Heme bioavailability and signaling in response to stress in yeast cells

Protoheme (hereafter referred to as heme) is an essential cellular cofactor and signaling molecule that is also potentially cytotoxic. To mitigate heme toxicity, heme synthesis and degradation are tightly coupled to heme utilization in order to limit the intracellular concentration of “free” heme. Such a model, however, would suggest that a readily accessible steady-state, bioavailable labile heme (LH) pool is not required for supporting heme-dependent processes. Using the yeast Saccharomyces cerevisiae as a model and fluorescent heme sensors, site-specific heme chelators, and molecular genetic approaches, we found here that 1) yeast cells preferentially use LH in heme-depleted conditions; 2) sequestration of cytosolic LH suppresses heme signaling; and 3) lead (Pb2+) stress contributes to a decrease in total heme, but an increase in LH, which correlates with increased heme signaling. We also observed that the proteasome is involved in the regulation of the LH pool and that loss of proteasomal activity sensitizes cells to Pb2+ effects on heme homeostasis. Overall, these findings suggest an important role for LH in supporting heme-dependent functions in yeast physiology.

Protoheme (iron protoporphyrin IX or heme b) is an essential cofactor and signaling molecule that is also potentially cytotoxic (1)(2)(3)(4). The molecules and mechanisms cells employ to utilize protoheme, while mitigating its inherent toxicity, are complex and poorly understood, especially during stress (1,2). Cells manage protoheme, which is hereafter referred to as heme, by controlling both its total concentration and bioavailability (1,2). Heme concentration is primarily governed by the relative rates of heme synthesis and degradation, which are well-understood processes (1,2). Indeed, all of the enzymes involved in eukaryotic heme synthesis and degradation have been structurally characterized to atomic resolution, and the mechanisms of action have been largely delineated (1)(2)(3)(4).
On the other hand, the factors that control heme bioavailability are poorly understood (1,2). Bioavailable heme can be operationally defined as a pool of chelatable, kinetically labile heme that can readily exchange between biomolecules and is accessible for heme-dependent processes. The identity of factors that buffer and traffic labile heme (LH) 3 and the mechanisms employed to mobilize it for utilization are largely unknown (1,2).
To mitigate heme toxicity, it is generally thought that heme is made on demand (i.e. heme synthesis and utilization are tightly coupled), and heme in excess of that required for metabolism is degraded (4 -6). As a consequence, it is unclear what role, if any, steady-state LH plays in supporting heme-dependent functions. Using genetically encoded ratiometric fluorescent heme sensors to probe LH and heme homeostatic mechanisms, we have previously demonstrated the existence of a highly dynamic steady-state pool of LH in the cytosol, spanning 20 -40 nM, and identified new heme trafficking factors (e.g. glyceraldehyde-3phosphate dehydrogenase (GAPDH)) and signaling molecules that mobilize LH (e.g. nitric oxide (NO)) (7). However, the physiological importance of LH remained unclear.
In the current work, using Saccharomyces cerevisiae as a model eukaryote and heme sensors and chelating agents, we have established the functional importance of LH in regulating heme signaling, demonstrated that LH is preferentially consumed when cells become heme depleted, and discovered that certain xenobiotics (e.g. lead ions (Pb 2ϩ )) can, rather paradoxically, both deplete total cellular heme, primarily through its inhibition of heme synthesis, and increase labile bioavailable heme. We further find that the proteasome is involved in the regulation of labile heme and its response to Pb 2ϩ stress. In total, our results establish a functional role for LH, indicate that total and labile heme pools can be decoupled in response to certain stressors, and provide evidence for heme-based signaling in response to heavy metal stress.

Sequestering labile heme impacts heme signaling
We first sought to determine whether cytosolic LH serves a functional role in heme signaling. Toward this end, we developed an approach to sequester LH and probe its effects on the heme-regulated transcription factor Hap1 (8 -11). To chelate cytosolic LH, we induced the expression of a high-affinity hemoprotein, cytochrome b 562 (Cyt b 562 ) (12-16), using a galactose (GAL)-inducible promoter (pGAL) (17). Cyt b 562 binds ferric heme at pH 7.0 with a dissociation constant (K D III ) of 10 nM (16). Given the reduction potential (E M ) of heme-Cyt b 562 , ϩ200 mV versus NHE at pH 7.0 (18), and the E M of free heme, Ϫ38 mV versus NHE (19,20), one can estimate the ferrous heme dissociation constant (K D II ) to be ϳ1 pM by completing a thermodynamic cycle (19).
First, we confirmed that Cyt b 562 overexpression could sequester heme by measuring changes in LH with the genetically encoded ratiometric fluorescent heme sensor, HS1-M7A (7). HS1 is a tridomain fusion protein consisting of a hemebinding domain, the His/Met coordinating 4-␣-helical bundle Cyt b 562 , fused to a pair of fluorescent proteins, enhanced GFP (eGFP) and mKATE2, that exhibit heme-sensitive and -insensitive fluorescence, respectively (7) (Fig. 1A). Heme binding to the Cyt b 562 domain results in the quenching of eGFP fluorescence via resonance energy transfer but has little effect on mKATE2 fluorescence (7). Thus, the ratio of eGFP fluorescence (excitation 488 nm, emission 510 nm) to mKATE2 fluorescence (excitation 588 nm, emission 620 nm) provides a readout of cellular heme independently of sensor concentration, with the eGFP/mKATE2 ratio inversely correlating with heme binding to the sensor (7). The high affinity of HS1 for ferric and ferrous heme, K D III ϭ 10 nM and K D II Ͻ 1 nM at pH 7.0, renders it fully saturated with heme in WT cells (7). On the other hand, a variant of HS1, HS1-M7A, in which the heme-coordinating Met ligand is mutated to Ala, exhibits ferrous and ferric affinities of K D II ϭ 25 nM and K D III ϭ 2 M at pH 7.0 and is 20 -50% bound in the yeast cytosol (7). The observed sensor eGFP/ mKATE2 fluorescence ratio (R expt ) can be used to determine the fractional heme saturation of the sensor and the concentration of LH, assuming a 1:1 heme/sensor binding model and previously established sensor calibration protocols that involve determining the sensor ratio when the sensor is 100% (R max ) and 0% (R min ) bound to heme (see "Experimental procedures") (7). Notably, heme binding to the sensor is reversible, and expression of the heme sensor in cells does not itself perturb heme homeostasis or otherwise affect viability (7).
WT cells expressing heme sensor HS1-M7A and GAL-inducible Cyt b 562 (pGAL-Cyt b 562 ) were cultured in 2% raffinose (RAF) with or without 1.0% GAL for 16 h. As demonstrated in Fig. 1B, induction with GAL results in an increase in the eGFP/ mKATE2 fluorescence ratio, from 1.90 Ϯ 0.09 to 2.82 Ϯ 0.04, consistent with less heme binding to the sensor and diminished LH. By comparison, cells that express an empty pGAL vector (EV) or those that are heme-depleted with the heme biosynthetic inhibitor, succinylacetone (SA) (7,21,22), are unaffected by induction of Cyt b 562 . Using the aforementioned sensor calibration protocols, which are described in detail under "Exper-imental procedures," the induction of Cyt b 562 results in a decrease in the fractional saturation of the heme sensor from ϳ60 to ϳ35% heme-bound, which corresponds to a change in LH from 35 to 14 nM.
Having established that LH can be sequestered by overexpression of Cyt b 562 , we next sought to determine whether this Model derived from the X-ray structures of mKATE (Protein Data Bank code 3BXB) and CG6 (Protein Data Bank code 3U8P). B, cells expressing the heme sensor, HS1-M7A, and an allele of Cyt b 562 on a galactose (GAL)-inducible promoter (pGAL-Cyt b 562 ) or empty vector (pGAL-EV) were cultured in 2% RAF with or without 1.0% GAL or 500 M SA for 16 h in SCE medium. After growth, HS1-M7A sensor eGFP (excitation 488 nm, emission 510 nm) and mKATE2 (excitation 588 nm, emission 620 nm) fluorescence emission ratios were recorded. C, Hap1 activity was measured in cells expressing pGAL-Cyt b 562 or pGAL-EV and cultured in 2% RAF with or without 1.0% GAL or 500 M SA for 16 h in SC medium using a transcriptional reporter consisting of an allele of eGFP driven by the CYC1 promoter (pCYC1-eGFP), a Hap1 target gene. All data represent the mean Ϯ S.D. (error bars) of triplicate cultures, and the statistical significance was assessed using a two-sample t test. *, p Ͻ 0.01; **, p Ͻ 0.001.

