Clustered Conserved Cysteines in Hyaluronan Synthase Mediate Cooperative Activation by Mg²⁺ Ions and Severe Inhibitory Effects of Divalent Cations

Abstract

Hyaluronan synthase (HAS) uses UDP-GlcUA and UDP-GlcNAc to make hyaluronan (HA). Streptococcus equisimilis HAS (SeHAS) contains four conserved cysteines clustered near the membrane, and requires phospholipids and Mg²⁺ for activity. Activity of membrane-bound or purified enzyme displayed a sigmoidal saturation profile for Mg²⁺ with a Hill coefficient of 2. To assess if Cys residues are important for cooperativity we examined the Mg²⁺ dependence of mutants with various combinations of Cys-to-Ala mutations. All Cys-mutants lost the cooperative response to Mg²⁺. In the presence of Mg²⁺, other divalent cations inhibited SeHAS with different potencies (Cu²⁺~Zn²⁺ >Co²⁺ >Ni²⁺ >Mn²⁺ >Ba²⁺ Sr²⁺ Ca²⁺). Some divalent metal ions likely inhibit by displacement of Mg²⁺-UDP-Sugar complexes (e.g. Ca²⁺, Sr²⁺ and Ba²⁺ had apparent K_i values of 2-5 mM). In contrast, Zn²⁺ and Cu²⁺ inhibited more potently (apparent K_i ≤ 0.2 mM). Inhibition of Cys-null SeHAS by Cu²⁺, but not Zn²⁺, was greatly attenuated compared to wildtype. Double and triple Cys-mutants showed differing sensitivities to Zn²⁺ or Cu²⁺. Wildtype SeHAS allowed to make HA prior to exposure to Zn²⁺ or Cu²⁺ was protected from inhibition, indicating that access of metal ions to sensitive functional groups was hindered in processively acting HA•HAS complexes. We conclude that clustered Cys residues mediate cooperative interactions with Mg²⁺ and that transition metal ions inhibit SeHAS very potently by interacting with one or more of these –SH groups.

The linear glycosaminoglycan hyaluronan (HA) is composed of unmodified disaccharide repeats: [-D-glucuronic acid-(β1,3)-D-N-acetylglucosamine-(β-1,4)]_n. HA is ubiquitous in vertebrate extracellular matrices, and it is a major component in tissues such as cartilage and dermis, and in synovial and vitreous fluids [1,2]. HA is also made by some prokaryotes and fungi [3-5]. HA plays an important role during fertilization, embryogenesis, development, wound healing, and differentiation [6,7] and in a variety of cellular functions such as migration and matrix assembly [8-10].

The streptococcal and eukaryotic Class I HASs are membrane proteins [11] containing 6 or 8 membrane domains, respectively, with topologies predicted to be similar to that of SpHAS (Figure 1A), which was determined experimentally [12]. The mammalian enzymes belong to three subfamilies, designated HAS1, HAS2 and HAS3, which are ~30% identical to the streptococcal HASs from pyogenes, equisimilis, and uberis. The molecular masses of the streptococcal synthases are ~49 kDa, which is relatively small to mediate the multiple functions required for HA synthesis [13]. HAS enzymes must bind the UDP-GlcUA and UDP-GlcNAc precursors, catalyze two distinct glycosyltransferase reactions, bind to the growing HA-UDP chain, translocate the growing HA polymer through the enzyme and thus the cell membrane (Figure 1B), and finally release the HA chain to the cell surface or medium; typically after >10,000 disaccharide units (>4 × 10⁶ Da) have been polymerized.

Figure 1. Conserved SeHAS Cys residues are clustered in the active sites and at the membrane surface.

A. The linear scheme illustrates the topology and organization of HAS domains within the membrane and the relative locations of the four Cys residues, indicated in green circles, conserved in the Class I HAS family. C226 and C367 are located in membrane domains and are thus at the membrane interface. Based on chemical inhibition studies (Kumari and Weigel, 2005), C281 is in close proximity to C367 and C262 is close to C281, as indicated by the dashed red arrows. B. The compact scheme for organization of HAS (thick black lines) in the membrane illustrates the growing HA chain (alternating blue and red circles) being assembled by addition of the UDP-GlcUA and UDP-GlcNAc substrates at its reducing end, while it simultaneously is translocated through an intraprotein pore to the cell exterior. The four conserved Cys residues are localized at the membrane surface and are within the active sites (blue oval). Each catalytic cycle of the novel processive mechanism for HA biosynthesis by Class I HAS enzymes alternately adds HA-UDP to free UDP-GlcNAc or UDP-GlcUA to create HA with a new extended reducing end attached to UDP [13].

