Author contributions: P.B., E.L.W.: conception and design; P.B., R.G., X.X., C.O., S.S.: collection and assembly of data; P.B., E.L.W.: data analysis and interpretation; P.B., D.M., E.L.W.: manuscript writing; P.B., R.G., X.X., C.O., S.S., D.M., E.L.W.: final approval of manuscript.
First published online in STEM CELLS EXPRESS May 28, 2009.
We have shown previously that prostatic stem/progenitor cells can be purified from isolated prostate ducts, based on their high expression of the Sca-1 surface antigen. We now report that high levels of aldehyde dehydrogenase (ALDH) activity are present in a subset of prostate epithelial cells that coexpress a number of antigens found on stem/progenitor cells of other origins (CD9, Bcl-2, CD200, CD24, prominin, Oct 3/4, ABCG2, and nestin). Almost all of these cells expressing high levels of ALDH activity also express Sca-1 and a third of them express high levels of this antigen. The cells with high levels of ALDH activity have greater in vitro proliferative potential than cells with low ALDH activity. Importantly, in an in vivo prostate reconstitution assay, the cells expressing high levels of ALDH activity were much more effective in generating prostatic tissue than a population of cells with low enzymatic activity. Thus, a high level of ALDH activity can be considered a functional marker of prostate stem/progenitor cells and allows for simple, efficient isolation of cells with primitive features. The elucidation of the role of ALDH in prostate stem/progenitor cells may lead to the development of rational therapies for treating prostate cancer and benign prostatic hyperplasia. STEM CELLS 2009;27:2220–2228
Aldehyde dehydrogenases (ALDHs) are a family of enzymes that oxidize aldehydes. Different ALDH family members play diverse roles in detoxification pathways and retinoic acid biosynthesis, as well as folate, amino acid, ethanol, and cyclophosphamide metabolism . Stem cells from a variety of tissues express high levels of ALDH activity, which may be a characteristic of “stemness” [2, 3]. Hematopoietic and neural stem cells are enriched in cells expressing high levels of ALDH activity (ALDH hi cells) [4–7] and ALDH hi cells with low side scatter are self-renewing and multipotent [6, 8]. The adipose-derived stem cell population also contains a subset of ALDH hi cells . The ALDH hi subpopulation of bone marrow cells also is enriched in hematopoietic, mesenchymal, and endothelial progenitor cells , indicating that high ALDH activity may be a general characteristic of stem/progenitor cells.
Normal stem cells and tumor cells have many common features , including expression of high levels of ALDH. Both normal and malignant human mammary [12, 13] and colon  stem/progenitor cells can be detected based on ALDH-1 activity. High ALDH activity has also been used as a marker for cancer stem/progenitor cells in prostate cancer cell lines . Tumor-initiating cells from colorectal cancer xenografts have also been isolated based on levels of ALDH activity  and a small fraction of cells from retinoblastoma tumors express the stem cell markers ABCG2, ALDH1, and Sca-1 . Thus, the expression of high levels of ALDH activity may identify both tumor-initiating and normal stem cells.
We have shown that prostatic stem cells are concentrated in the proximal region of prostatic ducts  and can be purified from this region by virtue of their high expression of the Sca-1 surface antigen . We now show that prostate stem/progenitor cells also express high levels of ALDH activity and that almost all of these cells coexpress Sca-1. Furthermore, the ALDH hi cells have higher in vitro and in vivo proliferative potential than cells expressing low levels of this enzyme.
MATERIALS AND METHODS
C57BL/6 mice, athymic nude mice, and CDIGS rats were housed in the animal research facilities of the University of Cape Town or New York University and all experiments were performed in compliance with institutional review board requirements.