Stress-induced heme signaling
would impact the activity of the heme-regulated transcription factor Hap1 (8 -11). Heme binding to Hap1 alters its ability to promote or repress transcription of a number of target genes, including CYC1, which Hap1 positively regulates (7)(8)(9)(10)(11). To probe Hap1 activity, we used a transcriptional reporter that employs the promoter of a Hap1 target gene, pCYC1, driving the expression of eGFP (7). As demonstrated in Fig. 1C, not only does heme depletion with SA decrease eGFP fluorescence, as expected, but Cyt b 562 induction with GAL also results in a decrease in eGFP fluorescence, both of which are consistent with reduced heme binding to Hap1 and diminished transcriptional activation of CYC1. Altogether, our results strongly suggest that overexpression of Cyt b 562 can sequester LH, and this results in diminished heme signaling and Hap1 activity.

LH is preferentially consumed relative to total heme during heme depletion
We next sought to determine whether LH is consumed preferentially relative to total heme during conditions of heme deficiency to ascertain whether LH is mobilized for heme-dependent functions when cells are confronted with defects in heme synthesis. Toward this end, we titrated the heme biosynthetic inhibitor SA in WT cells expressing the high-affinity heme sensor, HS1, or the medium-affinity sensor, HS1-M7A, and measured total and labile heme (Fig. 2). Titration of SA over a broad concentration range, up to 500 M, results in the depletion of both total (Fig. 2C) and labile heme (Fig. 2, A and B) as expected. However, most interestingly, LH is more sensitive than total heme to SA-mediated heme depletion (Fig. 2). When total heme is only modestly depleted by 25% using a relatively low 25 M dose of SA (Fig. 2C), there is a much larger diminution of LH, as measured by HS1-M7A and HS1 (Fig. 2, A and B); heme loading of the medium affinity sensor, HS1-M7A, shifts from ϳ26% bound to ϳ0% bound, and the high-affinity sensor, HS1, shifts in heme loading from ϳ100% bound to ϳ60% bound. In terms of LH concentration, based on the ferrous heme dissociation constants of HS1-M7A, K D II ϭ 25 nM (7), and HS1, assumed to be K D II ϳ 1 pM based on the estimated K D II for Cyt b 562 , we would estimate LH decreases from ϳ10 nM to Ͻ1 nM (likely ϳ2 pM based on the predicted K D II for HS1) in response to an SA concentration that modestly depletes total heme by 25%. Taken together, these data strongly suggest that heme deficiency mobilizes labile heme to nonexchangeable, and likely higher-affinity and/or buried, heme-binding sites.

Pb 2؉ depletes total heme but increases labile heme and heme-mediated signaling
A number of xenobiotics and environmental toxins negatively impact heme homeostasis (23), including N-methylprotoporphyrins (24,25), certain alkylating agents (e.g. 2-allyl-2isopropylacetamide) (26), various polyhalogenated biphenyls (e.g. tetrachlorodibenzo-p-dioxin) (27,28)), and heavy metals (29 -34). Pb 2ϩ , in particular, is a major public health concern, given the ubiquity of Pb 2ϩ -based paints, piping, and munitions (34). Among other targets, Pb 2ϩ is well-known to inhibit heme biosynthetic enzymes aminolevulinic acid (ALA) dehydratase (ALAD) and ferrochelatase, resulting in defects in heme synthesis, which in turn leads to anemia, cognitive decline, and other health problems (34). Further, Pb 2ϩ and other heavy metals are known to induce heme oxygenase (HO) (35,36), which may also contribute to decreased heme levels. However, whereas much is known about the role of Pb 2ϩ in impacting heme synthesis and degradation, virtually nothing is known about the effects of Pb 2ϩ on bioavailable labile heme. Given our Figure 2. LH is more sensitive to heme depletion using the heme biosynthetic inhibitor, SA, than total heme. Labile heme was measured in HS1-M7A-expressing (A) or HS1-expressing (B) cells cultured in SCE medium for 16 h with the indicated concentration of SA before measurement of the eGFP/ mKATE2 fluorescence ratios. C, total heme was measured in the cultures depicted in A or in heme-deficient hem1⌬ cells. All data represent the mean Ϯ S.D. (error bars) of triplicate cultures, and the statistical significance was assessed using a two-sample t test. Black asterisks, statistical significance between the indicated pairwise comparisons of conditions; red asterisks, statistical significance relative to 0 M SA. *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.001; n.s., not significant.