Cell-free HA biosynthesis was achieved in 1959 [3] and active HAS preparations were later obtained from detergent extracts of eukaryote [14,15] and prokaryote [16,17] membranes. Active purified recombinant enzymes were ultimately reported for Group A Streptococcus pyogenes HAS (SpHAS) and Group C Streptococcus equisimilis HAS (SeHAS) expressed in E. coli [18] and for mouse HAS1 expressed in COS-1 cells [19]. The streptococcal enzymes have been characterized kinetically [20,21], and shown to function as monomers [22] in complex with phospholipids [18], particularly cardiolipin (CL). The phosopholipid composition of E. coli membranes is typically (in mole %): 75% PG, 20% PE, and 5% CL. This CL content supports a high level of recombinant SeHAS expression (~10% of total membrane protein) with high activity [23]. Based on its’ lipid-dependence, multiple membrane domains and processive mechanism, we proposed that SeHAS contains an intraprotein pore (Figure. 1B) through which HA is synthesized and simultaneously translocated across the membrane to the cell exterior [18]. The Class I HASs mediate HA chain growth at the reducing end [24-26]. In this case, the UDP-sugars are acceptors and hyaluronyl-UDP chains are the donors in the two glycosyltransferase reactions required to make HA [13].

SeHAS is the smallest (417 amino acids) member of the Class I HAS family [27] and contains only four Cys residues (at positions 226, 262, 281, and 367) that are highly conserved within the Class I HASs, in particular in all the mammalian HASs. We previously created a panel of Cys-to-Ala mutants in order to study the role of these residues in HAS structure and function [28,29]. We found that the four Cys residues in active enzyme are not disulfide bonded and they are not required for activity, although Cys modification by sulfhydryl reagents (e.g. N-ethylmaleimide) can severely inhibit the enzyme .

Here we investigated the involvement of these four SeHAS Cys residues in the cooperative stimulation of enzyme activity by Mg²⁺, and the sensitivity of HAS activity to other ions, including transition metal ions. We found that the enzyme response to Mg²⁺ was sigmoidal, rather than hyperbolic, and abrogated in all Cys-mutants. Cys residues are also responsible for most of the severe inhibition by Cu²⁺, but not by Zn²⁺. The results indicate important though subtle roles for the four conserved Cys residues in the HAS catalyzed assembly of HA.

Materials and Methods

Materials, strains and plasmids

Reagents were supplied by Sigma-Aldrich, unless noted otherwise. Media components were from Difco (Fisher Scientific). The gene encoding HAS from S. equisimilis was inserted into the pKK223-3 vector (Amersham Pharmacia Biotech) and cloned into E. coli SURE™ cells [27]. The gene in this vector is driven by the tac promoter, which is regulated by the lac repressor and induced with 1 mM isopropyl-β-D-thiogalactoside. The SeHAS gene contains a C-terminal in-frame sequence encoding a 6-His tail to facilitate enzyme purification [18].

Cell growth and membrane preparation

E. coli SURE™ cells containing SeHAS-encoding plasmids were grown at 30°C in Luria broth to an A₆₀₀ ~1.6, induced for 3 hr, harvested and washed at 4°C, and stored frozen in PBS with 20% glycerol at -80°C. Membranes were prepared by osmotic shock, sonication and differential centrifugation as described previously [18]. Protein concentrations were determined with the Coomassie protein assay reagent (Pierce) using bovine serum albumin as the standard [30].

Preparation and characterization of SeHAS mutants

The generation of single, double, triple, and null Cys-to-Ala mutants of SeHAS by site directed mutagenesis, as well as their expression and kinetic characterization, were described previously [28]. All the mutants used here have substantial levels of activity, in some cases similar to wildtype.

