PAPER I
Progressive prostate hyperplasia in adult Mt-PRL transgenic mice is not dependent on elevated serum androgen levels
Transgenic mice overexpressing the rat PRL gene under control of the ubiquitous Mt-1 promoter develop a dramatic prostatic enlargement with parallel chronic hyperprolactinemia and elevated serum androgen levels.
Histologically the prostate enlargement is mainly characterized by an expansion of the stromal compartment and areas of glandular hyperplasia with an accumulation of secretory material [105]. In paper I, we aim to clarify the role of circulating androgen levels in the promotion of abnormal prostate growth in the adult Mt-PRL transgenic mouse prostate.
Separate groups of 12 weeks old animals (age-matched wild-type and Mt-PRL transgenic males) were surgically castrated followed by subcutaneous implantation of slow-release testosterone pellets containing 7.5 mg testosterone or placebo substance. The testosterone dose of 7.5 mg, was chosen to give as normophysiological levels of testosterone as possible, and was found to not significantly differ from the circulating testosterone serum levels of wildtype controls. After 8 weeks of hormone/placebo pellet treatment, animals were killed followed by serum sampling and prostate dissection. As an additional control, prostates from age-matched groups of non-treated wildtype and Mt-PRL transgenic mice were collected at both start and endpoint of the experiment. Results revealed that progression of prostate hyperplasia in adult Mt-PRL transgenic males was not affected by normalization of circulating testosterone levels. Immunohistochemical studies revealed a significantly increased proportion of AR positive epithelial cells in all prostate lobes of the Mt-PRL transgenic compared to wild-type. The increased distribution of epithelial AR remained high in the group of animals that were castration and substitution to normophysiological androgen levels.
In addition, the Mt-PRL transgenic males possess more prominent stromal AR positivity than wildtype controls.
The present study demonstrates that progressive prostate hyperplasia in adult Mt-PRL transgenic mice is not dependent on the elevated serum androgen levels present in the animals. In addition, our results suggest that prolonged hyperprolactinemia results in changes in prostate epithelial and stromal cell AR distribution. The increased AR distribution in both epithelial and stromal
- 48 -
cells in the Mt-PRL transgenic prostate lobes may increase the androgen sensitivity and thereby also influence the development of the observed prostate phenotype.
Prolonged androgen treatment has no significant effect on prostate growth in wildtype adult mice
The data about the importance of prolonged exposure to extraordinary high levels of androgens on prostate growth in rodents is conflicting, showing both unaffected prostate size [255] and induction of hyperplasia [256]. To determine the long-term effects of elevated circulating androgen levels on the prostate gland of wild-type male mice, a separate group of 12-week-old wildtype mice were sham-operated and subcutaneously implanted with 30 mg of testosterone slow-releasing pellet. The high dose of 30 mg testosterone was selected to give the treated group of wildtype animals a comparable levels of circulating testosterone as the Mt-PRL transgenic male mice have. After 8 weeks of treatment, prostates were dissected and serum samples obtained. On average, these animals displayed a 4-fold increase in serum testosterone levels compared to untreated wildtypes. These testosterone levels did not significantly differ from levels found in Mt-PRL transgenic males. Prostate wet weight in testosterone-treated wildtype did not significantly differ from that in untreated wildtype males neither as separate lobe weight nor as total organ weight. Histological appearance of the prostate lobes was not either different from that observed in wildtypes. These findings establish that prolonged androgen stimulation of adult male mice (C57BL/6JxCBA-strain) has no significant effects on prostate growth or histological appearance. This also supports the conclusions drawn from the results in castrated and androgen substituted Mt-PRL transgenic males, indicating that the hyperplastic process in transgenic prostate is not dependent on an elevated state of circulating androgens.
Like the Mt-PRL transgenic mice, wildtype males treated with 30 mg of testosterone exhibited significantly higher numbers of AR-positive epithelial cells compared to untreated wildtypes. However in contrast to the Mt-PRL transgenic mice, stromal AR content was unaffected in testosterone-treated wildtypes.
Taken together, these results show that prolonged androgen stimulation of young adult male mice has no significant effects on prostate growth or histological appearance. These data further support the findings in castrated and androgen substituted Mt-PRL transgenic males, that progression of prostate hyperplasia is not dependent on elevated levels of circulating androgens. Moreover, the results from the immunohistochemical analysis
- 49 -
suggest that the prostate hyperplasia of Mt-PRL transgenic mice is not primarily mediated via increased epithelial AR contents.
