G81 glycan in prostate cancer plasma sample #74 was higher compared to the level in control plasma sample #3 (Fig. 4A). These particular samples were chosen since their G81-binding auto-IgG values were similar to the group average. No bands were detected when the secondary antibody for human IgM was used for the Western blot analysis using nitrocellulose membrane (data not shown). The tryptic digestion database search of the 50–70 kDa bands resulted in tryptic peptides from 50 kDa IgG heavy chains. They are (a) Ig heavy chain (H) V-III region TRO (13 kDa, total IgH, 50 kDa) in prostate cancer, 9.08E+08, and control, 1.67E+08, (b) Ig heavy chain (H) V-III region BUT (12 kDa, total IgH, 50 kDa) in prostate cancer, 5.27E+08, and control, 0, and (c) Ig heavy chain (H) V-III region WEA (12 kDa, total IgH, 50 kDa) in prostate cancer, 4.63E+08, and control, 0. The IgG heavy chain V-III region BUT and WEA were identified only in the prostate cancer plasma sample (Table 3).
Fig. 4. G81 (Neu5Acα2–8Neu5Acα2–8Neu5Acα)-binding IgG and proteins in a prostate cancer (PCa) and a control plasma samples.
Samples #74 and #3 were selected for these analyses since their G81-binding auto-IgG values were similar to the group average. (A) Western blot analysis of G81-binding auto-IgG (50 kDa heavy chain): G-81 affinity column eluates obtained from 11 μl control and 11 μl prostate cancer plasma samples were separated by SDS-PAGE and transferred to nitrocellulose membrane and Western blot analysis was carried out using the secondary antibody for human IgG. (B) SDS-PAGE of papain digested proteins of G81 affinity column eluates obtained from 20 μl control and 20 μl prostate cancer plasma samples: G81 glycan-binding proteins were digested with papain, separated by SDS-PAGE and transferred to PVDF membrane. The 25 kDa species in the PCa plasma sample (arrow) was N-terminally sequenced and identified as FGF1 (Table 3). (C) Western blot analysis of G81-binding FGF1 protein (25 kDa): G81 glycan-binding proteins were digested with papain, separated by SDS-PAGE and transferred to nitrocellulose membrane and Western blot analysis was carried out using FGF1 antibody.
Table 3.
Identified molecules and amino acid sequences of proteins bound to biotinylated G81 glycan/streptavidin-agarose resin with or without treatment with papain to obtain 25 kDa and 50 kDa fragments, respectively
| Method | Result | Search Program | Identified Molecules |
|---|---|---|---|
| Tryptic digest/nanoLC-Orbitrap/MS analysis | Peptide sequences | Mascot | Ig heavy chain V-III region TRO, BUT and WEA (50 kDa). |
| Non-treated sample/ N-terminal sequencing | Gly-Val-Gln-Leu-Val-Glu-Ser | NCBI Blast® | IgG heavy chain variable region (50 kDa) |
| Papain digested samples/ N-terminal sequencing | Phe-Leu-Arg-X-Leu-Pro-Asp-X-Thr-Val- Asp | NCBI Blast® | 25 kDa fragment from FGF1 |
The proteins were separated by SDS-PAGE and Western blot analysis was carried out using anti-FGF1 IgG followed by a secondary antibody for rabbit IgG or a secondary antibody for human IgG, respectively. The 50 kDa PCa band was excised from the gel and subjected to tryptic digestion/nano LC-Orbitrap/MS analysis while, the 25 kDa or 50 kDa protein was transferred to PVDF membrane and was N-terminally sequenced.
Confirmation of variable region of the G81-binding auto-IgG and FGF1 by N-terminal amino acid sequencing:
The G81-affinity purified auto-IgG in the prostate cancer plasma sample was separated by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to PVDF membrane. The identity of the 50 kDa G81-binding auto-IgG band identified by Western blot analysis (Fig. 4A) was confirmed by N-terminal amino acid sequencing. The amino acid sequence of the 50 kDa band by the N-terminal sequence was Gly-Val-Gln-Leu-Val-Glu-Ser (major protein). NCBI BLAST® human protein search with the sequence matched with an IgG heavy chain variable region (100% match) (Table 3). Papain digestion of the G81-binding protein eluate produced a 25 kDa fragment (Fig. 4B) and the sequence, Phe-Leu-Arg-X-Leu-Pro-Asp-X-Thr-Val-Asp, matched with fibroblast growth factor-1 (FGF1) (Table 3). Western blot analysis using a specific antibody against FGF1 (Fig. 4C) confirmed the result obtained by the N-terminal sequencing of the 25 kDa fragment.
