(Received for publication, November 6, 1996, and in revised form, February 12, 1997)
From the Department of Pharmaceutical Chemistry, School of Pharmacy, University of California, San Francisco, California 94143-0446
The covalently bound prosthetic group of lactoperoxidase (LPO) has been obtained by hydrolysis of the protein and identified as a dihydroxylated heme. A baculovirus expression system has been developed for LPO and used to obtain protein in which the heme is only partially covalently bound. Reaction of the purified heme·apoLPO complex with H2O2 results in both autocatalytic modification of the heme and covalent attachment to the protein. Hydrolytic experiments establish that the autocatalytically incorporated heme is bound normally. Two monohydroxylated heme intermediates have been detected. The peroxidative activity of LPO increases in proportion to the extent of covalently bound heme. The LPO results provide a paradigm for autocatalytic incorporation of heme groups into the mammalian peroxidases, including myeloperoxidase and eosinophil peroxidase, all of which exhibit strong sequence similarity with LPO and have covalently-bound heme groups.
The mammalian peroxidases, including lactoperoxidase (LPO),1 myeloperoxidase (MPO), eosinophil peroxidase, and thyroid peroxidase, utilize H2O2 to catalyze a diverse set of reactions. LPO (1), MPO (2), and eosinophil peroxidase (3) contribute to the nonimmune host defense system by oxidizing chloride ion and the pseudohalide thiocyanate to the potent microbicidal agents hypochlorous and hypothiocyanous acids, respectively. LPO fulfills this function in exocrine secretions, including tears, milk, and saliva, while eosinophil peroxidase and MPO carry out this chemistry in the phagosomes of neutrophils and eosinophils during engulfment of microorganisms. Thyroid peroxidase, which is distinguished from the other peroxidases in that it is an intracellular membrane-bound protein, catalyzes the iodination and coupling of thyroglobulin moieties in the biosynthesis of the thyroid hormones thyroxine and triiodothyronine (4). The mammalian peroxidases also participate in the oxidative metabolism of xenobiotics responsible for hypersensitivity reactions and other toxic sequelae (5, 6).
The ease with which the mammalian peroxidases oxidize high potential substrates such as pseudohalides distinguishes them from enzymes such as horseradish peroxidase and yeast cytochrome c peroxidase. The oxidative potency of the mammalian enzymes may be related to one of their distinguishing features: the presence of a modified, covalently bound heme group. The prosthetic group of MPO has been identified by crystallographic and chemical studies as a 1,5-dihydroxymethyl-modified heme b attached by three bonds to the protein (7-9). Two of these are ester bonds that link Asp and Glu residues with the heme hydroxymethyl groups, and the third is a sulfonium ion link obtained by addition of a Met sulfur atom to the 2-vinyl moiety of the heme. The structure of the heme group in the other mammalian peroxidases and the mechanism by which the hemes are incorporated into the mature proteins remain unknown.
The cDNA for bovine LPO (10) was
subcloned into pAcGP67B using the EcoRI site of the vector.
Recombinant baculovirus was generated in Spodoptera
frugiperda (Sf9) cells by co-transfection of the
LPO-containing vector and wild type virus using the BaculoGold transfection kit (Pharmingen, San Diego, CA). The virus was purified and amplified according to established procedures (11). Expression of
rLPO was accomplished by infection of 1 liter of Trichoplusia ni (High FiveTM) cells (Invitrogen) (at a density of 2 × 106 cells/ml) with recombinant virus at a multiplicity of
infection of one. Ten ml of a sterile, neutralized solution of hemin
(0.5 mM) was added concurrently. The medium was harvested
65 h postinfection, and the pH brought to 7.0 with ammonium
hydroxide. The medium was then batch-treated with 50 ml of CG-50 resin
(equilibrated with 50 mM ammonium acetate, pH 8.0) for
4 h with stirring at 25 °C. The resin was then poured into a
column, washed with a 400-ml volume of the equilibration buffer,
followed by 200 ml of 0.25 M ammonium acetate buffer. rLPO
was then eluted with 0.5 M ammonium acetate. The fractions
containing 412 nm absorbing material were pooled and lyophilized. The
dried residue was dissolved in water and loaded onto a CG-50 column and
chromatographed as above. The pooled recombinant LPO (rLPO) fractions
were lyophilized and stored at 20 °C.