Comparison of post-castrational regression patterns in the prostate lobes of Mt-PRL transgenic mice compared to wildtype mice
Involution of the prostate after androgen-deprival by testicular castration is a well characterized process. Regression of the epithelial cell population, through an active process of apoptosis, occurs rapidly after castration. To establish the androgen dependency of this PRL transgenic prostate model, separate groups of 12-week-old male Mt-PRL transgenic and wildtype mice were castrated and subcutaneously implanted with placebo pellets. After 8 weeks, prostates were dissected and serum samples obtained. DLP and VP were significantly reduced after castration, in both Mt-PRL transgenic and wildtype prostates. In addition, similar histological appearance, with marked loss of both glandular epithelium and interductal stroma, were observed in both groups. Post-castrational VP weights in Mt-PRL transgenic males did not significantly differ from those of wildtype, whereas post-castrational DLP weight was significantly higher in Mt-PRL transgenic than in wildtype.
However, considering the small but significant weight difference already at 12 weeks of age, the relative rate of reduction in DLP weight after castration was similar in Mt-PRL transgenic and wildtype, -66% and -78%, respectively.
Altogether, these data show that androgens are clearly required for maintaining the transgenic phenotype as demonstrated by the similar patterns of prostatic regression seen in Mt-PRL transgenic and wildtype mice after androgen withdrawal. In the DLP, some weight differences were maintained after androgen-deprival; this difference may be attributable to the existing difference in glandular size at the time of castration. This finding could also partly be due to lobular differences in PRL responsiveness reported earlier [171, 175].
PAPER II
Isolation of differentially expressed transcripts in the enlarged prostates of Mt-PRL transgenic mice compared to controls
The objective of this study was to characterize the molecular mechanisms in the prostate of importance for the prostate hyperplasia seen in Mt-PRL
- 50 -
transgenic mice. Therefore, the method of cDNA representational difference analysis (cDNA RDA) was used which allow identification of novel genes that were differentially expressed in the enlarged prostates of the Mt-PRL transgenic mice compared to controls. The cDNA RDA was performed on prostatic tissue (DLP and VP) from mice of an age of four to six months. To generate samples as representative as possible, reflecting the prostate phenotype, RNA samples were pooled from four transgenic mice and five control littermates, respectively. Representations, generated from control and transgenic cDNA, respectively, were used as driver (control) and tester (hyperplastic), or vice versa, to generate both up- and down-regulated transcripts in the hyperplastic prostates compared to controls (see METHODOLOGICAL CONSIDERATIONS). Two successive rounds of subtraction and amplification were performed, generating libraries containing two sets of difference products, DP2-hyperplastic and DP2-control. Upon gel electrophoresis of the Sau3AI-cut RDA products of DP2-hyperplastic and DP2-control, six distinct bands were visualized. To subclone as many different products as possible from each library, each band were excised from the agarose gel and subcloned individually. 384 bacterial colonies, 192 from each RDA library, were picked, plasmid DNA was prepared followed by routine sequencing. After sequence alignments, the sequences were analyzed for homologies with published sequences in the non-redundant and EST divisions of the public databases of NCBI. This reduced the complexity of the RDA output so that the 384 clones sequenced, was reduced to 152 different unique sequences having a length longer than 50 base pairs. 69 of these, 37 DP2-hyperplastic and 32 DP2-control, were identified as previously annotated transcripts, whereas 83 were novel sequences not found in the public databases (referred to as unknowns) at the time when the study was performed.
Verification of the RDA output by using cDNA Microarray Analysis
To confirm that the obtained RDA difference products represented truly differentially expressed transcripts, the 152 non-redundant RDA products were selected for further verification using cDNA microarray analysis. 28 of the different RDA products were printed in duplicates, at different locations on the chip, to serve as “within slide” reproducibility controls (see METHODOLOGICAL CONSIDERATIONS). The RDA-derived microarrays were co-hybridized with labeled control and transgenic total RNA from a new set of animals. In order to eliminate dye specific effects caused by a labeling bias, dye-swap design of targeting labeling was used. The hybridizations were performed four independent times, twice Cy3-labeling the control and Cy5-labeling the hyperplastic total RNA, and twice with opposite colors
(dye-- 51 (dye--
swapped). Probes rendering weaker signals than 1.4 times the background were eliminated and not considered for further analysis. Using this criterion, 48 of the 152 uniquely printed RDA clones could be detected in the Mt-PRL transgenic and control prostatic total RNA. To identify the significant differentially expressed transcripts, the data from the four repeated microarray experiments were statistically analyzed using the SAM algorithm [231].
