**2. Heavy metals as carbonic anhydrase cofactors**

CAs catalyze the reversible hydration of carbon dioxide to bicarbonate and protons by means of a metal-hydroxide (Lig3M2+(OH)\_) mechanism, although the α-CAs possess other catalytic activities such as esterase, phosphatase, cyanate/cyanamide hydrase, etc. (Supuran et al., 2003; Supuran and Scozzafava 2007; Innocenti et al., 2008). In the α-, γ, and δ-CA classes, Lig3 is always constituted by three His residues. The metal (M) is ZnII for all classes. The zinc atom is in the +2 state and is located in a cleft near the center of the enzyme. The role of zinc in carbonic anhydrase is to facilitate the deprotonization of water with the formation of the nucleophilic hydroxide ion, which can attack carbonyl group of carbon dioxide to convert it into bicarbonate. This is obtained through the +2 charge of the zinc ion which attracts the oxygen of water, deprotonates water, thus converting it into a better nucleophile able to attack the carbon dioxide.

Water naturally deprotonates itself, but it is a rather slow process. Zinc deprotonates water by providing a positive charge for the hydroxide ion. The proton is donated temporarily to the surrounding amino acid residues, and then it is given to the environment, while allowing the reaction to continue. Zinc is able to help the deprotonation of water by lowering the pKa of water. Therefore, more water molecules are now able to deprotonate at a lower pH than normal, increasing the number of hydroxide ions available for the nucleophilic attack to carbon dioxide (Berg, 2007).

The affinity of carbonic anhydrase for zinc is in subpicomolar range, as assessed for studies on the α-class (Tripp et al., 2001). Cox et al. (2000) and Hunt et al. (1999) ascribed a role for hydrophobic core residues in human CA-II that are important for preorienting the histidine ligands in a geometry that favours zinc binding and destabilizes geometries that favour other metals. In particular, mutagenesis experiments demonstrated that substitutions of these amino acids at position 93, 95, and 97 decrease the affinity of zinc, thereby altering the metal binding specificity up to 104-fold. Furthermore, the free energy of the stability of native CAII, determined by solvent-induced denaturation, correlates positively with increased hydrophobicity of the amino acids at positions 93, 95, and 97 as well as with zinc affinity (Hunt et al., 1999).

β-CAs, present in green plants and cyanobacteria, contain also Zn2+ in the active site but are differentiated from α-CAs by virtue of the fact that the active site is coordinated by a pair of cysteine residues and a single histidine residue, whereas the fourth ligand may be either a water molecule/hydroxide ion, or a carboxylate from a conserved aspartate residue in some β-CAs (Type II β-CAs) [Trip et al., 2001; Xu et al., 2008]. The metal hydroxide catalytic mechanism seems to be also valid for these enzymes [Supuran, 2008].

Besides zinc, other metals have demonstrated to be physiologically relevant cofactor for some CAs. In fact, in the γ-CAs metal may also be FeII (Ferry et al., 2010). Cam, the prototypic γ-class carbonic anhydrase, from the anaerobic methane producing Archaea species *Methanosarcina thermophila,* contains zinc in the active site when overproduced in *Escherichia coli* and purified aerobically [Alber et al., 1996], while it has 3-fold greater carbonic anhydrase activity and contains Fe2+ in the active site (Fe-Cam) when purified anaerobically from *E. coli* or overproduced in the closely related species M. *acetivorans* and purified anaerobically. Soluble Fe2+ is abundant in oxygen free environments and available to anaerobic microbes. The different results obtained in aerobic and anaerobic conditions is explained by the fact that in aerobic conditions Fe3+ is oxidized and rapidly loss from CAM enzyme, substituted by Zn2+ contaminating buffers not treated with chelating agents. These results indicate Fe2+ as the physiologically relevant metal [MacAuley et al., 2009; Tripp et al., 2004] in the active site for CAM enzyme. Interestingly, evidence for the role of ferrous ion in CA has been obtained also for the α class. In fact carbonic anhydrase activity from duck erythrocytes is increased in the presence of iron in the incubation medium suggesting a role for iron in the active site (Wu et al., 2007).