Animals (6- to 8-week-old C57BL/6 mice) were sacrificed and the urogenital tract was removed en block into Hanks' balanced salt solution (HBSS), pH 7.4. The dorsal, ventral, lateral, and anterior prostates were dissected under a dissecting microscope using 25-gauge needles. In some instances, only the proximal region of prostatic ducts (those ducts nearest the urethra) was harvested [18, 19]. Cells were dissociated by incubation with 0.5% collagenase type II (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) in HBSS plus 7.5% fetal calf serum (FCS) for 45 minutes at 37°C, followed by digestion in 0.05% trypsin for 8 minutes at 37°C.
Separation of Cells with High and Low Levels of ALDH Activity
Cell digests were treated with lysing solution to lyse red blood cells (NH4Cl 0.15M, KHCO3 10 mM, EDTA 0.1 mM), washed with HBSS plus 5% FCS, resuspended in Aldefluor buffer (Aldagen, Durham, NC, http://www.aldagen.com) and passed through a 40-μm nylon cell strainer (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com). Cell viability was determined by trypan blue exclusion and cells were incubated with Aldefluor substrate for 30 minutes at 37°C, with and without the ALDH inhibitor, diethylaminobenzaldehyde (DEAB), according to the manufacturer's instructions. Aldefluor substrate, buffer, and DEAB was supplied in kit form by Aldagen. The substrate was converted by ALDH into a green fluorescent product that was retained in the cell and detected in the fluorescein isothiocyanate channel. After incubation, the cells were kept on ice and either separated by expression of high (ALDH hi) and low (ALDH lo) levels of ALDH activity by a FACS Vantage SE cell sorter (Becton Dickinson, San Jose, CA, http://www.bd.com) for in vitro and in vivo growth analysis or incubated with antibodies or isotype-matched immunoglobulins for fluorescence activated cell sorter (FACS) analysis, to determine the coexpression of ALDH activity with various other antigens (see below).
Cell digests were resuspended in Aldefluor buffer or in FACS buffer (phosphate-buffered saline containing bovine serum albumin 0.1%, sodium azide 0.01%, and aprotinin 20 μg/ml). Fc receptors were blocked with mouse CD16/32 antibodies and rat IgG. The cells were incubated with antibody or control IgG for 30 minutes on ice and washed with FACS buffer. In some experiments, the dye 7-aminoactinomycin D (1 μg/ml) was added 5 minutes prior to analysis so that dead cells could be excluded. Expression of intracellular antigens, such as Bcl-2, ALDH1/2, ALDH3A1, ABCG2, Oct 3/4, nestin, CK5, and CK8 were determined in paraformaldehyde-fixed cells, permeabilized with 0.2% Tween20 (Merck Schuchardt, Hohenbrunn, Germany, www.schuchardt.de) in phosphate-buffered saline. Cells were analyzed on a FACSCalibur flow cytometer (Becton Dickinson), using CellQuest software (Becton Dickinson). Sca-1+ cells with fluorescent intensities in the upper third were defined as Sca-1 hi cells.
Antibodies and Control Immunoglobulins (IgGs)
Rat anti-mouse CD9 biotin (clone KMC8) was obtained from BD Biosciences. Phycoerythrin (PE) conjugated rat anti-mouse Sca-1 (clone D7), rat anti-mouse Sca-1 biotin (clone D7), rat anti-CD24 biotin (clone CT-HAS), rat IgG2a biotin, rat IgG2b biotin, rat IgM biotin, rat IgG2a PE, rat IgG, mouse anti-mouse CD16/32, and streptavidin-conjugated allophycocyanin were from Caltag Laboratories (Burlingame, CA, http://www.caltag.com). Rabbit anti-nestin (clone H85), rabbit anti-ABCG2 (clone M-70), rabbit anti-Oct 3/4 (clone H134), rabbit anti-Bcl-2 (clone N-19), rabbit anti-ALDH1/2 (clone H-85) and rabbit anti-ALDH3A1 (clone M-76) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, http://www.scbt.com). Goat anti-EpCam biotin and rat IgG2a biotin (clone 54447) were from R&D Systems (Minneapolis, MN, http://www.rndsystems.com). Rat anti-prominin1 biotin (clone 13A4) and rat IgG1 biotin (clone eBRG1) were from eBioscience (San Diego, CA, http://www.ebioscience.com). Rat anti-mouse CD200 biotin (clone OX-90) was from AbD Serotec (Oxford, United Kingdom, www.abdserotec.com). Rabbit anti-mouse CK5 (clone AF138) was from Covance (Princeton, NJ, http://www.covance.com) and mouse anti-mouse CK8 (clone Ks8.7) was from Fitzgerald Industries International (Concord, MA, http://www.fitzgerald-fii.com). Mouse IgG1 and rabbit anti-mouse PE-Cy5 were obtained from DAKO (Glostrup, Denmark, http://www.dako.com); goat anti-rabbit APC from SouthernBiotech (Birmingham, AL, http://www.southernbiotech.com); and SA AlexaFluor 488 from Molecular Probes (Eugene, OR, http://probes.invitrogen.com).
