Multiple recognition assay reveals prostasomes as promising plasma biomarkers for prostate cancer

Abstract

Prostasomes are microvesicles (mean diameter, 150 nm) that are produced and secreted by normal and malignant prostate acinar cells. It has been hypothesized that invasive growth of malignant prostate cells may cause these microvesicles, normally released into seminal fluid, to appear in interstitial space and therewith into peripheral circulation. The suitability of prostasomes as blood biomarkers in patients with prostate cancer was tested by using an expanded variant of the proximity ligation assay (PLA). We developed an extremely sensitive and specific assay (4PLA) for detection of complex target structures such as microvesicles in which the target is first captured via an immobilized antibody and subsequently detected by using four other antibodies with attached DNA strands. The requirement for coincident binding by five antibodies to generate an amplifiable reporter results in both increased specificity and sensitivity. The assay successfully detected significantly elevated levels of prostasomes in blood samples from patients with prostate cancer before radical prostatectomy, compared with controls and men with benign biopsy results. The medians for prostasome levels in blood plasma of patients with prostate cancer were 2.5 to sevenfold higher compared with control samples in two independent studies, and the assay also distinguished patients with high and medium prostatectomy Gleason scores (8/9 and 7, respectively) from those with low score (≤6), thus reflecting disease aggressiveness. This approach that enables detection of prostasomes in peripheral blood may be useful for early diagnosis and assessment of prognosis in organ-confined prostate cancer.

Prostate cancer is the most frequent cancer among men older than 50 y in Western societies (1), and the incidence increases steadily in most countries. The aggressiveness of the disease is reflected by the degree of histological aberration recorded as the Gleason score.

Reliable plasma biomarkers for prostate cancer could prove of great value by enabling early diagnosis, distinction between indolent and clinically significant prostate cancer, and efficient follow-up of therapy. The most commonly used biomarker for prostate cancer is prostate specific antigen (PSA) (2, 3), but it has well known limitations in accuracy (4, 5). Besides malignancy of variable severity, increased plasma PSA levels can also reflect inflammatory conditions and benign hyperplasia of the prostate, necessitating the search for better biomarkers (2, 6).

A wide variety of cell types are able to release microvesicles of endocytic origin to the extracellular compartment. The first cellular system to be explored in this regard was the prostate acinar cells. Ronquist et al. showed more than 30 y ago that human prostatic fluid and seminal plasma contain membrane-surrounded, nanometer-sized microvesicles (7–9). Subsequent studies revealed that these extracellular microvesicles, now denoted prostasomes, corresponded to intracellular microvesicles inside another larger microvesicle, a so-called storage vesicle, equivalent to multivesicular bodies or late endosomes (10, 11). Accordingly, the release of prostasomes to the extracellular compartment is the result of a fusion between the membrane surrounding the storage vesicle and the plasma membrane of the secretory cell of the prostate gland (i.e., exocytosis) (10–12). The prostasomal membrane displays extraordinary properties with an unusually high content of the phospholipid sphingomyelin and a high cholesterol/phospholipid ratio, rendering the membrane highly ordered and stable (13), as reflected by resistance to detergents (14).

Prostasomes contain many different proteins (15, 16). Proteins present on the surface of prostasomes include aminopeptidase N (CD13) (17, 18) and tissue factor (CD142), a cell membrane-associated glycoprotein that serves as a receptor and essential cofactor for factors VII and VIIa of the coagulation cascade (19). The prostasomes seem to act as intercellular messengers between secretory cells of the prostate and sperm cells, transferring molecules propitious for fertilization by influencing, e.g., sperm motility (20) and exerting antibacterial, complement inhibitory, antioxidant, and immunosuppressive activities (21–24).

Also, neoplastic prostate cells—even poorly differentiated prostate cancer metastases—have the capacity to synthesize and export prostasomes, but the altered tissue architecture in malignancy facilitates the release of prostasomes to the interstitial space rather than to the lumen of the prostate (25–27).

