Synthetic amyloid beta does not induce a robust transcriptional response in innate immune cell culture systems.

BACKGROUND: Alzheimer's disease (AD) is a progressive neurodegenerative disease that impacts nearly 400 million people worldwide. The accumulation of amyloid beta (Aß) in the brain has historically been associated with AD, and recent evidence suggests that neuroinflammation plays a central role in its origin and progression. These observations have given rise to the theory that Aß is the primary trigger of AD, and induces proinflammatory activation of immune brain cells (i.e., microglia), which culminates in neuronal damage and cognitive decline. To test this hypothesis, many in vitro systems have been established to study Aß-mediated activation of innate immune cells. Nevertheless, the transcriptional resemblance of these models to the microglia in the AD brain has never been comprehensively studied on a genome-wide scale. METHODS: We used bulk RNA-seq to assess the transcriptional differences between in vitro cell types used to model neuroinflammation in AD, including several established, primary and iPSC-derived immune cell lines (macrophages, microglia and astrocytes) and their similarities to primary cells in the AD brain. We then analyzed the transcriptional response of these innate immune cells to synthetic Aß or LPS and INFγ. RESULTS: We found that human induced pluripotent stem cell (hIPSC)-derived microglia (IMGL) are the in vitro cell model that best resembles primary microglia. Surprisingly, synthetic Aß does not trigger a robust transcriptional response in any of the cellular models analyzed, despite testing a wide variety of Aß formulations, concentrations, and treatment conditions. Finally, we found that bacterial LPS and INFγ activate microglia and induce transcriptional changes that resemble many, but not all, aspects of the transcriptomic profiles of disease associated microglia (DAM) present in the AD brain. CONCLUSIONS: These results suggest that synthetic Aß treatment of innate immune cell cultures does not recapitulate transcriptional profiles observed in microglia from AD brains. In contrast, treating IMGL with LPS and INFγ induces transcriptional changes similar to those observed in microglia detected in AD brains.

Astrocytes and microglia differentially regulate trafficking of lymphocyte subsets across brain endothelial cells.

Feline brain endothelial cells (BECs), astrocytes, and microglia were combined in different configurations in a cell culture insert system to assess the effect of different cell types on the trafficking of peripheral blood mononuclear cell (PBMC) subsets in response to feline immunodeficiency virus (FIV). The addition of astrocytes to BECs significantly increased the adherence of PBMCs. This increase in adherence was suppressed by microglia, whereas microglia alone had no effect on PBMC adherence. FIV exposure of the glial cells did not alter PBMC adherence as compared to same configurations with untreated cells. All PBMC subsets showed some level of trafficking across the endothelial cell layer. The level of trafficking of monocytes and B cells was significantly increased if astrocytes were present. The presence of microglia with the astrocytes reduced transmigration across all PBMC subsets. FIV exposure of astrocytes significantly increased the percentage of CD8 T cell transmigration from 24% to 64% of the total CD4 and CD8 numbers. The presence of microglia significantly reversed the preferential trafficking of CD8 cells in the presence of astrocytes. The results suggested that interaction between the triad of endothelial cells, astrocytes, and microglia played an important, but varying, role in the trafficking of different PBMC subsets. In general, astrocytes had a positive effect on trafficking of PBMCs, while microglia had a suppressive effect. Effects of FIV on trafficking were largely restricted to increases seen in CD8 T cells and monocytes.

Development of neuronal sensitivity to toxins in cerebrospinal fluid from HIV-type 1-infected individuals.

HIV infection of the immature nervous system generally results in a rapid progression of neurological disease that cannot easily be explained by the severity of encephalitis, viral burden, systemic immune deficiency, or developmental changes in utero. Rather than the viral infection dictating disease progression, we explored the possibility that immature neurons might be particularly sensitive to toxins secreted in response to HIV. Primary cultures of rat cortical neurons were exposed to toxic cerebrospinal fluid (CSF) from HIV-infected individuals (CSF(tox)) and evaluated for changes in intracellular calcium and cell death. CSF(tox) had no detectable effect on early neurite outgrowth, calcium regulation, or cell death during the first few days in culture. Starting at Day 4, delayed increases in intracellular calcium appeared in response to CSF(tox). The magnitude of the delayed calcium rise and cell death increased with the age of the culture and correlated with the appearance of synaptophysin immunoreactive varicosities. A similar gradual development of sensitivity was seen during exposure of feline neurons to toxins generated by choroid plexus macrophages after exposure to feline immunodeficiency virus. The possibility that toxin sensitivity is dependent on the presence of synaptic activity is consistent with the rapid pathogenesis in the CNS seen during the first postnatal year. Emerging synaptic activity coupled with other factors such as high metabolic demand in the young nervous system may combine to increase the likelihood of calcium overload and neuronal dysfunction in response to HIV-associated toxins.

