A Role for Iron in an Ancient Carbonic Anhydrase*

Since 1933, carbonic anhydrase research has focused on enzymes from mammals (alpha class) and plants (beta class); however, two additional classes (gamma and delta) were discovered recently. Cam, from the procaryote Methanosarcina thermophila, is the prototype of the gamma class and the first carbonic anhydrase to be characterized from either an anaerobic organism or the Archaea domain. All of the enzymes characterized from the four classes have been purified aerobically and are reported to contain a catalytic zinc. Herein, we report the anaerobic reconstitution of apo-Cam with Fe2+, which yielded Cam with an effective kcat that exceeded that for the Zn2+-reconstituted enzyme. Mössbauer spectroscopy showed that the Fe2+-reconstituted enzyme contained high spin Fe2+ that, when oxidized to Fe3+, inactivated the enzyme. Reconstitution with Fe3+ was unsuccessful. Reconstitution with Cu2+, Mn2+, Ni2+, or Cd2+ yielded enzymes with effective kcat values that were 10% or less than the value for the Zn2+-reconstituted Cam. Cam produced in Escherichia coli and purified anaerobically contained iron with effective kcat and kcat/Km values exceeding the values for Zn2+-reconstituted Cam. The results identify a previously unrecognized biological function for iron.

Research in the intervening years has shown that CA is one of the most widely distributed enzymes in nature (2, 3) and continues to be intensely investigated. Amino acid sequence comparisons identify four classes (␣, ␤, ␥, and ␦) of independent origins (4). Isozymes of the ␣ class are found in virtually all mammalian tissues where they function in diverse essential processes. The ␤ class is ubiquitous in plants and algae, where it is indispensable for the acquisition and concentration of CO 2 for photosynthesis. CA plays a role in the sequestration of atmospheric CO 2 in carbonates, and the global cycles of silicon and carbon are linked by CA in diatoms (5); thus, CA plays an important role in major geochemical and atmospheric processes. Members of the ␤ and ␥ classes are wide spread in physiologically diverse procaryotes from both the Bacteria and Archaea domains. Indeed, the genome of Escherichia coli contains two ␥ class homologs and two ␤ class homologs (2).
Cam, from the procaryote Methanosarcina thermophila, is the prototype of the ␥ class and the first CA to be characterized from either an anaerobic organism or the Archaea domain (6). Sequence analyses approximate the evolution of the ␥ class at the estimated time of the origin of life (2). The crystal structure of Cam purified aerobically from E. coli reveals a homotrimer with a subunit fold composed of a left-handed ␤-helix motif followed by short and long ␣-helix structures (7). Each of the three active sites contain three histidines that coordinate a zinc ion. Two of the metal-binding histidines are donated by one monomer, and the third histidine from an adjacent monomer. Other residues in the active site of Cam are also donated from adjacent monomer faces and bear no resemblance to residues in the active site of the well characterized ␣ class CAs for which specific functions have been assigned.
Kinetic investigations of the ␣ class CAs reveal a "zinc hydroxide" mechanism for catalysis (8) that also extends to both the ␤ and ␥ classes (9,10). The overall enzyme-catalyzed reaction occurs in two mechanistically distinct steps, E-Zn 2ϩ -OH Ϫ ϩ CO 2^E -Zn 2ϩ -HCO 3 where E is enzyme, and B is buffer. The first step is the interconversion between CO 2 and bicarbonate (Equations 2 and 3) involving a nucleophilic attack of the zinc-bound hydroxyl on the CO 2 molecule. The second step is regeneration of the zinc-bound hydroxide, which involves intramolecular proton transfer from the zinc-bound water to a proton shuttle residue (Equation 4) and intermolecular proton transfer to an accepting buffer molecule in the surrounding media (Equation 5). Kinetic analyses of site-specific replacement variants of Cam have identified essential residues (11,12). All of the biochemically characterized CAs from the ␣, ␤, and ␥ classes are categorically described as zinc metalloenzymes; however, there are few reports investigating the role of metals other than zinc. Although in vitro replacement of zinc with cobalt in a few purified CAs from the ␣ and ␥ classes yields robustly active enzymes (10,13,14), it is unknown whether any cobalt-containing CAs are synthesized in vivo. A CA from the marine diatom Thalassiosira weissflogii contains cadmium (15), although there are no reports of a biochemical or structural characterization of the purified enzyme. In most cases, CAs are hyperproduced in E. coli cultured in medium supplemented with Zn 2ϩ and purified aerobically with buffers containing Zn 2ϩ . Clearly, these procedures have the potential to bias the incorporation of Zn 2ϩ in vivo during synthesis of the enzyme or by in vitro exchange of another metal for Zn 2ϩ during purification. In particular, a role for iron in any CA has not been adequately investigated. Herein we show that replacement of zinc in Cam with ferrous iron yields an enzyme with effective k cat values exceeding zinc-containing Cam, identifying a previously unrecognized biological function for iron.

