An interaction between carotene-15,15′-monooxygenase expression and consumption of a tomato or lycopene-containing diet impacts serum and testicular testosterone


  • Nikki A. Ford,

    1. Division of Nutritional Sciences, University of Illinois Urbana-Champaign, Urbana, IL
    Search for more papers by this author
  • Nancy Engelmann Moran,

    1. Department of Internal Medicine, Comprehensive Cancer Center, The Ohio State University, Columbus, OH
    2. Division of Medical Oncology, Department of Medicine, The Ohio State University, Columbus, OH
    Search for more papers by this author
  • Joshua W. Smith,

    1. Division of Nutritional Sciences, University of Illinois Urbana-Champaign, Urbana, IL
    Search for more papers by this author
  • Steven K. Clinton,

    1. Department of Internal Medicine, Comprehensive Cancer Center, The Ohio State University, Columbus, OH
    2. Division of Medical Oncology, Department of Medicine, The Ohio State University, Columbus, OH
    Search for more papers by this author
  • John W. Erdman Jr.

    Corresponding author
    1. Division of Nutritional Sciences, University of Illinois Urbana-Champaign, Urbana, IL
    2. Department of Food Science and Human Nutrition, University of Illinois Urbana-Champaign, Urbana, IL
    • Department of Food Science and Human Nutrition, University of Illinois Urbana-Champaign, Urbana, IL 61801, USA
    Search for more papers by this author
    • Fax: +217-333-9368


Lycopene, the red pigment of tomatoes, is hypothesized to reduce prostate cancer risk, a disease strongly dependent upon testosterone. In this study, mice lacking the expression of carotene-15,15′-monooxygenase (CMO-I−/−) or wild-type mice were fed either a 10% tomato powder (TP), lycopene-containing (248 nmol/g diet) or their respective control diets for 4 days, after which serum testosterone was measured. A significant diet × genotype interaction (p = 0.02) suggests that the TP reduces serum testosterone concentrations in CMO-I−/− mice but not in wild-type mice. Similarly, testicular testosterone was lowered in TP-fed CMO-I−/− mice (p = 0.01), suggesting that testosterone synthesis may be inhibited in this group. A similar pattern was also observed for lycopene fed mice. Interestingly, the CMO-I−/− mice showed a greater expression of the gene encoding the CMO-II enzyme responsible for eccentric oxidative carotenoid cleavage in the testes. Therefore, we hypothesize that serum testosterone is reduced by lycopene metabolic products of oxidative cleavage by CMO-II in the testes. Overall, these findings suggest that genetic polymorphisms impacting CMO-I expression and its interaction with CMO-II, coupled with variations in dietary lycopene, may modulate testosterone synthesis and serum concentrations. Furthermore, carefully controlled studies with tomato products and lycopene in genetically defined murine models may elucidate important diet × genetic interactions that may impact prostate cancer risk.

Epidemiological1 and laboratory evidence2 support a hypothesis that consumption of tomato products, including lycopene and its precursors,3 may impact risk of prostate cancer. Potential interactions between tomato phytochemicals and androgenic hormonal status may underlie this relationship, as testosterone is the critical hormone that impacts prostate development, biology, and promotes prostate carcinogenesis.2 We previously observed that changes in testosterone impacted lycopene metabolism, with castration increasing hepatic lycopene and higher testosterone leading to reduced lycopene accumulation.4 More recently, studies suggest that greater intake of tomato phytochemicals and higher serum lycopene may reduce serum testosterone. For example, in a phase II randomized-controlled trial, tomato extract supplementation reduced serum-free testosterone in men with clinically localized prostate cancer.5 Additionally, we previously found that serum testosterone concentrations were decreased in rats fed tomato powder (TP) or the tomato carotenoids lycopene or phytofluene for 4 days.6 Most critically, lycopene-fed rats exhibit alterations in steroid hormone metabolism in the prostate, including reduced 5-α-reductase gene expression, a critical step in the activation of testosterone to dihydrotestosterone, the most potent ligand for the androgen receptor.7 The pharmacologic agents, finasteride and dutasteride, that target 5-α-reductase, are the only drugs to have demonstrated efficacy as chemopreventive agents in prospective human prostate cancer trials.8, 9 Taken together, these observations suggest that tomato phytochemicals, including lycopene, impact testosterone production or signaling and may in part explain the inverse association between tomato consumption and prostate cancer risk.