The ζ-CA naturally uses Cd2+ as its catalytic metal in marine diatoms (Lane and Morel, 2000; Lane et al., 2005; Park et al., 2008). This cdmium-CA (CDCA1) consists of three tandem CA repeats (R1–R3), which share 85% identity in their primary sequences (Lane et al., 2005). Although CDCA1 was initially isolated as a Cd enzyme, it is actually a "cambialistic" enzyme since it can use either Zn or Cd for catalysis—and spontaneously exchanges the two metals (Xu et al., 2008). Kinetic data show that the replacement of Zn by Cd results nonetheless in a decrease in catalytic efficiency (Xu et al., 2008). In the active site, Cd is coordinated by three invariant residues in CDCA of all diatom species (Park et al., 2007): Cys 263, His 315 and Cys 325. The tetrahedral coordination of Cd is completed by a water molecule. The use of Cd in CDCA is thought to explain the nutrient-like concentration profile of Cd in the oceans, where the metal is impoverished at the surface by phytoplankton uptake and regenerated at depth by remineralization of sinking organic matter (Lane and Morel 2000). It is cycled in the water column like an algal nutrient. It is thought that the expression of a CDCA in diatoms, which are responsible for about 40% of net marine primary production, represents an adaptation to life in a medium containing vanishingly small concentrations of essential metals (Xu et al., 2008). As suggested by Xu et al. (2008) the remarkable ability to make use of cadmium, an element known for its toxicity, gave presumably a significant competitive advantage to diatoms in the oceans, poor in metals, with respect to other species, and could have contributed to the global ratiation of diatoms during the Cenozoic Era and to the parallel decrease in atmospheric CO2.

Moreover, Co(II) has been shown to replace Zn(II) in α-, β and γ-CA (Hoffmann et al., 2011). Cobalt ionic radius and polarizability are very similar to those of Zn(II). In contrast to Zn(II) (d10), the d7electron configuration of Co(II) is accessible to electronic spectroscopic methods (, yielding information about the interactions protein-metal. As a result, spectroscopy of Co(II) substituted CA isozymes has been used to probe the environment of the metal ions in the active sites and get information on the nature of the first coordination sphere of the metal (Hoffmann et al., 2011). The Co-containing form of the enzyme generally shows a marked decrease in activity compared with the native Zn form (Tu and Silverman, 1985). The demonstration that Zn can be extracted from a protein and replaced with Co *in vitro* does not demonstrate that such metal substitution takes place in vivo. The evidence for in vivo Co substitution in a CA was for the first time provided by Morel et al (1994) and Yee and Morel (1996) in the diatoms *T. weissflogii,* who demonstrated 65Zn and 57Co bands to comigrate with a single band of CA activity on a native gel of diatom proteins.

### **3. Heavy metals as inhibitors of carbonic anhydrase activity**

Several heavy metals were demonstrated to *in vitro* inhibit CA activity in a variety of organisms, including fishes, crabs, bovines, and humans.

explained by the fact that in aerobic conditions Fe3+ is oxidized and rapidly loss from CAM enzyme, substituted by Zn2+ contaminating buffers not treated with chelating agents. These results indicate Fe2+ as the physiologically relevant metal [MacAuley et al., 2009; Tripp et al., 2004] in the active site for CAM enzyme. Interestingly, evidence for the role of ferrous ion in CA has been obtained also for the α class. In fact carbonic anhydrase activity from duck erythrocytes is increased in the presence of iron in the incubation medium suggesting a role

The ζ-CA naturally uses Cd2+ as its catalytic metal in marine diatoms (Lane and Morel, 2000; Lane et al., 2005; Park et al., 2008). This cdmium-CA (CDCA1) consists of three tandem CA repeats (R1–R3), which share 85% identity in their primary sequences (Lane et al., 2005). Although CDCA1 was initially isolated as a Cd enzyme, it is actually a "cambialistic" enzyme since it can use either Zn or Cd for catalysis—and spontaneously exchanges the two metals (Xu et al., 2008). Kinetic data show that the replacement of Zn by Cd results nonetheless in a decrease in catalytic efficiency (Xu et al., 2008). In the active site, Cd is coordinated by three invariant residues in CDCA of all diatom species (Park et al., 2007): Cys 263, His 315 and Cys 325. The tetrahedral coordination of Cd is completed by a water molecule. The use of Cd in CDCA is thought to explain the nutrient-like concentration profile of Cd in the oceans, where the metal is impoverished at the surface by phytoplankton uptake and regenerated at depth by remineralization of sinking organic matter (Lane and Morel 2000). It is cycled in the water column like an algal nutrient. It is thought that the expression of a CDCA in diatoms, which are responsible for about 40% of net marine primary production, represents an adaptation to life in a medium containing vanishingly small concentrations of essential metals (Xu et al., 2008). As suggested by Xu et al. (2008) the remarkable ability to make use of cadmium, an element known for its toxicity, gave presumably a significant competitive advantage to diatoms in the oceans, poor in metals, with respect to other species, and could have contributed to the global ratiation of

diatoms during the Cenozoic Era and to the parallel decrease in atmospheric CO2.