Infection of the choroid plexus by feline immunodeficiency virus.

The human, simian, and feline immunodeficiency viruses rapidly penetrate into the brain and trigger an inflammatory process that can lead to significant neurologic disease. However, the mechanisms that permit efficient trafficking of macrophage-tropic and the more neurotoxic lymphocytotropic isolates are still poorly understood. One potential source of virus entry may be the blood-CSF barrier provided by the choroid plexus. Infected cells are often detected within the choroid plexus but it is unclear whether this reflects trafficking cells or infection of the large macrophage population within the choroidal stroma. To address this issue, we cultured fetal feline choroid plexus and evaluated the ability of feline immunodeficiency virus (FIV) to establish a primary infection. Significant provirus was detected in macrophage-enriched choroid plexus cultures as well as in the choroid plexus of cats infected in vivo. FIV p24 antigen production in vitro was very low but detectable. Addition of a feline T-cell line to macrophages inoculated with FIV resulted in a dense clustering of the T cells over macrophages with dendritic cell-like morphologies and a robust productive infection. The direct infection of choroid plexus macrophages with FIV, the efficient transfer of the infection to T cells indicate that the choroid plexus can be a highly efficient site of viral infection and perhaps trafficking of both macrophage-tropic and T-cell-tropic viruses into the CNS.

Choroid plexus macrophages proliferate and release toxic factors in response to feline immunodeficiency virus.

Recent observations have suggested that lentiviruses stimulate the proliferation and activation of microglia. A similar effect within the dense macrophage population of the choroid plexus could have significant implications for trafficking of virus and inflammatory cells into the brain. To explore this possibility, we cultured fetal feline macrophages and examined their response to feline immunodeficiency virus (FIV) or the T-cell-derived protein, recombinant human CD40-ligand trimer (rhuCD40-L). The rhCD40-L was the most potent stimulus for macrophage proliferation, often inducing a dramatic increase in macrophage density. Exposure to FIV resulted in a small increase in the number of macrophages and macrophage nuclei labeled with bromodeoxyuridine. The increase in macrophage density after FIV infection also correlated with an increase in neurotoxic activity of the macrophage-conditioned medium. Starting at 16-18 weeks postinfection, well after the peak of viremia, a similar toxic activity was detected in cerebrospinal fluid (CSF) from FIV-infected cats. Toxicity in the CSF increased over time and was paralleled by strong CD18 staining of macrophages/microglia in the choroid plexus and adjacent parenchyma. These results suggest that lentiviral infection of the choroid plexus can induce a toxic inflammatory response that is fueled by local macrophage proliferation. Together with the observation of increasing toxic activity in the CSF and increased CD18 staining in vivo, these observations suggest that choroid plexus macrophages may contribute to an inflammatory cascade in the brain that progresses independently of systemic and CSF viral load.

Destabilization of neuronal calcium homeostasis by factors secreted from choroid plexus macrophage cultures in response to feline immunodeficiency virus.

The choroid plexus contains a major reservoir of macrophages poised for efficient delivery of virus and neurotoxins to the brain after infection by lentiviruses such as human or feline immunodeficiency virus (FIV). However, their contribution to neurotoxicity is poorly understood. Medium from FIV-infected, choroid plexus macrophages applied to cultured feline cortical neurons induced a small acute calcium rise followed by either a delayed calcium deregulation (41%) or swelling and bursting (23%). NMDA glutamate receptor blockade prevented the acute calcium increase and antagonists to the IP(3) receptor, voltage-gated calcium channels and sodium channels suppressed both the acute and late increases. Analysis of intracellular calcium recovery in toxin-treated neurons after a brief exposure to glutamate, revealed a decrease in the rate and extent of recovery. The apparent diverse pharmacological contributions to intracellular calcium destabilization may be due to the ability of macrophage toxins to interfere with recovery of intracellular calcium homeostasis.

A role for the cytoskeleton-associated protein palladin in neurite outgrowth.