HA synthase assays

The activity of wildtype and Cys-mutant SeHAS in membranes was determined in 100 μl of 25 mM Na₂KPO₄, pH 7.0, containing 50 mM NaCl, 20 mM MgCl₂ (unless indicated otherwise), 0.1 mM EDTA, 2 M glycerol, 1.0 mM UDP-GlcUA, 1.0 mM UDP-GlcNAc and 0.69 μM UDP-[¹⁴C]GlcUA (380 mCi/mmol; New England Nuclear), a mixture of protease inhibitors, and the indicated divalent cation chloride salt to be tested. Membranes were added to initiate the enzyme reaction and the mixture was gently agitated in a MicroMixer E-36 (Taitec) at 30°C for 1 hr. For the conditions reported, the assays used limiting amounts of SeHAS protein and the kinetics of HA synthesis were linear. Reactions were terminated by the addition of SDS to a final concentration of 2% (w/v), and the incorporation of [¹⁴C] GlcUA into HA was assessed by descending paper chromatography and scintillation counting as described previously [31]. HAS activity was normalized to total membrane protein as described previously [27,28]. Each datum point is the average of duplicates or triplicates, and the individual values were typically within 10%. Results with a particular combination of SeHAS mutant and divalent cation were verified in 2-3 independent experiments. For experiments to assess inhibition of purified enzyme by divalent cations, wildtype SeHAS was purified and assayed as described previously [21,18] in the presence of 20 mM MgCl₂, 2 mM bovine cardiolipin, and the indicated concentration of divalent cation chloride salt to be tested. The cooperativity of enzyme kinetc responses to increasing MgCl₂, concentration were assessed by calculating the Hill coefficient [32].

Results

The effect of Mg²⁺ concentration on SeHAS activity

The Mg²⁺ requirement for SpHAS activity has been known for over five decades [3], and all other HAS enzymes, such as SeHAS [27], XlHAS [33], and mammalian HASs [34] also require Mg²⁺. This requirement for Mg²⁺ is considered to reflect the typical enzyme recognition of nucleotide-Mg²⁺ complexes. Surprisingly, the stimulation profile of membrane-bound wildtype SeHAS activity with increasing Mg²⁺ concentration was sigmoidal, with a Hill coefficient of 2, indicating a substantial cooperative binding behavior (Figures 2 and 3, open circles). Maximum stimulation of SeHAS activity occurred at 10-12 mM MgCl₂.

Figure 2. Effect of Mg⁺² on the activity of SeHAS single, double, and null Cys-mutants.

A. HAS activity was determined in the presence of increasing concentrations of MgCl₂, with no other divalent salts present, as described in Methods, using membranes containing wildtype or the indicated SeHAS double Cys-mutants. Panel B is a blow-up of the 0-1 mM range from A. The SeHAS variants are: wildtype, red circle; C(226,262)A, yellow triangle; C(226,281)A, green diamond; C(226,367)A, dark blue square; C(262,281)A, light blue inverted triangle; C(281,367)A, gray circle ; Cys-null, black square. C. Membranes containing single Cys-mutants were assayed as above: C(226)A, black triangle; C(262)A, black square; C(281)A, black circle; or C(367)A, black inverted triangle.

Figure 3. Effect of Mg⁺² on the activity of SeHAS triple Cys-mutants.

A. HAS activity was determined as in Figure 2 using membranes containing wildtype SeHAS or the four triple Cys-mutants. The single remaining Cys residue is designated for the indicated triple-Cys-mutants: wildtype, white circle; △3C(C226), ■; ; △3C(C262), ●; △3C(C281), ▲; △3C(C367), ▼. B shows the 0 - 1.0 mM region of panel A enlarged.

The cooperative response of membrane-bound SeHAS to Mg²⁺ is eliminated in single, double, triple or null Cys-mutants

To determine the possible involvement of Cys residues in the SeHAS response to Mg²⁺, we assessed the Mg²⁺-dependence of the single (Figure 2C), double or null (Figures 2A and 2B), and triple (Figures 3A and 3B) SeHAS Cys-mutants. The expression and activity characterization of these Cys-to-Ala mutants were previously reported [28].

