Heme oxygenase-2 (HO-2) binds and buffers labile ferric heme in human embryonic kidney cells

Abstract

Heme oxygenases (HOs) detoxify heme by oxidatively degrading it into carbon monoxide, iron, and biliverdin, which is reduced to bilirubin and excreted. Humans express two isoforms of HO: the inducible HO-1, which is upregulated in response to excess heme and other stressors, and the constitutive HO-2. Much is known about the regulation and physiological function of HO-1, whereas comparatively little is known about the role of HO-2 in regulating heme homeostasis. The biochemical necessity for expressing constitutive HO-2 is dependent on whether heme is sufficiently abundant and accessible as a substrate under conditions in which HO-1 is not induced. By measuring labile heme, total heme, and bilirubin in human embryonic kidney HEK293 cells with silenced or overexpressed HO-2, as well as various HO-2 mutant alleles, we found that endogenous heme is too limiting a substrate to observe HO-2-dependent heme degradation. Rather, we discovered a novel role for HO-2 in the binding and buffering of heme. Taken together, in the absence of excess heme, we propose that HO-2 regulates heme homeostasis by acting as a heme buffering factor that controls heme bioavailability. When heme is in excess, HO-1 is induced, and both HO-2 and HO-1 can provide protection from heme toxicity via enzymatic degradation. Our results explain why catalytically inactive mutants of HO-2 are cytoprotective against oxidative stress. Moreover, the change in bioavailable heme due to HO-2 overexpression, which selectively binds ferric over ferrous heme, is consistent with labile heme being oxidized, thereby providing new insights into heme trafficking and signaling.

Keywords: heme homeostasis; heme oxygenase-2; heme oxygenases; heme trafficking; labile heme.

Figure 1

Fractional heme occupancy and activity of HO-2 as a function of heme concentration. The heme occupancy of HO-2 (pink curve; right y-axis) was simulated using the 1-site heme-binding model (111, 112): $\frac{\left[\text{HO}-2-\text{Heme}\right]}{{\left[\text{HO}-2\right]}_{\text{Total}}}=\frac{\left\{{\left[\text{HO}-2\right]}_{\text{Total}}+{\left[\text{Heme}\right]}_{\text{Total}}+K_{\text{D}}\right\}-\sqrt{\left\{{\left[\text{HO}-2\right]}_{\text{Total}}+{\left[\text{Heme}\right]}_{\text{Total}}+K_{\text{D}}\right\}{}^2-4{\left[\text{HO}-2\right]}_{\text{Total}}{\left[\text{Heme}\right]}_{\text{Total}}}}{2{\left[\text{HO}-2\right]}_{\text{Total}}}$. The fraction of maximal velocity, V/V_max, (black curve; left y-axis) was simulated using the quadratic velocity equation for tight binding substrates (113): $\frac{\left[\text{V}\right]}{{\left[\text{V}\right]}_{\text{Max}}}=\frac{\left\{{\left[\text{HO}-2\right]}_{\text{Total}}+{\left[\text{Heme}\right]}_{\text{Total}}+K_{\text{M}}\right\}-\sqrt{\left\{{\left[\text{HO}-2\right]}_{\text{Total}}+{\left[\text{Heme}\right]}_{\text{Total}}+K_{\text{M}}\right\}{}^2-4{\left[\text{HO}-2\right]}_{\text{Total}}{\left[\text{Heme}\right]}_{\text{Total}}}}{2{\left[\text{HO}-2\right]}_{\text{Total}}}$. These simulations were parameterized using the previously determined heme-HO-2 K_D and K_M value of 3.6 nM (77) and 400 nM (37), respectively, and [HO-2]_Total = 10 nM, which was determined in the present study for HEK293 cells (vida infra). These models both assume the following mass balance: [Heme]_Total = [HO-2-Heme] + [Heme] and [HO-2]_Total = [HO-2-Heme] + [HO-2]. Moreover, the quadratic velocity equation assumes that the other HO-2 substrates, CPR and O₂, are not limiting, and that the HO-2-Heme Michaelis complex is generated as quickly as it is consumed, that is, the steady-state approximation. The light blue shaded region represents the span of values reported for buffered free heme in cells and the black arrow indicates the concentration of intracellular free heme in which HO-1 is induced. CPR, cytochrome P450 reductase; HO, heme oxygenase.