In Vitro Growth of ALDH Hi and ALDH Lo Cells
ALDH hi and ALDH lo populations were collected from the cell sorter into medium (HBSS, 25 mM Hepes, 1 mM EDTA, 2% bovine serum albumin) and assessed for viability using trypan blue exclusion. For each experiment, cells (1,000–5,000 viable cells/well) were seeded in triplicate or quadruplicate in collagen-coated 96-well plates, (PureCol; INAMED, Fremont, CA, http://www.advancedbiomatrix.com) in DMEM/F12 1:1 (Gibco; Invitrogen, Carlsbad, CA, http://www.invitrogen.com) containing 10% FCS, basic fibroblast growth factor 2 (10 ng/ml), glutamine (2 mM), penicillin (100 U/ml), and streptomycin (100 μg/ml). Half the medium was removed and replaced every 2–3 days and cell number was determined after approximately 13 days.
Implantation of Grafts under the Renal Capsule
FACS-sorted ALDH hi and ALDH lo cells (4 × 104) were combined with urogenital sinus mesenchyme (UGM) cells (2.5 × 105) and resuspended in 30 μl of type 1 collagen (BD Biosciences). The collagen grafts were inserted under the renal capsule . Each experiment contained grafts of UGM alone to ensure that tissue growth did not result from contaminating urogenital sinus epithelial cells. Grafts were harvested and weighed after 10 weeks. UGM was isolated from the urogenital sinus of 18-day old embryos from CDIGS rats [20–22].
Immunohistochemistry of Xenografts
Grafts were fixed in 3% paraformaldehyde and embedded in paraffin. Immunohistochemistry was performed as described previously . Antibodies used for immunocytochemistry were: (a) for staining of basal cells, rabbit anti-mouse CK5 (clone AF138, 1:200; Covance) followed by goat anti-rabbit AlexaFluor 594 (1:2000; Molecular Probes); (b) for luminal cells, mouse anti-mouse CK8 (clone Ks8.7 undiluted; Fitzgerald Industries International) followed by goat anti-mouse AlexaFluor 488 (1:2000; Molecular Probes); (c) for neuroendocrine cells, mouse anti-synaptophysin (clone SY38, 1:500; Chemicon, Temecula, CA, http://www.chemicon.com) followed by goat anti-mouse AlexaFluor 488 (1:2000; Molecular Probes); (d) for prostate secretions, rabbit polyclonal antibodies against dorsal prostate secretions (1:2500; a gift from Dr. C. Abate-Shen, Columbia University, New York, NY) followed by goat anti-rabbit HRP (1:1000; GE Healthcare UK, Little Chalfont, United Kingdom, http://www.gehealthcare.com) and detected using a SK4100 DAB detection kit (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com); (e) for prostate-specific proteins; Nxk3.1 (1:600; a gift from Dr. C. Abate-Shen) followed by ABC kit (Vector Laboratories). The specificity of staining was ascertained on sections using nonimmune serum or IgG in place of primary antibodies.