Stress-induced heme signaling
findings that LH is important for maintaining heme signaling to Hap1 and is preferentially consumed relative to total heme during heme deficiency, we sought to determine the effects of Pb 2ϩ on LH and heme-based signaling. Toward this end, we established a yeast model of Pb 2ϩ toxicity and probed the effects of Pb 2ϩ on total heme, LH, and heme signaling.
Due to the insolubility of Pb(NO 3 ) 2 in standard yeast growth media, we employed a Pb 2ϩ toxicity model similar to what was previously described in which yeast cells are subjected to an acute 3-h exposure to varying concentrations of Pb 2ϩ in 10 mM MES, pH 6.0, buffer, a medium in which Pb(NO 3 ) 2 is soluble and Pb 2ϩ is not complexed by the buffer (37). Following Pb 2ϩ exposure, cells were washed and allowed to recover for 4 h in an appropriate synthetic complete (SC) drop-out medium before subsequent analyses. Cell viability was measured by diluting cells into SC medium after the recovery period and monitoring growth for 20 h. Cell viability as measured by outgrowth is virtually identical to viability measurements using FUN-1, a fluorescent dye that exhibits red punctate emission in the vacuoles of metabolically active live cells and diffuse green cytosolic emission in dead cells (Fig. S1, A-C) (37). As shown in Fig. 3A, yeast viability decreases with increasing [Pb 2ϩ ]. The concentration of Pb 2ϩ that inhibits growth by 50%, lethal dose 50 (LD 50 ), varied between 25 and 500 M over the course of our studies with different batches of medium for reasons that are not completely understood. As a consequence, all of our analyses were done at the Pb 2ϩ LD 50 dose and not necessarily at a constant Pb 2ϩ concentration.
At the Pb 2ϩ LD 50 dose, elemental analysis using total reflection X-ray fluorescence (TXRF) spectroscopy confirms a significant enrichment of Pb 2ϩ in yeast cells (Fig. 3B). In addition, a host of other bio-elements are impacted as a consequence of Pb 2ϩ toxicity; phosphorus, sulfur, potassium, iron, and zinc are diminished, whereas calcium and copper are increased (Fig.  S2). These effects are consistent with prior studies in a number of organisms and cell types demonstrating the impact of Pb 2ϩ on various aspects of metal (37)(38)(39)(40)(41)(42), phosphate (43,44), and sulfur homeostasis (45)(46)(47).
Interestingly, whereas total heme is depleted due to Pb 2ϩ exposure, as expected due to its known effects on inhibiting heme synthesis (34) and up-regulating heme degradation (35,36), LH is completely maintained and, in fact, increases by 2-fold (Fig. 3C). In striking contrast, a dose of SA that depletes total heme to a similar degree as Pb 2ϩ treatment attenuates both LH and total heme (Fig. 3C). Titration of Pb 2ϩ over a broad concentration range indicates a dose-dependent increase in LH (Fig. 4A), decrease in total heme (Fig. 4B), and decrease in cell viability (Fig. 4C). The Pb 2ϩ -dependent depletion of total heme and increase in LH primarily occurs after the recovery period in SC medium; measurement of total heme and LH after the 3-h Pb 2ϩ exposure in MES buffer has minimal effects on LH (Fig. 4D) and total heme ( Fig. 4E) relative to cells allowed to Figure 3. Labile heme and heme signaling is increased, but total heme is attenuated in response to Pb 2؉ toxicity. A, yeast cell viability, as measured by solution turbidity at an optical density (O.D.) of 600 nm, is diminished by exposure to Pb 2ϩ in a dose-dependent manner. B, at the LD 50 dose of Pb 2ϩ (500 M), cells hyperaccumulate up to 10 mM Pb 2ϩ , as measured by TXRF. C, a dose of 500 M Pb 2ϩ diminishes total heme to levels similar to 500 M SA, but Pb 2ϩ increases labile heme 2-fold, whereas SA does not. D, Hap1p activity in response to heme depletion by SA or Pb 2ϩ is measured using the pCYC1-eGFP Hap1 reporter construct (pCYC1). To control for the Hap1-independent effects of Pb 2ϩ on eGFP expression, we measured eGFP fluorescence in response to Pb 2ϩ or SA using an allele of eGFP driven by the heme/Hap1-independent promoter, GPD (pGPD). The ratio of pCYC1 to pGPD eGFP expression (pCYC1/pGPD) is a measure of heme/Hap1-specific activation of CYC1. All data represent the mean Ϯ S.D. (error bars) of triplicate cultures, and the statistical significance relative to untreated cells was assessed using a two-sample t test. *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.001; n.s., not significant.

Stress-induced heme signaling
recover in SC medium (Fig. 4, A and B). The changes in HS1-M7A sensor ratio are not due to artifacts associated with Pb 2ϩdependent changes in sensor expression, given that the emission of mKATE2, the heme-insensitive fluorophore, is constant over a wide range of Pb 2ϩ doses (Fig. S3A). Furthermore, a variant of HS1 that cannot bind heme, HS1-M7A,H102A, which has the Met and His heme-coordinating residues mutated to Ala, does not exhibit Pb 2ϩ -dependent changes in fluorescence ratio (Fig. S3B). Altogether, these data indicate that the total and labile heme pools can be acted upon independently of each other in response to stress, and metabolically active cells are required for the observed changes in Pb 2ϩ -dependent LH and total heme. We next addressed whether the increase in LH in response to Pb 2ϩ translates to changes in heme signaling. Toward this end, we measured Hap1 activity using the pCYC1-eGFP transcrip- . Pb 2؉ only perturbs total and labile heme after a recovery period following exposure to Pb 2؉ . In cells expressing HS1-M7A, Pb 2ϩ increases labile heme (A), decreases total heme (B), and diminishes cell viability (C) in a dose-dependent manner if cells are allowed to recover for 4 h in SCE medium following Pb 2ϩ exposure in MES buffer. However, labile (D) and total heme (E) in cells expressing HS1-M7A immediately after the exposure to Pb 2ϩ in MES buffer are not significantly affected. All data represent the mean Ϯ S.D. (error bars) of triplicate cultures, and statistical significance was assessed using a two-sample t test. Black asterisks, statistical significance between the indicated pairwise comparisons of conditions; red asterisks, statistical significance relative to 0 M Pb. *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.001; n.s., not significant.

Stress-induced heme signaling
tional reporter in cells conditioned with the LD 50 dose of Pb 2ϩ . To account for the effects of Pb 2ϩ on eGFP expression independently of Hap1 activation, we also expressed eGFP under control of the GPD promoter (pGPD), which is not a transcriptional target of Hap1. As shown in Fig. 3D, exposure to Pb 2ϩ results in a ϳ33% decrease in the eGFP fluorescence of the Hap1 reporter, whereas SA exposure results in a 6-fold decrease, despite a similar depletion in total heme.
Given that Pb 2ϩ also affects eGFP fluorescence independently of Hap1, as evidenced by a ϳ60% decrease in fluorescence of parallel cultures harboring the pGPD-eGFP construct, we normalized the fluorescence from pCYC1-driven expression of eGFP to the fluorescence from pGPD-driven expression of eGFP, giving a pCYC1/pGPD ratio. Exposure to Pb 2ϩ results in a nearly 2-fold increase, from 0.16 to 0.26, in the pCYC1/ pGPD ratio of eGFP fluorescence, an indicator of Hap1/ pCYC1-specific activation. In striking contrast, cells conditioned with a dose of SA that results in a similar concentration of intracellular heme as the LD 50 dose of Pb 2ϩ exhibit a nearly ϳ10-fold decrease in the pCYC1/pGPD ratio of eGFP fluorescence, with a nearly 10-fold decrease in fluorescence of the pCYC1-eGFP construct and minimal perturbation to fluorescence from the pGPD-eGFP construct.
Taken together, at minimum, these results suggest that, in response to Pb 2ϩ toxicity, Hap1 activity is largely maintained, despite a Ͼ10-fold decrease in total heme. At maximum, these results suggest that "heme-specific" Hap1 activity, as recorded from the pCYC1/pGPD ratio, actually increases in response to Pb 2ϩ stress. The maintenance of or increase in Hap1 activity, depending on the interpretation of the data, is presumably due to the Pb 2ϩ -induced increase in LH.

