Phosphodiesterase 11A (PDE11A) Genetic Variants May Increase Susceptibility to Prostatic Cancer

Abstract

Context: Among the genomic loci harboring potential candidate genes for prostatic cancer (PCa) is the 2q31-33 chromosomal region that harbors the gene encoding phosphodiesterase 11A (PDE11A). In addition, the combined cancer genome expression metaanalysis datasets included PDE11A among the top 1% down-regulated genes in PCa.

Objective: In the present study, we screened 50 unrelated PCa patients of Brazilian descent for PDE11A coding defects.

Design: The study consisted of PDE11A sequencing, in vitro functional assays, and immunostaining analysis.

Results: We identified eight different sequence alterations in 15 patients (30%): one stop-codon and seven missense mutations. Three of the variants (R202C, Y658C, and E840K) were novel, and the remaining five (Y727C, R804H, R867G, M878V, and R307X) have been associated with predisposition to adrenal or testicular tumors. The overall prevalence of PDE11A-inactivating sequence variants among PCa patients was significantly higher than in 287 healthy controls (0.16 vs. 0.051, respectively, P < 0.001, odds ratio 3.81, 95% confidence interval 1.86–7.81) and the R202C, Y658C, and E840K substitutions were not found in controls. All missense mutations led to decreased PDE11A activity in human embryonic kidney 293 and PC3M cells and immunostaining of PCa samples with sequence changes showed decreased PDE11A protein expression.

Conclusion: Our data suggest that, like in the adrenal cortex and the testicular germ cells, PDE11A-inactivating genetic alterations may play a role in susceptibility to PCa.

PDE11A, that is polymorphic in the general population and conveys risk to adrenal and testicular tumors, may also be a susceptibility gene for prostatic cancer.

Prostatic cancer (PCa) is the most common noncutaneous cancer among males in the United States, with an estimated incidence in excess of 200,000 new cases per year. In addition, PCa causes approximately 30,000 deaths per year, making PCa the second leading cause of cancer-related death after lung cancer among males in the United States (10% of all cancer deaths in men) (1). Genetic predisposition to PCa is suggested by familial clustering, younger age of diagnosis in families, and increased incidence in certain populations, such as African-American males (2,3). However, identifying specific high-risk susceptibility variants has been challenging due to extreme genetic heterogeneity (4,5). Recently a number of regions harboring potential susceptibility genes for PCa have been located in the genome (2,3,4,5,6,7).

Chromosome 2q31-33, containing the gene encoding phosphodiesterase 11A (PDE11A), is among the most highly linked to PCa genomic regions (2,3). In addition, PDE11A is among the top 1% of mRNAs that are down-regulated in PCa compared with normal prostate, according to expression metaanalyses (8). PDE11A is a dual-specificity phosphodiesterase (PDE) that catalyzes the hydrolysis of both cAMP and cGMP (9). PDE11A protein-truncating mutations were first identified in a series of patients with rare forms of adrenocortical hyperplasias (10); however, other milder PDE11A-inactivating mostly missense mutations were later implicated in the predisposition to a variety of endocrine neoplasms, including adrenocortical tumors, such as adenomas and cancer, and testicular germ-cell tumors (10,11,12,13).

Thus, the PDE11A gene by virtue of its genetic linkage and its involvement in other endocrine tumors appears to be a good candidate for PCa. Additional lines of evidence suggesting PDE11A involvement in PCa included: 1) the 2q chromosomal region has been reported in cytogenetic rearrangements that were identified in PCa (14); 2) PDE11A is highly expressed in normal prostate, and its expression is decreased in PCa as demonstrated by immunostaining (4,15); and 3) cAMP signaling involvement in PCa cell growth and modification of androgen receptor effects (16).

In the present study, we analyzed the coding sequence of PDE11A in patients with PCa. We then assessed the functional effects of the identified genetic variants in vitro and studied PDE11A protein expression in tumor tissues from PCa patients with and without PDE11A sequence defects. Data suggest that germline PDE11A-inactivating variants may confer a certain risk for development of PCa.

Materials and Methods

Subjects and protocol

The PDE11A gene was analyzed in 50 patients of Caucasian descent with PCa (45 sporadic and five familial) (Supplemental Table 1). Written informed consent was obtained from all participants, and the study was approved by the institutional review boards of participating centers. Frozen prostate tumors tissues were collected by the A. C. Camargo Hospital Biobank. For diagnosis and scoring, tumors were fixed in formalin and embedded in paraffin, and 4-μm sections were cut and stained with hematoxylin and eosin. Two control groups were used for this study: 1) 192 previously described individuals (100% Caucasians) with a negative family history of endocrine disorders (designated the endo-negative group) (10,13), and 2) 95 DNA samples from unselected individuals that were negative for the most common adult diseases (designated the Coriell group). Group ii samples were obtained from the commercially available Coriell Institute database. Eighty percent of the Coriell group was of Caucasian descent, and the rest were with mixed or unknown origin. For all individuals, patients and controls, the complete PDE11A-coding and surrounding intronic sequence was analyzed, as previously described (10,11,12).