migrate with a single band of CA activity on a native gel of diatom proteins.

**3. Heavy metals as inhibitors of carbonic anhydrase activity**

organisms, including fishes, crabs, bovines, and humans.

Moreover, Co(II) has been shown to replace Zn(II) in α-, β and γ-CA (Hoffmann et al., 2011). Cobalt ionic radius and polarizability are very similar to those of Zn(II). In contrast to Zn(II) (d10), the d7electron configuration of Co(II) is accessible to electronic spectroscopic methods (, yielding information about the interactions protein-metal. As a result, spectroscopy of Co(II) substituted CA isozymes has been used to probe the environment of the metal ions in the active sites and get information on the nature of the first coordination sphere of the metal (Hoffmann et al., 2011). The Co-containing form of the enzyme generally shows a marked decrease in activity compared with the native Zn form (Tu and Silverman, 1985). The demonstration that Zn can be extracted from a protein and replaced with Co *in vitro* does not demonstrate that such metal substitution takes place in vivo. The evidence for in vivo Co substitution in a CA was for the first time provided by Morel et al (1994) and Yee and Morel (1996) in the diatoms *T. weissflogii,* who demonstrated 65Zn and 57Co bands to co-

Several heavy metals were demonstrated to *in vitro* inhibit CA activity in a variety of

for iron in the active site (Wu et al., 2007).

The early work of Christensen and Tucker (1976) demonstrated carbonic anhydrase inhibition by heavy metals for the first time in fish. The study was carried out on red blood cells CA of the teleost *Oncorhynchus mykiss*. Erythrocyte CA, which represents the most abundant pool of the enzyme in fish, appeared significantly in *vitro* inhibited by several heavy metals cations, such as Cd2+, Cu2+, Ag+, and Zn2+ (Tab1).

In the intestine and gills of the European eel, *Anguilla anguilla*, Lionetto et al. (1998; 2000) found cadmium to significantly inhibit carbonic anhydrase activity. The inhibition appeared tissue specific (Lionetto et al., 1998; Lionetto et al., 2000). The gill CA was much more sensitive to the heavy metal as compared to the enzyme activity in the intestine, as observed by comparing the IC50 values (Tab1). In particular in the intestine the inhibitory effect of cadmium was more pronounced on the cytosolic than the membrane-bound CA, which revealed only a partial inhibition at high concentrations. Moreover CA activity inhibition showed a certain time-dependence, with a delay of at least 10 min and 30 min for the cytosolic isoform and the membrane bound isoform respectively. The authors attributed this behaviour to the time required by cadmium for displacing the metal (zinc) associated with the enzyme, giving an inactive Cd-substituted carbonic anhydrase. Cadmium is a bivalent metal, similar in many respects to zinc: both are in the same group of the periodic table, contain the same common oxidation state (+2), and when ionized have almost the same size. Due to these similarities, cadmium can replace zinc in many biological systems. Moreover, the delayed inhibition of membrane-bound CA with respect to the cytosolic isoform was explained by a more difficult access of cadmium to the active site of the enzyme bound to the membrane. In fact, it has to be considered that the membrane-bound CA is stabilised by disulfide bonds (Whitney and Briggle, 1982) which could contribute to a less sensitivity of the membrane bound CA to cadmium.