Pb 2؉ -mediated heme depletion is largely due to a block in heme synthesis
We next sought to determine the mechanism by which intracellular heme is depleted in response to Pb 2ϩ toxicity. The intracellular concentration of heme is governed by the relative rates of heme synthesis and degradation or export. To parse apart the effects of Pb 2ϩ on these opposing processes, we tested the effects of Pb 2ϩ on cells that had a defect in the ability to synthesize heme. First, we determined that Pb 2ϩ -dependent heme depletion occurs within the first 1.5 h of the 4-h recovery period in SC medium (Fig. 5A, 0 M SA). Second, we found that in cells that had a defect in heme synthesis due to conditioning with 500 M SA, Pb 2ϩ had an attenuated effect on the loss of total heme. For instance, after 2.7 h, cells that could biosynthesize heme (Fig. 5A, 0 M SA) exhibited a 2-fold and 6-fold decrease in total heme when treated with 125 and 250 M Pb 2ϩ , respectively. On the other hand, in cells with ablated heme biosynthesis (Fig. 5A, 500 M SA), both 125 and 250 M Pb 2ϩ resulted in a relatively modest ϳ30% decrease in total heme. The diminished effect of Pb 2ϩ on total heme in cells conditioned with the heme biosynthetic inhibitor SA suggested that Pb 2ϩ depletes total heme by largely suppressing heme biosynthesis.
We next sought to further validate the observation that Pb 2ϩ depletes heme by primarily affecting heme synthesis and not heme degradation or export. Toward this end, we utilized a hem1⌬ strain, which lacks the first enzyme in the heme synthesis pathway, ALA synthase (ALAS), to test the effects of Pb 2ϩ on the degradation of mitochondrially derived de novo synthesized heme or exogenously supplied heme. hem1⌬ cells can only acquire heme by stimulating mitochondrial heme synthesis via supplementation with ALA, the product of ALAS, or by heme uptake via hemin supplementation. We supplemented hem1⌬ cells with ALA or heme during a 16-h preculture in SCE medium and subjected them to Pb 2ϩ exposure in MES buffer for 3 h, followed by a 4-h recovery in SCE medium and then analysis of total heme (Fig. 5B). WT and hem1⌬ cells supple- Figure 5. Pb 2؉ -dependent attenuation of total heme is largely dependent on heme synthesis. A, WT cells untreated (left) or treated (right) with 500 M SA and the indicated concentrations of Pb 2ϩ demonstrate that Pb 2ϩ has a larger effect on depleting total heme in cells that can properly synthesize heme. B and C, endogenous heme, but not exogenous heme, is degraded in a Pb 2ϩ -dependent manner. Total (B) and labile heme (C) were measured in HS1-expressing WT or hem1⌬ cells conditioned with the indicated Pb 2ϩ concentration and/or 200 M ALA or 50 M hemin chloride during the preculture and/or post-Pb 2ϩ recovery period. All data represent the mean Ϯ S.D. (error bars) of triplicate cultures, and statistical significance was assessed using a two-sample t test. In A, black asterisks represent the statistical significance between the indicated pairwise comparisons of conditions, and red asterisks represent the statistical significance relative to 0 M Pb for each time point. In B, black asterisks represent the statistical significance relative to 0 M Pb for each test condition. In C, the black asterisks represent the statistical significance between the indicated pairwise comparisons of strains. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant.

Stress-induced heme signaling
mented with ALA both in the preculture and during the recovery accumulated similar amounts of intracellular heme, underwent a Pb 2ϩ -dependent depletion of total heme (Fig. 5B, black and green columns), and exhibited Ͼ80% heme loading of the high-affinity heme sensor HS1 (Fig. 5C, black and green columns). hem1⌬ cells treated with ALA only during the preculture, but not during the recovery, also exhibited a Pb 2ϩ -dependent attenuation in intracellular heme (Fig. 5B, red columns). However, the total amount of heme was ϳ5-fold lower (Fig. 5B, red columns), and the heme-loading of HS1 was considerably reduced, ϳ10% bound, compared with hem1⌬ cells supplemented with ALA in both the preculture and the recovery medium (Fig. 5C, red columns), presumably because ALA was limiting because it is not supplied during the recovery phase. In striking contrast to ALA supplementation, hem1⌬ cells supplemented with exogenous heme during the preculture, which contributed to a 2-fold increase in intracellular heme relative to ALA supplemented hem1⌬ cells, did not exhibit the Pb 2ϩ -dependent depletion of heme (Fig. 5B, yellow columns). Exogenously supplied heme is still available to bind HS1, and in fact, its fractional saturation is similar to ALA-treated cells after normalizing for the change in total intracellular heme concentration (Fig. 5C, yellow columns). The observation that Pb 2ϩ depletes endogenously synthesized heme and not exogenously supplied heme strongly suggests that Pb 2ϩ primarily inhibits heme synthesis, with minor effects on heme degradation.
Given that heme synthesis is O 2 -dependent and our finding that Pb 2ϩ largely inhibits heme synthesis, we next sought to test our prediction that Pb 2ϩ should have minimal impact on total heme in the absence of O 2 . To test the O 2 dependence of Pb 2ϩinduced heme depletion, we subjected aerobically grown cells to Pb 2ϩ exposure in MES buffer in an anoxic or normoxic environment, followed by recovery in anoxic or normoxic culture conditions and measurement of total heme, LH, and Pb 2ϩ toxicity (Fig. 6). Quite strikingly, despite accumulating similar levels of Pb 2ϩ between normoxic and anoxic Pb 2ϩ exposures (Fig.  S4), in the absence of O 2 , cells do not exhibit Pb 2ϩ -dependent heme depletion (Fig. 6A). On the other hand, the increase in LH in response to Pb 2ϩ is similar between both normoxic and anoxic Pb 2ϩ exposures (Fig. 6B). Notably, Pb 2ϩ -mediated cell toxicity is entirely O 2 -dependent, given that viability is not affected in anoxic cultures (Fig. 6C). Taken together, our data suggest that Pb 2ϩ -induced depletion of total heme only occurs in cells that are synthesizing heme de novo, further supporting the notion that Pb 2ϩ largely inhibits heme synthesis.
We next sought to determine whether a block in heme degradation or export could preserve total heme in the face of Pb 2ϩ toxicity and affect labile heme. One established mechanism of heme degradation is via HO, an enzyme that oxidatively degrades heme into bilirubin, carbon monoxide (CO), and ferrous iron (Fe 2ϩ ) (36). To test the role of HO in Pb 2ϩ -dependent heme depletion, we compared WT and hmx1⌬ cells, which lack heme oxygenase 1 (48,49), for Pb 2ϩ -dependent heme depletion. As shown in Fig. 7A, in the absence of Pb 2ϩ , hmx1⌬ cells exhibit a ϳ25% increase in total heme, consistent with prior studies in yeast and the known role of Hmx1 in heme degradation (48). However, both WT and hmx1⌬ cells exhibit the Pb 2ϩinduced depletion of heme to similar degrees, indicating that HO does not play a role in the loss of heme during Pb 2ϩ toxicity. Both WT and hmx1⌬ cells exhibited a Pb 2ϩ -dependent increase in LH (Fig. 7B). For reasons that are not entirely clear at this time, hmx1⌬ cells appeared to be more resistant to Pb 2ϩ toxicity (Fig. 7C). In contrast, prior work has demonstrated that Hmx1p confers resistance to various oxidative insults, includ-
An alternative mechanism for cellular heme depletion is heme export. Many metazoans express heme exporters (e.g. FLVCR1) to aid in heme detoxification. S. cerevisiae expresses a porphyrin-heme exchanger, PUG1, which uptakes protoporphyrin IX and expels heme (50). To test the role of heme export via Pug1 during Pb 2ϩ toxicity, we determined the extent to which heme is depleted in WT and pug1⌬ cells exposed to increasing doses of Pb 2ϩ . As with hmx1⌬ cells, pug1⌬ cells exhibit a similar degree of Pb 2ϩ -induced heme depletion as WT cells (Fig. 7D), indicating that Pug1 does not affect the loss of heme during Pb 2ϩ toxicity. Both WT and pug1⌬ cells exhibit an increase in LH in response to Pb 2ϩ (Fig. 7E) and similar sensitivities to Pb 2ϩ toxicity (Fig. 7F).
Altogether, our data indicate that 1) Pb 2ϩ depletes total heme primarily through its ability to inhibit heme synthesis, and 2) heme degradation and export pathways do not affect Pb 2ϩ -dependent changes in total and labile heme.