The outgrowth of neurites is a critical step in neuronal maturation, and it is well established that the actin cytoskeleton is involved in this process. Investigators from our laboratory recently described a novel protein named palladin, which has been shown to play an essential role in organizing the actin cytoskeleton in cultured fibroblasts. We investigated the expression of palladin in the developing rat brain by Western blot and found that the E18 brain contained a unique variant of palladin that is significantly smaller (approximately 85 kDa) than the common form found in other developing tissues (90-92 kDa). Because the expression of a tissue-specific isoform suggests the possibility of a cell type-specific function, we investigated the localization and function of palladin in cultured cortical neurons. Palladin was found preferentially targeted to the developing axon but not the dendrites and was strongly localized to the axonal growth cone. When palladin expression was attenuated by transfection with antisense constructs in both the B35 neuroblastoma cell line and in primary cortical neurons, a reduction in the expression of palladin resulted in a failure of neurite outgrowth. These results implicate palladin as a critical component of the developing nervous system, with an important role in axonal extension.

Choline deficiency induces apoptosis in primary cultures of fetal neurons.

Treatment of rats with choline during brain development results in long-lasting enhancement of spatial memory whereas choline deficiency has the opposite effect. Changes in rates of apoptosis may be responsible. We previously demonstrated that choline deficiency induced apoptosis in PC12 cells and suggested that interruption of cell cycling due to a decrease in membrane phosphatidylcholine concentration was the critical mechanism. We now examine whether choline deprivation induces apoptosis in nondividing primary neuronal cultures of fetal rat cortex and hippocampus. Choline deficiency induced widespread apoptosis in primary neuronal cells, indicating that cells do not have to be dividing to be sensitive to choline deficiency. When switched to a choline-deficient medium, both types of cells became depleted of choline, phosphocholine and phosphatidylcholine, and in primary neurons neurite outgrowth was dramatically attenuated. Primary cells could be rescued from apoptosis by treatment with phosphocholine or lysophosphatidylcholine. As described previously for PC12 cells, an increase in ceramide (Cer) was associated with choline deficiency-induced apoptosis in primary neurons. The primary neuronal culture appears to be an excellent model to explore the mechanism whereby maternal dietary choline intake modulates apoptosis in the fetal brain.

Pharmacokinetic studies in Tg.AC and FVB mice administered [14C] benzene either by oral gavage or intradermal injection.

Chronic benzene toxicity has been demonstrated to result in either aplastic anemia or acute myelogenous leukemia, a form of granulocytic leukemia, in exposed people (Snyder and Kalf, Crit. Rev. Toxicol. 24, 177-209, 1994). Aplastic anemia has been demonstrated in animal models following benzene exposure but, heretofore, it has not been possible to replicate benzene-induced granulocytic leukemia in animals. The Tg.AC mouse appears to be the first animal model in which a granulocytic leukemia was produced by treatment with benzene (Tennant et al., The Use of Short- and Medium-Term Tests for Carcinogenic Hazard Evaluation, 1999; French and Saulnier, J. Toxicol. Environ. Health 61, 377-379, 2000). Leukemia was observed in Tg.AC mice to which benzene was administered dermally. Neither orally dosed Tg.AC mice or mice of the parental FVB strain treated by either route of exposure developed leukemia. It is well established that benzene metabolism is required to produce benzene toxicity. To determine whether metabolic differences arising from differences in route of exposure or strain of mouse directed the development of leukemia, the pharmacokinetics of benzene were compared between the two strains and between the two routes of administration. Regardless of the route of exposure or the strain of mouse, seven major metabolites plus unmetabolized benzene were detected in most samples at most time points. Few differences were observed between the two strains following either route of administration. These results suggest that the genetic modification in the Tg.AC mouse, i.e., insertion of the v-Ha-ras construct into the genome, did not disrupt any major pathways involved in determining the pharmacokinetics of benzene. Two significant differences were observed between the two routes of exposure: first, benzene was absorbed more slowly after intradermal injection than after oral gavage, and second, the intradermally dosed mice produced more conjugates of hydroquinone than did the orally dosed mice. These differences in metabolism may be involved in the previously observed differences in hematotoxicity between the two routes of exposure.

Oral bioavailability of lead and arsenic from a NIST standard reference soil material.