Although most of the attention has focused upon lycopene as the predominant bioactive phytochemical in tomatoes, the carotenoids phytoene and phytofluene should be considered,3 along with polyphenols10 and other components.11 A new consideration regarding the lycopene-prostate cancer relationship has emerged as a result of recent rapid advances in the field of carotenoid metabolism.10, 11 Indeed, we now postulate that cleavage products of lycopene, phytoene and/or phytofluene may mediate some of the observed anticancer activities.3 Provitamin A carotenoids like β-carotene are clearly metabolized by β,β-carotene-15,15′-monooxygenase I (CMO-I) through central chain cleavage to form vitamin A, and further metabolism results in the formation of retinoic acid and other retinoids.12 The carotene-9′,10′-monooxygenase (CMO-II) is responsible for eccentric cleavage of acyclic carotenoids, including lycopene, to form aldehyde metabolites.13 Because these genes show polymorphisms in humans and are inducible by dietary variables,14 we can now postulate a genetic component into the relationship between tomato carotenoid, androgen metabolism and prostate cancer risk.

As predicted, CMO-II−/− mice accumulate lycopene in the liver and serum,13 clearly supporting the hypothesis that this gene is critical in lycopene cleavage. Somewhat surprisingly, knockout of CMO-I, an enzyme that does not appear to directly metabolize lycopene, alters tissue lycopene accumulation.15 We have reported that CMO I−/− mice fed lycopene accumulate significantly less hepatic lycopene but more lycopene in the prostate, seminal vesicles and testes compared to wild-type mice.15 Perhaps, compensatory changes in other enzyme systems in CMO-I−/− mice may impact lycopene metabolism.

The above evidence supports a complex interrelationship between dietary tomato products and lycopene with androgen metabolism and action, a process potentially mediated by genetic variation in genes impacting carotenoid metabolism. Thus, we examined the impact of CMO-I genotype (wild type or CMO-I−/−) in combination with dietary intake of TP or lycopene on specific aspects of testosterone production in mice.


AIN, AIN-93G diet; CMO-I, carotene-15,15′-monooxygenase; CMO-II, carotene-9′10′-monooxygenase; LYC, lycopene beadlet diet; PBC, placebo beadlet control diet; PBS, phosphate buffered saline; SNP, single nucleotide polymorphism; TP, tomato powder diet

Material and Methods

Mouse husbandry and diets

The generation of CMO-I−/− mice was previously described.16 C57BL/6 × 129/SvJ (F1) mice (wild-type) were purchased from The Jackson Laboratory (Bar Harbor, ME). Genomic DNA from mouse tail biopsies and the Sigma Extract-N-Amp Tissue PCR Kit (St., MO) were used to confirm genotype. Mice were housed in shoebox cages with free access to water and weighed and handled daily. Mice consumed an AIN-93G diet for 3 weeks before randomization onto experimental diets. Fresh diet was provided every 48 hr, and new diet was made monthly and stored in the dark at 4°C.