Pb 2؉ -dependent increases in labile heme correlate with protein degradation
Most heme is associated with nonexchangeable binding sites in high-affinity hemoproteins. As such, we predicted that the increase in labile heme in response to Pb 2ϩ may be associated with the degradation of hemoproteins and the release of heme into the labile heme pool. To test this hypothesis, we conducted one-dimensional PAGE analysis of cells conditioned with and without Pb 2ϩ at the LD 50 dose immediately after the 3-h exposure to Pb 2ϩ in MES buffer and after the 4-h recovery phase  (Fig. 8A). We found that immediately following exposure to Pb 2ϩ , protein expression is unaffected, whereas after the 4-h recovery, a number of proteins are degraded (Fig. 8A). The diminished protein expression during the recovery phase correlates with when we observe increased LH, suggesting that protein turnover and an increase in LH are linked. Notably, using tandem MS, we found that GAPDH, a component of the LH buffer (7), is retained during Pb 2ϩ exposure (Fig. 8, A and B, and Table S2). Another glycolytic protein, enolase, is also maintained during Pb 2ϩ exposure. On the other hand, Pb 2ϩ has the capacity to degrade high-affinity hemoproteins. Ctt1, which is a heme-containing catalase enzyme, is degraded in a Pb 2ϩ -dependent manner (Fig. 8B). Altogether, our data are consistent with a model in which Pb 2ϩ -induced degradation of hemoproteins releases heme that contributes to the LH pool.

Stress-induced heme signaling
To test the hypothesis that Pb 2ϩ -dependent protein turnover releases heme from hemoproteins and increases LH, we assayed Pb 2ϩ -dependent changes in LH in yeast mutants defective in various protein degradation pathways, including vacuolar, autophagic, and proteasomal degradation. We found that a  (Table S2). B, the expression and activity of the high-affinity hemoprotein Ctt1p, a heme-catalase enzyme, is downregulated in response to Pb 2ϩ conditioning, whereas GAPDH expression, a constituent of the labile heme buffer, is maintained. All gels are representative of at least three independent cultures. C and D, effects of atg1⌬ (C), rpn10⌬ (C), and pep4⌬ (D) deletion on Pb 2ϩ -dependent changes in labile heme were assessed. E and F, Pb 2ϩ -dependent changes in total heme (E) and growth (F) were assessed in WT and rpn10⌬ cells. All labile and total heme measurements represent the mean Ϯ S.D. (error bars) of triplicate cultures, whereas the growth data represent the mean Ϯ S.D. of duplicate cultures. Statistical significance was assessed using a two-sample t test. Black asterisks, statistical significance between the indicated pairwise comparisons of conditions; red asterisks, statistical significance relative to 0 M Pb. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant; n.d., not detectable.

Stress-induced heme signaling
proteasome mutant, rpn10⌬ (51), exhibited Pb 2ϩ -dependent and -independent changes in LH (Fig. 8C), whereas mutants defective in vacuolar (pep4⌬) (52) (Fig. 8D) or autophagic (atg1⌬) (53) (Fig. 8C) pathways did not have altered LH. Under nonstressed conditions, WT cells exhibited an LH concentration of ϳ5 nM, corresponding to an HS1-M7A fractional saturation of 17%. On the other hand, HS1-M7A was ϳ0% bound to heme when expressed in rpn10⌬ cells, which corresponds to an LH concentration of Ͻ1 nM. However, Pb 2ϩ induced a greater increase in LH (Fig. 8C) and decrease in total heme (Fig. 8E) in rpn10⌬ cells relative to WT. Moreover, the growth of rpn10⌬ cells was more sensitive to Pb 2ϩ toxicity than WT (Fig. 8F). In total, these results indicate that the proteasome positively regulates LH, and loss of proteasomal function sensitizes cells to Pb 2ϩ -dependent effects on heme homeostasis.