DISCUSSION
Early diagnosis of prostate cancer could improve patient outcomes by providing care at the earliest possible stage [1–3]. Serum or plasma PSA is a biomarker easy to measure for prostate cancer diagnosis and is widely used for evaluation of cancer treatment [1–3]. However, many studies suggest that the clinically used value of ≥4.0 ng PSA/ml might be unreliable due to the high incidence of false positives and negatives. These problems may result in unnecessary prostate biopsies or could delay the beginning of treatment [2, 16]. In this study, we present strong evidence that measurements of NAG-1 (GDF-15) and glycan-binding auto-IgG could improve the accuracy of prostate cancer diagnosis.
Apart from prostate cancer, other prostate complications such as benign prostatic hyperplasia and bacterial prostatitis may display elevated PSA levels [4, 16]. False positive diagnosis, typically detected using the clinically used value of ≥4.0 ng PSA/ml, induces unnecessary prostate biopsies, generating anxiety, discomfort and undue costs [1–3]. However, recent studies have reported that 25% of prostate cancer patients had PSA levels between 2.1–4.0 ng/ml [16, 17]. Our result demonstrated that using the value of ≥2.1 ng PSA/ml for prostate cancer improved accuracy of prostate cancer diagnosis. Use of a lower cut-off value of PSA could increase the probability of patient survival. However, use of new prostate cancer biomarkers will more likely improve the precision of diagnosis.
NAG-1 (GDF-15) is expressed at low levels in the blood and in most tissues and cells, with the exception of the placenta and macrophages [18, 19]. Use of increased serum GDF-15/NAG-1 levels improved prostate cancer diagnosis [20, 21]. However, patients with BPH had higher serum levels of NAG-1 compared to levels of NAG-1 in serum samples obtained from prostate cancer patients [22]. Our study found that the plasma level of NAG-1 combined with the PSA level improved diagnosis of prostate cancer by 4% compared to the PSA level alone. The level of G81 auto-IgG combined with the PSA level improved diagnosis of prostate cancer by 6% compared to the PSA level alone by correction of some of false-positive diagnosis of 21 of 54 healthy subjects obtained by using PSA level alone. Indeed, the use of the NAG-1 and G81 auto-IgG scores in combination with the PSA level (≥2.1 ng PSA/ml) improved prostate cancer diagnosis by 8%.
Levels of EGFR increased in plasma samples obtained from prostate cancer patients. However, an increase in EGFR is not a prostate cancer-specific biomarker [23, 24]. When plasma EGFR level was combined with levels of ICAM-1, RANTES and PSA, the specificity of prostate cancer detection improved by ~4%, suggesting that measurement of NAG-1, G81-auto-IgG and PSA levels could be a better tool for improving prostate cancer diagnosis.
Anti-tumor auto-antibodies may be involved in cancer cell cytotoxicity and inhibition of metastasis [14, 25]. A high-throughput profiling of glycan-binding auto-antibodies revealed that blood type A or AB patient was related with a poorer prognosis after PROSTVAC-VF prostate cancer vaccine treatment to induce immunity to PSA compared to blood type B or O patient [6, 8]. Serum auto-IgG for Fuc1–2Gal1–3GlcNAc-PAA [Led (H type 1), Globo H] (G37) was identified to be a breast cancer biomarker [12]. Our results support the idea that the Led (H type 1, Globo H) (G37)-binding auto-IgG could be a biomarker specific for breast cancer, since no changes were found between control subjects and prostate cancer patients regarding the binding of this glycan with IgG. Our results are the first to suggest that, in addition to the Neu5Acα2–8Neu5Acα2–8Neu5Acα (G81) auto-IgG, an auto-IgG binding to GlcNAc (G09) or Fucα1–3GlcNAcβ (G24) could also be a biomarker for prostate cancer. In addition, these glycan/auto-IgG biomarker studies could lead to a therapeutic opportunity for development of a prostate cancer-specific vaccine as suggested by other studies [6, 8].