EXPERIMENTAL PROCEDURES
Enzyme Purification-Hyperproduction of Cam in E. coli BL21(DE) was as described (16) except where indicated the growth medium was supplemented with either 0.01% ferric ammonium citrate or 0.5 mM ZnSO 4 . Cam was purified as described (11) except, where indicated, anaerobically utilizing an inert atmosphere glove bag (Coy Laboratory Products, Ann Arbor, MI) and buffers that were rendered O 2 -free by vacuum degassing and replacement with N 2 . After the final purification step, the fractions containing Cam were pooled, dialyzed against 50 mM MOPS (pH 7.5), placed in aliquots, frozen in liquid N 2 , and stored at Ϫ80°C until further use. When purified anaerobically, no metals were added to the buffers, and all solutions were treated with Chelex-100 resin (Bio-Rad). In addition, all of the glassware was acid-washed and rinsed with Chelex-treated water.
Preparation of apo-Cam and Metal Reconstitution-In contrast to ␣ class carbonic anhydrases (14), the Cam active site metal cannot be readily removed by dialysis with chelators such as dipicolinate. Thus, Cam (10 -50 mg/ml) was unfolded to allow removal of existing metals by chelation and then refolded to produce apo-Cam that was subsequently reconstituted with the indicated metals. Purified Cam (10 -50 mg/ml) was incubated in 50-ml polyethylene centrifuge tubes at 4°C for 16 h in 20 mM MOPS (pH 7.5) containing 4.0 M spectroscopic grade guanidine hydrochloride (Fisher) and 50 mM dipicolinate. All of the buffers were treated with Chelex-100 resin to remove transition metal ions. The metal-free unfolded protein was then transferred into metal-free Spectra/Por 7 MWCO8000 tubing (Spectrum Laboratories, Rancho Dominguez, CA) and dialyzed three times for 24 h at 4°C against 25 mM MOPS (pH 7. Analytical-Cam preparations were analyzed for metals (aluminum, boron, barium, calcium, cadmium, cobalt, chromium, copper, iron, potassium, magnesium, manganese, molybdenum, sodium, nickel, phosphorus, lead, silicon, strontium, and zinc) by inductively coupled plasma emission spectrophotometry using a Jarrell-Ash 965 ICP instrument at the Chemical Analysis Laboratory of the University of Georgia. Circular dichroism spectra were collected on an AVIV model 62 DS spectrophotometer (Lakewood, NJ) at 25°C using a 2-mm path length quartz cuvette and a monomer concentration of 10 M in 50 mM, MOPS (pH 7.5) containing 50 mM Na 2 SO 4 . The protein solutions were scanned over 200 -260 nm using a bandwidth of 2.0 nm and a data acquisition time of 5 s. The Fe 2ϩ content of Cam was determined using the ferrouschelating chromophore ferrozine (3-(2-pyridyl)-5,6-bis(4-phenylsulfonic acid)-1,2,4-trizine) as previously described (17) except the assay mixture also contained 6.1 M guanidine hydrochloride to unfold the protein and allow the otherwise inaccessible ligated Fe 2ϩ to bind ferrozine. Carbon dioxide hydrating activity was measured by stopped flow spectroscopy using the changing pH indicator method as described previously (10), except for enzymes purified anaerobically, in which case all of the solutions were anaerobic. The protein concentration determinations were performed as described (10). The Mössbauer spectra were recorded on a spectrometer (WEB Research, Edina, MN) operating in the constant acceleration mode in a transmission geometry. During the measurement the sample was kept inside a SVT-400 Dewar from Janis (Wilmington, MA) at a temperature of 4.2 K in a magnetic field of 40 milliteslas applied parallel to the ␥-beam. Data analysis was performed using the program WMOSS (WEB Research).

