Sampangine Inhibits Heme Biosynthesis in both Yeast and Human

Abstract

The azaoxoaporphine alkaloid sampangine exhibits strong antiproliferation activity in various organisms. Previous studies suggested that it somehow affects heme metabolism and stimulates production of reactive oxygen species (ROS). In this study, we show that inhibition of heme biosynthesis is the primary mechanism of action by sampangine and that increases in the levels of reactive oxygen species are secondary to heme deficiency. We directly demonstrate that sampangine inhibits heme synthesis in the yeast Saccharomyces cerevisiae. It also causes accumulation of uroporphyrinogen and its decarboxylated derivatives, intermediate products of the heme biosynthesis pathway. Our results also suggest that sampangine likely works through an unusual mechanism—by hyperactivating uroporhyrinogen III synthase—to inhibit heme biosynthesis. We also show that the inhibitory effect of sampangine on heme synthesis is conserved in human cells. This study also reveals a surprising essential role for the interaction between the mitochondrial ATP synthase and the electron transport chain.

INTRODUCTION

The plant-derived alkaloid sampangine has broad and potent antiproliferation activities against fungal pathogens, human cancer cell lines, malaria parasites, and mycobacteria (19, 20, 25, 31). In particular, its potency against various human fungal pathogens in vitro is comparable to the potencies of existing antifungal drugs (1). Other analogs of the same aporphine family of alkaloids have also been shown to exhibit antiproliferation activities against viruses, bacteria, fungi, parasites, and tumor cell lines (7, 8, 17, 24, 34, 36). However, the mechanism of action exhibited by sampangine and its analogs in both cancer cells and human pathogens remains unclear. Sampangine was shown to induce apoptosis in cancer cells by stimulating the generation of reactive oxygen species (ROS) (20). A closely related alkaloid, ascididemin, was shown to inhibit the growth of Mycobacterium tuberculosis through iron depletion (5). We have previously also shown that sampangine inhibits fungal growth, likely by interfering with heme metabolism (1). This was supported by the observations that multiple mutants with mutations affecting the Saccharomyces cerevisiae yeast heme biosynthesis pathway were comparatively more sensitive to sampangine than a wild-type strain and that exogenously supplied hemin partly suppressed the inhibitory activity of the drug (1). However, exogenous hemin failed to completely reverse the inhibitory effect of sampangine. Overexpressing genes in the heme biosynthesis pathway also failed to confer sampangine resistance in a wild-type strain background (unpublished results). These results raise the question of whether heme synthesis is a primary target of the drug and also whether there is any molecular relationship between the many different cellular effects of sampangine.

In this study, we took an unbiased functional genomic approach by systematically screening the yeast genome-wide deletion mutant libraries to identify mutants exhibiting hypersensitivity or resistance toward the drug. Although none of the mutants tested conferred obvious resistance, we identified 132 mutants that were hypersensitive. Among these, the most sensitive ones affected mitochondrial functions, especially subunits of the ATP synthase. We next performed genome-wide synthetic lethality analyses with strains with two representative mutations (atp1Δ and yme1Δ) that conferred the highest drug sensitivity to identify common genes or pathways that, when disrupted, cause severe growth defects or lethality. This identified mutations that affect the electron transport chain (ETC) and heme synthesis (hem14Δ) to be among the most significant ones. Comparison of synthetic lethality profiles suggested that the hem14Δ mutation reproduced the effects of sampangine treatment much better than mutations in ETC (cox17Δ or cyc3Δ), suggesting that the heme biosynthesis pathway is a primary target of the drug. Consistent with this model, we found that sampangine inhibits heme production in both yeast and human cells. Moreover, our studies on porphyrin profiles and enzyme activity levels indicate that sampangine inhibits heme synthesis likely by hyperactivating Hem4, the fourth step within the pathway. We also found that the growth defect of a hem14Δ yeast mutant was partially suppressed by the antioxidant N-acetyl cysteine (NAC), indicating that heme deficiency likely accounts for the increased ROS levels observed in sampangine-treated cells (1, 20).

MATERIALS AND METHODS

Yeast media and chemicals.

The haploid selection synthetic complete (SC) medium without Leu, His, and Arg and with G418 and l-canavanine (Can) contained dextrose (20 g/liter), yeast nitrogen base without amino acids and ammonium sulfate (1.7 g/liter), SC medium Leu-His-Arg dropout mix (2 g/liter), sodium glutamate (1 g/liter), G418 (200 mg/liter), l-canavanine (60 mg/liter), and agar (2%). Sodium glutamate substituted for ammonium sulfate as the nitrogen source to make the G418 selection more reliable on the minimal medium. A version of this that lacked uracil was used to select for double mutants during synthetic lethality analyses. SC medium contained dextrose (20 g/liter), yeast nitrogen base without amino acids and ammonium sulfate (1.7 g/liter), SC medium mix (2 g/liter), and sodium glutamate (1 g/liter) with or without agar (2%). A similar medium that lacked uracil was also used. Solid sporulation medium contained potassium acetate (10 g/liter), zinc acetate (0.05 g/liter), and agar (2%).