Discussion
Heme synthesis and degradation are tightly coordinated so as to minimize the accumulation of potentially cytotoxic "free" or labile heme (LH) (4 -6). As a consequence, the concentration and physiological role of LH in biology has been controversial (1, 2). Whereas estimates for the concentration of LH across various cell types have spanned subpicomolar to micromolar quantities (1, 2, 5, 54, 55), the use of genetically encoded heme sensors in yeast and various nonerythroid human cells lines has established a consensus range for cytosolic LH that spans 10 -100 nM, representing up to 10% of the total heme concentration (7,56). However, few studies have directly probed the contribution of steady-state LH as a heme source for hemoproteins and heme signaling. Herein, we established a functional role for LH in regulating heme signaling and demonstrated that LH is preferentially consumed relative to total heme when cells become heme-depleted.
We find that modestly depleting total heme by 25% with succinylacetone, an inhibitor of the second enzyme in the heme biosynthetic pathway, ALAD, results in a much larger perturbation to LH, depleting it more than 10-fold, from 10 nM to Ͻ1 nM, possibly as low as ϳ2 pM (Fig. 2). The sensitivity of the LH pool toward heme depletion relative to total heme is an indication that LH is mobilized to support heme-dependent processes when cells are confronted with diminished heme synthesis. In other words, LH re-equilibrates with other high-affinity and/or poorly exchangeable heme binding sites that may become vacant when heme is depleted. These results are consistent with prior pulse-chase experiments that utilized radiolabeled ALA and/or heme sources to demonstrate that P450 enzymes equilibrate with LH (23). Altogether, these observations may have significant implications for conceptualizing new heme-based therapies for porphyrias, a family of inherited disorders associated with defects in heme biosynthesis. For instance, treatment methods designed to increase LH via supplementation with appropriate heme complexes may form the basis for alleviating the symptoms of heme scarcity (57).
To address the role of LH in heme-dependent functions, we induced a site-specific heme chelator in the cytosol via the overexpression of Cyt b 562 and assessed its effect on heme signaling through the yeast heme-regulated transcription factor Hap1 (Fig. 1). We found that sequestration of LH inhibited Hap1 activity, demonstrating that LH can control heme-signaling processes. Importantly, these results indicate that increasing heme synthesis, a metabolically demanding process, is not the only means to activate heme signaling, as was previously suggested for activation of Hap1 (10,58) and metabolic cycling in yeast (59,60), as well as heme regulation of the circadian clock through the nuclear receptors, Rev-erb-␣ and Rev-erb-␤, in mammals (61).
To understand the relationship between total heme, LH, and heme signaling during adaptation to cellular stress, we subjected yeast cells to an environmentally relevant toxicant well known to affect heme homeostasis, Pb 2ϩ (34). In yeast, Pb 2ϩ toxicity suppresses metabolic activity and proliferation through a mechanism that requires new protein synthesis (37). In addition, mitochondria are a major target of Pb 2ϩ toxicity and are the source of Pb 2ϩ -induced reactive oxygen species production (62). Heavy metal sequestration by vacuoles and binding to GSH and metallothioneins are important detoxification pathways for Pb 2ϩ (63).
A number of previously established yeast models for Pb 2ϩ toxicity employ exposures spanning 0.1-10 mM, which is high relative to the 240 nM [Pb 2ϩ ] threshold in patient blood samples used to initiate public health actions (64), the 1-5 M blood [Pb 2ϩ ] associated with inhibition of heme biosynthesis in humans (65,66), or the 0.1-250 M [Pb 2ϩ ] used in mammalian cell culture models of Pb 2ϩ toxicity (67,68). However, it is important to note that the biochemical features of Pb 2ϩ toxicity at these high concentrations in yeast phenocopy many of the hallmarks of Pb 2ϩ toxicity in mammalian cell lines at lower concentrations (37, 62, 63, 69 -71), including inhibition of heme synthesis (current work). The higher Pb 2ϩ exposure concentrations required for yeast may reflect differences in Pb 2ϩ uptake, efflux, or intracellular bioavailability.
Whereas Pb 2ϩ has the capacity to deplete cellular heme through its ability to inhibit two enzymes in the heme biosynthetic pathway (ALAD, also the target of SA, and ferrochelatase) and up-regulate the expression of the heme-degrading enzyme HO (34 -36), we found that, in yeast, Pb 2ϩ significantly attenuates total heme primarily due to the inhibition of heme synthesis (Figs. 5 and 6). However, rather surprisingly and paradoxically, we found that Pb 2ϩ significantly increases LH (Fig.  3C). Given our previous findings that thiol-specific alkylating agents impact the NO-mediated mobilization of LH (7) and the well-documented interactions between Pb 2ϩ and cysteine residues (72), we propose that Pb 2ϩ may liberate heme in cells from certain thiol-containing heme-binding sites. Alternatively, because Pb 2ϩ induces the degradation of a large fraction of the proteome, the increase in LH could simply be a result of heme that is released from hemoproteins that are being turned over. Consistent with this model, we found that a mutant that has a defect in proteasomal function, rpn10⌬, has less LH than WT cells (Fig. 8C). However, it is unclear why rpn10⌬ cells exhibit a greater magnitude increase in LH (Fig. 8C) and decrease in total heme (Fig. 8E) in response to Pb 2ϩ stress. One explanation that could account for the former is that Pb 2ϩ is directly or indirectly inducing the release of heme from a hemoprotein target that is stabilized due to the attenuation in proteasome function in rpn10⌬ cells. Given that previous studies

Stress-induced heme signaling
have demonstrated that Pb 2ϩ positively regulates the proteasome (73) and heme and ALAD act as negative regulators of the proteasome (74,75), there is a complex mosaic of competing effects that might account for the Pb 2ϩ -dependent changes in heme homeostasis in WT and rpn10⌬ cells. We are currently elucidating the molecular mechanisms underlying heme regulation of LH and the proteasome.
The Pb 2ϩ -induced increase in LH seems to impact heme signaling via Hap1 (Fig. 3D). Indeed, despite the attenuation in total heme concentration due to Pb 2ϩ toxicity (Fig. 3C), we still observe a level of Hap1 activity that is much greater relative to SA-mediated heme depletion (Fig. 3D), which attenuates both LH and total heme (Fig. 3C).
The observation of an increased LH pool in response to Pb 2ϩ may have implications for the pathology of Pb 2ϩ toxicity. For instance, the increase in LH in response to Pb 2ϩ may contribute to a cytotoxic heme pool (5, 6) that leads to the oxidative stress associated with Pb 2ϩ toxicity (76,77). Alternatively, given that there are a number of heme-regulated transcription factors, kinases, and ion channels (1,2), the increase in LH in response to Pb 2ϩ may be required to activate heme-based signaling pathways important for adaptation to heavy metal stress. The specific physiological consequences of increased LH in response to Pb 2ϩ stress remain to be fleshed out.
Altogether, our studies demonstrate the functional importance of LH in heme utilization, especially during stress associated with heme depletion and Pb 2ϩ toxicity. Our future work will involve probing the specific mechanisms by which LH can be coupled to heme utilization as well as the mechanisms underlying the Pb 2ϩ -mediated increase in LH and the physiological consequences of this action.

Labile heme and total heme depletion
To sequester labile heme in the cytosol, we generated a WT yeast strain expressing an episomal plasmid (p316-GAL) containing an allele of the high-affinity hemoprotein Cyt b 562 driven by the galactose-inducible promoter (pGAL); the plasmid is referred to as pGAL-Cyt b 562 . A control strain expressing the EV (pGAL-EV) was also generated. For labile heme measurements, the strains expressing pGAL-Cyt b 562 or pGAL-EV also co-expressed the previously described heme sensor, HS1-M7A, using an episomal plasmid (pRS415) that drives sensor expression with the GPD promoter (7). For Hap1 activity measurements, the strains expressing pGAL-Cyt b 562 or pGAL-EV also co-expressed the previously described pCYC1-eGFP Hap1 reporter, which is an episomal plasmid (pRS415) that drives eGFP expression using the CYC1 promoter, a transcriptional target of Hap1 (7). To induce Cyt b 562 expression, cells were cultured in SC-URA-LEU medium to maintain selection of both Cyt b 562 and the heme sensor, HS1-M7A, or pCYC1-eGFP. Instead of using 2% glucose, which will repress the expression of the GAL-inducible promoter, we cultured cells in 2% raffinose and either 1.0% galactose (inducing conditions) or vehicle (sterile water, noninducing conditions). Parallel control cultures were treated with 500 M SA to deplete intracellular heme. All cultures were seeded at an initial optical density of A 600 nm ϭ 0.005 and cultured until cells reached a final density of A 600 nm ϳ 1.0, which typically took 14 -16 h. Following growth, cells were harvested, washed, and resuspended in PBS, and sensor or eGFP fluorescence was measured as described under "Labile heme quantification" and "Hap1 activity." To deplete total heme, cells were cultured with the indicated concentrations of the heme biosynthetic inhibitor, SA, in an appropriate SCE drop-out medium. All cultures were seeded at an initial optical density of A 600 nm ϭ 0.005 and cultured until cells reached a final density of A 600 nm ϳ 1.0, which typically took 14 -16 h. Following growth, cells were harvested, washed, and processed for determination of labile or total heme as described under "Labile heme quantification" and "Total heme quantification."