Functional studies

For the transfection experiments, the PDE11A gene-open reading frame was cloned into pCR3.1, and all missense mutations (including the new ones, R202C, Y658C, E840K) were introduced by overlapping PCR, as previously described (10). Primers used for vector generation are described in Supplemental Table 2. The human embryonic kidney (HEK) 293 and PC3M (derived from prostate adenocarcinoma) cell lines were transiently transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) following the manufacturer’s protocol. Cells were transfected with 6 μg of plasmid DNA expressing either the wild-type or the mutated form of PDE11A, harvested 48 h after the transfection, and subjected to PDE activity assay, as previously described (11). The in silico analysis of these single-nucleotide polymorphisms were done using the program PolyPhen (http://coot.embl.de/polyphen).

To assess PDE11A protein expression in affected tumor tissues, we applied immunohistochemistry, as described previously (10).

Results

We identified eight different variants in a total of 15 of our patients (30%): seven nonsynonymous substitutions and one stop-codon mutation. Three of the nonsynonymous substitutions were novel (R202C, Y658C, E840K). We have reported before three nonsynonymous substitutions (R804H, R867G, M878V) that are PDE11A-inactivating, and the stop-codon mutation c.171TdelFS41X, all in patients with adrenocortical and testicular tumors, and, rarely, in the general population (10,11,12). One nonsynonymous substitution (Y727C) was previously known as a polymorphic variant and was also found to significantly impair the ability of the protein to degrade cAMP in vitro (11).

We compared the frequency of PDE11A nonsense and missense variants in the PCa patients with those in 287 healthy control individuals of predominantly Caucasian descent: 192 endo-negative individuals, and 95 Coriell controls (Table 1). Two previously missense variants (p.A349T, p.D609N) reported by our group were identified only among the control individuals (11). Analyzed individually, only the Y727C and E840K variants showed significantly higher frequency in patients with PCa compared with the controls (χ² = 6.02, P = 0.014, and χ² = 5.75, P = 0.017, respectively). The cumulative frequency of the coding variants in the patients with PCa was also significantly higher (χ² = 14.67, P < 0.001, odds ratio 3.81, 95% confidence interval 1.86–7.81). The newly identified substitutions among patients (R202C, Y658C, E840K) were not seen among the controls and have not been seen before in any of our other studies. They were also not present in the extended set of 625 New York Cancer Project control individuals that were genotyped for all identified so far PDE11A variants.

Table 1.

Allele frequency (minor allele) of sequence variations in PDE11A in PCa patients and control individuals

Sequence change	PCa patients (100 alleles)	Controls			PC patients vs. controls
		Endo-negative (384 alleles)	Coriell (190 alleles)	Total (574 alleles)	χ2	P
c.171Tdel FS41X	1 (0.01)	0 (0.00)	2 (0.01)	2 (0.003)	0.01	Ns
R202Ca	1 (0.01)	0 (0.00)	0 (0.00)	0 (0.00)	0.98	Ns
A349T	0 (0.00)	1 (0.002)	1 (0.005)	2 (0.003)	0.16	Ns
D609N	0 (0.00)	0 (0.00)	1 (0.005)	1 (0.002)	0.98	Ns
Y658Ca	1 (0.01)	0 (0.00)	0 (0.00)	0 (0.00)	0.98	Ns
Y727C	5 (0.05)	0 (0.00)	6 (0.03)	6 (0.01)	6.02	0.014
R804H	3 (0.03)	5 (0.01)	2 (0.01)	7 (0.012)	0.83	Ns
E840Ka	2 (0.02)	0 (0.00)	0 (0.00)	0 (0.00)	5.75	0.017
R867G	1 (0.01)	5 (0.01)	5 (0.03)	10 (0.017)	0.01	Ns
M878V	2 (0.02)	0 (0.00)	1 (0.005)	1 (0.002)	2.95	Ns
Total	16 (0.16)	11 (0.029)	18 (0.094)	29 (0.05)	14.67	<0.001

χ² is calculated after Yates correction for continuity (Yates correction was applied for all calculations having a number less than 10 in any cell of the contingency table). Ns, Nonsignificant. 

^aNewly identified in the PCa patients only variations. 

All of the mutations were found in the heterozygote state; one patient (PCa no. 23DT) was a carrier of two mutations: the stop codon variant c.171TdelFS41X and the missense change Y727C. So far, the c.171TdelFS41X and Y727C sequence defects have always been observed separately and on different alleles (10,11,12), although lack of tissue prevented us from studying allelic expression in the patient with PCa.