The oral bioavailability of soil contaminants is measured using in vitro or in vivo techniques. Current efforts in our laboratory are focused on the comparisons of in vitro methods for bioavailability estimation with the presently employed in vivo techniques, such as animal models. We present a comparison of two techniques for oral bioavailability estimation: in vitro dissolution and in vivo rat feeding using a standard reference soil. Lead (Pb) and arsenic (As) were chosen because of the range of concentration in this soil as well as the large historical database of bioavailability values for these metals. Metal solubility was measured using a sequential soil extraction in synthetic analogues of human saliva, gastric and intestinal fluids. The soluble metal was defined as the bioaccessible fraction. Oral bioavailability of Pb and As was measured in Sprague Dawley rats by determining metal levels in the major organs and urine, feces, and blood at 1-, 2-, and 3-day time points. Extractions to determine bioaccessibility yielded a gastric component of 76.1% and 69.4% for Pb and As, respectively, and intestinal components were 10.7% and 65.9%. The oral bioavailability of the standard reference soil was 0.7% and 37.8% for Pb and As, respectively. Bioaccessibility was greater than bioavailability for both metals in both gastrointestinal compartments. Although Pb had the highest soil concentration of the selected metals, it was the least bioavailable, while As was highly available in both the in vitro and in vivo method. These types of data allow for an in vitro-in vivo comparison of a soil whose metal concentrations have been certified and validated.

Inhibition of hippocampal kindling by metabotropic glutamate receptor antisense oligonucleotides.

Recent work has shown that metabotropic glutamate receptors (mGluRs) increase in response to seizure activity and can contribute significantly to the expression and progression of partial seizures. Using the kindling model of temporal lobe seizures, we evaluated the ability of local hippocampal injections of mGluR1 antisense or mGluR3 antisense oligonucleotides to suppress receptor expression and alter hippocampal kindling. Daily antisense injections in the hippocampus resulted in a significant decrease in mGluR1 or mGluR2/3 immunoreactivity. Rats injected with mGluR3 antisense showed a brief suppression of afterdischarge duration when compared to matched rats injected with a nonsense-oligonucleotide. Rats injected with a mGluR1 antisense oligonucleotide had a dramatic suppression of the rate of seizure progression with no significant effect on afterdischarge duration. Suppression of mGluR1 synthesis by local antisense inhibition may provide a new therapeutic approach for the control of epileptogenesis.

Osmotic and glutamate receptor regulation of c-Jun NH(2)-terminal protein kinase in neuroendocrine cells.

Expression of a c-Jun NH(2)-terminal protein kinase (JNK), also known as stress-activated protein kinase (SAPK) in rodents, has been implicated in the ability of cells to respond to a variety of stressors. In nonmammalian cells, JNK participates in the regulation of cell volume in response to hyperosmotic stress. To explore the possibility that JNK may participate in the transduction of osmotic information in mammals, we evaluated the expression of JNK immunoreactivity in neuroendocrine cells of the supraoptic nucleus. Low basal expression of JNK-2 (SAPK-alpha) and JNK-3 (SAPK-beta) was seen in vivo and in vitro. During water deprivation, JNK-2 increased in the supraoptic nucleus but not in the cortex. Osmotic or glutamate receptor stimulation in vitro also resulted in an increase in JNK-2 that was tetrodotoxin (TTX) insensitive and paralleled by increased nuclear phospho-c-Jun immunoreactivity. A TTX-sensitive increase in JNK-3 was seen in smaller neurons. Thus different JNK pathways may mediate individual cellular responses to osmotic stress, with JNK-2 linked to osmotic and glutamate receptor stimulation in magnocellular neuroendocrine cells.

Microglial proliferation in cortical neural cultures exposed to feline immunodeficiency virus.

Microglia are thought to play an important role in neurodegenerative changes due to infection with human or animal immunodeficiency viruses. Using feline immunodeficiency virus and cat neural cultures, we observed a dramatic increase in the accumulation of microglia from a basal rate of 5-7% day(-1) to 25-126% day(-1). Both live virus and heat-inactivated virus induced proliferation. Negligible proliferation was seen in purified microglial cultures. Conditioned medium from astrocytes or mixed neural cultures treated with feline immunodeficiency virus stimulated the proliferation of purified microglia. Disease progression may be facilitated by early non-infectious interactions of lentiviruses with neural tissue that promote the activation and proliferation of microglia.

Neurotoxicity of CSF from HIV-infected humans.