The results are reported as means and standard deviations of each set of data. Comparisons between groups were made using the two-tailed, paired Student's t test or the Mann Whitney U test. A p value of < .05 was considered statistically significant.
A Minor Population of Murine Prostate Cells Express High Levels of ALDH Activity
As stem cells from other origins express high levels of ALDH activity [4–10, 12], we examined the activity of this enzyme in prostate cells. Cells from prostate digests were incubated with Aldefluor substrate (this substrate is converted into a green fluorescent product by ALDH) in the presence and absence of DEAB (10 μM), a potent inhibitor of ALDH, in order to determine the fraction of cells that express this enzyme. We found that 8.1 ± 2.1% of cells display high levels of ALDH activity, with the ALDH hi population having a mean fluorescence intensity (MFI) 5.2-fold higher than the ALDH lo population (MFI = 348 ± 96 vs. 67 ± 34; p < .001; n = 7; Fig. 1B), indicating that a subset of prostate cells express high levels of ALDH enzymatic activity. The majority of ALDH hi cells have medium to low side-scatter properties (Fig. 1B), indicating that the cells have medium to low granularity or cellular complexity. Phase contrast microscopy of DEAB-treated (Fig. 1C) or FACS-sorted (Fig. 1E) ALDH hi cells show that they are both composed of small round cells with an approximate diameter of 10 μm. Similarly, there is no difference in the sizes of the ALDH hi and ALDH lo cells (supporting information Fig. 1). Under fluorescence microscopy, cells treated with DEAB exhibit no fluorescence (Fig. 1D), whereas in the absence of inhibitor ALDH hi sorted cells show bright green fluorescence (Fig. 1F), indicating that they contain high levels of ALDH activity.
As we have shown that cells with stem cell features (quiescence, high proliferative potential) are concentrated in the proximal region of prostatic ducts , we determined the activity of ALDH in proximal cells compared with cells isolated from the whole prostate. We found that the proximal region contained 2.2-fold more cells (p < .01) expressing high levels of ALDH activity than cells isolated from the whole prostate (proximal region: 17.6 ± 3.6%, n = 4; whole prostate: 8.1 ± 2.1%, n = 7), indicating that ALDH hi cells are concentrated proximally (supporting information Fig. 2). However, as we obtained more ALDH hi cells from digests of the whole prostate than from only the proximal region of ducts, we used cells from the whole prostate for phenotypic and growth assays, in order to maximize the total yield of ALDH hi cells.
ALDH Hi Cells Are of Epithelial Origin and Express Antigens Present on Stem/Progenitor Cells of Other Origins
To determine if ALDH hi cells were of epithelial origin and contained markers expressed by stem cells of other origins, we examined these cells for the expression of EpCam, a marker of epithelial cells , as well as for antigens present in stem cells. We found that almost all (98.1 ± 1.5%, n = 3) ALDH hi cells coexpressed EpCam, confirming the epithelial nature of these cells (Fig. 2B, 2C). In addition, 70.2 ± 10.0% (n = 5) of ALDH hi cells expressed the basal cell cytokeratin, CK5, indicating that most of the ALDH hi cells were of basal origin , while 31.0 ± 20.8% (n = 3) expressed the luminal cell marker, CK8 (Fig. 2C). These data show that 2.3-fold more cells with high ALDH activity expressed CK5 than CK8 (p = .03).
We found that 93.7 ± 2.4% (n = 8) of ALDH hi cells expressed Sca-1 (Fig. 2C, 2E). Sca-1 is present on prostate stem cells [19, 26], as well as on stem cells from a variety of tissues, such as hematopoietic, cardiac, mammary gland, skin, muscle, and testis . Importantly, 36.3 ± 5.4% (n = 8) of ALDH hi cells expressed high levels of Sca-1 (Sca-1 hi; Fig. 2C, 2G). We also found that 34.6 ± 3.6% (n = 3) of Sca-1 hi cells expressed high levels of ALDH activity. Because prostatic cells with high levels of Sca-1 have high in vivo prostate regenerating potential , this indicates that the ALDH hi population is considerably enriched in stem cells.