Native LPO (nLPO) (3.2 mg) from Sigma was digested with Pronase (0.2 mg/mg of enzyme) in 50 mM Tris-HCl bicarbonate buffer, pH 8.0, by incubation for 4 h at 37 °C. The sample was chromatographed by direct injection on a 250 × 4.6-mm C4 reverse phase column (Vydac). The prosthetic group was eluted with a linear gradient of 20-50% acetonitrile in water (containing 0.1% trifluoroacetic acid) over 25 min with detection at 400 nm. The major peak representing the Pronase-released prosthetic group (retention time = 11.5 min) was collected and the sample lyophilized to dryness. The prosthetic group was then dissolved in methanol and subjected to mass spectrometric analysis by both MALDI and electrospray methods. The accurate mass of the sample was obtained using electrospray ionization and detection with an AutoSpec (Micro Mass Laboratories) magnetic sector instrument.
Relation of Prosthetic Heme Covalent Binding to the Peroxidative Activity of rLPO and nLPOC4 HPLC and activity assays were
carried out on nLPO, rLPO, and
H2O2-preincubated rLPO. Preincubation of rLPO
with H2O2 was accomplished by adding 10 eq of
H2O2 to the enzyme and incubating for 5 min at
25 °C prior to analysis by HPLC and measurement of activity. HPLC
was performed as described above for characterization of the prosthetic
group. The proportion of covalently bound heme in the chromatographed
rLPO samples corresponded to the percent of the total 400 nm
integration represented by the holoprotein peak (retention time = 26.7 min). The activity assays contained ABTS
(2,2-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (1.7 mM) in 50 mM potassium acetate buffer, pH 4.5, and enzyme (1 fmol) and were initiated with 2.5 mM
H2O2. Formation of the ABTS radical cation was
monitored at 416 nm. Specific activities were calculated as activity
per Soret absorbance unit of the ferric enzyme added to each assay.
rLPO (3 µM) in 0.1 M ammonium acetate buffer, pH 7.0, was incubated with H2O2 (3.3-34.5 µM) for 5 min at 25 °C. Excess H2O2 was then consumed by incubation with catalase (5 units) for 5 min. Samples were chromatographed as described for prosthetic group characterization, with detection and integration at 400 nm. The sum of the integration values for the polar, free, and covalent heme peaks represents "total heme" for each chromatographic run.
As part of an effort to elucidate the structure-function relationships that define diverse peroxidase activities, we have proteolytically digested native LPO with Pronase as done previously by Nichol et al. (12) to release the heme group. The MALDI spectrum of the heme thus isolated has a molecular ion at m/z 648 (i.e. heme plus 32 mass units). This was interpreted in the original studies as evidence for incorporation of a sulfur atom, but determination of the accurate mass of the heme derivative by electrospray mass spectrometry with an Autospec detector yields a value of m/z 648.1685. This value fits within 2 ppm of the expected value for incorporation of two oxygen atoms (calculated m/z 648.1671) but deviates by 30 ppm from the value for the incorporation of a sulfur atom (calculated m/z 648.1494). The heme isolated hydrolytically from LPO is thus a dihydroxy derivative rather that the originally postulated thiol species. Rae and Goff recently came to a similar conclusion based on the finding that the high- and low-spin ferric forms of the isolated heme(-peptide) display only two NMR methyl proton resonances and two new methylene resonances, as expected for hydroxylation of two of the heme methyl groups (13). The LPO structure is thus consistent with that of the heme in MPO except that the third link between a methionine and a heme vinyl is not formed. Failure to form this link is not surprising, because sequence comparisons indicate that in LPO a Gln replaces the critical Met found in MPO (10, 14). The prosthetic heme modification and binding, like their primary protein sequences, are thus very similar in LPO and MPO.