The double Cys-mutants seemed most revealing, showing no sigmoidal activation in the presence of increasing Mg²⁺, although at high concentrations all the mutants were substantially active (Figure 2A). In contrast, the sigmoidal behavior of wildtype SeHAS was very evident at lower MgCl₂ (Figure 2B). At concentrations ≥ 1 mM Mg2+, the mutants displayed several types of kinetic behavior. The null and C(226,367)A Cys-mutants, and to a lesser degree the C(262,281) mutant, were efficiently activated at ≤0.2 mM, and then showed nearly hyperbolic increases from 0.2-1.0 mM MgCl₂ (Figure 2B). In contrast, the C(226,281)A and C(281,367)A mutants were activated at low MgCl₂ just slightly more than wildtype. The C(226,262)A mutant and wildtype enzyme were virtually identical and were only minimally activated at low MgCl₂ (Figure 2B), consistent with a sigmoidal response. As the MgCl₂ concentration increased above 1 mM, all the mutants were stimulated in a nearly linear and parallel manner (Figure 2A). All the double Cys-mutants and the Cys-null enzyme were active at 15 mM MgCl₂ (35% to 85% of wildtype; Figure 2A). Wildtype enzyme activity was saturated at 15 mM MgCl₂, whereas activities of the mutants were still increasing. The results show that modifying any pair of Cys residues can drastically alter, but not eliminate, the response to increasing Mg²⁺ concentration.

Similarly, each of the four single Cys-mutants showed loss of the characteristic sigmoidal increase in enzyme activity with increasing MgCl₂ (Figure 2C). The initial enzymatic rates for all four mutants at low Mg²⁺ were faster than wildtype and essentially hyperbolic, although the activities of each mutant varied. Results with the four triple Cysmutants were also similar in that none showed a cooperative response to increasing Mg²⁺ concentration (Figure 3). These mutants were substantially less active than wildtype enzyme (Figure 3A) and showed only modest hyperbolic profiles with increasing MgCl₂ concentration (Figure 3B).

Many divalent metal ions inhibit purified wildtype SeHA

The Mg²⁺ requirement for HAS activity has been known for decades, but had not yet been determined for a purified HAS enzyme. As with the membrane-bound enzyme, the stimulation profile of SeHAS activity with increasing Mg²⁺ concentration was sigmoidal, indicating a cooperative binding behavior with a Hill coefficient of 2 (Figure. 4A; dashed line). Maximum stimulation of SeHAS activity occurred at 10 mM MgCl₂ with an apparent K_m of ~5 mM. The MgCl₂ concentration was then kept constant at 20 mM in order to examine the potential of other divalent metal ions to affect HA polymerization. An earlier report indicated that SpHAS has a small amount of activity with Mn²⁺ and Co²⁺ [3], but we found for SeHAS that none of the eight divalent metal ions tested (Zn²⁺, Cu²⁺, Co²⁺, Ni²⁺, Mn²⁺, Ba²⁺, Sr²⁺ or Ca²⁺) could effectively substitute for Mg²⁺ (not shown). Despite their inability to support HA synthesis, all of these divalent cations inhibited SeHAS activity in a dose-dependent manner (Figure 4A). Several transition metal ions, particularly Zn²⁺ and Cu²⁺ were inhibitory in the sub-mM range. The order of inhibitory effect (greatest-to-least) was: Cu²⁺~ Zn²⁺>Co²⁺ >Ni²⁺ >Mn²⁺ >Ba²⁺ >Sr²⁺ Ca²⁺. The apparent K_i values for Zn²⁺ and Cu²⁺ were essentially identical at 100-200 μM, whereas the K_i for Ca²⁺ was ~12 mM. To assess the generality of the potent inhibition of SeHAS by Cu²⁺, we also examined its effect on purified SpHAS, which contains six Cys residues - two others in addition to the four conserved Cys residues in SeHAS. Both enzymes showed virtually identical sensitivity to inhibition by CuCl₂ (Figure 4B).

Figure 4. Mg²⁺-dependence of purified SeHAS and the effects of other divalent cations on enzyme activity.

A. Purified SeHAS was incubated with the indicated concentration of MgCl₂ in the absence of any other divalent metal ions (dashed line, black circles). In separate experiments, purified SeHAS in the presence of 20 mM MgCl₂ was incubated with the indicated concentrations of one of the eight divalent cation chloride salts noted in the symbol legend (right of figure), and HA synthase activity was quantified as described in Methods. The divalent cations salts tested were: MnCl₂ (black triangle); CuCl₂ (blue inverted triangle); CoCl₂ (green hexagon); CaCl₂ (white circle); BaCl₂ (white square); NiCl₂ (white triangle); ZnCl₂ (yellow hexagon); SrCl₂ (black square). B. Purified SeHAS (○) and SpHAS (●) were assayed in the presence of 20 mM MgCl₂ and the indicated concentration of CuCl₂.