Exosomes corresponding to prostasomes exist also in other cell types. They are actively secreted by most, if not all, nucleated cells. Their physiological functions are diverse, generally related to the functions of the parent cell (28). Exosomes are produced by tumors in vivo and they are often abundant in malignant effusions for example in advanced ovarian carcinoma (29) and pleural malignant mesothelioma (30). Exosomes released from malignified epithelial cells that no longer respect their basal membranes but grow invasively may be useful biomarkers in blood plasma (31–34) and urine (35) in some cancer types.

To evaluate the suitability of prostasomes as biomarkers for prostate cancer, we developed a modified proximity ligation assay (PLA) to detect these microvesicles in blood plasma from patients with prostate cancer and control subjects. PLA is an assay mechanism for sensitive and specific detection of proteins in which target molecules must be recognized by antibodies, carrying short DNA strands. Upon binding to their targets, these DNA strands can be joined by ligation. The method allows amplified detection of the DNA reporter molecules on antibodies bound in proximity, offering high specificity and sensitivity (36, 37). In the 4PLA technique presented herein, prostasomes are first captured on a solid support by one monoclonal antibody, and signal generation then requires simultaneous binding by another four antibodies, each carrying a specific DNA strand. A total of three monoclonal antibodies and two aliquots of a polyclonal antibody are used to recognize five different binding sites on the prostasomes. Only when all four of these PLA probes are brought in proximity by virtue of having recognized the target molecular complex can the DNA strands on two of the PLA probes be joined by enzymatic ligation, thereby giving rise to a new, amplifiable DNA sequence. This reporter DNA molecule is then quantified by real-time quantitative PCR (qPCR) as a measure of detected target complexes (Fig. 1).

Fig. 1.

Mechanism of 4PLA. Target molecules are captured by antibodies immobilized on the walls of a reaction vessel (A), the four PLA probes are added (B), and the probes are allowed to bind different epitopes on the target structure. The four oligonucleotides attached to the antibodies hybridize to each other (C) and guide hybridization of a further oligonucleotide (D). This oligonucleotide that is added together with ligation/amplification mix is joined by enzymatic DNA ligation to oligonucleotides attached to two of the antibodies, templated by oligonucleotides on the two other antibodies. Finally, the newly formed DNA template is amplified and quantified by qPCR.

By using this assay, we successfully established that prostasomes are present in blood plasma, and we observed increased levels in samples from patients with prostate cancer, also correlating with tumor aggressiveness. Accordingly, prostasomes measured via a multiple-recognition assay are promising blood biomarkers for cancer of the prostate.

Results

Detection of Prostasomes Using 4PLA and PLA.

To determine if prostasomes in blood plasma could serve as markers for prostate cancer, we established a dedicated assay. Detection of these multiprotein microvesicles depends on simultaneous recognition of five different epitopes on at least four different proteins by using four different mono- or polyclonal antibodies. The proteins must be located within an estimated distance of a few tens of nanometers on the surface of the prostasomes. After capture by an anti-CD13 monoclonal antibody immobilized on the surfaces of reaction tubes, four oligonucleotide-conjugated antibodies are added [two monoclonal and one polyclonal (×2)] directed against four different epitopes on the surface of the prostasomes. Two aliquots of a polyclonal antibody directed against tissue factor were modified with two different oligonucleotides, and two prostasome-specific monoclonal antibodies called mAb78 and mAb8H10 each carried unique oligonucleotides. After washes, all four oligonucleotides contributed to the creation of an amplifiable DNA strand via two enzymatic ligation events. Finally, the amount of target dependent ligation products was measured by qPCR (Fig. 1).

Purified prostasomes were detected with a lower limit of detection (LOD) of 0.032 ng protein/mL. Using a more conventional PLA protocol involving one capture antibody and only two PLA probes, the LOD was 4.83 ng/mL (Fig. 2A). Thus, 4PLA exhibited an approximately 150-fold lower LOD and it had a dynamic range that extended by two further orders of magnitude compared with PLA reactions, in which only two antibody–oligonucleotide conjugates were used. Similar sensitivities and dynamic ranges were observed whether prostasomes were detected by 4PLA in 10% human blood plasma or in buffer (Fig. 2B).

Fig. 2.

