Prototype of a Heme Chaperone Essential for Cytochrome c Maturation

See allHide authors and affiliations

Science  21 Aug 1998:
Vol. 281, Issue 5380, pp. 1197-1200
DOI: 10.1126/science.281.5380.1197


Heme, the iron-containing cofactor essential for the activity of many enzymes, is incorporated into its target proteins by unknown mechanisms. Here, an Escherichia coli hemoprotein, CcmE, was shown to bind heme in the bacterial periplasm by way of a single covalent bond to a histidine. The heme was then released and delivered to apocytochrome c. Thus, CcmE can be viewed as a heme chaperone guiding heme to its appropriate biological partner and preventing illegitimate complex formation.

In c-type cytochromes, heme is bound covalently by way of two thioether bonds to the conserved CXXCH motif of the apoprotein in a posttranslational process referred to as cytochrome c maturation (1–3). Heme synthesis in the mitochondrial matrix (4) or bacterial cytoplasm (2), and the stereospecific, covalent heme attachment in the intermembrane space (3) or periplasm (5, 6) are spatially separated processes requiring heme trafficking. Heme addition in mitochondria has been attributed to the enzyme cytochrome c heme lyase (3,7), although neither the mode of heme binding to that enzyme nor the mechanism of the ligation reaction has been elucidated. In Escherichia coli, eight ccm genes encode membrane proteins that are essential for cytochrome c maturation (8, 9).

Escherichia coli genes ccmABCDEFGH were overexpressed from a plasmid to stimulate cytochrome c maturation. Analysis of the membrane fraction by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) revealed an 18-kD protein that retained peroxidase activity of c-type cytochromes with covalently bound heme. However, the size of this protein corresponded best to one of the products of the ccm genes, CcmE. A chromosomal in-frame deletion mutant, which was constructed by removing 92ccmE-internal codons (Ile3 to Ser94) (10), was unable to produce mature c-type cytochromes (Fig. 1A). When ccmE was expressed in the ΔccmE background from the arabinose-inducible promoter para (11), membranes of the complementing strain contained both endogenous holocytochromes c and high levels of proposed heme-binding CcmE (Fig. 1A), as confirmed by immunoblot (Fig. 1B). The heme-protein association was SDS-resistant, as demonstrated by labeling ccmE-expressing E. coli cells with the heme precursor [14C]-δ-aminolevulinic acid (δ-ALA) followed by SDS-PAGE of trichloroacetic acid (TCA)–precipitated cell extracts (12). Cells expressing the eight ccm genes plus the two naturally adjacent structural genes for the endogenous c-type cytochromes NapB and NapC on a multicopy plasmid (6) were transformed with a second plasmid containing an additional, arabinose-inducible ccmE gene and analyzed for heme-binding proteins (Fig. 1C). When ccmE expression from para was repressed by the addition of glucose, endogenous E. coli c-type cytochromes such as NapB and NapC and the 18-kD CcmE protein were labeled to a similar extent. WhenccmE was overexpressed by arabinose induction, however, most of the [14C] label was incorporated into the 18-kD protein, confirming that CcmE contains a covalently bound tetrapyrrole. We conclude that the peroxidase activity (Fig. 1A) resulted from the presence of bound heme.

Figure 1

Identification of CcmE as a heme-binding protein. (A) Heme stain of proteins separated by 15% SDS-PAGE. For membrane fractions, we applied (from left to right) 150, 150, and 50 μg of membrane proteins from cells grown anaerobically in the presence of 5 mM nitrite for expression of endogenous c-type cytochromes: wild-type (WT) cointegrate (ccmABCDEFGH +), ΔccmE mutant cointegrate, and ΔccmE mutant cointegrate complemented with ccmE overexpressed on a plasmid. The endogenous c-type cytochromes of 18 and 24 kD are most likely the NrfB and NapC proteins, respectively (9). Wild-type CcmE migrates slightly faster than NrfB. For purified proteins, 8, 10, and 10 μg of purified, histidine-tagged CcmE derivatives were loaded. Membrane-bound CcmE-H6 migrates at ∼20 kD; the soluble, H6-tagged periplasmic (p-CcmE-H6) and cytoplasmic (c-CcmE-H6) versions of CcmE have apparent molecular masses of 15 kD. Molecular size standards are indicated on the left. (B) Protein immunoblot of a gel similar to that in (A) developed with anti-CcmE (17, 23). For membrane fractions, lanes were loaded with (from left to right) 20, 20, and 3 μg of protein, and for purified proteins, 1, 10, and 10 μg. Migration of heme-charged p-CcmE-H6 and heme-free c-CcmE-H6 was indistinguishable in the gel. (C) Radiolabeling of CcmE with the heme precursor [14C]-δ-ALA. The ΔccmA-H strain EC06 was complemented with plasmid pEC66 carrying thenapBCccmABCDEFGH genes (6, 8) plus an arabinose-inducible ccmE gene on a second plasmid. Whole cells labeled with 42.9 μM [14C]-δ-ALA for 1 hour (12) in the presence of either glucose or arabinose (repressing or inducing ccmEexpression from para, respectively) were precipitated with TCA, and proteins were separated by 15% SDS-PAGE. Radiolabeled proteins were visualized on a PhosphorImager. Expression of the ccm genes from pEC66 resulted in four major labeled bands corresponding to the NrfA (50 kD), NapC (24 kD), NrfB/CcmE (18 kD), and NapB (16 kD) c-type cytochromes (9). Upon addition of 0.8% arabinose, cells overexpressed CcmE into which most of the radiolabel was incorporated.