As suggested by Lionetto et al (2000), the observed *in vitro* inhibition of cadmium on CA activity could be useful in the understanding of the toxic effects that the heavy metal can elicits on fish physiology *in vivo*. The inhibitory effect on gill CA activity suggests that the heavy metal might interfere with a number of physiological functions in which gill CA is involved as gas exchanges (Randall and Daxbaeck, 1984), acid–base balance (Heisler, 1984), osmoregulation (Henry, 1984) and clearance of the waste products from nitrogenous metabolism (Evans and Cameron, 1986). Morgan et al (2004) directly demonstrated in *in vivo*  expoxure experiments on rainbow trout that inhibition of branchial CA was able to induce an early decline in the gill Na+ and Cl uptake. With regards to the intestine, the physiological role of the cytosolic CA is that of generating HCO3 \_ from metabolic CO2 while the role of the CA enzyme associated to the brush-border membrane should be that of mediating the environmental HCO3\_ uptake (Maffia et al., 1996). Therefore, the inhibitory effect of cadmium on intestinal CA isoforms should interfere with bicarbonate balance and in turn with systemic acid–base balance and osmoregulation in fish. In fact, as previously shown (Schettino et al., 1992), the HCO3- entry via the membrane-bound CA in the cell across the luminal membrane of the enterocytes seems to be essential for maintaining a steady intracellular HCO3 concentration and/or pHi; as a consequence the salt transport in eel intestine occurs at a highest rate and the passive water loss is recovered, so solving in part the osmoregulatory problem in marine fish. Therefore, inhibition of CA enzymes by cadmium could alter [HCO3\_]i and/or pHi leading to a reduction of salt absorption and consequently impairing the osmoregulation of marine fish.

More recently, Soyut et al (2008) demonstrated Co2+, Cu2+, Zn2+, Ag+, and Cd2+ to be potent inhibitor for brain CA enzyme activity in Rainbow trout (*Oncorhynchus mykiss),* with the following sequence Co2+ >Zn2+ >Cu2+>Cd2+>Ag+. They also demonstrated that Co2+, Ag+, and Cd2+ inhibit the enzyme with competitive manner, Cu2+ inhibits with noncompetitive manner, and Zn2+ with uncompetitive manner.

Ceyhun et al., 2011 *in vitro* demonstrated Al+3, Cu+2, Pb+2, Co+3, Ag+1, Zn+2 and Hg+2 to exert inhibitory effects on fish liver CA. Metal ions inhibited the enzyme activity at low concentrations. Al+3 and Cu2+ resulted the most potent inhibitors of the CA enzyme. All the metals inhibited CA in competitive manner and aluminium showed to be the best inhibitor for fish liver CA. Concerning the mechanism of inhibition, the authors argued a possible interaction of the metal with the histidines exposed on the surface of the molecule and/or other aminoacids around the active site.

In invertebrates Vitale et al (1999)demonstrated cadmium, copper and zinc to *in vitro* inhibit CA activity in the gills of the estuarine crabs *Chasmagnathus granulate* (Tab.1*). The* inhibitory potentials of the three metals on CA was in the following sequence: Cu2+ > Zn2+ > Cd2+. The observed inhibitory effect *in vitro* was confirmed by a corresponding inhibitory effect *in vivo.* 

In the euryhaline crabs *Callinectes sapidus and Carcinus maenas* Skaggs et al (2002) also documented a significant *in vitro* inhibition of gill CA by Ag+, Cd2+, Cu2+ and Zn2+. The binding affinities of the metals were one thousand times weaker for cytoplasmic CA from the gills of *C. maenas* than that from *C. sapidus*. The large differences in Ki values (Tab.1) suggests the presence of two different CA isoforms in the gills of these species, with *Callinectes sapidus* possessing a highly metal-sensitive CA isoform and *Carcinus maenas*  having a metal-resistant isoform. Interestingly, heavy metal inhibition of CA from the gills of another euryhaline crab, *Chasmagnathus granulata*, (as reported by Vitale et al., 1999, see above) appears to be intermediate between that found in the other two species. Moreover, in *Callinectes sapidus* CA isolated from the cytoplasmic pool of gill homogenates was much more sensitive to heavy metal inhibition than was CA from the microsomal fraction, which is believed to be anchored to the basolateral membrane, and as such, it exists within a lipidrich environment. The authors argued that metal could be sequestered in the lipid component of the microsomal fraction and, therefore, higher amounts of metals are required to achieve an effective concentration of free metals available for CA inhibition. However, the authors did not considered the time-dependence of the inhibition, which can be an important aspect to be taken into account (see Lionetto et al., 2000) in the analysis of membrane bound vs cytosolic isoform CA inhibition.