Microarrays represent a powerful technology that can be used for the detection of a large number of biomarkers for different diseases in clinical settings as previously reported [26, 27]. Key features of analytical protein microarrays include high-throughput and relatively low costs due to minimal consumption of reagent, multiplexing, fast kinetics and measurements and the possibility of functional integration. For many years, high-throughput profiling of glycan-binding serum auto-antibodies has been used to identify the immunologic response to malignant cells and to decide treatment strategies [6–14]. However, few proteomic analyses of glycan-binding antibodies and proteins have been explored. Using G81-affinity chromatography, a G81-binding auto-IgG in a prostate cancer plasma sample was isolated. The 50 kDa heavy chain of the auto-IgG was separated by SDS-PAGE. The N-terminal amino acid sequence of the IgG heavy chain was obtained and identified as the variable region of the 50 kDa heavy chain of the IgG. In addition, the analysis of the N-terminal amino acids of the 25 kDa fragment obtained after papain digestion of the G81 column eluate strongly suggest that fibroblast growth factor-1 (FGF1), a heparin- or heparan sulfate-binding protein, could bind to the G81 glycan in the prostate cancer plasma sample. Interestingly, the FGF1 receptor has been a plausible candidate as a cancer therapy target since it is overexpressed in multiple types of tumors, such as breast, prostate and lung cancer [28–30].
In summary, our results demonstrated that a multiplex biomarker diagnosis consisting of PSA (≥2.1 ng/ml), NAG-1 and G81 glycan-binding auto-IgG could increase the specificity and sensitivity of prostate cancer diagnosis. This could improve the early diagnosis of prostate cancer thereby allowing the prompt delivery of prostate cancer treatment. In addition, interaction of FGF1 with Neu5Acα2–8Neu5Acα2–8Neu5Acα (G81) might be involved with development or metastasis of prostate cancer.
AKNOWLEDGEMENT
This study was supported by NIH/NCI SBIR Phase I grant, CA159721 (HK).
Footnotes
CONFLICT OF INTEREST
Benjamin A. Rybicki, German Perez Bakovic and Shannon L. Servoss declare that they have no competing interests. Hyesook Kim is the president of Detroit R&D, Inc. and has a commercial interest. David J. Kaplan, Aby Joiakim, Julia M. Santos and David A. Putt also work for Detroit R&D, Inc. Alan A. Dombkowski is a board member for Detroit R&D, Inc.
REFERENCE
- 1-.Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F (2012) Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN. Int J Cancer;136(5):E359–86. doi: 10.1002/ijc.29210. [DOI] [PubMed] [Google Scholar]
- 2-.Adhyam M, Gupta AKA (2012) Review on the clinical utility of PSA in cancer Prostate. Indian J Surg Oncol;3(2):120–129. doi: 10.1007/s13193-012-0142-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3-.Kawakami J, Siemens DR, Nickel JC (2004) Prostatitis and prostate cancer: implications for prostate cancer screening. Urology;64:1075. doi: 10.1016/j.urology.2004.07.017. [DOI] [PubMed] [Google Scholar]
- 4-.Masahiko N, Hisanori K, Tomihiro T, Oyama R, Imanaka K, Yoshizawa K (2016) Association of early PSA decline and time to PSA progression in abiraterone acetate-treated metastatic castration-resistant prostate cancer; a post-hoc analysis of Japanese phase 2 trials. BMC Urol;16(1):27. doi: 10.1186/s12894-016-0148-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5-.Lerner L, Tao J, Liu Q, Nicoletti R, Feng B, Krieger B, Mazsa E, Siddiquee Z, Wang R, Huang L, Shen L, Lin J, Vigano A, Chiu MI, Weng Z, Winston W, Weiler S, Gyuris J. (2016) MAP3K11/GDF15 axis is a critical driver of cancer cachexia. J Cachexia Sarcopenia Muscle;7(4):467–82. doi: 10.1002/jcsm.12077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6-.Campbell CT, Gulley JL, Oyelaran O, Hodge JW, Schlom J, Gildersleeve JC (2013) Serum antibodies to blood group A predict survival on PROSTVAC-VF. Clin Cancer Res;19(5):1290–9. doi: 10.1158/1078-0432.CCR-12-2478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7-.Griffin ME, Hsieh-Wilson LC (2016) Glycan engineering for cell and developmental biology. Cell Chem Biol; 23(1):108–121. doi: 10.1016/j.chembiol.2015.12.