ATP, N-acetyl cysteine, and antimycin were purchased from Sigma-Aldrich and dissolved in water as stock solutions of 1 M, 100 mg/ml, and 100 mg/ml, respectively. Sampangine was isolated as described previously (31) and dissolved in dimethyl sulfoxide (DMSO) as a stock solution of 2 mg/ml.

Yeast strains and plasmids.

Yeast strains used in this study were haploid-convertible heterozygous diploid yeast deletion mutants (mutants YSC4035 and YSC4428; Open Biosystems) and their haploid MATa convertants and were used after sporulation and haploid selection as previously described (29). The genotype of a typical haploid strain was MATa ura3Δ0 leu2Δ0 his3Δ0 met15Δ0 can1Δ::LEU2-MFA1pr-HIS3 goiΔ::kanMX, where goiΔ stands for deletion of any gene of interest. A hoΔ::kanMX mutant was used as a surrogate wild-type control in most experiments because the HO gene is already mutated in the parent strain.

Wild-type genes HEM3, HEM4, and HEM12 were PCR amplified and cloned into the vector YEplac195 (2μm, URA3) (13) to construct the overexpression plasmids.

Screening for and validation of sampangine-hypersensitive haploid YKOs.

Screening for and validation of sampangine-hypersensitive haploid yeast knockouts (YKOs) were carried out essentially as previously described (28, 29). For the screen, a pool of haploid-convertible heterozygote diploid yeast deletion mutants was sporulated. Pools of isogenic MATa haploid cells were derived by growth on a haploid selection medium (SC medium without Leu, His, and Arg and with G418 and canavanine) that either contained (experiment) or lacked (control) sampangine at 0.5 μg/ml. Relative representation of each YKO in drug-treated and untreated pools was compared by bar code microarray analysis. For validation, individual haploid convertible heterozygous diploid mutants were sporulated, spotted onto haploid selection medium that either contained or lacked sampangine at the indicated concentrations, and incubated at 30°C for 3 days. The confirmed results are reported in Table S1 in the supplemental material.

Genome-wide synthetic lethality analyses and validation.

Genome-wide synthetic lethality analyses and validation were carried out essentially as previously described (28, 29). The query constructs used in this study were atp1Δ::URA3, yme1Δ::URA3, hem14Δ::URA3, cox17Δ::URA3, and cyc3Δ::URA3, each of which contained ∼1.5-kb flanking sequences that allow highly efficient and accurate disruption of the target gene. Genome-wide synthetic lethality screens were performed with the atp1Δ::URA3 and yme1Δ::URA3 constructs. The other constructs were tested individually against the list of 132 mutations that conferred sampangine hypersensitivity.

Cellular component enrichment analysis.

Gene lists identified from the genome-wide sampangine hypersensitivity mutant profiling and synthetic lethality analyses were subject to Database for Annotation, Visualization, and Integrated Discovery (DAVID) analysis as previously described (16). Significantly enriched cellular components were selected on the basis of a cutoff false discovery rate (Q) value of 0.05, and the fold enrichment was plotted.

Measuring heme levels in yeast.

An overnight culture of a yeast strain carrying an empty plasmid or overexpressing one of the heme biosynthesis genes was seeded into 50 ml of fresh SC medium without Ura at a 0.1 optical density at 600 nm (OD₆₀₀)/ml and incubated at 30°C for about 2 doublings. The culture was split into two, with one exposed to sampangine (1 μg/ml) and the other one exposed to DMSO. Both cultures were subsequently incubated at 30°C for 3 h. We note that sampangine at this concentration did not obviously reduce growth of the wild-type strain during this treatment period. Similar numbers of cells from each culture were harvested, washed with ice-cold water, and homogenized in 500 μl of lysis buffer as previously described (27). Two microliters of each cell lysate was used to measure the amount of heme with a hemin assay kit (BioVision, CA) according to the manufacturer's instructions. The protein concentration of each lysate was also measured and used to calculate the heme concentration as fmol/μg protein. Three independent experiments were performed for each culture condition, and the results were averaged. The average heme level in the wild-type strain containing the empty vector that was grown in the absence of sampangine was set at 100%.

Measuring heme levels in human cancer cell lines.

Exponentially growing cultures (2 × 10⁵) of the acute T cell leukemia Jurkat and non-small cell lung cancer NCI-H1299 cell lines were seeded into 6-well plates containing RPMI 1640 supplemented with fetal bovine serum (10%; Sigma-Aldrich), glutamine (2 mM), penicillin (100 U/ml), and streptomycin (100 μg/ml) and incubated at 37°C in humidified air with 5% CO₂. After an overnight incubation, cells were treated with sampangine (at 0.01 μg/ml or 0.1 μg/ml) or DMSO as the vehicle control and incubated for another 24 h, harvested, and washed softly with germfree phosphate-buffered saline. About 5,000 cells from each culture were used to measure the amount of hemin present using a hemin assay kit (BioVision, CA) according to the manufacturer's instruction. At least three independent experiments were performed for each culture condition, and the results were averaged. The average heme level for each cell line grown in the absence of sampangine was set at 100%.