RESULTS
Unfolding Cam with denaturant in the presence of a metal chelator and subsequent refolding by removal of the denaturant yield trimeric apo-Cam that can be reconstituted with nearly a full complement of Zn 2ϩ or Co 2ϩ producing highly active enzymes (10). The crystal structures of Zn 2ϩ -and Co 2ϩreconstituted Cam show the metals coordinated to the nitrogen atoms of histidines in all three active sites (18). The properties of Cam reconstituted with Zn 2ϩ , Co 2ϩ , and several other metal ions are shown in Table I. Incorporation of Zn 2ϩ or Fe 2ϩ was nearly quantitative with respect to the three available metal binding sites. Although substoichiometric, reconstitution was substantial for Co 2ϩ , Cu 2ϩ , Mn 2ϩ , Ni 2ϩ , and Cd 2ϩ . Analysis for a suite of metals showed that the sum of contaminating metals (other than the indicated reconstitution metal) was less than 0.18/trimer for each reconstituted enzyme. An exception was the 57 Fe 2ϩ -reconstituted Cam, for which a significant amount of contaminating zinc was incorporated. Attempts to reconstitute with Fe 3ϩ or Mg 2ϩ yielded enzymes with negligible amounts of the respective metals.
The coordination of metal by histidines from adjacent subunits contributes to the stability of the trimer (11); thus, apo-Cam is susceptible to subunit dissociation during size exclusion column chromatography. Recovery of the trimeric form of the reconstituted enzymes from the size exclusion column was proportional to the metal content (data not shown), a result indicating that the metals were coordinated in the active site. Attempts to reconstitute apo-Cam with Fe 3ϩ or Mg 2ϩ resulted in a nearly complete loss of the trimeric form consistent with the metal content, indicating that the active site of Cam is unable to accommodate these metals.
The effective k cat and k cat /K m values (Table I) for the Zn 2ϩand Co 2ϩ -reconstituted enzymes were similar to published values (10). Notably, the Fe 2ϩ -reconstituted Cam had effective k cat values that exceeded those for the Zn 2ϩ -and Co 2ϩ -reconstituted enzymes. Circular dichroism spectra for the Zn 2ϩ -and Fe 2ϩreconstituted enzymes were nearly identical, indicating a similar secondary structure (Fig. 1). These results indicate that iron functions at least as well as zinc in catalysis and also for the stability of the trimer. The divalent metals Cu 2ϩ , Mn 2ϩ , Ni 2ϩ , and Cd 2ϩ also replaced Zn 2ϩ ; however, the low effective k cat values of these reconstituted enzymes relative to the value for the Fe 2ϩ -reconstituted Cam suggest that they are not physiologically relevant.