Divalent cation inhibition of SeHAS Cys-mutants reveals multiple sensitivity components depending on the ion

The inhibition by Cu²⁺, Zn²⁺ and the other ions could be due to a direct effect on the enzyme or to an indirect effect, in particular the displacement and substitution of Mg²⁺ ions in UDP-sugar complexes. To examine the potential involvement of the conserved Cys residues in the inhibition of SeHAS by the divalent cations noted above, we tested their effects on the activity of the membrane-bound Cys-mutants in the presence of 20 mM MgCl₂ (Figure 5). The double Cys and Cys-null mutants showed very different inhibition patterns with CaCl₂ (Figure 5A) and CuCl₂ (Figure 5B), chosen as the least and most potent inhibitors, respectively. The mutants and wildtype showed essentially the same sensitivities to Ca²⁺, with most effects occurring between 0-1 mM and maximal inhibitions of 10-25% between 5-15 mM (Figure 5A). Thus, whether any Cys residues were present or not, all the variants retained roughly 80% of their activity in the presence of CaCl₂. In contrast, the double and null Cys-mutants displayed distinctly different inhibition patterns with CuCl₂ (Figure 5B), indicating that Cys residues are responsible for a substantial portion of the severe inhibition that essentially inactivates wildtype SeHAS. The C(226,367)A and C(281,367)A mutants were most similar to wildtype in their sensitivity to Cu²⁺, whereas the C(262,281)A and Cys-null mutants were the most resistant to Cu²⁺. The C(226,262)A and C(226,281)A mutants showed an intermediate sensitivity, retaining 20-40% of their activity between 1-10 mM CuCl₂ (while wildtype had <1% activity). Similar results were obtained with CdCl₂ (not shown).

Figure 5. The activities of SeHAS wildtype, double and null Cys-mutants are differentially affected by Ca⁺², Ni⁺² or Cu⁺².

Membranes containing SeHAS wildtype, null, or double Cys-mutants (coded as noted in the Panel A insert) were incubated in the presence of 20 mM MgCl₂ and the indicated concentrations of either CaCl₂ (A) or CuCl₂ (B) and HA synthase activity was determined as described in Methods. The SeHAS variants in A and B are: wildtype, red circle; Cys-null, black square, C(226,262)A, green triangle; C(226,281)A, blue diamond; C(226,367)A, yellow hexagon; C(262,281)A, light blue inverted triangle; C(281,367)A, gray circle. C. Membranes containing wildtype (○, □) or Cys-null (●, ■) SeHAS were incubated in the presence of 20 mM MgCl₂ and the indicated concentrations of CuCl₂ (●, ○) or NiCl₂ (■, □) and HA synthase activity was determined as described in Methods.

Figure 5C shows the sensitivity of wildtype and Cys-null SeHAS to inhibition by <1 mM CuCl₂ or ZnCl₂. The Cys-null mutant retained 80% activity between 0.2-1.0 mM Cu²⁺, whereas the wildtype enzyme was <1% active. In contrast, there was not much difference in relative resistance to ZnCl₂ between wildtype and Cys-null enzyme. These results indicate that the mechanisms by which Cu²⁺ and Zn²⁺ inhibit are different; that there are multiple mechanisms by which individual divalent cations can severely inhibit HAS. The sensitivity of the triple Cys-mutants to low concentrations (0-0.2 mM) of CuCl₂ illustrates this as well (Figure 6), since each mutant containing a different single Cys residue was much more active than wildtype enzyme, yet the mutants themselves showed different levels of activity (e.g. the mutant with C226 free was only 20% inhibited at 0.2 mM Cu²⁺, while the mutant with C367 free was ~55% inhibited).

Figure 6. The activities of SeHAS triple Cys-mutants are differentially affected by Cu⁺².

Membranes containing SeHAS wildtype or triple Cys-mutants were incubated in the presence of 20 mM MgCl₂ and the indicated concentrations of CuCl₂ and HA synthase activity was determined as described in Methods. The SeHAS variants (with the single free Cys indicated) were: wildtype, red circle; △3C(C226), green square; △3C(C262), dark blue triangle; △3C(C281), yellow inverted triangle; △3C(C367), light blue diamond.