Determination of porphyrin profile in yeast.

An overnight culture of wild-type yeast grown in liquid SC medium was used to inoculate 100 ml fresh medium at 0.1 OD₆₀₀/ml. After one doubling, either sampangine at the 50% inhibitory concentration (IC₅₀; 1.17 μg/ml) or DMSO (0.25%) was added to the cultures. The cells were allowed to grow for 14 h after treatment, and equal numbers of cells from the corresponding treated and untreated cultures were harvested by centrifugation. The cells were washed once with sterile distilled water, and the cell pellets were flash frozen in liquid nitrogen. The experimental conditions, including the medium, temperature, aeration, and concentration of sampangine, were identical to the conditions used in a previously reported study (1). Three independent experiments were performed on independently grown cultures. The samples were processed for high-performance liquid chromatography (HPLC) analysis at Frontier Scientific Inc. (Logan, UT) as described previously (4).

In vitro assays of porphobilinogen deaminase (PBGD), uroporphyrinogen III synthase (UIIIS), and uroporphyrinogen III decarboxylase (UroD). (i) PBGD assay.

A 100-μl mixture contained 330 μM porphobilinogen (PBG), 70 mM Tris (pH 7.65), 10 mM dithiothreitol (DTT), 6 μg purified recombinant human porphobilinogen deaminase (rPBGD), and 6 μl of stock DMSO that contained various amounts of sampangine. The mixture was then incubated at 37°C for 30 min in the dark. The reaction was terminated by adding 100 μl of 3 M HCl. The uroporphyrinogen I produced in the assay mixture was then oxidized to uroporphyrin I by exposure to UV light for 30 min. The acidified sample was then centrifuged in a microcentrifuge. The supernatant was analyzed by ultraperformance liquid chromatography (UPLC) as described below.

(ii) UIIIS assay.

The UIIIS assay is performed as described above with the addition of 2 μg of purified recombinant UIIIS (rUIIIS). The porphyrin isomers were resolved by a method derived from those described previously (35, 39). An HPLC system that consisted of a Waters 2795 separations module (Alliance HT), a Waters 474 scanning fluorescence detector, a Waters 2996 photodiode array detector and a Phenomenex Gemini C₁₈ column (250 mm by 4.6 mm by 5 μm) was used. The solvents were 1 M NH₄OH in water adjusted to pH 5.16 with glacial acetic acid (solvent A), methanol (solvent B), acetonitrile (solvent C), and 10% (vol/vol) methanol in water (solvent D). All gradients were linear, and the flow rate was always 1 ml/min for the duration of the 40-min run. Solvent C was set at 10% throughout the run. The gradient conditions were set at 75% solvent A and 15% solvent B at 0 min, 60% solvent A and 30% solvent B at 8 min, 40% solvent A and 50% solvent B at 13 min, 28% solvent A and 62% solvent B at 17 min, and 70% solvent B and 20% solvent D at 20 min and 25 min and then went back to initial conditions at 28 min.

(iii) UroD assay.

The UroD assay was carried out essentially as previously described (32). Uroporphyrinogen I (30 to 35 nmol) for a single assay was enzymatically produced in a 100-μl mixture from PBG using 6 μg PBGD and 2 μg rUIIIS as described above.

(iv) Separation and detection of porphyrins.

About 10 μl of the supernatant from the assay described above was injected into a reverse-phase Waters Acquity UPLC system with an Acquity fluorescence detector that was set at 404 nm excitation and 618 nm emission. A BEH phenyl 1.7-μm column (2.1 by 50 mm) was used to resolve the various porphyrins for a 3.5-min total run time at 60°C. The flow rate was set at 0.7 ml per min. Two solvents were used: solvent A consisted of 0.1% formic acid in water, while solvent B was pure acetonitrile. The gradient conditions were set at 60% solvent A at 0.0 min, 35% solvent A at 2.0 min, 10% solvent A at 2.1 min, and 1% solvent A at 2.6 min and then went back to initial conditions at 2.7 min. Except at 2.0 min, where the gradient was set at convex 5, all the rest were set at linear 6. The millivolt signals detected for the various porphyrins were individually compared with those in a standard solution that contained 5 pmol each of 8-, 7-, 6-, 5-, 4-, and 2-carboxylporphyrin per 10-μl injection. The chromatograms were processed using Waters Empower Pro software.

RESULTS

Mutants of mitochondrial ATP synthase are hypersensitive to sampangine.