The ferrozine assay indicated that iron present in the Fe 2ϩreconstituted enzyme was ferrous. Exposure to air diminished the k cat with a half-life of less than 5 min (Fig. 2), indicating the Fe 2ϩ -reconstituted Cam is oxygen-sensitive. The Mössbauer spectrum of 57 Fe 2ϩ -reconstituted Cam is shown in Fig. 3B. The spectrum has two broad absorption lines at Ϫ0.5 mm/s and ϩ2.9 mm/s. The spectrum collected over a narrower range of Doppler velocities (Fig. 3A) reveals the presence of at least two overlapping quadrupole doublets. The solid line overlaid with the experimental data in Fig. 3A is a theoretical simulation assuming two quadrupole doublets having the following parameters: isomer shift ␦ 1 ϭ 1.19 Ϯ 0.03 mm/s, quadrupole splitting parameters ⌬E Q1 ϭ 3.00 Ϯ 0.03 mm/s (47% of relative intensity); and ␦ 2 ϭ 1.15 Ϯ 0.03 mm/s, ⌬E Q2 ϭ 3.54 Ϯ 0.03 mm/s (53% of relative intensity). These parameters are typical of high spin Fe 2ϩ sites coordinated by five or six nitrogen/ oxygen ligands. The occurrence of two (or more) different iron species may reflect binding of different ligands (e.g. OH Ϫ , H 2 O, or HCO 3 Ϫ ) to the iron center. Exposure of Zn 2ϩ -Cam to H 2 O 2 had little effect on activity; however, the same treatment of Fe 2ϩ -reconstituted Cam rapidly produced a brown color with vigorous bubbling and reduced the k cat value to less than 10% consistent with oxidation of Fe 2ϩ to Fe 3ϩ (Fig. 4). The Mössbauer spectrum of H 2 O 2 -inactivated 57 Fe 2ϩ -reconstituted Cam (Fig. 3C) exhibited a broad, magnetically split, six-line spectrum that is typical of high spin Fe 3ϩ in the slow relaxation limit, which unambiguously demonstrates that exposure to H 2 O 2 results in complete (Ͼ98%) oxidation of the high spin Fe 2ϩ center. These results establish that the enzyme contains high spin ferrous iron and that oxidation to ferric iron inactivates the enzyme. The residual activity after oxidation of 57 Fe 2ϩ -Cam is attributed to zinc-Cam (Table I) produced by reconstitution of apo-Cam with zinc contaminating the 57 Fe 2ϩ .
Anaerobic size exclusion chromatography of air-oxidized Fe 2ϩ -Cam showed nearly complete loss of the trimeric form (not shown), indicating subunit dissociation as a consequence of the loss of Fe 3ϩ from the active site. These results are consistent with the inability of Fe 3ϩ to bind to apo-Cam (Table  I) and suggest that only Fe 2ϩ is able to coordinate in the active site. These results also suggest that Fe 2ϩ -Cam synthesized in vivo could lose iron during aerobic purification because of oxidation of Fe 2ϩ to Fe 3ϩ with oxygen. All published descriptions of unreconstituted Cam report zinc as the only metal; however, the enzyme was produced in E. coli cultured in medium supplemented with Zn 2ϩ and purified aerobically with buffers amended with Zn 2ϩ (6,16). The low yields obtained both for cell mass and purified enzyme have precluded a reliable metal analysis of Cam produced in M. thermophila; thus, Cam was produced in E. coli that was cultured in unsupplemented medium or media supplemented with either Fe 2ϩ , Zn 2ϩ , or both and purified anaerobically without supplementation of buffers with metals. Table I shows the metal content and kinetic parameters. Each of the unreconstituted Cam preparations had effective k cat and k cat /K m values that approximated those for Fe 2ϩ -reconstituted Cam, a result consistent with the metal   (Table I) were assayed by the stopped flow method using metal-free buffers at the times indicated after exposure to 1.0 atmosphere of air at 25°C. A control incubated anaerobically retained 100% activity over the time period shown (data not shown).