A precedent for physiological covalent attachment of heme to a protein
is provided by cytochrome c. Cytochrome c
maturation in bacteria, fungi, and yeast requires an enzyme, cytochrome
c lyase, that catalyzes covalent attachment of the heme to
the protein (15-18). However, it is not known whether a comparable
enzyme is required for heme modification and attachment in the
mammalian peroxidases. To address this question, we have successfully
expressed bovine LPO in T. ni insect cells by infection with
a recombinant baculovirus containing the LPO gene. The recombinant
protein (rLPO) thus obtained in yields of ~5 mg/liter is
indistinguishable from the native protein (nLPO) by a number of
criteria. The electronic absorption spectra of rLPO and nLPO are
identical, with Soret maxima at 412 nm and Soret bandwidths of 45 nm.
rLPO co-migrates with nLPO on both SDS-polyacrylamide gel
electrophoresis and C4 reverse phase HPLC and cross-reacts with a
polyclonal antibody raised to nLPO (19) (not shown). To examine the
extent of heme attachment in the baculovirus-expressed rLPO, we
developed an HPLC system in which noncovalently bound heme dissociates
from the protein and is separated from the covalently bound heme. This HPLC system has been shown to determine whether a heme protein contains
acid-dissociable or covalently bound heme. The pH of the eluting
solvent is less than 1, which leads to protonation of the proximal
histidine and the consequent rupture of its bond to the heme iron. This
causes otherwise noncovalently bound hemes to dissociate from the
apoprotein. Thus, both myoglobin and HRP yield a free heme species that
is separable from the apoprotein, as evidenced by comparison of the
chromatograms at 400 and 215 nm. On the other hand, nLPO yields a heme
species that co-elutes with the 215 nm absorbing protein peak. As shown
in Fig. 1, only 60% of the prosthetic heme is
covalently bound in rLPO, in contrast to 100% in nLPO. The finding
that 40% of the heme is associated noncovalently with rLPO yet the
chromophore of the protein is indistinguishable from that of the native
protein indicates that the unusual red-shifted Soret band of nLPO does
not result from heme peripheral substitution and attachment to the
protein (as proposed previously (20)), but from noncovalent
interactions within the protein milieu. Of the noncovalently bound
heme, the majority is unmodified iron protoporphyrin IX (Fig. 1A,
peak II). The minor polar peak (peak I) representing
7% of the total displays chromatographic behavior consistent with the
addition of a single hydroxyl group to iron protoporphyrin IX. The
presence of heme and a monohydroxylated derivative of heme in purified
rLPO suggests that they are intermediates in the biosynthesis of the
native diesterified heme.
The possibility that heme oxidation and incorporation result from an
autocatalytic process was tested by preincubating rLPO with
H2O2 and determining the extent of covalent
heme binding by HPLC. Indeed, the proportions of both the free heme and
the polar heme product decrease, while the amount of protein-bound heme
increases, as a function of the H2O2
concentration (Fig. 2). The covalent binding reaction is
H2O2-dependent, rapid (complete in < 1 min), and yields a maximum of 90% covalently bound heme with excess H2O2 (9 µM) (Fig. 2).
The spectrum of the protein is little altered by reaction with small
amounts of H2O2 if higher valent iron species
are reduced with ascorbate. When the oxidation of ABTS by rLPO is
measured after preincubation with H2O2, the specific activity as a percent of the activity of nLPO is found to
increase in parallel with the extent of covalent heme binding (Table
I). LPO is thus active in the oxidation of exogenous
substrates only after the heme is autocatalytically modified and
covalently bound.
|
To establish that the covalent heme in rLPO is the same as that in the
native enzyme, the hemes released by Pronase digestion were compared
using reverse phase HPLC. As already discussed, nLPO yields as the
major peak the dihydroxymethyl-modified heme (Fig.
3A). This species is also isolated from rLPO
(Fig. 3B), confirming that the autocatalytic modification in
rLPO is identical to that in the native protein. Furthermore, a new
polar heme peak with a retention time of 17.5 min, in addition to the
polar peak at 16 min found also in rLPO (peak I in Fig. 1),
is obtained by proteolysis of rLPO. This peak, like peak I, appears to
be a monohydroxylated heme because it elutes with authentic
8-hydroxymethylheme isolated from an incubation of horseradish
peroxidase with phenylhydrazine (21). Given that the heme in MPO is a
1,5-dihydroxymethyl heme derivative (8, 9), the 16- and 17.5-min
peaks probably correspond to the 1- and 5-monohydroxymethyl
derivatives, although which is which is not clear. These polar
hemes may derive from the monoester intermediates in the pathway to the
final diester product because they disappear (along with part of the
unmodified heme) in parallel with the increase in the dihydroxymethyl
derivative when rLPO reacts with H2O2 (Fig.