To identify the primary target of sampangine in a eukaryote, we first systematically screened the yeast genome-wide heterozygous diploid deletion mutants (11) and found that the mutant with the TOM40/tom40Δ mutation, which affects mitochondrial protein import (37), was the only one exhibiting significant hypersensitivity. However, this hypersensitivity was likely due to the intrinsic growth defects exhibited by this mutant (data not shown). We also screened a genome-wide open reading frame overexpression library that we constructed (Z. Huang and X. Pan, unpublished data) and found that overexpressing the multidrug ABC transporter SNQ2 (33) confers sampangine resistance (data not shown), possibly due to decreased drug accumulation inside yeast cells. We next screened the genome-wide haploid deletion mutants with complete loss of gene functions for increased drug sensitivity, reasoning that such mutants could define cellular functions closely related to the drug's target, if not the target itself. Upon individual validation, we identified 132 haploid deletion mutants that were significantly more sensitive to sampangine at 0.5 μg/ml than a wild-type strain (see Table S1 in the supplemental material). Highly enriched among these were mutants with mutations affecting the mitochondrial ATP synthase, histone modification and chromatin remodeling, peroxisome biogenesis, and endoplasmic reticulum-Golgi functions (Fig. 1A; see Table S1 in the supplemental material). Mutations affecting oxidative stress response (e.g., lys7Δ) and DNA damage repair (e.g., rad50Δ) were also identified (Fig. 1B; see Table S1 in the supplemental material), consistent with previous observations that the drug causes oxidative damage (1, 20). A hem14Δ mutant lacking the only nonessential gene of the heme biosynthesis pathway in yeast was also sensitive to the drug (Fig. 1B; see Table S1 in the supplemental material). Among all these mutants, those with mutations affecting multiple subunits of the mitochondrial ATP synthase (e.g., Atp1p) and several other mitochondrial proteins, such as Yme1p, the catalytic subunit of the mitochondrial inner membrane i-AAA protease involved in the turnover of unfolded or misfolded mitochondrial membrane proteins (22, 26, 30, 38), exhibited the highest sensitivity (Fig. 1B and C; see Table S1 in the supplemental material). The sampangine-hypersensitive phenotype of the atp1Δ mutant was partly suppressed both by exogenously supplied ATP and by the antioxidant NAC (Fig. 1C). These results together indicated that sampangine affects mitochondrial functions, resulting in reduced energy and elevated ROS production. Interestingly, the sampangine-hypersensitive phenotype of the yme1Δ mutant was suppressed by ATP but was barely suppressed by NAC under the same conditions (Fig. 1C), suggesting that the increase in ROS levels may not represent the most fundamental challenge faced by cells treated with sampangine.

Fig. 1.

Sampangine-hypersensitive mutants identified from genome-wide fitness profiling analysis. (A) Sampangine-hypersensitive mutants define several highly enriched cellular components. In parentheses are the numbers of the Gene Ontology (GO) terms. ESCRT II, endosomal sorting complex required for transport II. (B) Representative mutants with differing sensitivities to sampangine (SMP). The hoΔ mutant served as a surrogate wild-type (WT) control in this experiment and in experiments whose results are presented in all other figures. (C) Suppression of sampangine hypersensitivity by exogenously supplied ATP (1 mM) and/or NAC (0.1 μg/ml) in atp1Δ and yme1Δ mutants.

The ATP synthase and Yme1p become essential when the ETC or heme synthesis is defective.

To identify the primary function(s) affected by sampangine, we next performed genome-wide synthetic lethality analyses with the atp1Δ and yme1Δ mutations using a high-throughput technology that we previously described (28, 29). Given that both mutations conferred hypersensitivity to sampangine yet the corresponding proteins seem to have distinct biological functions, we expected that genome-wide synthetic lethality analyses with these two mutations would identify common as well as distinct genetic interactions. Some of the common synthetically lethal interactions would likely define the pathway(s) targeted by sampangine. Upon individual validation, we identified 172 and 190 synthetically lethal or sick interactions for atp1Δ and yme1Δ, respectively, with the majority of them being surprisingly common (Fig. 2A). Most of these common interactions affected mitochondrial functions, particularly the mitochondrial ribosome and the ETC (Fig. 2B and C; see Table S2 in the supplemental material). While mutations of the mitochondrial ribosome typically cause severe growth defects on their own and tend to exhibit synthetic lethality interactions with other mutations, single mutations affecting mitochondrial ETC cause only modest growth defects (data not shown), and their synthetic lethality interactions with atp1Δ and yme1Δ were thus deemed more specific and significant. In addition, mutations of mitochondrial ribosome could affect expression of some of the ETC components. These results suggested that the atp1Δ and yme1Δ mutants both need a functional ETC to survive, and this was further corroborated by the observation that they were both highly sensitive to the ETC inhibitor antimycin (Fig. 2D). However, the ETC itself is unlikely a primary target of sampangine because it is dispensable for yeast cell survival, whereas the drug completely inhibits cell growth at >2 μg/ml. Instead, sampangine likely targets a process whose inhibition severely compromises the function of ETC.

Fig. 2.