analysis showing incorporation of Fe 2ϩ . The sum of metals other than iron or zinc for each growth condition was less than 0.03/trimer, except for the enzyme purified from cells cultured in unsupplemented medium, which contained copper albeit in much lower amounts than iron or zinc ( Table I). The enzyme purified from cells cultured in unsupplemented medium contained less than 0.01 cobalt/trimer. Substoichiometric amounts of metals were present in the enzymes purified from E. coli cultured in Zn 2ϩ -supplemented and unsupplemented media, indicating the presence of apo-protein in both of these enzyme preparations, results suggesting that iron and zinc were limiting for both culture conditions. The iron content of the enzyme purified from cells cultured in Fe 2ϩ -supplemented medium exceeded the zinc content by 9-fold. In contrast to the ␣ class CAs (14), the Cam active site metal cannot be readily removed by dialysis with metal chelators unless the enzyme is unfolded with a chemical denaturant (10, 11). Thus, it is highly unlikely that E. coli produced predominantly Zn 2ϩ -Cam, and the Zn 2ϩ was exchanged for Fe 2ϩ during anaerobic purification. These results establish that E. coli synthesizes Fe 2ϩ -Cam in vivo.
Remarkably, Cam from E. coli cells cultured in medium supplemented with only Zn 2ϩ (0.5 mM) contained iron in amounts approximately equal to that of zinc (Table I). This result is in stark contrast to that predicted by the Irving-Williams series, for which the stability of complexed Fe 2ϩ is at least an order of magnitude less than that for Zn 2ϩ , Co 2ϩ , and Cu 2ϩ ligated with nitrogen (19); indeed, anaerobic reconstitution of apo-Cam with equimolar amounts of Fe 2ϩ , Zn 2ϩ , and Co 2ϩ produced enzyme with predominantly zinc (Table I). Furthermore, E. coli cultured in unsupplemented medium produced Cam with 7-fold more iron than zinc, although metals analysis of the medium showed it contained 1.6 ppm of iron and 3.2 ppm of zinc. Finally, Cam purified from cells cultured in medium supplemented with only Zn 2ϩ contained substoichiometric amounts of iron plus zinc in contrast to Cam purified from cells cultured in medium supplemented with Fe 2ϩ . These results establish that E. coli incorporates Fe 2ϩ into Cam with greater efficiency compared with Zn 2ϩ .

DISCUSSION
The results presented here establish a previously unrecognized biological role for iron. It is reported that Cam from M. thermophila aerobically purified from E. coli cells cultured in Zn 2ϩ -supplemented medium contains substoichiometric amounts of zinc and no appreciable amounts of other metals, which lead to the conclusion that Cam is a zinc metalloenzyme (16). However, as reported here, Cam contained approximately equal amounts of iron and zinc when purified anaerobically from E. coli cultured in medium supplemented with Zn 2ϩ . Based on the results reported here, it is likely that the previously reported Zn 2ϩ -Cam (16) also contained Fe 2ϩ in vivo that was oxidized to Fe 3ϩ and lost from the enzyme during aerobic purification. This result is not without precedent. Peptide deformylase from E. coli was first reported to contain Zn 2ϩ when purified aerobically and later shown to utilize Fe 2ϩ for catalysis (20). Oxygen inactivates the deformylase by oxidation of Fe 2ϩ to Fe 3ϩ , and exclusion of oxygen favors the incorporation of Fe 2ϩ over Zn 2ϩ (21). Use of strictly anaerobic conditions to prevent oxidation also lead to the identification of Fe 2ϩ as the active site metal in the previously reported Zn 2ϩ -containing methionyl aminopeptidase from E. coli (22).