Detection of prostasomes by using 4PLA and PLA. (A) Comparison of 4PLA (circles) and solid-phase PLA (squares) for measuring purified prostasomes. For 4PLA the SD of 0.021 and for solid-phase PLA the SD of 0.056 for negative controls were used to calculate the LOD. (B) 4PLA was used to detect serial dilutions of purified prostasomes, spiked in 4PLA buffer (squares) and in 10% human plasma (circles). (C) The 4PLA mechanism was investigated by omitting each of the four antibodies used in the probe mix in separate reactions while still adding the corresponding oligonucleotide. The omission of any antibody resulted in reduction of the detection signals to background levels. The y axes show the C_T average and the x axes indicate the concentration of prostasomes. Error bars indicate SDs from the mean for triplicate reactions.

To ascertain that the detection reaction in fact depends on binding by all four PLA probes, we replaced one probe at a time with the corresponding concentration of the free oligonucleotide, normally attached to that particular antibody. Fig. 2C shows that omission of any of the antibodies resulted in background level signals, confirming the requirement for binding by all four antibodies.

Detection of Prostasomes in Plasma Samples from Prostate Cancer Patients.

We investigated levels of prostasomes in blood plasma samples from patients with prostate cancer and control subjects by using the 4PLA technology. Significantly increased average levels of prostasomes were observed in blood plasma from 20 patients with prostate cancer [median, 7.7 ng/mL; range 1.1–34.9 ng/mL; 95% confidence interval (CI)] compared with 20 age-matched controls (median, 1.1 ng/mL; range <1.1–12.4 ng/mL; 95% CI; P < 0.001; Fig. 3A). In a separate, blinded validation experiment, prostasome levels again were elevated in another constellation (the subgroup) of 13 patients (median, 2.9 ng/mL; range 1.3–4.6 ng/mL; 95% CI) compared with 11 age-matched controls (median, 0.5 ng/mL; range 0.5–1.8 ng/mL; 95% CI; P < 0.001; Fig. 3B). No prostasomes were detected in blood plasma samples from 12 prepubertal boys (Fig. S1), who had no prostasome production (38).

Fig. 3.

4PLA was used to measure levels of prostasomes in plasma samples from patients and control subjects. (A) The analysis of samples from prostate cancer patients (n = 20) revealed, on average, significantly higher concentrations of prostasomes than those observed in samples from age-matched control subjects (n = 20; P < 0.001). (B) The higher concentrations of prostasomes in samples from patients were confirmed in a blind-test validation experiment examining the subgroup of 13 patients and 11 age-matched control samples (P < 0.001). The results are shown as box plots in which the dashed lines extend between the minimum and maximum values, boxes extend between the lower and higher quartiles, and the horizontal black bars indicate the medians for the patients and controls, respectively. (C) Plasma samples from five patients with prostate cancer were pooled, and the level of prostasomes in the supernatant after ultracentrifugation (open bar) was compared with the level in pooled plasma that had not been ultracentrifuged (gray bar). Error bars indicate SDs from the mean for triplicate reactions.

To further establish that the 4PLA test indeed detects entities with properties of microvesicles rather than individual protein molecules, pooled plasma from five patients with prostate cancer with plasma PSA values between 10 and 120 ng/mL was ultracentrifuged at 200,000 × g for 2 h to quantitatively pellet prostasomes. Prostasomes were not detected in the supernatant of the plasma sample subjected to ultracentrifugation, whereas they were readily detected by 4PLA in the sample not subjected to ultracentrifugation (Fig. 3C).