Genes with average fold changes of more than 50% (correspond to a fold change of 1.5) were counted as differentially expressed. With an estimated FDR of less than 2%, 15 out of the 48 detected RDA products were identified as differentially regulated (of which 5 were unknowns). In terms of fold regulation, previous results from our laboratory have shown that this level can be reproduced, as shown by independent validation using RNase protection assay [257, 258]. Overall, the complexity of the RDA output could be largely reduced by: i) the annotation step (from 384 to 152), ii) the detection limitation of cDNA microarray technique (from 152 to 48), and iii) verification step (from 48 to 15) – resulting in 15 significantly differentially regulated transcripts, by an average fold-change of least 1.5, between the Mt-PRL transgenic and control prostates.
One might reflect on the low number of significantly differentially regulated transcripts that were identified in this study. It has to be clarified that the final outcome of RDA differential cloning method depends largely on how extensive one makes the cloning. In our study, we cloned a number of 384 transcripts and certainly the more bacterial colonies that are cloned the higher probability there is to isolate and cover the complete set of differentially expressed transcripts that there are between the two groups that are compared. This may contribute to missing out important transcripts.
Another reflection might be the obvious detection limitations of cDNA microarray analysis. There are several possible explanations for the relatively small number of detected transcripts. First, the ideal length of the cDNAs to be printed on cDNA microarrays is approximately 1000 base pairs. The lengths of our RDA products were between 200-500 base pairs, which probably contribute to reduce the sensitivity for detection by decreasing the hybridization. Second, a classical hybridization-based method like that of cDNA microarray depends on the specific activity of probes.
Third, the PCR amplification steps in the RDA makes small expression differences greater as well as enables detection of low expressed transcripts.
To enable verification of low expressed transcripts or small expression differences, an alternative method including PCR amplification steps, such as real-time RT-PCR, might need to be used. A further reflection is the relatively few significantly differentially regulated transcripts that were found at last. Most likely this is a consequence of the detection limitations of the method of cDNA microarray, but there are also other possible
- 52 -
explanations. A relative large proportion of the RDA output was expected to be so-called “false positive” clones as the RDA procedure was haltered after two rounds to. The reason for stopping the RDA subtraction and amplification steps at an early round is to diminish the loss of differentially expressed cDNAs and thereby maintain the RDA output diversity [259].
The verified differentially expressed RDA clones in hyperplastic versus control prostates
In the present study, 152 non-redundant transcripts were differentially cloned in the Mt-PRL transgenic prostate compared to control. Although, not all of these transcripts could be detected and/or verified using cDNA microarray analysis, we still think they together may contribute to an interesting result as many of them are new transcripts to be cloned in the prostate. Therefore, several of the 152 differentially expressed transcripts most likely hold information of differentially expressed transcripts between the Mt-PRL transgenic and control prostates which will be found to be significantly differentially regulated if another verification method than cDNA microarray is used.
Regarding the 10 annotated and verified differentially expressed transcripts, a number of them gave interesting information of possible molecular mechanisms involved in the development/ progression of the prostate hyperplasia of Mt-PRL transgenic mice. Of particular interest were the up-regulation of vimentin and the down-up-regulation of cytokeratin 8 which may indicate the importance of the “embryonic reawakening theory” in the development of the prostate phenotype of the Mt-PRL transgenic mice.
Furthermore, the down-regulation of aldose reductase may be a sign of involvement of reduced apoptosis for development of the hyperplasia of the Mt-PRL transgenic mice. In addition the down-regulation of the candidate tumor-suppressor, the transcript coding for the RIL protein may further contribute to the prostate phenotype of Mt-PRL transgenic mice.
In summary, the identified differentially expressed transcripts supports molecular similarities between the prostate hyperplasia of the Mt-PRL-transgenic mice and human BPH. Furthermore, the finding of new prostate hyperplasia related transcripts, both previously annotated and unknown transcripts, might be of large use both as potential biomarkers and to understand the underlying cause of benign growth of the prostate gland.
- 53 -
PAPER III
Generation of transgenic mice overexpressing the PRL transgene specifically in the prostate under normophysiological androgen levels
To address the role of local PRL action in the prostate, a new transgenic mouse model (Pb-PRL) was generated using the prostate-specific rat probasin (Pb) minimal promoter to drive expression of the rat PRL gene. Pb-PRL transgenic males developed a significant enlargement of both the DLP and VP lobes evident from 10 weeks of age and increasing throughout animal life span. In addition, the DNA content was measured in the prostate gland at 20 weeks of age, showing a significant three-fold increase in the DLP and VP, respectively, indicating a true hyperplasia with increased number of cells.