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8-.Muthana SM, Gulley JL, Hodge JW, Schlom J, Gildersleeve JC (2015) ABO blood type correlates with survival on prostate cancer vaccine therapy. Oncotarget; 6(31):32244–56. doi: 10.18632/oncotarget.4993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9-.Ohyama C, Hosono M, Nitta K, Oh-eda M, Yoshikawa K, Habuchi T, Arai Y, Fukuda M. (2004) Carbohydrate structure and differential binding of prostate specific antigen to Maackia amurensis lectin between prostate cancer and benign prostate hypertrophy. Glycobiology;14(8):671–679. doi: 10.1093/glycob/cwh071 [DOI] [PubMed] [Google Scholar]
- 10-.Tang X, Mahajan SS, Nguyen LT, Béliveau F, Leduc R, Simon JA, Vasioukhin V (2014) Targeted inhibition of cell-surface serine protease Hepsin blocks prostate cancer bone metastasis. Oncotarget;5(5):1352–62. doi: 10.18632/oncotarget.1817. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11-.Totten SM, Adusumilli R, Kullolli M, Tanimoto C, Brooks JD, Mallick P, Pitteri SJ. (2018) Multi-lectin affinity chromatography and quantitative proteomic analysis reveal differential glycoform levels between prostate cancer and benign prostatic hyperplasia Sera. Sci Rep;8(1):6509. doi: 10.1038/s41598-018-24270-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12-.Wang CC, Huang YL, Ren CT, Lin CW, Hung JT, Yu JC, Yu AL, Wu CY, Wong CH. (2008) Glycan microarray of Globo H and related structures for quantitative analysis of breast cancer. Proc Natl Acad Sci U S A; 105(33):11661–6. doi: 10.1073/pnas.0804923105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13-.Meany DL, Zhang Z, Sokoll LJ, Zhang H, Chan DW (2009) Glycoproteomics for prostate cancer detection: changes in serum PSA glycosylation patterns. J. Proteome Res; 8:613–619. doi: 10.1021/pr8007539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14-.Wang X, Yu J, Sreekumar A, Varambally S, Shen R, Giacherio D, Mehra R, Montie JE, Pienta KJ, Sanda MG, Kantoff PW, Rubin MA, Wei JT, Ghosh D, Chinnaiyan AM (2005) Autoantibody signatures in prostate cancer. N Engl J Med; 353(12):1224–1235. doi: 10.1056/NEJMoa051931. [DOI] [PubMed] [Google Scholar]
- 15-.Seurynck-Servoss SL, White AM, Baird CL, Rodland KD, Zangar RC (2007). Evaluation of surface chemistries for antibody microarrays. Analytical biochemistry; 371(1):105–115. doi: 10.1016/j.ab.2007.07.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16-.Horninger W, Cheli CD, Babaian RJ, Fritsche HA, Lepor H, Taneja SS, Childs S, Stamey TA, Sokoll LJ, Chan DW, Brawer MK, Partin AW, Bartsch G (2002) Complexed prostate-specific antigen for early detection of prostate cancer in men with serum prostate-specific antigen levels of 2 to 4 nanograms per milliliter. Urology;60 (4 Suppl 1):31–5. doi: 10.1016/s0090-4295(02)01693-x. [DOI] [PubMed] [Google Scholar]
- 17-.Liu X, Tang J, Fei X, Li QY (2015) Prostate-specific antigen (PSA) density and free to total PSA ratio in diagnosing prostate cancer with prostate-specific antigen levels of 4.0 ng/ml or less. Iran J Public Health; 44(11):1466–72. [PMC free article] [PubMed] [Google Scholar]