Sampangine inhibits heme synthesis and mitochondrial ETC. (A) Genome-wide synthetic lethality analyses with atp1Δ and yme1Δ identified a list of common mutations. The number of interactions for each is shown in parentheses. The genes are listed in Table S2 in the supplemental material. (B) Common genetic interactions of atp1Δ and yme1Δ mostly affect mitochondrial protein synthesis and the respiration chain. (C) Synthetic lethality between atp1Δ and cox17Δ or hem14Δ and between yme1Δ and cox17Δ or hem14Δ revealed by tetrad analysis. (D) Hypersensitivity of the atp1Δ and yme1Δ mutants toward the ETC inhibitor antimycin (0.1 μg/ml). (E) Inability of wild-type yeast cells to use glycerol as the sole carbon source in the presence of sampangine (0.1 μg/ml). The cox17Δ and hem14Δ mutants were included as controls.

We next considered the essential pathway(s) critical for the function(s) of ETC as a potential sampangine target(s) and focused on heme biosynthesis for three main reasons. First, our previous studies have implicated heme in the antifungal activity of sampangine. Second, heme is a physical component of ETC as a cofactor of the cytochromes. Third, synthetic lethality interactions were observed between heme deficiency (caused by hem14Δ mutation) and both the atp1Δ and yme1Δ mutations (Fig. 2C; see Table S2 in the supplemental material). Inhibition of heme synthesis thus explains why the atp1Δ and yme1Δ mutants were sensitive to the drug. Consistent with the model that sampangine inhibits the heme biosynthesis pathway and subsequently causes defects in ETC, treating yeast with a low concentration of the drug blocked cell growth on glycerol as the sole carbon source. A similar blockade in cell growth on glycerol was observed with deletions in HEM14 or COX17, which encodes the copper metallochaperone required for the assembly of cytochrome c oxidase (3, 15) (Fig. 2E). Moreover, mutants lacking ETC components were no more sensitive to sampangine than a wild-type strain (Fig. 1B and data not shown), and a synthetic lethality interaction was not observed between hem14Δ and mutations of the ETC (cox17Δ) (Fig. 3B), further indicating that sampangine treatment, hem14Δ, and cox17Δ all affect the same pathway.

Fig. 3.

The heme biosynthesis pathway is a primary target of sampangine. (A) Venn diagram representation of genetic interaction profiles of sampangine treatment (0.5 μg/ml) and a hem14Δ or cox17Δ mutation. The number of interactions for each is shown in parentheses. These results were derived from Table S1 in the supplemental material. (B) Synthetic sick interaction between hem14Δ and mutations affecting DNA repair (rad52Δ), peroxisome biogenesis (pex13Δ), and oxidative stress response (lys7Δ). Haploid-convertible heterozygous diploid deletion mutants that harbored an hem14Δ::URA3 mutation and those of the indicated genotypes (kanMX as the marker) were sporulated, spotted on three haploid selection media, and evaluated for growth. The medium lacking both uracil and G418 allowed growth of the hem14Δ::URA3 mutant as well as the double mutant. The medium that lacked uracil but contained G418 allowed growth of double mutants only. The medium that contained both uracil and G418 allowed growth of the double mutants as well as single mutants of the indicated genotypes. (C) Suppressing the growth defect of a hem14Δ mutant with N-acetyl cysteine (0.1 μg/ml). (D) Exogenously supplied hemin (65 μg/ml) partly suppressed the lethality caused by deleting heme biosynthesis genes but failed to rescue the inhibitory effect of a lethal dose of sampangine (2 μg/ml).

The hem14Δ mutation phenotypically mimics sampangine treatment.

We further investigated whether the heme biosynthesis pathway represents a primary target of sampangine. In such a case, we expected that most if not all of the sampangine-sensitive mutants would exhibit lethality or a growth defect when heme synthesis is reduced. Consistent with this model, ∼58% (77 of 132) of the mutations that conferred sensitivity toward sampangine (at 0.5 μg/ml) were indeed synthetically lethal or caused sick interactions with the hem14Δ mutation (Fig. 3A; see Table S1 in the supplemental material). These included mutations affecting the ATP synthase, histone modification, chromatin remodeling, peroxisome functions, oxidative stress response, and DNA repair (Fig. 3B; see Table S1 in the supplemental material). For most of the mutations that did not exhibit synthetically lethal or sick interactions with hem14Δ, the corresponding single mutants were also not sensitive to a lower dose of the drug (0.2 μg/ml) (see Table S1 in the supplemental material), reflecting the possibility that the defect caused by a hem14Δ mutation was likely milder than that caused by treatment with sampangine at 0.5 μg/ml. The rest of the hem14Δ noninteractors likely either affect drug metabolism or reflect additional functions targeted by sampangine that are not related to heme synthesis. In contrast, only ∼22% (29 of 132) of mutants with sampangine-sensitive mutations exhibited synthetic lethal or sick interactions with the cox17Δ mutation, and all these were included within the interaction profile of hem14Δ (Fig. 3A; see Table S1 in the supplemental material). These results together indicated that the heme biosynthesis pathway is a primary target of sampangine and that the respiration defect represents only one of many possible secondary effects of heme deficiency due to sampangine treatment.