Yeast model of Pb 2؉ toxicity
Due to the insolubility of Pb(NO 3 ) 2 in yeast medium, we subjected exponential phase yeast cells to varying doses of Pb 2ϩ in MES buffer for 3 h, a medium in which Pb(NO 3 ) 2 is soluble. Cells were precultured in an appropriate SC drop-out medium to a final density of A 600 nm ϳ 1.0. Following washing with sterile water, cells were resuspended in 10 mM MES buffer containing varying concentrations of Pb(NO 3 ) 2 at a density of 1 OD/ml and mixed every 15 min at 25°C for 3 h. Following exposure to Pb 2ϩ in MES buffer, cells were thoroughly washed with sterile water, resuspended in an appropriate SC drop-out mixture to a final density of A 600 nm ϳ 1.0, and allowed to recover for 4 h, while shaking at 220 rpm at 30°C. Cell viability was measured by diluting the cells to an initial density of A 600 nm ϭ 0.01 in an appropriate SC drop-out medium after the recovery phase, and solution turbidity was recorded by measuring A 600 nm after 20 h of growth shaking at 220 rpm and 30°C.
To assess the effects of Pb 2ϩ toxicity in an anaerobic environment, the Pb 2ϩ toxicity model described above was modified as follows: the Pb 2ϩ exposure in MES buffer was accomplished in degassed MES buffer with varying concentrations of Pb 2ϩ in a COY anaerobic chamber maintained with an inert atmosphere of 95% N 2 and 5% H 2 . Following Pb 2ϩ exposure,

Stress-induced heme signaling
cells were washed with sterile degassed water and allowed to recover in degassed SC medium, all anaerobically in the COY anaerobic chamber. Cell viability was measured by diluting the anaerobic cultures into in an appropriate SC drop-out medium after the recovery phase, and solution turbidity was recorded by measuring A 600 nm after 16 h of growth shaking at 220 rpm and 30°C in air.
All analytical analyses, including for labile or total heme, were conducted immediately after the 3-h Pb 2ϩ exposure in MES buffer (pre-recovery) or after the 4-h recovery in SC medium (post-recovery).

Viability measurements using FUN-1
Cells were grown as indicated under "Yeast model of Pb 2ϩ toxicity." After conditioning cells in MES buffer with or without Pb 2ϩ , 1 OD of cells were pelleted and washed once with 1 ml of sterile MilliQ water and once with 500 l of 10 mM HEPES with 2% glucose (w/v) (HG buffer). Cells were pelleted again, resuspended in 300 l of HG buffer, and then treated with 24.5 l of 200 M FUN-1 (Thermo Fisher Scientific) to a final concentration of 15 M. The 200 M FUN-1 working stock solution was prepared by diluting a 10 mM DMSO FUN-1 stock solution into HG buffer. Cells were allowed to incubate in the dark at 30°C for 30 min and then washed three times in HG buffer. Cells were imaged on glass slides with coverslips using the Cytation 3 imaging plate reader (Biotek) with eGFP and Texas Red Filter Cubes. In stained cells, the observation of red puncta was used to score viable cells, and the observation of diffuse green fluorescence was used to score dead cells. On average, ϳ100 cells were analyzed per sample.

Labile heme quantification
Measurements of labile heme were accomplished as described previously (7). For all heme sensor fluorescence measurements, following cell growth, cells were washed in water and resuspended in PBS at a density between 3 and 5 A 600 nm / ml, or 6 ϫ 10 7 to 1 ϫ 10 8 cells/ml. Fluorescence was recorded on a Synergy Mx multimodal plate reader using black Greiner Bio-one flat-bottom fluorescence plates. eGFP and mKATE2 fluorescence was recorded using excitation and emission wavelength pairs of 488 and 510 nm and 588 and 620 nm, respectively. Background fluorescence of cells not expressing the heme sensors was recorded and subtracted from the eGFP and mKATE2 fluorescence values. The sensor eGFP/mKATE2 fluorescence ratio (R expt ) is a qualitative indicator of labile or bioavailable heme, with a low ratio indicating a high concentration of labile heme and a high ratio indicating a low concentration of labile heme.
For quantitative labile heme monitoring, we can convert R expt values to the fractional heme saturation of the sensor (% heme bound) (Equation 1) or, if the sensor heme dissociation constant is known (K D ), the concentration of labile heme (Equation 2). Both metrics require that the eGFP/mKATE2 sensor fluorescence ratio is known when the sensor is 100% bound (R max ) or 0% bound (R min ) (7). Based on a 1:1 hemebinding model, the fractional saturation, % Bound, of the sensor can be calculated according to Equation 1 (7).
The labile heme concentration can be calculated according to Equation 2 (7,80).
R expt is the eGFP/mKATE2 fluorescence ratio under any given experimental conditions, R min is the eGFP/mKATE2 fluorescence ratio when 0% of the sensor is bound to heme, R max is the eGFP/mKATE2 fluorescence ratio when 100% of the sensor is bound to heme, F min mKATE2 is the mKATE2 emission intensity when 0% of the sensor is bound to heme, and F max mKATE2 is the mKATE2 emission intensity when 100% of the sensor is bound to heme. The F min mKATE2 /F max mKATE2 ratio is typically taken to be 1, given that mKATE2 fluorescence emission is not significantly perturbed upon heme binding to the sensor (7). Determination of R max and F max mKATE2 involves recording eGFP and mKATE2 fluorescence after digitonin permeabilization of cells and incubation with 50 M heme. Briefly, 3-5 A 600 nm /ml of cells are resuspended in PBS with 100 g/ml of digitonin, 1 mM ascorbate, and 50 M hemin chloride. After a 30-min incubation at 30°C, cells are harvested, washed, and resuspended in PBS buffer before recording of fluorescence. Given that the high affinity heme sensor, HS1, is quantitatively saturated with heme and its fluorescence properties are virtually identical to HS1-M7A, we can also determine R max and F max mKATE2 from parallel WT cultures expressing HS1 (7). Determination of R min and F min mKATE2 involves recording eGFP and mKATE2 fluorescence after cells are treated with the heme biosynthesis inhibitor SA (21) or from hem1⌬ cells cultured in parallel (7).

Total heme quantification
Measurements of total heme were accomplished using a fluorimetric assay designed to measure the fluorescence of protoporphyrin IX upon the release of iron from heme as described previously (81). For all total heme measurements, following cell growth, 2 ϫ 10 8 cells were harvested, washed in sterile water, and resuspended in 500 l of 20 mM oxalic acid and stored in a closed box at 4°C overnight (16 -18 h). Next, an equal volume (500 l) of 2 M oxalic acid was added to the cell suspensions in 20 mM oxalic acid. The samples were split, with half the cell suspension transferred to a heat block set at 95°C and heated for 30 min and the other half of the cell suspension kept at room temperature (ϳ25°C) for 30 min. All suspensions were centrifuged for 2 min on a table-top microcentrifuge at 21,000 ϫ g, and the porphyrin fluorescence (excitation 400 nm, emission 620 nm) of 200 l of each sample was recorded on a Synergy Mx multimodal plate reader using black Greiner Bio-one flat-bottom fluorescence plates. Heme concentrations were calculated from a standard curve prepared by diluting 500 -1500 M hemin chloride stock solutions in 0.1 M NaOH into MilliQ water, which was then added back to extra cell samples as prepared above. To calculate heme concentrations, the fluorescence of the unboiled sample (taken to be the background level of protoporphyrin IX) is subtracted from the fluorescence of the boiled sample (taken to be the free base porphyrin generated upon the release of heme iron). The cellular concentration of heme is determined by dividing the moles of heme deter-Stress-induced heme signaling mined in this fluorescence assay and dividing by the number of cells analyzed, giving moles of heme per cell, and then converting to a cellular concentration by dividing by the volume of a yeast cell, taken to be 50 fl (7). This fluorescence assay gives similar qualitative trends between samples as an HPLC assay for heme we employed previously (7), but the absolute concentrations tend to be consistently 3-5-fold higher (data not shown).