Finally, we investigated the relation between blood plasma prostasome levels before surgery and histological evidence of aggressiveness of prostate cancer in a cohort of 59 patients whose tumors were histologically classified after radical prostatectomy as having Gleason scores from 5 to 9. A higher Gleason score means the tumor deviates more from normal prostate glandular tissue by being less well differentiated. The patients were divided in three groups with low Gleason scores (≤6; n = 20), medium scores (score of 7; n = 19), and high scores (8 or 9; n = 20). For each of the groups, blood plasma levels of prostasomes were measured by 4PLA and compared with levels for the same group of control individuals. The difference in levels of prostasomes for the same 20 control samples run with each of the three patient groups illustrates the current interassay variation. This variation obscures some of the differences of prostasome levels among samples and needs to be addressed by further optimization of assay precision and reproducibility. Patients with prostate cancer with Gleason scores of 6 or lower had prostasome levels similar to those of control subjects (median, 1.0 ng/mL; range, 0.9–2.3 ng/mL for patient samples vs. median, 1.0 ng/mL; range, 0.9–4.5 ng/mL for control samples; P = 0.94), whereas the levels in patient groups with Gleason scores of 7 and 8/9 (median, 2.2 and 2.9 ng/mL; range, 0.5–17.3 and 0.3–7.2 ng/mL, respectively) were both significantly elevated compared with their respective controls (median, 0.9 and 1.2 ng/mL; range, 0.5–3.5 and 1.2–3.0 ng/mL; P < 0.001 for both groups; Fig. 4). The medians for prostasome levels in patient samples displaying medium and high Gleason scores were between 2.5 and sevenfold higher than for background prostasome level in control samples.

Fig. 4.

Plasma levels of prostasomes in samples from patients with prostate cancer classified in three groups according to histological Gleason scores. Each patient group was analyzed in a separate experiment together with the same control group. The levels of prostasomes in plasma samples from patients with Gleason score 7 (medium) and 8/9 (high) were both significantly elevated compared with those with Gleason score of 6 or lower and those of controls (P = 0.001). The levels of prostasomes in samples from patients with Gleason scores of 6 or lower were similar to those in samples from controls (P = 0.94).

The PSA test failed to distinguish patients with Gleason score of 7 from those with scores of 6 or lower (P = 0.87), whereas PSA levels differed significantly in samples from patients with Gleason scores of 7 compared with those with Gleason scores of 8/9 (P = 0.01; Fig. 5A). The prostasome and PSA plasma levels did not correlate in any of the three patient groups (Fig. 5B).

Fig. 5.

(A) Box plots showing differences in levels of prostasomes and PSA, respectively, in patient groups with different Gleason scores. The P values were calculated by using a two-sample Wilcoxon rank-sum test. (B) Scatter plots illustrating the correlation between plasma prostasome and PSA levels for patients divided in three groups with Gleason scores of 6 or lower (n = 20), 7 (n = 19) or 8/9 (n = 20). ρ values are Spearman rank correlation coefficients, measuring the statistical dependence of PSA and prostasome levels as a function of Gleason scores.

Discussion

The acini of the prostate glandular epithelium consist of luminal epithelial cells on a layer of aligned basal cells. Surrounding these basal cells, there is a basement membrane, which separates acini from the stroma. The luminal epithelium consists of tall columnar cells that are highly polarized both morphologically and functionally, actively secreting products such as PSA and prostasomes into the glandular lumen. During progress of malignancy, the columnar cells are successively transformed into cuboidal cells that can appear in small conglomerates with loss of polarity, increasingly extending beyond the basal membrane to invade stromal tissue (39). These invasive malignant cells thus are confined to produce and export prostasomes to the interstitial space (27), increasing the probability of uptake in circulating blood. By using an extremely sensitive and specific method we demonstrate here that prostasomes can be detected at elevated levels in blood plasma from prostate cancer patients. Furthermore, this assay seems to distinguish patients with prostate cancer with medium and high Gleason scores from those with low Gleason scores and thus more benign disease. This clinically useful property of prostasomes as biomarkers may be related to loss of polarity of the secretory cells, as well as their composite structural arrangement in comparison with the lower molecular weight PSA molecules. The greater size and lower diffusion capacity of prostasomes compared with the PSA molecules may reduce the risk that they would occur at elevated concentrations in blood in nonmalignant states such as hyperplasia or inflammation. More studies will be needed, however, to establish any such benefits of this assay.

The 4PLA test, which depends on simultaneous recognition of targets by four different antibody preparations recognizing five different epitopes, exhibited improved sensitivity and specificity for prostasomes compared with solid-phase PLA, in which three recognition events are required for detection. This improvement may be explained in part by the lower risk of target-independent proximity of all the binders and thus nonspecific background, but 4PLA also provides particular advantages for specific detection of multiprotein complexes as diagnostic targets while avoiding interference from individual proteins or complexes containing only some of the targeted proteins. To our knowledge, this is the first assay that depends on simultaneous binding to as much as five different epitopes for detection.