Nine- to 12-week-old CMO-I−/− or wild-type male mice were randomly assigned to one of two experimental carotenoid-containing diets: 10% TP (FutureCeuticals, Momence, IL; TP) or lycopene (DSM, Basel Switzerland, 10% water-soluble lycopene beadlet; LYC). For statistical analysis, the TP group was compared to an AIN-93G dietary group while the LYC group was compared to a placebo beadlet (DSM, 10% water-soluble lycopene beadlet; PB) dietary group; n = 25 per genotype/diet. Mice consumed the experimental diets for 4 days, ad libitum. At the conclusion of the study, mice were fasted for 3 hr, asphyxiated by CO2, and blood was collected by cardiac puncture. Cohorts of mice were killed daily in the afternoon within a 3-hr time frame to avoid diurnal alterations in serum testosterone. Testes were collected, snap frozen in liquid nitrogen and stored at −80°C. All animal procedures were approved by the University of Illinois Institutional Animal Care and Use Committee.

The TP diet contained 204 nmol lycopene/g diet while the LYC diet provided 248 nmol lycopene/g diet. The TP diet also contained additional carotenoids: 10.1 nmol phytoene, 2.6 nmol phytofluene and 0.8 nmol β-carotene/g diet. All diets were balanced for total fiber, nitrogen and calories.13 Additionally, vitamin A levels were reduced in all diets to 1500 IU retinyl palmitate per kilogram diet to ensure adequate absorption of carotenoids without resulting in a vitamin A deficiency.15

Serum and testicular testosterone

Testicular testosterone was extracted as previously described, with minor modifications.17 Briefly, half of a testis (20–60 mg) was homogenized in PBS (125 μL) at room temperature for 20 sec. Diethyl ether (2.5 ml) was added, and the sample was vortexed for 2 min and then centrifuged at 183g for 1 min. The upper phase was removed and saved. Another 1.4 ml diethyl ether was added, and the sample was vortexed and placed on dry ice for 10 min. The upper phase was again removed and combined with the previous extract. Samples were allowed to evaporate at room temperature and stored at −80°C until use. Samples were diluted 1:10 in PBS before quantification. Serum and testicular testosterone were quantified with a radioimmunoassay kit (DSL-4000 ACTIVE Testosterone Coated-Tube Radioimmunoassay Kits; Diagnostic Systems Laboratories) as per the manufacturer's instructions.

Real-time PCR analysis

Relative testicular gene expression was evaluated by qRT-PCR. Briefly, testicular RNA was isolated using 2.0 ml Trizol (Invitrogen, Carlsbad, CA) per the manufacturer's instructions, including treatment with DNase I (New England Biolab, Ipswich, MA). The concentrations and quality of mRNA were determined by spectrophotometry and agarose gel electrophoresis. Complimentary DNA was synthesized using Superscript II Reverse Transcriptase (Invitrogen) and random hexamers (Applied Biosystems, Foster City, CA). Primer pairs were selected to measure CMO-II (NM_1332217): Forward-5′-GTTATCTACTTCGAGTTGGACCTGG and Reverse-5′-AAGCAACGCCATTCCATCA and 18S (Internal Control, Forward-5′-GATCCATTGGAGGGCAAGTCT and Reverse-5′ ACTGCAGCAACTTTAATATACGCTATT). Real-time PCR was performed using a 7900HT Fast Real-Time PCR detection system (Applied Biosystems) with the SYBR green fluorescence dye (Invitrogen). Statistical analysis was conducted on the ΔCt value using two-way ANOVA (detailed below), and data are presented as fold change (2−ΔΔCt) ± the standard error of the mean (SEM) of ΔCt for testis tissue of wild-type mice that consumed the control diet (AIN or PBC).

Statistical analysis

A factorial design with two genotypes (CMO-I−/−, wild-type) and two diets (AIN vs. TP or LYC vs. PBC) was used. This design allowed us to investigate the impact of the main effects (diet and genotype) and the interaction between main effects on study outcomes. All parameters were analyzed by two-way analysis of variance (ANOVA) using SAS 9.2 (Cary, NC) with α = 0.05. Although it is possible to test for differences in the levels of main effects by post hoc analysis, it is not generally advisable if interactions are present. When assumptions for ANOVA were violated, data were natural log transformed. For the analysis of serum and testicular testosterone, all data points outside of two standard deviations were considered extreme outliers and removed from the dataset. Results were expressed as the mean ± SEM.