HA synthesis protects SeHAS from divalent metal ion inhibition

Inhibition by Zn²⁺ or Ca²⁺, but not Cu²⁺, was partially rescued by EGTA (not shown), indicating that the interactions between some metal ions and SeHAS may have a higher affinity and not be readily reversible. If Zn²⁺ or Cu²⁺ ions interact with the active sites of the enzyme, then SeHAS might be protected from inhibition if it was already catalytically engaged in HA biosynthesis. Protection by substrates is particularly likely for HAS, since it functions by a processive mechanism; HAS•HA-UDP complexes do not dissociate during biosynthesis, until the final product polymer is released [35,36]. To test this, the effects of the two ions were assessed after first incubating purified enzyme in the absence (no HA) or presence of UDP-sugars to allow HA synthesis for 5, 10 or 15 min (Figure 7). The subsequent sensitivity of SeHAS to inhibition by either Cu²⁺ (Figure 7A) or Zn²⁺ (Figure 7B) was substantially diminished, in a time-dependent way, by allowing the enzyme to make HA before exposure to these ions. After 15 min of HA synthesis, the levels of enzyme inhibition by Zn²⁺ and Cu²⁺ were similar to that of Ca²⁺ (Figures 4A and 5A), indicating a residual moderate effect on Mg²⁺-UDP-sugar complex formation rather than an interaction with Cys residues. Thus, the inhibition by both Zn²⁺ and Cu²⁺ was substantially decreased when wildtype SeHAS was actively engaged in HA synthesis, consistent with hindered access to the clustered Cys residues in the active sites of HAS•HA-UDP complexes.

Figure 7. Purified SeHAS synthesizing HA is resistant to inhibition by ZnCl₂ or CuCl₂.

SeHAS was purified as described in Methods and, in a pretreatment, samples were incubated without (no HA; black circles), or with UDP-sugars for 5 (yellow triangles), 10 (green inverted triangles) or 15 (red squares) min, after which the indicated concentration of CuCl₂ (A) or ZnCl₂ (B) was added and the samples were incubated an additional 30 min under normal assay conditions to assess relative enzyme activities.

Discussion

An important point to note for the experiments performed in this study is that the E. coli SURE cells used for SeHAS expression do not support HA synthesis, since these cells lack the UDP-GlcUA precursor [31,37]. This was a key discovery that aided our initial successful cloning of SpHAS and, importantly, it means that the results presented here were obtained with membrane-bound or purified SeHAS that was free and not already in complex with HA-UDP (until UDP-sugars were added in vitro).

Many enzymes require divalent metal ions, with Mg²⁺ being one of the most prevalent, and SeHAS is no exception. Since glycosyltransferase substrates are UDP-sugars, it is expected that HASs utilize Mg²⁺ ions to help coordinate the pyrophosphoryl group of the substrates. Many X-ray structure examples are known in which Mg²⁺, water, and amino acid side chains (e.g. carboxylates) interact to bind and orient the nucleotide in an active site, making it ready for catalysis [38]. The cooperative response to increasing Mg²⁺ concentration indicates that SeHAS contains multiple low affinity Mg²⁺ binding sites and that the initial binding of Mg²⁺ to the enzyme facilitates binding of additional Mg²⁺ ions. Interestingly, the original report of a Mg²⁺ requirement for SpHAS also showed a slight sigmoidal response [3]. A cooperative response for Mg²⁺ activation has not yet been tested for in any other purified Class I HAS.

We previously showed [23] that the four Cys residues in SeHAS are clustered together at the membrane-protein junction and are either part of, or very near, UDP-sugar binding sites (Figure 1B). The present results show that alteration of one, two, three, or all four Cys residues in SeHAS eliminates the Mg²⁺-dependent cooperative activation of the enzyme, indicating a requirement for all four Cys residues in the sigmoidal response to Mg²⁺. However, Cys-null SeHAS shows no sigmoidal behavior, but is still Mg²⁺-dependent and active. This finding indicates that either the Cys -SH groups are not involved directly in coordination of Mg²⁺ in wildtype SeHAS or that if they do directly coordinate Mg²⁺, then this function is not required for activity in vitro. Although sulfhydryl groups can directly coordinate with some divalent metal ions, a cooperative Mg²⁺-binding response might also be indirectly mediated if the cluster of Cys residues facilitates a HAS conformation that is able to bind multiple Mg²⁺ ions in a cooperative way.