Hap1 activity
After growth, cells expressing p415-CYC1-eGFP, or eGFP driven by the Hap1-regulated CYC1 promoter, were washed in sterile water and resuspended in PBS to a concentration of 1 ϫ 10 8 cells/ml, and 100 l was used to measure eGFP fluorescence (excitation 488 nm, emission 510 nm). Background autofluorescence of cells not expressing eGFP was recorded and subtracted from the p415-CYC1-eGFP-expressing strains. To account for heme/Hap1-independent changes in eGFP expression/fluorescence, we also cultured cells expressing p415-GPD-eGFP, a plasmid expressing eGFP under control of the heme/ Hap1-independent GPD promoter.

Immunoblotting
After culturing, cells were harvested, washed in ice-cold Mil-liQ water, and lysed in two pellet volumes of phosphate buffer supplemented with protease inhibitors as described previously (7,82). Lysis was achieved at 4°C using one pellet volume of zirconium oxide beads and a bead beater (Bullet Blender, Next Advance) on a setting of 8 for 3 min (7). Lysate protein concentrations were determined by the Bradford method (Bio-Rad), and 14% Tris/glycine gels (Invitrogen) were employed for SDS-PAGE (7). ␣-GAPDH rabbit polyclonal antibodies (Genetex, GTX100118) and a goat ␣-rabbit secondary antibody conjugated to a 680-nm emitting fluorophore (Biotium) were used to probe for GAPDH. Yeast cytosolic catalase, Cttp1, was probed using a custom antibody generated by Genscript's custom antibody service (Poly Express Premium Service, SC1676). An ␣-Ctt1p antibody was raised in rabbits against an amino acid 1-320 fragment of Ctt1p. The ␣-Ctt1p antibody was validated in yeast by comparing immunoreactivity between WT, ctt1⌬, and cta1⌬ cells, the latter being a deletion mutant of a peroxisomal/mitochondrial catalase, Cta1p, unrelated to Ctt1p (Fig. S5). All gels were imaged on a LI-COR Odyssey IR imager (7,82).

Catalase activity
After culturing, cells were harvested and lysed in phosphate buffer, and 10 g of protein lysate were subjected to native PAGE on a 10% Tris/glycine gel (Invitrogen). After electrophoresis, an in-gel activity stain was utilized to measure catalase activity (7,83). Briefly, a catalase-staining solution containing 1 part dopamine (20 mg/ml) in pH 8 0.2 M KP i buffer, 1 part para-phenylenediamine (3.5 mg/ml) in pH 8 0.2 M KP i , 1 part 15% H 2 O 2 , and 2 parts DMSO were mixed in the order listed. The staining solution was added directly to the gel and allowed to stain for 2 min, followed by rinsing in MilliQ water and imaging.

Plasmids
All yeast expression plasmids used in this study are listed in Table S1 and were described previously, except for pDH039, the plasmid expressing cytochrome b 562 (Cyt b 562 ) driven by a GAL-inducible promoter, pGAL-Cyt b 562 . This plasmid was constructed by amplifying the coding sequence of Cyt b 562 from the HS1 sensor (7). The following primers were utilized to amplify a BamHI/XbaI fragment of Cyt b 562 : prDH025, 5Ј-CCTTTGGTGGCTCTGGATCCATGGCAGATCTGGAAG-ACAACATGG-3Ј; prDH026, 5Ј-GGTCAGTTTGCCACC-TCTAGATCATCTGTATTTCTGATGATATGC-3Ј.
Following amplification and digestion with BamHI and XbaI, the Cyt b 562 coding sequence was ligated into BamHI-and XbaI-digested p316-GAL1 (17). The amino acid sequence of Cyt b 562 is as follows, with the heme-coordinating residues in boldface type: ADLEDNMETLNDNLKVIEKADNAAQV-KDALTKMRAAALDAQKATPPKLEDKSPDSPEMKDFRHGF-DILVGQIDDALKLANEGKVKEAQAAAEQLKTTRNAYHQ-KYR.

TXRF spectroscopy
Elemental analysis of cells, and in particular metal analysis of lead, first row transition elements, phosphorus, and sulfur, was accomplished by TXRF on a Bruker S2 Picofox TXRF as described previously (7). Briefly, following growth, cells were washed sequentially in ice-cold Tris-EDTA, pH 8.0, buffer and MilliQ water and then finally resuspended in MilliQ water to a density of 2 ϫ 10 9 cells/ml. 2 l of a cell suspension, spiked with 1 ppm of a gallium internal standard, were spotted onto a quartz sample disc, and atomic fluorescence emission spectra were collected according to the manufacturer's recommendations (7). The cellular metal concentrations were determined by assuming a yeast cell volume of 50 fl and that a solution turbidity of A 600 nm ϭ 1.0 is equivalent to 2 ϫ 10 7 cells/ml (7).

In-gel protein digestion
In-gel protein digestion was conducted as described previously (84), with modifications. Briefly, selected protein bands were excised from the Coomassie-stained gel, diced into small pieces, and then destained with HPLC-grade water (Avantor) and 1:1 acetonitrile (ACN)/ammonium bicarbonate (ABC) (Sigma-Aldrich). The destained gel pieces were then dehydrated with multiple ACN washes until rock hard, followed by air drying for ϳ15 min. The gel pieces were rehydrated for 30 min with 10 mM DTT (Sigma-Aldrich) to reduce disulfide bonds, followed by replacement with 55 mM iodoacetic acid (Sigma-Aldrich) and 45 min shaking at 750 rpm in the dark to alkylate the reduced thiols. After reduction/alkylation, the gel pieces were once again washed and dehydrated as before and then chilled on ice for 15 min. To the chilled gel pieces, 30 l of 20 g/ml sequencing grade modified trypsin (Promega) was added for 30 min on ice. Excess trypsin solution was then removed and replaced with 100 l of 50 mM ABC, and the pieces were incubated overnight at 37°C with shaking at 750 rpm. Resultant proteolytic peptides were extracted by two rounds of dehydration using 100 l of ACN and collection of the resulting extract into low-retention microcentrifuge tubes, which were frozen solid at Ϫ80°C and then sublimated by cen-

Stress-induced heme signaling
tri-vapping. The dried peptides were reconstituted by sonication in 5% ACN, 0.1% formic acid and stored at Ϫ80°C before analysis.

Mass spectrometry
LC-MS analysis of peptides produced by in-gel digestion was carried out with an UltiMate TM 3000 RSLCnano UPLC system (Dionex) with Acclaim PepMap RSLC column (75 m ϫ 25-cm nanoViper C18 2 m, 100 Å) coupled to a Q-Exactive Plus Orbitrap mass spectrometer (Thermo Scientific) run in datadependent acquisition mode (top 15). Resultant RAW files were analyzed using Proteome Discoverer version 2.1 with embedded SEQUEST search algorithm operating with an allowable 1% false discovery rate, wherein the S. cerevisiae protein sequence database was used as the target for spectral matching. Only high-confidence peptide spectral matches were used for protein identification, and they are reported in Table S2.