Analysis of the expression of ALDH1/2 (74.8 ± 6.8%, n = 4) and ALDH3A1 (31.3 ± 17.0%, n = 3; Fig. 2C) confirms that cells isolated based on high levels of ALDH enzymatic activity expressed various aldehyde dehydrogenase isozyme proteins (the ALDH1/2 antibody detects ALDH1A1, ALDH1A2 and ALDH1A3, as well as ALDH2). ALDH3A1 and ALDH2 are involved in detoxification processes, ALDH1A1 is active in both detoxification and the retinoic acid pathway, while ALDH1A2 and ALDH1A3 participate in the retinoic acid signaling pathway .
The majority of ALDH hi cells also expressed high levels of CD9 (97.5 ± 2.1%, n = 3; Fig. 2C), a protein present on embryonic  and spermatogonial stem cells , as well as neural progenitor cells . Bcl-2, an antiapoptotic protein  that is expressed by prostate, hematopoietic, keratinocyte, and colon stem cells [19, 32–35], was present in 82.7 ± 5.8% (n = 4) of ALDH hi cells (Fig. 2C). CD200, an antigen expressed by hair follicle [36, 37] and mesenchymal stem cells , was found on 70.5 ± 5.7% of ALDH hi cells (n = 4; Fig. 2C). Almost half of the ALDH hi cells expressed Oct 3/4 (47.7 ± 2.5%, n = 2), whereas 42.9 ± 13.5% (n = 5) expressed ABCG2, both of which are present in primitive cells [39–42]. Approximately a third (37.3 ± 10.4%, n = 3) of ALDH hi cells coexpressed CD24, an antigen present on murine mammary stem cells . Prominin, a protein present on hematopoietic  and neural stem cells  and primitive prostate cells [46, 47], was expressed by 32.3 ± 5.6% of ALDH hi cells (n = 3; Fig. 2C). A small percentage of ALDH hi cells (6.8 ± 4.9%, n = 4; Fig. 2C) expressed nestin, a protein present on stem cells of prostate and neural origin [48, 49].
We next compared the expression of the various antigens found on ALDH hi cells with those on cells with low ALDH activity (Table 1). The expression of ALDH1/2 was increased 1.8-fold (p < .05) in the ALDH hi population compared with the ALDH lo population, demonstrating that ALDH 1/2 protein levels correlate with ALDH enzymatic activity (Table 1). CK5 expression was also increased 1.8-fold (p < .05) in the ALDH hi population, showing that this population is enriched in basal cells (Table 1). The expression of Sca-1 was increased 1.6-fold (p < .01), while the fraction of cells with high levels of Sca-1 was increased 1.8-fold (p < .01) in the ALDH hi samples (Table 1), confirming that primitive Sca-1 hi cells are concentrated in the population with high levels of ALDH activity. The ALDH hi population was also enriched in cells expressing Bcl-2 (2.0-fold; p < .05), Oct 3/4 (2.1-fold; p < .01), ABCG2 (2.5-fold; p < .05), CD24 (2.7-fold; p < .05) and CD200 (2.9-fold; p < .001) compared with the ALDH lo population (Table 1). Representative dot plots of the expression of surface antigens together with the levels of Aldefluor present in ALDH hi versus ALDH lo cells, as well as isotype controls, are presented in supporting information Fig. 3. For FACS analysis of the intracellular antigens, the ALDH hi and ALDH lo populations were first separated by FACS sorting, with subsequent permeabilization of the cells prior to antigen visualization (supporting information Fig. 4). This was done as permeabilization releases the Aldefluor substrate from the cells. Thus, it is not possible to simultaneously visualize ALDH activity in conjunction with intracellular antigens by FACS.