Next we characterized the spectroscopic features of the 18-kD hemoprotein. A hexa-histidine tag (H6) was fused to the NH2-terminus of CcmE. The protein was overexpressed in a ΔccmE background and purified from membranes (13). Pure CcmE-H6 protein migrated as a prominent 20-kD heme-staining band (Fig. 1, A and B). Visual light spectroscopy of dithionite-reduced and ammonium persulfate–oxidized CcmE-H6 protein was performed and the difference spectrum recorded (Fig. 2A). Maxima of α, β, and γ peaks were detected at 554.5, 524.5, and 426.5 nm, respectively, which is characteristic of c-type cytochromes (14). The pyridine hemochrome spectrum (14) showed peaks at 550.5, 520, and 420.5 nm, again typical for covalently bound heme. A stoichiometry of one molecule heme per three molecules CcmE polypeptide was calculated. This suggested that possibly not every CcmE molecule was charged with heme. After classical heme extraction with acidic acetone (14), no heme was detected in the organic solvent. Neither boiling of the protein for 30 min in 2% SDS and 6 M urea, nor overnight treatment with 50 mM HgCl2–6 M urea (pH 2) removed heme from the polypeptide. Thus, heme appeared to be bound covalently to the CcmE polypeptide.

Figure 2

Spectroscopic characterization of CcmE. (A) Dithionite-reduced minus ammonium persulfate–oxidized difference spectrum of purified, H6-tagged CcmE (0.6 mg ml–1). (B) Tandem mass spectrum of the parent ion (m/z = 1038; z = 2) from the tryptic CcmE heme peptide. The recorded masses are indicated on the right, the charges on the left of the peaks.

In order to overproduce CcmE in its natural cellular compartment, we first examined its topology by CcmE-PhoA fusion analysis (15). Alkaline phosphatase lacking its own signal sequence was fused to the last amino acid of CcmE. PhoA activity was expected only if CcmE was able to translocate the fused PhoA reporter to the periplasm. Cells containing the CcmE159-PhoA fusion plasmid produced 216.2 ± 0.4 phosphatase units, compared with 0.6 units of control cells containing the same plasmid lacking thephoA coding sequence. Thus, CcmE was anchored in the membrane by a hydrophobic, NH2-terminal domain and its bulk was oriented toward the periplasm.

A soluble, H6-tagged CcmE was produced as both a cytoplasmic (c-CcmE-H6) and a periplasmic (p-CcmE-H6) protein (Fig. 1B) (16) and tested for heme binding (Fig. 1A). Only the periplasmic version was heme stainable, suggesting that heme binding occurs in the periplasm.

The soluble p-CcmE-H6 protein was purified (13), then digested with trypsin, and a heme-binding peptide with an α absorption peak at 554 nm was isolated (17). Edman degradation resulted in the amino acid sequence XDENYTPPEVEZ (18) that matched the tryptic peptide HDENYTPPEVEK, except at the first position (X), where an extremely hydrophobic, nonidentifiable residue was found, and at the last position (Z), which is not conserved in the known CcmE homologs (2) and, therefore, is not likely the heme-binding residue. To unambiguously characterize the heme-bound tryptic peptide, we performed ion spray mass spectrometry (19) and obtained a molecular weight of 2073. This corresponded well to the sum of the theoretical molecular weights of the HDENYTPPEVEK peptide (1457.65) plus heme (616.55). Tandem mass spectrometry (MS/MS) of the parent ion [mass-to-charge ratio (m/z) = 1038; z = 2] resulted in two dominant daughter ions with molecular weights of 617 and 1459 corresponding to heme and the tryptic peptide, respectively (Fig. 2B). Various minor daughter ions were also obtained, among them ions with molecular weights of 1112, 1275, and 1377 corresponding to heme-HDEN, heme-HDENY, and heme-HDENYT, respectively. This result implied that heme was bound to one of the first four amino acids of the tryptic peptide. Consistent with the Edman degradation data we inferred that His130 of CcmE is the residue that is derivatized with covalently bound heme. To confirm this, we changed His130 to alanine by site-directed mutagenesis. The mutated CcmE protein was produced in the ΔccmE strain and tested for heme binding. As expected, only apo-CcmE was detected, and cytochrome c maturation was not restored. Therefore, His130is essential for both heme binding and cytochrome c maturation. It resides near the center of a seven–amino acid stretch (VLAKHDE) that is strictly conserved in all CcmE-like proteins (2) and may thus represent a previously undescribed heme-binding motif.