Serum and testicular testosterone

Serum testosterone was measured in wild-type and CMO-I−/− mice that consumed a TP or AIN diet for 4 days (Fig. 1). Neither diet nor genotype alone significantly impacted serum testosterone, but a significant interaction (p = 0.02) suggests that the TP reduced serum testosterone levels in CMO-I−/− mice (Fig. 1a). Serum testosterone levels are largely derived by synthesis of testosterone in the testis. To determine if changes in testicular production of testosterone contribute to the observed serum testosterone changes, testicular testosterone was measured. Again, genotype and diet significantly interacted to alter testicular testosterone (p = 0.01; Fig. 1c). Although not significant, testicular testosterone appears to be slightly elevated in CMO-I−/− mice, but the LYC and TP diets appeared to modulate testosterone levels. To determine if lycopene specifically contributed to the observed dietary TP by CMO-I genotype interaction effect on serum and testicular testosterone, we examined lycopene-fed wild-type and CMO-I−/− mice. The lycopene diet significantly reduced serum testosterone (p = 0.03; Fig. 1b), and the expression of CMO-I and lycopene significantly interacted to alter testicular testosterone (p = 0.03; Fig. 1d).

Figure 1.

CMO-I−/− mice and wild-type mice were fed AIN-93G diet or diets supplemented with tomato powder, water-soluble lycopene beadlets or placebo beadlets for 4 days. Serum (a, b) and testicular (c, d) testosterone were quantified with a radioimmunoassay kit (DSL-4000 ACTIVE Testosterone Coated-Tube Radioimmunoassay Kits; Diagnostic Systems Laboratories). Bars represent mean ± SEM; serum n = 10–15, testis n = 8–10. p < 0.05 was considered statistically significant. (a) Genotype, p = 0.08; diet, p = 0.68; Interaction, p = 0.02. (b) Gentoype, p = 0.37; diet, p = 0.03; interaction, p = 0.35. (c) Genotype, p = 0.85; diet, p = 0.04; interaction, p = 0.01. (d) Genotype, p = 0.77; diet, p = 0.45; interaction, p = 0.02.

Testis RNA expression

We hypothesized that the second carotenoid oxygenase enzyme, CMO-II, may have played a role in altering testosterone levels through the production of bioactive lycopene metabolites.13, 14, 18 The expression of CMO-II was significantly upregulated in testicular tissues of CMO-I−/− mice compared to wild-type mice (Fig. 2). The lycopene or TP diets did not directly impact the expression of CMO-II in mouse testes. The mRNA expression of testicular CMO-I was not altered in wild-type mice by the TP or LYC diets compared to their respective controls (data not shown).

Figure 2.

The relative mRNA expression of testicular CMO-II compared to 18S expression of wild-type and CMO-I−/− mice fed (a) an AIN-93G, tomato powder and (b) a placebo beadlet or lycopene diet. Statistical analysis was conducted on the ΔCt value using two-way ANOVA, and data are presented as fold change (2−ΔΔCt) ± SEM of ΔCt for testis tissue of wild-type mice that consumed the control diet (AIN-93G or placebo beadlet). Bars represent mean ± SEM; n = 9–11. p < 0.05 was considered statistically significant. (a) Genotype, p = 0.02; diet, p = 0.09; interaction, p = 0.68. (b) Genotype, p < 0.01; diet, p = 0.26; interaction, p = 0.73.

Tissue carotenoid concentrations

Testicular lycopene accumulation was not altered by genotype in this short-term feeding study (data not shown). Phytoene and phytofluene did not accumulate in detectable levels in testicular tissue of either wild-type or CMO-I−/− mice (data not shown). As predicted, β-carotene significantly accumulated in testicular tissue of CMO-I−/− mice compared to wild-type mice (p < 0.0001; data not shown), confirming earlier reports.15, 16 The TP diet contained very little β-carotene and testis tissue accumulated ∼10,000 times more lycopene than β-carotene.