Table 1. Expression of the indicated antigens in cells with high and low ALDH activity
These data indicate that ALDH hi cells are epithelial in nature, predominantly basal, and express a number of surface and intracellular antigens that are present on stem/progenitor cells of various origins. In addition, approximately a third of ALDH hi cells expressed high levels of Sca-1, a marker of prostate stem cells  and about a third of Sca-1 hi cells expressed high levels of ALDH. Compared with ALDH lo cells, the ALDH hi population was enriched in basal cells, in cells expressing high levels of Sca-1, ALDH 1/2, as well as a number of antigens expressed by stem/progenitor cells, such as Bcl-2, ABCG2, CD200, Oct 3/4, and CD24. Importantly, the results show for the first time that ALDH activity is a functional marker of prostate epithelial cells that also express antigens associated with a primitive phenotype.
ALDH Hi Cells Have Greater Proliferative Potential In Vitro than ALDH Lo Cells
Because an important feature of stem cells is their high proliferative potential, we compared the growth potential of ALDH hi and ALDH lo cells. Prostate cells were isolated by FACS sorting based on the expression of high or low levels of ALDH activity (Fig. 1) and their growth on collagen films was compared. ALDH hi cells showed greater growth potential (3.3-fold; p = .012; n = 8) than ALDH lo cells (Fig. 3). However, as the difference between the in vitro growth of the two ALDH populations could be due to a difference between the rate of proliferation rather than to a difference in growth potential, we compared the ability of these two populations to regenerate prostate tissue in vivo.
ALDH Hi Cells Have Greater Prostate Regeneration Capacity than ALDH Lo Cells
As the ability to regenerate tissue in vivo is a characteristic of stem cells, we determined the in vivo growth potential of ALDH hi cells compared with that of cells with low levels of this enzyme. ALDH hi and ALDH lo cells were FACS sorted and these populations were combined with UGM cells (inductive mesenchyme for prostatic tissue [20–22]) and inserted under the renal capsule of male animals. Each experiment also contained grafts of UGM alone to control for the possibility that the UGM was not completely depleted of urogenital sinus epithelial cells. The amount of prostatic tissue generated was measured after 10 weeks.
ALDH hi cells formed significantly more prostatic tissue (29.8 ± 11.0 mg: 2.5-fold) than ALDH lo cells (12.0 ± 1.8 mg: p = .006) and 2.3-fold more than UGM cells alone (12.9 ± 3.1 mg; p = .018; n = 2: six kidneys for each population; Fig. 4A, 4B). The prostatic tissue obtained from ALDH hi cells had normal prostatic morphology comprising basal and luminal cells lining ducts containing secretory products, whereas the sparse tissue arising from ALDH lo cells had minimal ductal structures lacking secretory material (Fig. 4C). UGM cells formed tissue that contained only stromal elements. Interestingly, whereas the prostatic tissue derived from ALDH hi cells contained both basal (CK5+) and luminal (CK8+) cells (Fig. 5A), the ducts formed by ALDH lo cells contained only luminal cells (Fig. 5B). Moreover, synaptophysin expressing cells were present only in grafts arising from ALDH hi cells (Fig. 5C) and not those formed from ALDH lo cells (Fig. 5D), indicating that only ALDH hi cells gave rise to cells of neuroendocrine origin. In addition, although the ductal tissue formed from ALDH hi cells contained abundant secretory products (Fig. 5E), little or no secretory material was present in ducts from ALDH lo grafts (Fig. 5F). However, tissue generated from ALDH hi and ALDH lo cells both show expression of Nkx3.1, indicating their prostatic origin (Fig. 5G, 5H). The data reveal that the ALDH hi population has the ability to regenerate functional prostatic tissue containing basal, luminal, and neuroendocrine cells, whereas ALDH lo cells have little growth and limited differentiation potential, indicating that cells with the greatest prostate regenerating ability reside within the compartment that contains high levels of ALDH.