The genetic interactions between a hem14Δ mutation and mutations affecting DNA repair, peroxisome biogenesis, and the oxidative stress response (Fig. 3B) also suggested that heme deficiency might lead to oxidative stress, possibly through crippling the ETC. Consistent with this model, the growth defect of the hem14Δ mutant was partly suppressed by the antioxidant NAC (Fig. 3D). Thus, heme deficiency is at least partly responsible for the increase in ROS levels observed in sampangine-treated cells (1, 20).

However, exogenously supplied hemin failed to completely reverse the inhibitory effect of sampangine at relatively high concentrations (Fig. 3D). The reason for this is currently not clear. Possibly, exogenous hemin is not readily taken up by yeast cells and/or efficiently transported to the right cellular compartment (e.g., mitochondrion) once inside the cells. This is consistent with the observation that exogenous hemin only weakly suppressed the lethality caused by deleting components of the heme biosynthesis pathway even in the absence of sampangine (Fig. 3D). It is also possible that sampangine inhibits cell proliferation by targeting heme biosynthesis as well as other yet-to-be identified pathways. In this case, exogenously supplying hemin will unlikely restore all functions inhibited by sampangine.

Sampangine inhibits heme biosynthesis.

Given the phenotypic similarity caused by sampangine treatment and a hem14Δ mutation, as discussed above, we further tested the model in which sampangine inhibits heme biosynthesis. As expected, the hem14Δ mutant showed greatly reduced heme levels compared to a wild-type strain (Fig. 4A). A similar effect was observed when a wild-type yeast strain was treated with sampangine at the IC₅₀ after one round of cell division (Fig. 4A). Such an inhibitory effect of sampangine on heme levels was also observed in human cells. Treatment with sampangine significantly reduced the intracellular levels of heme in two different human cancer cell lines tested (Fig. 4B). These results together indicate that sampangine inhibits heme synthesis in both yeast and human cells.

Fig. 4.

Sampangine inhibits heme biosynthesis in vivo. (A) Sampangine treatment reduces heme levels in wild-type yeast. The heme concentration in the hoΔ mutant not treated with the drug, calculated as fmol per μg of total protein, was set at 100%. Values shown are means ± standard deviations from assays performed in triplicate. The absolute levels of heme in the hem14Δ mutant were close to the background levels and the detection limit of the assay. (B) Sampangine treatments (0.01 μg/ml and 0.1 μg/ml) reduce heme levels in two human cancer cell lines, as indicated. The heme concentration in each cell line not treated with the drug, calculated as fmol per μg of total protein, was set at 100%. Values shown are means ± standard deviations from assays performed in triplicate. (C) Heme biosynthesis pathway in yeast. CoA, coenzyme A. (D) Effects of sampangine treatment on porphyrin profiles in a wild-type yeast strain. Uro, uroporphyrin III; Hepta, heptaporphyrin III; Hexa, hexaporphyrin III; Penta, pentaporphyrin III; Copro, coproporphyrin III; Proto, protoporphyrin IX. Values shown are means ± standard deviations from assays performed in triplicate. (E) Effects of sampangine treatment (1 μg/ml) on protein levels of each of the eight enzymes of the yeast heme biosynthesis pathway analyzed by Western blotting. Enzymes were expressed as tandem affinity purification (TAP) tag fusion proteins from the endogenous loci. The fusion proteins are indicated with black arrows. Tub2 was used as a loading control.

We next investigated whether sampangine inhibits a specific step in the heme biosynthesis pathway (Fig. 4C) (14). We compared the porphyrin intermediate metabolite profiles of yeast cells grown in the presence and absence of the drug. In particular, we measured the levels of uroporphyrin III, coproporphyrin III, and protoporphyrin IX, the oxidized intermediate products of the fourth, fifth, and sixth steps of this pathway, respectively, because sampangine treatment leads to accumulation of a red pigmentation, which we thought to be porphyrin (1). Similar to what was reported previously (21), wild-type yeast cells not treated with the drug mostly accumulated coproporphyrin III and protoporphyrin IX, whereas the level of uroporphyrin III was extremely low and beyond detection (Fig. 4D). Treatment with the drug increased the total amount of porphyrins by 5- to 20-fold in three independent experiments (data not shown). More importantly, the porphyrin profile in the drug-treated cells was drastically altered, with the relative level of uroporphyrin greatly increased and the levels of coproporphyrin III and protoporphyrin IX decreased (Fig. 4D). The levels of the hepta, hexa, and penta intermediates of uroporphyrin III metabolism were also greatly increased (Fig. 4D). These results were similar to those observed in mutants with mutation of the uroporphyrinogen decarboxylase (UroD) encoded by HEM12 (21), suggesting that sampangine either inhibits Hem12 or hyperactivates enzymes in the preceding steps, or both. Interestingly, sampangine treatment also modestly yet reproducibly caused reduced protein levels of Hem3 and Hem4 and increased levels of Hem13 and Hem14 (Fig. 4E), possibly due to feedback regulation of expression by the accumulated intermediate metabolites. In contrast, the expression levels of Hem12 itself and Hem1, Hem2, and Hem15, which lie farther away from Hem12 along the pathway, were not significantly affected under similar conditions (Fig. 4E).