Figure 2

Heme sensor 1 (HS1) can sense labile heme in HEK293 cells.A, representative violin plots depicting the distribution of eGFP/mKATE2 fluorescence ratios in single HEK293 cells expressing the high affinity heme sensor, HS1, the moderate affinity heme sensor, HS1-M7A, and the heme-binding incompetent control scaffold, HS1-M7A, H102A. The heme responsiveness of the sensors was established by culturing HEK293 cells in regular media (reg. med.; DMEM with 10% v/v FBS), heme deficient (HD) media supplemented with succinylacetone (SA) (HD + SA; DMEM with heme depleted 10% v/v FBS and 500 μM SA), or in regular media with 10 μM hemin chloride (10 μM heme). To saturate the sensors, the cells were permeabilized with digitonin and incubated with 100 μM hemin chloride (heme saturation; serum-free DMEM with 1 mM ascorbate, 40 μM digitonin, and 100 μM hemin chloride). B, median heme sensor eGFP/mKATE2 fluorescence ratio values derived from flow cytometry experiments (as in A) from triplicate HEK293 cultures. C, median heme sensor eGFP/mKATE2 fluorescence ratio values derived from flow cytometry experiments from triplicate HEK293 cultures grown in HD + SA media or in regular media supplemented with the indicated concentrations of hemin chloride or 5-aminolevulinic acid (ALA) for 24 h. Representative violin plots of eGFP/mKATE2 fluorescence ratio distributions from single cell analysis of HEK293 cultures are shown in Fig. S8A. D, measurements of total heme in triplicate HEK293 cultures grown in HD + SA media or in regular media supplemented with the indicated concentrations of hemin chloride or ALA for 24 h. E, relationship between sensor heme occupancy for HS1 and HS1-M7A and buffered-free heme, depending on weather heme is oxidized (ferric) or reduced (ferrous). See main text and Experimental procedures for details. The statistical significance is indicated by asterisks using one-way ANOVA for multiple comparisons using Tukey’s range test. ∗p = 0.0140, ∗∗∗∗p < 0.0001, ns = not significant. eGFP, enhanced green fluorescent protein; FBS, fetal bovine serum.

Figure 3

HO-2 silencing increases labile heme but does not affect total heme or bilirubin levels.A, representative immunoblot of HO-1 (HMOX1), HO-2 (HMOX2), and GAPDH expression in HEK293 cells treated with or without 350 μM 5-aminolevulinic acid (ALA) and scrambled (Ctrl siRNA) or targeted (HMOX2 siRNA) siRNA against HO-2 (left). Immunoblot analysis from six independent trials demonstrates that siRNA against HMOX2 results in ∼80 to 90% silencing of HO-2 protein expression (right). B, median heme sensor eGFP/mKATE2 fluorescence ratio values derived from flow cytometry experiments from 9 to 13 replicates of HEK293 cultures grown in HD + SA media, regular media, or regular media supplemented with 350 μM ALA or control or targeted siRNA against HMOX2. Representative violin plots of eGFP/mKATE2 fluorescence ratio distributions from single cell analysis of HEK293 cultures are shown in Fig. S8B. C, measurements of total heme in quintuplicate HEK293 cultures grown in regular media supplemented with control or targeted siRNA against HMOX2. D, measurements of total bilirubin in 3 to 6 replicates of HEK293 cultures grown in in HD + SA media, regular media, or regular media supplemented with 350 μM ALA or control or targeted siRNA against HMOX2. The statistical significance is indicated by asterisks using one-way ANOVA for multiple comparisons using Tukey’s range test. ∗p = 0.0376, ∗∗p = 0.0022, ^†p = 0.0156, ∗∗∗∗p < 0.0001, ns = not significant. eGFP, enhanced green fluorescent protein; HD, heme deficient; HO, heme oxygenase; SA, succinylacetone.