Approximately 15-20% of individuals infected with the human immunodeficiency virus will develop severe neurological disease. This may be due in part to virus-induced release of a number of putative neurotoxins. However, there is little information to predict which individuals will progress to dementia or the precise mechanisms that drive pathogenesis. In an effort to identify early markers of neurological disease progression we used an in vitro bioassay with rat cortical neurons to test for the presence of toxins in CSF from 40 HIV-infected humans with mild, minimal or no neurological disease. A subset of HIV-infected individuals was found to have significant toxic activity in CSF indicating that toxic factors may be circulating prior to the development of dementia. The toxicity was concentration dependent and due to a factor with a molecular mass of less than 30 kDa. Only a small proportion of the cell death appeared to be due to apoptosis. Neuronal toxicity was associated with a gradual accumulation of intracellular calcium in a subset of cortical neurons over a period of 1-2 h and in the absence of a significant acute response. Individuals with both high viral burden and high CSF toxicity were significantly more likely to have neurological symptoms. These initial analyses indicate that toxic factors are present in the CSF of HIV-infected patients that could serve as useful markers of neurological disease progression and provide insights into pathogenic mechanisms in vivo.

Functional activation of punch-cultured magnocellular neuroendocrine cells by glutamate receptor subtypes.

To provide a simple means to isolate and study the cellular functions of small groups of neurons, we developed a modified 'punch' culture procedure that facilitates acute and long-term in vitro physiological studies. Primary 'punch' cultures of magnocellular neuroendocrine cells (MNCs) from the supraoptic nucleus (SON) were established and the basic physiological effects of subtype-specific glutamate receptor agonists were characterized. MNCs from the punch cultures established a mature morphology in culture with extensive outgrowth of axons and varicosities. After 8 days, a single cultured SON punch produced an average of 10.0 +/- 2.1 pg AVP and contained an average of 222 +/- 53 vasopressin-neurophysin immunoreactive cells. Patch clamp recordings from MNCs demonstrated the presence of N-methyl-D-aspartate (NMDA)-sensitive and DL, alpha-amino-3-hydroxy-5-methylisoxazole propionic acid (AMPA)-receptors. Stimulation of metabotropic receptors with 1S,3R ACPD induced acute or gradual increases in intracellular calcium. NMDA, AMPA and metabotropic receptors all contributed to the secretion of vasopressin from the punch cultures with an agonist rank order potency of: NMDA (10 microM) > AMPA (1 microM) = 1S,3R ACPD (100 microM) > kainate (10 microM). This culture preparation should be useful for cellular studies of small groups of neuroendocrine and other cells.

Neurotoxicity of FIV and FIV envelope protein in feline cortical cultures.

The neurotoxic effects of the feline immunodeficiency virus (FIV) and FIV envelope proteins were measured in primary cultures of feline cortical neurons. Envelope protein from the FIV-PPR strain promoted neuronal swelling and death, whereas envelope protein from the FIV-34TF10 isolate produced intermediate or negligible toxicity. No effect was observed in control cultures treated with envelope protein from the Epstein-Barr virus. A concentration-effect curve showed that FIV-PPR protein produced maximal toxicity at 200 pM protein and decreased toxicity at higher concentrations, which is consistent with previous reports of the HIV-1 surface glycoprotein, gp120. These effects required the presence of low concentrations of glutamate. Using the natural host cells as targets, the effects of envelope protein and infectious virions were directly compared. All of the toxic activity could be attributed to non-infectious interactions between the viral envelope and target cells. Addition of 1 microM tetrodotoxin failed to block the effects of FIV-PPR in the presence of 20 microM glutamate. Toxicity would appear to involve two steps in which the envelope protein first sensitizes neurons through non-synaptic interactions (TTX insensitive) thereby setting the stage for enhanced synaptic activation via glutamate receptors (TTX sensitive).

Kindled seizures increase metabotropic glutamate receptor expression and function in the rat supraoptic nucleus.