Sampangine likely hyperactivates Hem4 to inhibit heme synthesis.

We next tested the possibility that sampangine inhibits UroD. Using a well-characterized in vitro assay with a recombinant human enzyme (32), we surprisingly found that the activity of purified UroD was not significantly affected by excess sampangine at three concentrations tested (Fig. 5A). On the contrary, sampangine treatment significantly increased the activities of both porphobilinogen deaminase (PBGD) (P = 0.0009) and uroporphyrinogen III synthase (UIIIS) (P = 0.01), enzymes preceding UroD in the heme biosynthesis pathway (Fig. 5A). That sampangine stimulates the activities of these two enzymes could explain the increased levels of uroporphyrinogen III observed in drug-treated yeast cells (Fig. 4D). However, it was not clear how this might be related to the blockade of heme synthesis by sampangine treatment observed in vivo (Fig. 4A). It is possible that hyperactivation of PBGD and/or UIIIS disrupts the balance within the entire heme synthesis pathway and leads to decreased heme levels. To test this hypothesis, we measured cellular levels of hemin in yeast strains that overexpress HEM3, HEM4, and HEM12, genes encoding PBGD, UIIIS, and UroD, respectively (2, 10, 18), from high-copy-number plasmids. Overexpressing both HEM3 and HEM12 increased heme levels in yeast cells, and these effects were blocked by sampangine treatment (Fig. 5B). On the contrary, HEM4 overexpression reproducibly caused a reduction in heme levels even in the absence of sampangine (P = 0.005) (Fig. 5B). Consistent with this observation, overexpression of HEM4 caused a growth defect in both a wild-type strain and an atp1Δ mutant (Fig. 5C). This inhibitory effect was further exacerbated by sampangine treatment (Fig. 5C). These results together suggested that a higher level of activity of Hem4 due to sampangine treatment or genetic overexpression inhibits heme biosynthesis. However, reducing the dosage of HEM4 in a diploid yeast strain by half did not confer sampangine resistance (data not shown). An attempt with targeted random mutagenesis also failed to produce sampangine-resistant mutants, despite screening about 50,000 distinct alleles of the HEM4 gene (data not shown), possibly because potential resistance alleles are also nonfunctional.

Fig. 5.

Sampangine likely hyperactivates uroporphyrinogen III synthase to inhibit heme synthesis. (A) Effects of sampangine on the relative activities of uroporphyrinogen III synthase, porphobilinogen deaminase, and uroporphyrinogen decarboxylase in vitro. The activity of each enzyme in the absence of the drug was set at 100%. Values shown are means ± standard deviations from assays performed in triplicate. (B) Effects of overexpressing HEM3, HEM4, or HEM12 on cellular heme levels in the presence or absence of sampangine (1 μg/ml). Values shown are means ± standard deviations from assays performed in triplicate. (C) HEM4 overexpression inhibits yeast growth and causes sampangine hypersensitivity. Sampangine was used at 1 μg/ml and 0.05 μg/ml for the hoΔ and atp1Δ mutants, respectively.

DISCUSSION

By using a combination of genomic, genetic, and biochemical approaches, we have demonstrated that the plant alkaloid sampangine antagonizes cellular proliferation mainly by inhibiting heme biosynthesis, a function essential for cell survival. This mechanism of action is apparently conserved in yeast and human cells, further proving yeast to be a valuable system for investigating the mechanisms of action exhibited by bioactive small molecules. In order to discover the pathway(s) most significantly affected by sampangine treatment, we used a combination of genome-wide deletion mutant fitness profiling and synthetic lethality analysis. We first identified mutations that confer the highest sensitivity toward the drug. Cells with representative mutations were then used in genome-wide synthetic lethality analyses to reveal the cellular functions or pathways most likely affected by the drug. This approach will likely be useful in revealing the mechanisms of action exhibited by other drugs. Apparently, this was not the most straightforward approach for identifying drug targets in yeast cells. However, it was necessary with sampangine because both screening the genome-wide heterozygous diploid deletion mutants for drug-induced haploinsufficiency and analyzing a genome-wide gene overexpression library for drug-resistant clones, which could presumably directly reveal drug targets (6, 12, 23), had failed. We note that some heterozygous diploid mutants of the heme synthesis pathway were comparatively more sensitive to the drug than a wild-type strain (1) (data not shown). However, their defects were too subtle to be detected by the genome-wide screen that we performed.