Human and experimental studies suggest that consuming tomato products reduces serum testosterone action, which is positively associated with prostate cancer risk.1, 5–7, 19 Additionally, evidence suggests that testosterone status impacts lycopene metabolism and that the CMO-I and CMO-II genes are involved in lycopene homeostasis. In the present study, we found that serum testosterone levels were influenced by the expression of CMO-I and the intake of tomato carotenoids. The results of this study suggest that an interaction between these variables determines the resulting testosterone level. The significant genotype × diet interactions suggest that dietary TP reduces serum testosterone in CMO-I−/− mice. In parallel, dietary TP and LYC reduce testicular testosterone concentrations in CMO-I−/− mice but have no effect in wild-type mice. Data from the LYC-fed mice suggest that lycopene is largely responsible for the dietary TP's effects on testosterone levels. Additionally, because the testis is the primary site of testosterone production, alterations in testicular production of testosterone are likely responsible for the changes in serum testosterone.

At present, the CMO-I gene appears to encode the enzyme primarily responsible for central cleavage of β-carotene, providing two molecules of vitamin A, which can then interface with its receptor to signal critical downstream events essential for development and health. Lycopene does not appear to be a substrate for CMO-I, yet is clearly cleaved by CMO-II.13, 14 Interestingly, the CMO-I−/− mice show changes in lycopene metabolism,15 suggesting an indirect effect. For reasons currently unknown, the CMO-II gene mRNA expression is upregulated in testes of CMO-I−/− mice (Fig. 2). The fact that CMO-II gene expression is increased may in turn result in the production of various lycopene metabolites. It is probable that CMO-II action is a key step in the degradation and excretion of lycopene. However, we postulate that lycopene metabolites may be agonists or antagonists to retinoic acid receptors, peroxisome proliferator-activated receptors or other members of the steroid receptor family that are key regulators of cell biology.20

Our findings, comparing wild-type mice to CMO-I−/− mice, also suggest that polymorphisms in the CMO-I or CMO-II genes may impact the relationship between dietary tomato products, lycopene and prostate cancer risk. Indeed, there is a person-to-person heterogeneity in serum carotenoid concentrations, even when fed similar dietary concentrations in controlled studies.21 Thus, polymorphisms in CMO-I and CMO-II may account for the inconsistent results and conclusions drawn from various epidemiologic studies. Recently, two single-nucleotide polymorphisms (SNPs) existing in the human CMO-I gene were reported. These SNPs are found in high frequency within the European, Chinese and Japanese populations.22 Several SNPs have also been identified in the coding region of the CMO-I gene.23 Certainly, genetic variations impacting carotenoid bioavailability and metabolism may occur at many steps in the process. When genetic variation is coupled to the profound impact of diet composition and inherent measurement error in many diet assessment tools, the poor correlation between estimated intake and serum or tissue concentrations of tomato carotenoids is not surprising.

Thus far, there are no published human studies examining the impact of CMO-II gene SNPs on tissue or serum lycopene concentrations. Alternatively, mutations in CMO-II result in altered carotenoid tissue deposition in domestic livestock.24, 25 We propose that alterations in the function of the CMO-I and CMO-II enzymes, due to SNPs or epigenetic regulation, may explain the effects of tomato carotenoids on endocrine processes related to prostate cancer risk. Future studies are required to unravel these intriguing interrelationships.


We thank Future Ceuticals for donating the tomato powder and DSM for providing the lycopene and placebo beadlets. We thank Dr. Adrian Wyss at DSM Nutritional Products and Dr. Johannes von Lintig at Case Western Reserve University for providing the original breeding pairs of CMO-I−/− and CMO-II−/− mice for our studies.