To assess the potential of the newly identified PDE11A mutations (R202C, Y658C, E840K) to impair protein function, we performed transfection experiments in HEK293 cell line and the human prostate cancer cell line PC3M (Fig. 1, A and B). As with the other PDE11A missense variants found in patients with adrenal and testicular tumors, all three substitutions led to decreased PDE activity in both HEK293 and PC3M cells. This effect is likely to be different in various organs and cells. An example of this tissue dependence of PDE11A activity was the observation that the effects of the constructs with mutations were different in the two cell lines, HEK293 and PC3M. We have observed different cell line-dependent effects of PDE11A sequence variants in other settings too: for example, between HEK293 and Hela cells, with regard to cAMP and cGMP handling of this dual-specificity PDE (11). Overall, the most significant inhibitory effect on PDE activity on both cell lines was observed after transfection with the Y658C-harboring construct. This observation is consistent with the in silico analysis that predicted a high potential of this sequence change to impair protein function (Supplemental Table 3).

Figure 1.

PDE activity and immunohistochemistry for PDE11A. PDE activity after transfection of HEK293 cells (A) and PC3M cells (B) with wild-type (WT) and mutant PDE11A expression vectors. C, Expression of PDE11A in surrounding normal prostate cells (magnification, × 20). D, Expression of PDE11A in tumor tissue from the same PCa patient. Magnification of ×40 of PDE11A immunostaining in surrounding normal prostate cells (E) and in tumor tissue (F).

Immunohistochemistry in tumor tissue showed lower PDE11A protein expression compared with the adjacent normal tissue sections for all available for the analysis 16 PC patients: two with no PDE11A mutations and 14 PDE11A mutation carriers; representative immunostaining is shown in Fig. 1, C–F.

Discussion

PCa links to more than 30 distinct genomic regions and is thus considered one of the most genetically heterogeneous malignancies in humans. Fine mapping and sequencing of these regions, however, has not always confirmed the initial association and often produced contradictory results. Failure to identify single high-penetrance disease-causing variants accounting for a reasonable number of cases led the researchers to suggest an interactive network of mild to moderate genetic changes as a likely predisposition model of PCa (8,17). In this study, we observed a significantly higher frequency of PDE11A coding changes in patients with PCa (30%) compared with healthy controls (10%, P < 0.001, odds ratio 3.81, 95% confidence interval 1.86–7.81). When compared with other genes with a potential predisposition to PCa, the prevalence of PDE11A variations in our PCa group among controls was similar or higher.

Other defects that have been recently implicated in predisposition to PCa include macrophage scavenger receptor 1 gene (MSR1), located in the genetic locus 8p22-24, most consistently associated with PCa and considered one of the strongest candidate genes, showed comparable frequencies of deleterious variations in patients and controls and was not identified as a major PCa susceptibility gene (18). ELAC2, extensively studied for association with PCa, was also found to eventually have similar genetic variation in disease cases and controls and to play an apparently weak role in susceptibility to PCa (19).

Considering the possible additive effect of different PDE11A genetic variants in PCa, it is noteworthy that our patient PCa no. 23 carrier of two different PDE11A coding changes, Y727C and the stop codon mutation c.171TdelFS41X and was most likely a compound heterozygote; he was one of five patients with a familial history of PCa (affected brother), relatively early disease onset (58 yr) and Gleason score 7 (see Supplemental Table 1). Despite the segregation model, however, the association of a more severe phenotype with a stop codon PDE11A mutation favors its involvement in the PCa pathogenesis, among the PCa patients, only PCa no. 23 carried a PDE11A protein-truncating variant.

Incomplete penetrance of PDE11A variations is not clearly understood but likely relates to the versatility and redundancy of the PDE family: a PDE11A defect can easily be compensated by other cAMP- or cGMP-specific PDEs. The presence of a defect almost certainly predisposes to mild effects at least in endocrine tissues, as our studies demonstrate. This suggests that the endo-negative group has a rate of PDE11A sequencing variants lower than the general unscreened population (13). Our previous experiments with inactivating PDE11A mutations in the adrenal cortex are suggestive of partial loss of heterozygosity in tumor tissues; a similar mechanism may take place in prostate cancer tumors (10). However, other mechanisms responsible for the reduction in PDE11A level expression in tumor tissue may be present.

In conclusion, after the adrenal cortex and the testes, prostate is the third gland in which tumors develop in patients with PDE11A-inactivating sequence variants. These tissues are extremely sensitive to cAMP levels and have the highest expression of PDE11A (10). In fact, compared with patients with adrenocortical and testicular tumors, patients with PCa display a higher frequency of PDE11A defects (with an allelic frequency of 0.16 vs. 0.10 and 0.13 in the adrenocortical and testicular, respectively, tumor groups, P < 0.05) (Supplemental Table 4). Whether PDE11A sequence defects predispose to other endocrine tumors remains unclear. It should also be noted that PDE11A is inhibited by some of the PDE inhibitors used for erectile dysfunction in patients with a similar age to those at risk for PCa. One of the several related trials assessed the relation between daily use of tadalafil and benign prostatic hyperplasia (20). Future studies should address the frequency of prostatic side effects taking into consideration the PDE11A genotypes.