We show, for the first time, that a subset of normal murine prostatic cells displays high levels of ALDH activity. Hematopoietic, neural, mesenchymal, endothelial, and mammary stem/progenitor cells are enriched in ALDH hi cells [4, 5, 10, 12] and it has been proposed that ALDH activity may be a functional marker of “stemness” [2, 3]. The ALDH hi cells from the prostate express EpCam, indicating that they are epithelial cells . They also express the cytokeratins, CK5 and CK8, present in basal (CK5) and luminal (CK8) epithelial cells, in a ratio of 2.3 to 1, indicating that the majority of ALDH hi cells are of basal origin.
Significant numbers of ALDH hi cells coexpress antigens that are present on stem/progenitor cells of other origins, such as CD9 [28–30], Bcl-2 [19, 32–35], and CD200 [36–38]. A third to a half of ALDH hi cells express CD24, prominin, Oct 3/4, and ABCG2—antigens found on mammary, hematopoietic, neural, and prostate stem cells, as well as by primitive cells exhibiting a side population [39–47]. A small subset of ALDH hi cells express nestin, an antigen also expressed by primitive cells [48, 49]. Compared with the ALDH lo population, ALDH hi cells are significantly enriched in antigens that are expressed by prostate basal epithelium, such as CK5 and Bcl-2, and also in antigens involved with protective and detoxifying functions, such as Bcl-2, ABCG2, and ALDH1/2. This may serve to protect stem/progenitor cells from toxic substances and ensure their long-term survival. Compared to ALDH lo cells, the ALDH hi cells are also enriched in antigens present in stem cells of other origins, such as CD200, Oct 3/4, and CD24. In addition, the ALDH hi population contains more cells that express Sca-1, as well as cells with high levels of Sca-1, previously shown to be a feature of prostate stem cells , than the population with low ALDH activity. Furthermore, ALDH hi cells are enriched in the proximal region of prostatic ducts, a region in which stem cells are concentrated [18, 19, 50].
Interestingly, we did not find an increase in the expression of prominin/CD133 in the ALDH hi population compared with the ALDH lo population. We also noted no increase in prominin expression in the stem cell containing Sca-1 hi population compared with the Sca-1+ population. Prominin expression was not concentrated in the stem cell-enriched proximal region of ducts compared with the remaining regions of ducts (unpublished observations). Although prominin has been described as a marker for hematopoietic, neural, and prostate stem cells [44–47], as well as colon cancer initiating cells [51, 52], it has been suggested that its expression is not restricted to stem cells [53, 54]. Normal and malignant human colonic stem cells have recently been shown to express both high ALDH protein expression and high ALDH activity  and both these properties are more specific for colon stem cells than CD133 expression. Another stem cell marker that has recently been identified on adult murine prostate stem cells is c-kit (CD117). A whole prostate can be reconstituted from a single cell that, among other surface antigens, also expresses c-kit . It will be interesting to determine the incidence of c-kit expression by ALDH hi cells compared to cells with low ALDH activity.
We show that cells with high ALDH activity have greater in vitro proliferative potential than cells expressing low levels of this enzyme. Importantly, ALDH hi cells have a greater capacity than ALDH lo cells to regenerate prostate tissue in an in vivo transplantation assay. Examination of the tissue regenerated by ALDH hi cells showed the presence of basal, luminal, and neuroendocrine cells, as well as prostatic secretory material and the prostate-specific protein, Nxk3.1. These data confirm that ALDH hi cells are able to differentiate into multiple cell types. In contrast, the small amount of tissue derived from ALDH lo cells shows positivity for Nxk3.1 and for the presence of luminal cells, indicating a lack of pluripotency. These observations confirm that the cell population expressing high levels of ALDH enzyme activity contains stem/progenitor cells. ALDH activity may therefore be used as a new functional marker for the isolation of primitive prostate cells.