This study thus expands the scope for biomarker diagnostics by illustrating the possibility to measure in blood plasma (and possibly in other body fluids) previously inaccessible classes of markers, such as complexes of interacting proteins, aggregates, and multiprotein structures, e.g., microvesicles. The available technologies for sensitive protein detection typically use pairs of affinity reagents in a sandwich configuration. This is true for the popular sandwich ELISA (40), and for more recent highly sensitive techniques such as the biobarcode assay (41), single-molecule counting assays (42–44), and homogeneous PLA (36, 37, 45).

The specificity of our method was illustrated by the lack of signals in the supernatant of ultracentrifuged blood plasma sample from patients with prostate cancer. It was further confirmed by the failure to detect prostasomes in plasma from prepubertal boys, in agreement with previous investigations, demonstrating the androgen dependence of prostasome production and secretion (38, 46). In contrast to measurements of PSA, the prostasome test in blood plasma successfully distinguished prostate cancers with low from those with medium and high Gleason scores, reflecting disease severity.

Due to its high sensitivity and specificity for prostasomes in blood samples the assay described herein is promising as a diagnostic and prognostic test for prostate cancer, and potentially analogous assays may be developed for exosomes released from other cancers. We continue to optimize the assay to further increase sensitivity, precision, and reproducibility via, e.g., automation and alternative means for readout of reporter DNA strands. Potential applications include early diagnosis, monitoring under active surveillance, selection of therapy in localized disease, and monitoring of responses to treatment.

Materials and Methods

Plasma Samples from Patients and Control Subjects.

Blood plasma was obtained from two groups of patients with prostate cancer and compared with age-matched controls. A first group included samples from 20 patients (age, 52–69 y; PSA, 94–2,706 ng/mL) and 20 age-matched controls with PSA levels lower than 2.5 ng/mL. In a second group, samples from 59 patients (age, 53–73 y; PSA, 1.1–39.1 ng/mL) were compared with 20 age-matched controls (age, 53–75 y; PSA, 1.7–14.8 ng/mL) with benign results from transrectal ultrasound-guided biopsy. Thirteen patients (PSA, 4.3–22.2 ng/mL) and 11 control subjects (PSA, 2.7–14.8 ng/mL) were recruited from the second group to constitute a subgroup for a blinded validation experiment (Results). All analyses were approved either by the ethical committee of Uppsala University or by the internal review board of the University of Münster in accordance with practices and ethical standards of the committee on ethical issues of the university and the Declaration of Helsinki, including informed consent by the patients.

Antibodies.

Monoclonal antibodies mAb78 and mAb8H10, directed against seminal prostasomes, were produced in mice by intrasplenic immunization, and the antibodies were biotinylated as described (17, 47, 48). The mAb78 was previously used to recognize prostasomes in benign and neoplastic cells of the prostate gland (26, 47) and PC-3 cells (25), indicative of its functionality in a cell-biological context. The monoclonal anti-CD13 antibody was from AbD Serotec. Biotinylated, polyclonal anti-human coagulation factor III/tissue factor antibodies were purchased from R&D Systems (cat. no. BAF2339).

Preparation of Prostasomes.

Fresh semen samples were obtained from normospermic men according to the World Health Organization Laboratory Manual during evaluation for in vitro fertilization. Semen samples were centrifuged for 20 min at 1,000 × g at 21 °C to pellet spermatozoa and any other cells from the seminal plasma.

Seminal prostasomes were prepared from pooled seminal plasmas by procedures including differential centrifugation, preparative ultracentrifugation, and separation by gel chromatography as previously described (17). The protein content of the prostasome calibrator was determined using a total protein assay (Protein Assay ESL; Roche Diagnostics). All preparatory procedures were carried out at 0–4 °C if not otherwise stated.

Preparation of Probes.