Does the cytochrome c maturation pathway in E. coliinvolve a step where CcmE binds heme tightly and transfers it subsequently to the apocytochrome c? An E. coli strain was constructed that expressed ccmE constitutively from a plasmid carrying all of the ccm genes downstream of atet promoter. In addition, the Bradyrhizobium japonicum cycA gene encoding the soluble B. japonicumcytochrome c550 was expressed from the inducible para promoter (20). Figure 3A shows the kinetics of holo-CcmE and holocytochrome c550 formation upon induction of apocytochrome c550 expression. Although CcmE contained heme by the time of induction, the signal intensity decreased with time. Holocytochrome c550 appeared 2 to 3 min after induction and accumulated at later time points. Immunoblotting of parallel samples demonstrated that they all contained constant amounts of CcmE polypeptide throughout the experiment (Fig. 3B). When expression of a heme–binding site mutant apocytochrome c (CLASH instead of the wild-type CLACH) (20) was induced in the presence of CcmE, heme remained bound to CcmE, which indicated that only wild-type apocytochrome c was able to trigger the release of heme from CcmE (Fig. 3C, upper and middle panels). Likewise, release of heme from CcmE was blocked in the absence of the CcmFGH proteins (Fig. 3C, lower panel), indicating that transfer of heme from CcmE to apocytochrome c requires at least one of these proteins.

Figure 3

Heme transfer from CcmE to apocytochrome c. (A) [14C]-δ-ALA–labeled cells (12) expressing ccmE constitutively were induced for expression of the B. japonicum apocytochrome c550. Samples (0.5 ml) were precipitated with TCA. Proteins were separated by 15% SDS-PAGE and visualized after 2 weeks of exposure on a PhosphorImager. The faint band migrating above holocytochrome c represents most likely the uncleaved, heme-binding precursor–cytochrome c, supporting the finding that cleavage of the signal sequence from the precursor is not obligate before heme attachment (6). The signal intensities were quantified by equalizing them such that the counts of each lane added up to 100%. (▴) CcmE, (○) unprocessed cytochrome c550 precursor, and (•) mature cytochrome c550. (B) Immunoblot of a gel similar to that in (A) developed with anti-CcmE immunoglobulins. (C) Heme stains of TCA-precipitated samples prepared as in (A) except that cells were not labeled. (Top) Induction was performed in the presence of all ccm genes. (Middle) A heme–binding site mutant (20) of cytochrome c550 was expressed. (Bottom) Wild-type apocytochrome c550 was expressed in the presence of ccmABCDE, but in the absence of ccmFGH (24).

CcmE represents a prototype of hemoprotein that binds heme covalently by a single bond to a histidine. Our MS data suggest that this binding does not involve a major loss of atoms from either the porphyrin or the histidine, which is likely due to an addition of the histidine imidazole to the double bond of one of the vinyl side chains of heme. The chemical nature of the bond may have implications for the precise mechanism of heme attachment to apocytochrome c.

CcmE appears to capture heme in the periplasm and promote transfer to newly synthesized apocytochrome c. The binding of heme to CcmE is a prerequisite for the subsequent heme ligation. CcmE may serve as periplasmic heme chaperone by (i) preventing heme from precipitation, (ii) hindering its nonspecific association with the membrane owing to its high hydrophobicity, (iii) presenting it in a sterically suitable conformation to the apocytochrome, or (iv) activating it for subsequent attachment to apocytochrome.

A ccmE gene has so far been identified only in representatives of the α and γ subdivisions of proteobacteria (2) and in the Arabidopsis plant (21). Other cytochrome c– forming organisms (for example, yeast,Bacillus subtilis, Helicobacter pylori,Synechocystis sp., Archaeoglobus fulgidus) seem to use a ccm-independent mechanism (22). Perhaps distinct types of heme chaperones will be identified in alternative cytochrome c maturation pathways.

  • * To whom correspondence should be addressed. E-mail: lthoeny{at}


View Abstract

Navigate This Article