In humans Ekinci et al (2007) demonstrated the inhibition of two human carbonic anhydrase isozymes *in vitro*, the cytosolic HCA I and II by lead, cobalt and mercury. Lead was a noncompetitive inhibitor for HCA-I and competitive for HCA-II, cobalt was competitive for HCA-I and noncompetitive for HCA-II and mercury was uncompetitive for both HCA-I and HCA-II. Lead was the best inhibitor for both HCA-I and HCA-II.

In tab.1 the Ki, IC50 values and the type of inhibition for several heavy metals on CA from different vertebrate and invertebrate species is summarized. A great variability among species, tissues and metals can be observed. This suggests that the inhibitory mechanisms through which heavy metals exert their effect on carbonic anhydrase activity could be different for different isoenzymes and that also small structural differences between CA isoforms could result in different metal binding affinities.

More recently, Soyut et al (2008) demonstrated Co2+, Cu2+, Zn2+, Ag+, and Cd2+ to be potent inhibitor for brain CA enzyme activity in Rainbow trout (*Oncorhynchus mykiss),* with the following sequence Co2+ >Zn2+ >Cu2+>Cd2+>Ag+. They also demonstrated that Co2+, Ag+, and Cd2+ inhibit the enzyme with competitive manner, Cu2+ inhibits with noncompetitive

Ceyhun et al., 2011 *in vitro* demonstrated Al+3, Cu+2, Pb+2, Co+3, Ag+1, Zn+2 and Hg+2 to exert inhibitory effects on fish liver CA. Metal ions inhibited the enzyme activity at low concentrations. Al+3 and Cu2+ resulted the most potent inhibitors of the CA enzyme. All the metals inhibited CA in competitive manner and aluminium showed to be the best inhibitor for fish liver CA. Concerning the mechanism of inhibition, the authors argued a possible interaction of the metal with the histidines exposed on the surface of the molecule and/or

In invertebrates Vitale et al (1999)demonstrated cadmium, copper and zinc to *in vitro* inhibit CA activity in the gills of the estuarine crabs *Chasmagnathus granulate* (Tab.1*). The* inhibitory potentials of the three metals on CA was in the following sequence: Cu2+ > Zn2+ > Cd2+. The observed inhibitory effect *in vitro* was confirmed by a corresponding inhibitory effect *in vivo.*  In the euryhaline crabs *Callinectes sapidus and Carcinus maenas* Skaggs et al (2002) also documented a significant *in vitro* inhibition of gill CA by Ag+, Cd2+, Cu2+ and Zn2+. The binding affinities of the metals were one thousand times weaker for cytoplasmic CA from the gills of *C. maenas* than that from *C. sapidus*. The large differences in Ki values (Tab.1) suggests the presence of two different CA isoforms in the gills of these species, with *Callinectes sapidus* possessing a highly metal-sensitive CA isoform and *Carcinus maenas*  having a metal-resistant isoform. Interestingly, heavy metal inhibition of CA from the gills of another euryhaline crab, *Chasmagnathus granulata*, (as reported by Vitale et al., 1999, see above) appears to be intermediate between that found in the other two species. Moreover, in *Callinectes sapidus* CA isolated from the cytoplasmic pool of gill homogenates was much more sensitive to heavy metal inhibition than was CA from the microsomal fraction, which is believed to be anchored to the basolateral membrane, and as such, it exists within a lipidrich environment. The authors argued that metal could be sequestered in the lipid component of the microsomal fraction and, therefore, higher amounts of metals are required to achieve an effective concentration of free metals available for CA inhibition. However, the authors did not considered the time-dependence of the inhibition, which can be an important aspect to be taken into account (see Lionetto et al., 2000) in the analysis of

In humans Ekinci et al (2007) demonstrated the inhibition of two human carbonic anhydrase isozymes *in vitro*, the cytosolic HCA I and II by lead, cobalt and mercury. Lead was a noncompetitive inhibitor for HCA-I and competitive for HCA-II, cobalt was competitive for HCA-I and noncompetitive for HCA-II and mercury was uncompetitive for both HCA-I and

In tab.1 the Ki, IC50 values and the type of inhibition for several heavy metals on CA from different vertebrate and invertebrate species is summarized. A great variability among species, tissues and metals can be observed. This suggests that the inhibitory mechanisms through which heavy metals exert their effect on carbonic anhydrase activity could be different for different isoenzymes and that also small structural differences between CA

manner, and Zn2+ with uncompetitive manner.

other aminoacids around the active site.

membrane bound vs cytosolic isoform CA inhibition.