3C). Heme and the two polar hemes are present not only in
rLPO but also in nLPO, as detailed examination of the chromatogram of
the native Pronase-digested protein shows they are present at levels of
between 0.1 and 1% of the total heme absorption (not shown). The
correspondence between the native and recombinant proteins argues that
free heme and the polar hemes derived from the heme-rLPO monoesters are
intermediates in the normal autocatalytic maturation of LPO.
LPO can thus be expressed in vitro, and the recombinant protein is indistinguishable from the native protein except for the degree of covalent heme modification (Fig. 1). Incomplete covalent incorporation of heme into the isolated protein suggests that heme associates with the protein intracellularly and that there is insufficient exposure to H2O2 in the insect cell prior to its secretion across the cell membrane in the present efficient secretory expression system.
Formation of the covalently bound heme-protein complex is the result of an autocatalytic process, as indicated by the increase in the extent of covalent binding when the protein is incubated with H2O2 in the absence of other factors. Furthermore, consumption of the monohydroxylated heme species derived from heme·rLPO monoesters during the H2O2-supported reaction (Figs. 1 and 3) supports their identification as intermediates in the process. The fact that the in vitro attachment plateaus when ~90% of the heme is covalently bound may reflect either binding of a portion of the noncovalently bound heme somewhere other than in the active site, the presence of a subpopulation of autocatalytically incompetent protein, or self-inflicted oxidative damage to a fraction of the enzyme during the autocatalytic process.
Autocatalytic heme incorporation into LPO is likely to result from
protein-mediated heme co-oxidation. A mechanistic precedent for this is
provided by the formation of 8-hydroxymethyl heme in the reaction of
HRP with phenylhydrazine, in which the phenyl radical functions as the
co-oxidizing species (21). Thus, reaction with the first equivalent of
H2O2 produces a 2-electron-oxidized Compound I
intermediate, with one oxidizing equivalent present as a heme ferryl
oxo (FeIV=O) species and the other residing on the protein
as a free radical (Scheme 1) (22). Hydrogen abstraction
from a heme methyl by the protein radical produces a methyl-centered
radical, which is oxidized to the cation by intramolecular electron
transfer to the ferryl oxygen. The benzylic carbocation is then trapped by the carboxylate side chain of an aspartate or glutamic acid residue
to generate the protein-linked monoester (pathway a). A second cycle
produces the second ester link. Alternatively, the cation may be
trapped by water to yield a hydroxylated heme as a "dead end"
product or, less likely, as an intermediate that undergoes
esterification with the carboxylate side chains (pathway b). It is of
interest in this context that maximal heme binding requires ~8 eq (9 µM) of H2O2/heme group (Fig. 2),
four times as much as predicted by Scheme 1, suggesting that the
autocatalytic reaction is inefficient. One intriguing aspect of the
mechanism is the requirement for oxidation of two methyl groups
separated by a distance of >15 Å. Either there is enough
conformational flexibility for one amino acid radical to swing from one
site to the other, or two distinct amino acid radicals function as co-oxidizing species. Including the two amino acids involved in formation of the diester bonds, at least three amino acids appear to be
intimately involved in covalent binding of the heme.
The discovery that heme modification and attachment in LPO results from an autocatalytic process indicates that heme attachment in MPO, eosinophil peroxidase, and thyroid peroxidase is likely to be mediated by a similar mechanism rather than, as in the case of cytochrome c, by the action of one or more auxiliary enzymes. The role that heme modification and attachment plays in the function of the mammalian peroxidases is unclear, but the finding that the ABTS-oxidizing activity of LPO increases in parallel with the degree of heme covalent binding indicates that it is important. Covalent heme attachment may influence the redox potential, substrate binding, susceptibility to autoinactivation, or other parameters that govern productive catalytic turnover.
We thank J. F. Seilhamer for the bovine LPO cDNA (10) and P. Vilja for the gift of the anti-LPO antibody (19).