Expression of the transgene was restricted to the prostate (DLP, VP, and AP) and present from 4 weeks of age. Also, a weak expression of the transgene could be observed in seminal vesicles at this age. Moreover, transgenic rPRL was detectable at low levels in the circulation of transgenic animals from 10 weeks of age, most likely associated with the continuing increase in prostate size. In contrast to the ubiquitous Mt-PRL transgenic mice, serum androgen levels did not significant differ from that of wild-type mice at any time point.
The Pb-PRL prostate is histologically characterized by a significant stromal hyperplasia and secretion-filled distended ducts and focal areas of epithelial dysplasia. The glandular dysplastic foci had several morphological characteristics in common with low-grade prostatic intraepithelial neoplasia (PIN) lesions previously reported in other genetically engineered mouse models [260]. No high-grade PIN or prostate tumor formation were detected in Pb-PRL transgenic prostate. In addition, focal areas of mild to moderate chronic inflammation, exhibiting stromal mononuclear (primarily lymphocytes and macrophage) infiltrate, were frequently observed in both VP and LP lobes in Pb-PRL transgenic mice.
Furthermore, immunohistochemical analysis revealed a significant increase in stromal cell distribution of androgen receptors (AR) and estrogen receptors alpha (ERα). In contrast, distribution of estrogen receptor beta (ERβ) was nearly uniform in both Pb-PRL transgenic and wildtype prostate.
- 54 -
Comparative analysis of prostate ductal branching morphogenesis and quantitative analysis of prostate cellular composition in Mt-PRL and Pb-PRL transgenic mice compared to controls
To reveal possible phenotypic differences in ductal architecture due to different onset of transgenic rPRL expression, microdissection technique was used to examine branching morphogenesis of individual lobes in Mt-PRL and Pb-PRL transgenic prostate. Quantification was made by counting primary urethral ducts as well as duct branchpoints and terminal ductal tips at 12 weeks of age. In 12-weeks-old Pb-PRL prostate, no statistically significant differences were detected in the number of branch points per duct and the number of ductal tips present in each lobe compared to wild-type controls.
However, marked ductal dilation and elongation was seen in the Pb-PRL from an early age, and complete microdissection was not achievable in animals over 20 week of age due to the formation of a densely fibrous interductal stroma that abrogated its normally high susceptibility to collagenase. In contrast, counting of ducts and tips in Mt-PRL VP and LP lobes at the same age demonstrated a significant increase, with approximately a doubling in the number of branching points and terminal tips compared to wildtype, whereas the number of main urethral ducts remained unchanged. Like the Pb-PRL transgenic prostate, the ducts were elongated and more dilated compared to controls and microdissection was also prevented by formation of a densely fibrous stroma in prostate lobes of older Mt-PRL animals.
Quantitative analysis of prostatic tissue cellularity demonstrated a marked increase in the stromal to epithelial ratio in all lobes of both Mt-PRL and Pb-PRL transgenic prostates compared to controls. In wild-type controls, the lobe-specific stromal:epithelial ratio varied between 1:2.5 and 1:10, whereas in all lobes of Mt-PRL and Pb-PRL, transgenic prostate stromal and epithelial cells were present in approximately equal numbers.
Overall, the Pb-PRL transgenic represents a new model for the study of PRL effects in the prostate. Most significantly, the development of Pb-PRL hyperplasia occurs mainly post-pubertally and in a setting of normal androgen levels, thereby resembling the situation in the adult human prostate. This study indicates the ability of PRL to promote, directly or indirectly, ductal morphogenesis in the developing prostate and further to induce abnormal growth primarily of the stroma in the adult gland in a setting of normal androgen levels.
- 55 -
PAPER IV
Global analysis of gene expression in the enlarged prostate lobes of Pb-PRL transgenic mice
The objective of this study was to characterize the molecular mechanisms involved in the prostate hyperplasia seen in the Pb-PRL transgenic mice overexpressing the PRL gene specifically in the prostate. Global changes of gene expression were analyzed by using a cDNA microarray chip containing
The objective of this study was to characterize the molecular mechanisms involved in the prostate hyperplasia seen in the Pb-PRL transgenic mice overexpressing the PRL gene specifically in the prostate. Global changes of gene expression were analyzed by using a cDNA microarray chip containing