Figure 4

Heme oxygenase-2 overexpression depletes cytosolic and nuclear labile heme but not total heme, bilirubin levels, or mitochondrial labile heme.A, representative immunoblot of HO-1 (HMOX1), HO-2 (HMOX2), and GAPDH expression in untransfected (UT) or HO-2 overexpressing HEK293 cells treated with or without 350 μM ALA. B, measurements of total heme in triplicate untransfected (-) or HO-2 overexpressing (OE) (+) HEK293 cultures grown in HD + SA or regular media. C, measurements of total bilirubin in replicate untransfected (-) or HO-2 overexpressing (OE) (+) HEK293 cultures grown in regular media with or without 350 μM ALA. D, median cytosolic, nuclear, or mitochondrial (mito)-targeted HS1 eGFP/mKATE2 fluorescence ratio values derived from flow cytometry experiments from triplicate untransfected (-) or HO-2 overexpressing (OE) (+) HEK293 cells grown in HD + SA or regular media. Representative violin plots of eGFP/mKATE2 fluorescence ratio distributions from single cell analysis of HEK293 cultures are shown in Fig. S8C. The statistical significance is indicated by asterisks using one-way ANOVA for multiple comparisons using Tukey’s range test. ∗p = 0.0136, ∗∗p = 0.0078, ^†p = 0.0145, ^††p = 0.0083, ^‡p = 0.0292, ∗∗∗∗p < 0.0001, ns = not significant. ALA, 5-aminolevulinic acid; eGFP, enhanced green fluorescent protein; HD, heme deficient; HO, heme oxygenase; HS1, heme sensor 1; SA, succinylacetone.

Figure 5

Heme sequestration and buffering requires the HO-2 heme binding pocket but not catalytic activity or heme regulatory motifs (HRMs).A, schematic of HO-2 and residues of interest, including heme-binding ligand H45, a residue that accommodates heme binding within the heme-binding pocket, G159, Cys residues within Cys-Pro (CP; stars) dipeptides in the HRMs, C265 and C282, and the ER membrane-spanning region spanning residues 288 to 316 (blue). B, median cytosolic HS1 eGFP/mKATE2 fluorescence ratio values derived from flow cytometry experiments from triplicate HEK293 cells that were untransfected (UT) or overexpressing WT or the indicated mutant HO-2 alleles. The cells were cultured in HD + SA, treated with or without 50 μM hemin chloride, or regular media, treated with or without 350 μM ALA. Representative violin plots of eGFP/mKATE2 fluorescence ratio distributions from single cell analysis of HEK293 cultures are shown in Fig. S9. C, representative immunoblot demonstrating overexpression of the indicated HO-2 (HMOX2) alleles relative to untransfected (UT) cells. D, measurements of total heme in triplicate cultures of HEK293 cells that were untransfected (UT) or overexpressing WT or mutant HO-2 grown in regular media. The statistical significance is indicated by asterisks using one-way ANOVA for multiple comparisons using Tukey’s range test. ∗p = 0.0162, ∗∗p = 0.0012, ∗∗∗p = 0.0001, ^††p = 0.0068, ^†††p = 0.0003, ∗∗∗∗p < 0.0001, ns = not significant. ALA, 5-aminolevulinic acid; eGFP, enhanced green fluorescent protein; ER, endoplasmic reticulum; HD, heme deficient; HO, heme oxygenase; HS1, heme sensor 1; SA, succinylacetone.

Figure 6

The effects of HO-2 depletion or overexpression on HS1 heme occupancy are consistent with labile heme being oxidized.A, simulation of HS1 heme occupancy as a function of HO-2 expression assuming that heme is oxidized. B, simulation of HS1 heme occupancy as a function of HO-2 expression assuming that heme is reduced. Simulations assume the following mass balance terms: [H]_Total = [H] + [HO-2-H] + [S-H]; [S]_Total = [S] + [S-H]; [HO-2]_Total = [HO-2] + [HO-2-H], where H is unbound free heme, S is apo heme sensor, HO-2 represents apo HO-2, S-H is the heme bound sensor, and HO-2-H is heme bound HO-2. The heme sensor oxidized and reduced heme K_D values were assumed to be 3 nM and 1 pM, respectively. The HO-2-H oxidized and reduced heme K_D values were assumed to be 3.6 nM and 320 nM, respectively. The simulations were parameterized by fixing sensor concentration [S]_Total = 10 nM, and total heme being present at 1 nM (black), 10 nM (green), or 100 nM (purple). The blue patch represents the span in cellular HO-2 concentration accessed upon its silencing (∼1 nM), endogenous expression (∼10 nM), and overexpression (∼100 nM). Simulations were conducted using ChemEQL (v. 3.2.1) (85, 86). HO, heme oxygenase; HS1, heme sensor 1.