Homologs of Cam genes are present in the genome of E. coli (YrdA and CaiE) consistent with synthesis of Cam homologs in this organism. The finding that E. coli synthesizes Fe 2ϩ -Cam with higher efficiency relative to Zn 2ϩ -Cam establishes that Fe 2ϩ -Cam is the predominant in vivo metal form. Furthermore, the results suggest that E. coli synthesizes Fe 2ϩ -Cam in nature. This proposition is supported by the preponderance of Fe 2ϩ -Cam synthesized by cells cultured in unsupplemented medium that more closely mimic conditions in the native environment. Furthermore, unsupplemented medium contained twice the amount of zinc relative to iron, and the Irving-Williams series predicts a greater affinity for Zn 2ϩ over Fe 2ϩ , as was shown for anaerobic reconstitution of apo-Cam with competing amounts of these metals. Other considerations argue for the synthesis of Fe 2ϩ -Cam by a diversity of anaerobes in their native anoxic environments. In addition to M. thermophila and E. coli, homologs of the gene encoding Cam are found in many strict and facultative anaerobes (2) that proliferate in anoxic environments where Fe 2ϩ availability is severalfold greater than Zn 2ϩ (23). Indeed, the metabolic pathways of anaerobes including M. thermophila (24) evolved to utilize an abundance of diverse oxygen-sensitive Fe 2ϩ -containing proteins (25). Finally, sequence analyses approximates the origin for the ␥ class   Table I. (2) to the time when Fe 2ϩ -containing proteins are thought to have played a major role in the origin and early evolution of life in an anoxic and Fe 2ϩ -rich aqueous environment (26,27). Although reconstitution of apo-Cam with Co 2ϩ yielded an enzyme with a catalytic efficiency between Fe 2ϩ -and Zn 2ϩ -reconstituted Cam (10), the significant synthesis of Co 2ϩ -Cam in nature is unlikely. This proposition is supported by the results showing that cobalt was negligible in unreconstituted Cam that was anaerobically purified from cells cultured in unsupplemented medium. Finally, the intracellular concentration of cobalt reported for M. thermophila is 32-fold less than for iron, reflecting the relative availability of these metals in the native environment (28).
It was found that Fe 2ϩ is incorporated in vivo with a greater efficiency than Zn 2ϩ in contrast to that predicted by the Irving-Williams series; however, the mechanism by which E. coli preferentially incorporates Fe 2ϩ is unknown. If incorporation is by unassisted binding of uncomplexed metals, then the Irving-Williams series requires a concentration of free Fe 2ϩ severalfold greater than Zn 2ϩ in the cytosol. This scenario is supported by the report that free Zn 2ϩ in the E. coli cytosol is less than one atom/cell (29). Furthermore, Cam purified anaerobically from cells cultured in Zn 2ϩ -supplemented medium contained significant amounts of apo-Cam, suggesting that the cells were limited for Zn 2ϩ , whereas Cam purified anaerobically from cells cultured in Fe 2ϩ -supplemented medium contained a nearly full complement of iron. On the other hand, if the intercellular concentrations of free Fe 2ϩ and Zn 2ϩ are comparable, it is likely that a chaperonin assists Fe 2ϩ incorporation into Cam. The presence of genes in E. coli with homology to the Cam gene from M. thermophila presents the possibility of a chaperonin in E. coli that also functions to incorporate Fe 2ϩ in Cam.
The results reported here raise a question regarding the extent to which Fe 2ϩ functions in CAs from other classes. There are no previous reports of the anaerobic purification or assay of any CA, and most often Zn 2ϩ is included in buffers used for aerobic purification. The anaerobic purification of CA from bovine erythrocytes yields an enzyme with 0.95 atoms of zinc, 2 suggesting that Fe 2ϩ -containing ␣ class CAs are inconsequential in mammalian cells. These results are not surprising considering the low abundance of Fe 2ϩ in mammalian tissue. However, the report that CAs are up-regulated in anoxic cancer cells (30) warrants an investigation of the metal content for these enzymes. Clearly, the results reported here call for a re-evaluation of the role of Fe 2ϩ in isozymes from all classes of CA utilizing anaerobic methods to exclude oxidation of Fe 2ϩ . Indeed, CAs from the ␤ class are found in strictly anaerobic microbes (2,31,32) that proliferate in Fe 2ϩ -rich environments and are highly dependent on Fe 2ϩ -containing proteins. Clarification of the in vivo metal form in CAs will have a profound influence on the biological understanding of this ubiquitous enzyme. The results also raise a caution for other enzymes characterized under aerobic conditions and reported to utilize zinc.