Our results suggest that heme biosynthesis is a primary target of sampangine in yeast cells and likely also in human cells. This was supported by the evidence that heme deficiency due to a hem14Δ mutation largely mimicked the effects of sampangine treatment and that sampangine treatment inhibited heme synthesis in vivo. We also showed that heme deficiency is at least partly responsible for the increased levels of ROS observed in sampangine-treated cells (1, 20), possibly due to a defect in the ETC. In particular, we observed synthetic sick interactions between heme deficiency (due to the hem14Δ mutation) and oxidative stress response defects (due to the lys7Δ mutation) (Fig. 3C). In addition, the growth defect of a hem14Δ mutant was partly suppressed by the antioxidant NAC (Fig. 3C). Both sampangine treatment and heme deficiency were also shown to downregulate expression of multiple genes involved in iron uptake in yeast (1, 9). This could deprive cells of iron and further contribute to heme deficiency. This also points to the possibility that heme deficiency might also be responsible for the iron depletion observed in mycobacteria treated with the structurally related ascididemin (5). Taken together, inhibition of heme biosynthesis is likely the primary cause of the various biological effects of sampangine and its analogs.

Our results have also provided a glimpse into how sampangine might inhibit heme biosynthesis. It most likely involves hyperactivation of UIIIS (encoded by HEM4), which is by far the most active enzyme within the heme biosynthesis pathway (14). This model is based on the observations that sampangine stimulates the activity of human UIIIS in vitro and that overexpressing Hem4 inhibits heme synthesis and causes growth defects in yeast cells (Fig. 5). In addition, yeast cells treated with sampangine also accumulate higher levels of uroporphyrinogen III, the product of UIIIS, and its decarboxylation derivatives (Fig. 4A). However, it still remains to be investigated how sampangine treatment and hyperactivation of UIIIS might inhibit heme synthesis. It certainly does not involve reduced levels of expression of any of the downstream enzymes of the pathway (Fig. 4E). Overexpressing none of the downstream enzymes suppressed the detrimental effects of HEM4 overexpression and sampangine treatment (data not shown). We speculate that Hem4 hyperactivation might somehow cause irreversible damage to the pathway. This might help to explain the puzzling observation that heterozygous diploid mutants with mutations of Hem1 and Hem2, which lie upstream to Hem4 in the heme synthesis pathway, are hypersensitive to sampangine (1). Possibly, reducing the gene dosage of the upstream components in these mutants has little effect on Hem4 activation by the drug. However, once Hem4 is hyperactivated and the pathway is damaged, reducing the flux of the pathway further exacerbates the defects in heme synthesis.

Given that sampangine inhibits heme biosynthesis in both yeast and human cells and possibly other organisms, it might lack the needed specificity as a therapeutic agent against human pathogens and cancer cells. However, high specificity could still be achieved by exploiting unique genetic interaction networks within particular pathogens or cancer cell types. The list of sampangine-hypersensitive mutations identified in this study could aid in identifying pathogens or human cancer types that might be particularly sensitive to this drug and be a guide for selecting drug combinations that exhibit higher therapeutic potency and/or broader therapeutic indexes. Regardless of these, sampangine could serve as a chemical tool for investigating the physiological functions of heme metabolism in various biological systems and for studying the molecular mechanisms of human diseases related to heme deficiency, because its effect in yeast was largely mimicked by that of heme deficiency (Fig. 3A; see Table S1 in the supplemental material). However, for such applications, one should take into consideration possible off-target effects of sampangine because heme deficiency is probably not the only biological effect of sampangine treatment in yeast. This is because exogenous hemin partly rescued the growth inhibition caused by deleting heme biosynthesis genes yet completely failed to suppress sampangine's lethal effects at high concentrations (Fig. 3D). Possibly, inhibition of the heme biosynthesis pathway by sampangine confers additional cellular defects that cannot be overcome simply by supplying cells with heme. For example, the accumulation of uroporphyrinogen III and potentially increased levels of other metabolic products derived from uroporhyrinogen III might also contribute to sampangine's cytotoxic effect. It is also possible that, in addition to heme biosynthesis, sampangine directly inhibits an additional yet-to-be discovered pathway.

Our study also revealed a surprising synthetic lethality interaction relationship between mutations in the mitochondrial ATP synthase and those in the ETC (Fig. 2B and C; see Table S2 in the supplemental material). Mitochondrial ATP synthase and ETC are together required for producing ATP via respiration, which is thought to be dispensable for yeast cell survival under normal growth conditions because this organism gets enough energy from glycolysis alone. We also found that the lethality of an atp1Δ cyc3Δ double mutant was not suppressed by exogenously supplied ATP (data not shown), even though the same amount of ATP partly suppressed the sampangine hypersensitivity of an atp1Δ mutant (Fig. 1D). NAC, when used alone or together with ATP, also failed to restore growth in the atp1Δ cyc3Δ double mutant (data not shown), suggesting that the lethality of this mutant simply due to decreased ATP production and increased ROS levels is unlikely. Therefore, in addition to ATP production, mitochondrial ATP synthase and the ETC together likely play an additional undefined role that is essential for yeast cell survival. Understanding the molecular mechanism of the lethality observed in the atp1Δ cyc3Δ double mutant could provide insights into such a novel function of mitochondrial ETC and ATP synthase.