The spread of experimentally kindled seizures in rats results in sustained increases in plasma vasopressin (VP) and VP mRNA in the supraoptic nucleus (SON). These increases provide an excellent example of the pathological plasticity that can develop in normal cells exposed to recurrent seizure activity. To test whether this plasticity might be due in part to changes in metabotropic glutamate receptors (mGluRs), we examined mGluR mRNA expression in the SON 1 month after stage 5 amygdala kindling. Three mGluR subtypes were detected by in situ hybridization in the SON in the following relative levels: mGluR3 > mGluR1 > mGluR7. Both mGluR1 and mGluR3 mRNAs were significantly increased in the SON (+28-61%) and cortex (+27-42%) after kindling. Immunoreactivity for mGluR1 but not mGluR2/3 was significantly increased in vivo in the SON. Receptor protein expression and intracellular calcium accumulation in response to the mGluR agonist, 1S,3R ACPD, were evaluated after in vitro "kindling" of neuroendocrine cells by Mg2+ deprivation. Increased immunoreactivity for mGluR1 and mGluR2/3 was seen in all cultures 3 days after a brief exposure to Mg2+-free medium. 1S,3R 1-aminocyclopentane-1,3-dicarboxylic acid (ACPD) induced rapid peak responses and gradual accumulations of intracellular Ca2+ in neurons. Both responses were increased in the "kindled" cells. Increases in the expression of functional mGluR1 and perhaps mGluR3 receptors may contribute to the development of long-lasting plastic changes associated with seizure activity.

Transcriptional suppression of estrogen receptor gene expression by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD).

TCDD, the most potent congener of the polychlorinated dioxins, has been shown to be an antiestrogen. The mechanisms of TCDD-induced antiestrogenicity are still under investigation. In this study, we investigated the effects of TCDD on the expression of the estrogen receptor (ER) gene. We studied the levels of un-spliced ER transcript (hnRNA) as well as the ER mRNA in ovary, uterus and liver of TCDD-treated mice with different genetic backgrounds. To quantitate the ER hnRNA levels, the intron and exon boundary of ER hnRNA was amplified by competitive RT-PCR. The ER mRNA from these mice was quantitated by competitive RT-PCR amplifying exons separated by an intron. ER hnRNA and ER mRNA levels were quantitated 4 days after a single i.p. dose of TCDD (5 microg/kg) in female C57BL/6J (B6) mice, which carry the responsive allele to TCDD. TCDD treatment significantly (p < 0.05) suppressed the levels of ER hnRNA in the ovary (27.4%) and uterus (21.9%). The decreases in ER hnRNA were coordinated with significant (p < 0.01) decreases in ER mRNA in ovary (57.7%) and uterus (37.6%). There was a significant decrease (20.3%, p < 0.05) in liver ER mRNA, however, the changes of ER hnRNA in liver were not significant. The coordinated decreases in ER hnRNA and mRNA in TCDD-treated mice suggest a suppression of transcription of the ER gene. We performed the same study on DBA/2J (D2) mice, which possess the "non-responsive" allele of the aryl hydrocarbon receptor (AhR). These mice demonstrated no significant decrease in either the ER mRNA or hnRNA after TCDD treatment. Overall, these results suggest that TCDD suppresses the gene expression of the ER receptor by decreasing its transcription, and the AhR plays an important role in mediating this response.

Clinical and biologic activity of an estrogenic herbal combination (PC-SPES) in prostate cancer.

BACKGROUND: Herbal mixtures are popular alternatives to demonstrated therapies. PC-SPES, a commercially available combination of eight herbs, is used as a nonestrogenic treatment for cancer of the prostate. Since other herbal medicines have estrogenic effects in vitro, we tested the estrogenic activity of PC-SPES in yeast and mice and in men with prostate cancer. METHODS: We measured the estrogenic activity of PC-SPES with transcriptional-activation assays in yeast and a biologic assay in mice. We assessed the clinical activity of PC-SPES in eight patients with hormone-sensitive prostate cancer by measuring serum prostate-specific antigen and testosterone concentrations during and after treatment. RESULTS: In complementary yeast assays, a 1:200 dilution of an ethanol extract of PC-SPES had estrogenic activity similar to that of 1 nM estradiol, and in ovariectomized CD-1 mice, the herbal mixture increased uterine weights substantially. In six of six men with prostate cancer, PC-SPES decreased serum testosterone concentrations (P<0.05), and in eight of eight patients it decreased serum concentrations of prostate-specific antigen. All eight patients had breast tenderness and loss of libido, and one had venous thrombosis. High-performance liquid chromatography, gas chromatography, and mass spectrometry showed that PC-SPES contains estrogenic organic compounds that are distinct from diethylstilbestrol, estrone, and estradiol. CONCLUSIONS: PC-SPES has potent estrogenic activity. The use of this unregulated mixture of herbs may confound the results of standard or experimental therapies and may produce clinically significant adverse effects.