Although the interactions between Aldefluor substrate and various ALDH isoforms have not been systematically characterized, the substrate is recognized by ALDH1A1 and ALDH3 isoforms but not by mitochondrial forms of ALDH (Aldefluor kit information). It is unknown at present which isozyme (ALDH1A1, ALDH1A2, ALDH1A3, and/or ALDH3A1) is responsible for the ALDH activity exhibited by stem cells and whether different isozymes function within various types of stem/progenitor cells (see  for review). ALDH family members play roles in diverse metabolic processes and a number of ALDH isoforms are involved in retinoic acid biosythesis . Retinoids play an important role in the normal prostate. Lack of vitamin A during rat embryogenesis inhibits mature prostate development . Retinoic acid initiates prostatic bud formation ; retinoids are necessary during early postnatal development for normal prostate branching  and for mature prostate homeostasis . Conversely, it is thought that aberrant retinoic acid metabolism is involved in the development of prostate cancer. Levels of retinoic acid are 5–8 fold lower in prostate carcinoma compared with normal prostate or benign prostatic hyperplasia , retinoic acid receptor (RAR) gamma knockout mice have abnormal prostates , cellular retinoic acid binding protein 2 mRNA and protein are downregulated in prostate carcinoma cells compared with normal tissue , and inactivation of RXR ∝ leads to preneoplastic lesions in the prostate . In addition, ALDH1A2 is expressed in normal prostate epithelium but not in prostate cancer and is a candidate tumor suppressor gene in prostate cancer . Furthermore, differences in the cellular distribution of RAR isoforms exist between normal, benign prostatic hyperplasia and prostate cancer epithelial cells . These studies confirm the importance of the role of retinoid metabolism in the prostate. Various strategies are currently being pursued to identify useful agents for the treatment of prostate cancer, including ALDH inhibitors  and retinoic acid metabolism blocking agents .
The presence of high levels of ALDH activity in a variety of stem cells acts as a functional rather than a phenotypic marker for stem cells. Hess et al.  proposed that, because the cell phenotype can change during the cell cycle, the isolation of hematopoietic stem cells based on a conserved stem cell function such as ALDH activity may be preferable to that based on a changing surface phenotype. Some ALDH isoforms act as detoxifying enzymes, which may serve to protect stem cells from toxic compounds. Other ALDH isozymes, involved in retinoic acid biosynthesis, may have a role in stem cell self-renewal or differentiation. Inhibition of ALDH promotes the self-renewal of human hematopoietic stem cells through inhibition of retinoic acid signaling  and studies with RAR ∝−/− and RAR δ−/− mice show that RAR δ is crucial for the regulation of HSC self-renewal and differentiation . In the hair bulge (stem cell niche), ALDH1A2 is expressed throughout anagen; it is proposed that retinoic acid is involved in stem cell maintenance in the hair follicle . Our data indicate that the murine prostate contains a subset of cells that express high levels of ALDH activity, that these cells have prostate-regenerating activity, and that these cells express many molecules previously identified in stem cells of other origins. ALDH activity can therefore be used as a new functional marker for the isolation of prostate stem/progenitor cells. Further research is necessary to elucidate the role of this functional marker in these cells.
We show that a subset of murine epithelial prostate cells has high ALDH activity, a functional marker of stem cells of other origins. The majority of ALDH hi cells express antigens associated with stem/progenitor cells, including several that serve a protective role, such as Bcl-2 and ABCG2. The ALDH hi population has greater in vitro and in vivo proliferative capacity than cells with low ALDH activity, indicating that this population contains prostate stem/progenitor cells. The expression of high ALDH activity by prostate cells is a novel finding that will assist with the isolation and further characterization of prostate stem cells and the development of therapeutic strategies for the treatment of benign prostatic hyperplasia and prostate cancer.
We thank Ronnie Dreyer and Peter Lopez for help with FACS sorting. This work was supported by the University of Cape Town Staff Research Fund, the South African Medical Research Council, National Institutes of Health CA132641, New York State Department of Health, Amgen Inc., and the Helen L. and Martin S. Kimmel Center for Stem Cell Biology at the New York University School of Medicine.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.