Sequences for all oligonucleotides are shown in Table S1. All oligonucleotide–streptavidin conjugates used to prepare PLA probes for 4PLA and solid-phase PLA tests were combined with free streptavidin and briefly heated to obtain streptavidin tetramers containing reduced number of oligonucleotides as described (49). Oligonucleotide–streptavidin conjugates (100 nM) were incubated with 100 nM biotinylated antibodies in 1× PBS solution for 1 h at 21 °C. The conjugated probes were diluted in 4PLA buffer [10 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 0.025% Tween 20 (Sigma-Aldrich), 1 mM D-biotin (Invitrogen), 1% BSA and 100 mM NaCl], incubated for 20 min at 21 °C, and stored at 4 °C for a maximum of 2 mo. For analyses of blood plasma samples, the oligonucleotides SLC1 and SLC2 were conjugated to mAb78 and mAb8H10, respectively, and the oligonucleotides Acc1 and Acc2 were conjugated to polyclonal anti-human coagulation factor III/tissue factor antibodies, respectively.

The pair of antibodies used to prepare probes for PLA according to a previously published protocol (50) were mAb78 and mAb8H10, and the four PLA probes in 4PLA used mAb78, mAb8H10, and two aliquots of polyclonal anti-tissue factor antibodies. In 4PLA, the oligonucleotides attached to two of the PLA probes were designed to hybridize to hairpin-loop structured blocking oligonucleotides to prevent hybridization to the other PLA probes in the absence of target binding. When bound in proximity, these hybrids are outcompeted by hybrids between oligonucleotides on different antibodies. After washes, a short bridging oligonucleotide is added to join the oligonucleotides on two PLA probes via two enzymatic ligation reactions. Oligonucleotides on PLA probes having failed to bind in proximity become ligated to the blocking oligonucleotides, thus preventing nonspecific generation of amplifiable DNA strands.

Detection of Prostasomes.

Prostasomes were detected by PLA essentially as described (50) or as adapted for 4PLA. Briefly, capture antibodies were immobilized in 0.2 mL reaction tubes (AJ Roboscreen) by using 50 μL of 1 ng/μL anti-CD13 in coating buffer (0.05 M sodium bicarbonate, 0.05 M sodium carbonate, 0.015 M sodium azide, pH 9.6) at 4 °C overnight. Excess antibodies were removed by three washes in 1× Tris-buffered saline solution and 0.01% (vol/vol) Tween 20.

Five microliters of purified prostasomes (1 mg/mL) or blood plasma samples from patients or controls diluted tenfold in 4PLA buffer to a final volume of 50 μL were added to the tubes and incubated without agitation for 2 h at 37 °C, followed by three washes as described earlier. Next, 50 μL of 500 pM PLA probe mixtures were added, and the tubes were incubated at 37 °C for 2 h and unreacted probes were removed as described earlier.

Thereafter, a 50-μL ligation/amplification mix was added, containing 1× PCR buffer (Invitrogen), 2.5 mM MgCl₂ (Invitrogen), 0.2 μM of each Biofwd and Biorev primer, 0.4 μM TaqMan probe, 0.08 mM ATP, 0.2 mM of each deoxynucleoside triphosphate (containing dUTP), 1.5 U platinum Taq DNA polymerase (Invitrogen), 0.5 Weiss units of T4 DNA ligase (Fermentas), 0.1 U uracil-DNA glycosylase (Fermentas), and 40 nM cassette oligonucleotide for 4PLA and 200 nM connector oligonucleotide for PLA. The reactions were incubated for 5 min at 21 °C and qPCR was then performed in an Mx-3000 or Mx 3005 instrument (Stratagene), with an initial incubation for 2 min at 95 °C, followed by 45 cycles of 95 °C for 15 s and 60 °C for 1 min. The results are presented as threshold cycle (C_T) values, reflecting the amount of PLA ligation products. For clinical samples the C_T values were converted into mass quantities by using data from a standard curve using seminal prostasomes, run in same experiment. Prostasomes were expressed as nanograms of prostasomal protein per milliliter blood plasma.

Data Analysis.

The qPCR data were analyzed with MxPro software (Stratagene), and the recorded C_T values were exported and further analyzed using the R statistics software package. Logistic regression models were calculated by using the drc package in R. The statistical significance of the difference in levels of prostasomes between patient and control groups as determined by 4PLA was calculated by using a two-sample Wilcoxon rank-sum test in R. The LOD was determined as the concentrations that resulted in a signal two SDs greater than the mean background levels.