HCA-II. Lead was the best inhibitor for both HCA-I and HCA-II.

isoforms could result in different metal binding affinities.



Table 1. Ki, IC50 and type of inhibition for several heavy metals in different species and tissues as assessed in *in vitro* studies.

Concerning the mechanisms of inhibition some heavy metals are believed to bind to CA not at the specific catalytic site of CO2 hydration but nearby in a pocket, the so called 'proton

**inhibition Tissue Species Ref** 

*mykiss* 

*labrax* 

*Ictalurus punctatus* 

*granulata*

*Callinectes sapidus* 

*Carcinus maenas* 

*mykiss* 

*labrax* 

*labrax* 

*labrax* 

*labrax* 

(CAI) *Homo sapiens* Ekinci et al.,

(CAII) *Homo sapiens* Ekinci et al.,

(CAI) *Homo sapiens* Ekinci et al.,

(CAII) *Homo sapiens* Ekinci et al.,

(CAI) *Homo sapiens* Ekinci et al.,

(CAII) *Homo sapiens* Ekinci et al.,

Soyut et al., 2008

Christensen and Tucker, 1976

Vitale et al., 1999

Skaggs and Hery, 2002

Skaggs and Hery, 2002

Soyut et al., 2008

Ceyhun et al., 2011

2007

2007

Ceyhun et al., 2011

Ceyhun et al., 2011

2007

2007

2007

2007

Ceyhun et al., 2011

Ceyhun et al., 2011

competitive brain *Oncorhynchus* 

cells

Gills

Gills

(cytoplasmic isofom)

(cytoplasmic isofom)

Erytrocytes

Erytrocytes

**IC50 (M) Type of** 

n.d. 6.50 10-5 n.d. Red blood

1.75 10-5 7.15 10-5 Competitive liver *Dicentrarchus* 

n.d. 3.75 10-6 n.d. gills *Chasmagnathus* 

Co2+ 5 10-5M 1.40 10-5 competitive brain *Oncorhynchus* 

5.32 10-4 3.16 10-4 competitive liver *Dicentrarchus* 

competitive

Al3+ 1.48 10-4 6.92 10-5 competitive liver *Dicentrarchus* 

Pb2+ 2.42 10-4 1.13 10-4 competitive liver *Dicentrarchus* 

competitive

Hg2+ 7.68 10-4 4.48 10-4 competitive liver *Dicentrarchus* 

Table 1. Ki, IC50 and type of inhibition for several heavy metals in different species and

Concerning the mechanisms of inhibition some heavy metals are believed to bind to CA not at the specific catalytic site of CO2 hydration but nearby in a pocket, the so called 'proton

3.91 10-3 n.d. competititve Erytrocytes

5.6 10-5 n.d. uncompetitive Erytrocytes

1.42 10-3 n.d. uncompetitive Erytrocytes

3.12 10-4 n.d. uncompetitive Erytrocytes

**Metal** 

**Average value of Ki (M)** 

Cu2+ 27.6 10-3M 3.00 10-2 Non

3.60 10-7 n.d. n.d.

6.0 -25.0 10-4 n.d. n.d.

1.7 10-3 n.d. non

9.90 10-4 n.d. Non

tissues as assessed in *in vitro* studies.

shuttle' as demonstrated for human CAII (Tu et al., 1981). His-64 is a proton shuttle in catalysis, where it accepts the proton product (via the bridging solvent molecules) from zinc-bound water as zinc-bound hydroxide is regenerated; subsequently, the proton product is passed along to buffer (Liang et al, 1988; Tu et al., 1989; Vedani et al., 1989). The mechanism of inhibition of heavy metals on proton shuttle has been elucidated for copper on human CA II. Cu2+ is believed to competitively inhibit CAII by binding to the imidazole side chain of His-64, blocking its role in proton transfer from the zinc-bound water molecule to buffer molecules located outside of the active site region [Tu et al., 1981]. However, the knowledge of the mechanism of action of other metals on different CA isoforms is lacking. It cannot be excluded the CA binding to other different parts of the protein, possibly cysteine residues, as demonstrated in studies with other enzymes for silver and mercury.