- 18-.Bruzzese F, Hägglöf C, Leone A, Sjöberg E, Roca MS, Kiflemariam S, Sjöblom T, Hammarsten P, Egevad L, Bergh A, Ostman A, Budillon A, Augsten M (2014) Local and systemic protumorigenic effects of cancer-associated fibroblast-derived GDF15. Cancer Res; 74(13):3408–17. doi: 10.1158/0008-5472.CAN-13-2259. [DOI] [PubMed] [Google Scholar]
- 19-.Vaňhara P, Hampl A, Kozubík A, Souček K (2012) Growth/differentiation factor-15: prostate cancer suppressor or promoter? Prostate Cancer Prostatic Dis; 15(4):320–8. doi: 10.1038/pcan.2012.6. [DOI] [PubMed] [Google Scholar]
- 20-.Bauskin AR, Brown DA, Kuffner T, Johnen H, Luo XW, Hunter M, Breit SN (2006) Role of macrophage inhibitory cytokine-1 in tumorigenesis and diagnosis of cancer. Cancer Res; 15;66(10):4983–6. doi: 10.1158/0008-5472.CAN-05-4067. [DOI] [PubMed] [Google Scholar]
- 21-.Brown DA, Stephan C, Ward RL, Law M, Hunter M, Bauskin AR, Amin J, Jung K, Diamandis EP, Hampton GM, Russell PJ, Giles GG, Breit SN (2006) Measurement of serum levels of macrophage inhibitory cytokine 1 combined with prostate-specific antigen improves prostate cancer diagnosis. Clin Cancer Res; 12:89–96. doi: 10.1158/1078-0432.CCR-05-1331. [DOI] [PubMed] [Google Scholar]
- 22-.Stephan C, Xu C, Brown DA, Breit SN, Michael A, Nakamura T, Diamandis EP, Meyer H, Cammann H, Jung K (2006) Three new serum markers for prostate cancer detection within a percent free PSA-based artificial neural network. Prostate; 1;66(6):651–9. doi: 10.1002/pros.20381. [DOI] [PubMed] [Google Scholar]
- 23-.Jiang T, Su C, Ren S, Cappuzzo F, Rocco G, Palmer JD, van Zandwijk N, Blackhall F, Le X, Pennell NA, Zhou C (2018); written on behalf of the AME Lung Cancer Collaborative Group. A consensus on the role of osimertinib in non-small cell lung cancer from the AME Lung Cancer Collaborative Group. J Thorac Dis;10(7):3909–3921. doi: 10.21037/jtd.2018.07.61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24-.Tysarowski A, Nasierowska-Guttmejer A (2018) Quality and practical aspects of pathological and molecular diagnostics in metastatic colorectal cancer (mCRC). Contemp Oncol (Pozn);22(2):75–85. doi: 10.5114/wo.2018.77047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25-.Reuschenbach M, von Knebel Doeberitz M, Wentzensen N (2009) A systematic review of humoral immune responses against tumor antigens. Cancer Immunol Immunother; 58(10):1535–1544. doi: 10.1007/s00262-009-0733-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26-.Sauer U (2017) Analytical protein microarrays: Advancements towards clinical applications. Sensors (Basel); 17(2):256. doi: 10.3390/s17020256. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27-.Srinivasan S, Clements JA, Batra J (2016) Single nucleotide polymorphisms in clinics: Fantasy or reality for cancer? Crit Rev Clin Lab Sci; 53(1):29–39. doi: 10.3109/10408363.2015.1075469. [DOI] [PubMed] [Google Scholar]
- 28-.He XX, Du S, Gao SQ, Chen JY, Cao RJ, Xing ZK, Kazim ARS, Yu HL, Zheng QC, Zhu XJ (2018) Humanization of fibroblast growth factor 1 single-chain antibody and validation for its antitumorigenic efficacy in breast cancer and glioma cells. J Cell Mol Med; 22(6):3259–3263. doi: 10.1111/jcmm.13547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29-.Li J, Chen H, Li X, Wang L, Gao A, Zhang P, Lin W, Gao W, Yang D, Guo X, Liu J, Dang Q, Sun Y (2017) Co-expression of podoplanin and fibroblast growth factor 1 predicts poor prognosis in patients with lung squamous cell carcinoma. Mol Med Rep; 16(2):1643–1652. doi: 10.3892/mmr.2017.6830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30-.Udayakumar TS, Nagle RB, Bowden GT (2004) Fibroblast growth factor-1 transcriptionally induces membrane type-1 matrix metalloproteinase expression in prostate carcinoma cell line. Prostate; 58(1):66–75. doi: 10.1002/pros.10293. [DOI] [PubMed] [Google Scholar]