Although purified SeHAS had little or no activity when Ca²⁺, Sr²⁺, Ba²⁺, Mn²⁺, Ni²⁺, Co²⁺, Zn²⁺, or Cu²⁺ were substituted for Mg²⁺, these same metal ions in the presence of MgCl₂ inhibited the synthesis of HA to varying degrees. The Group 2A elements tested (Ca²⁺, Sr²⁺ and Ba²⁺) inhibited SeHAS less effectively (K_i values of 2-5 mM) than the transition elements (row 4: Mn²⁺, Co²⁺, Ni²⁺, Cu²⁺, and Zn²⁺) tested (K_i values of 0.2-2 mM), indicating that these two cation groups inhibit by different mechanisms. Ni²⁺, Co²⁺, Cu²⁺ and Zn²⁺ caused the greatest inhibition of SeHAS activity and generally have larger equilibrium binding constants and smaller atomic radii. Although ionic radii are variable properties of ions that can be influenced by factors such as spin, coordination number, and local environment, a general trend is that ionic radii increase within a descending periodic group. Ionic radii (in pm) for the ions used here are: Cu²⁺ (73); Zn²⁺ (75); Co²⁺ (65); Ni²⁺ (70); Mn²⁺ (82); Ba²⁺ (142); Sr²⁺ (125); Ca²⁺ (112). The observed correlation for these ions, which are listed in order of decreasing potency to inhibit SeHAS, is that ionic radius increases as ion inhibition ability decreases. Ionic size, however, does not solely explain overall inhibitory potency, e.g. among the four most potent ions.

Several mechanisms could explain the inhibition of SeHAS activity by divalent metal ions. One mechanism is that some ions displace Mg²⁺ in coordination with the pyrophosphoryl substrates, UDP-GlcUA and UDP-GlcNAc [39] and HA-UDP (in the case of synthesis at the reducing end of HA). A second known mechanism is that some metal ions (e.g. Ag⁺) can bind covalently to -SH or other groups in proteins [40]. A third established mechanism is the formation of high affinity coordination complexes between a metal ion and unpaired electrons in the orbitals of specific, spatially well positioned atoms in proteins [39,41,42]. Metal ions have known characteristic preferences for binding with particular amino acid R groups and atoms [43]. For example, Mg²⁺, Mn²⁺, Ca²⁺, Cr³⁺ and Co³⁺ are bound mainly to O-containing ligands, whereas Cu⁺, Ag⁺, Au⁺, Ti⁺, Pd²⁺, Pt²⁺, Cd²⁺, and Hg⁺ are generally bound to S-containing ligands. In contrast, Zn²⁺, Cu²⁺, Ni²⁺, Fe²⁺, and Co²⁺ preferentially bind to N-containing ligands.

Our results with purified SeHAS and the panel of membrane-bound Cys-mutants lead us to conclude that the first mechanism noted above (displacement of Mg²⁺ in UDP-substrate complexes) accounts for approximately 20-30% inhibition for any SeHAS variant, and is the sole mechanism of action of the least potent Group 2 ions (Ca²⁺, Sr²⁺ and Ba²⁺). This is also evident with the Cys-null SeHAS, for which the effect of the most potent inhibitor (Cu²⁺) on wildtype enzyme was similar to that of Ca²⁺ (Figure 5). The inhibition patterns of the more potent ions, particularly Cu²⁺, were distinct depending on the combination of the four Cys residues mutated in each SeHAS variant. The results indicate that whether directly or indirectly, all four Cys residues are involved in and required for ligation and high affinity binding of inhibitory ions to the protein. The third mechanism, coordination complex formation, noted above seems the most consistent with our results. One possibility is that the different inhibitory ions may be coordinated not only by one or more -SH groups in the Cys residues but also by other nearby preferred ligation atoms in the active sites. Alternatively, the Cys residues are required for a subtle conformation change that brings a set of coordinating atoms together that allows high affinity binding to appropriate ions (e.g. inhibition of SeHAS by Zn²⁺, Cu²⁺, Ni²⁺ and Co²⁺ could be due to the metal ion binding to N-containing amino acids in or near the active site).

Finally, we present a possible explanation for the cooperative response to Mg²⁺ that is speculative, but may be important since it addresses a novel aspect of HA synthase function, namely the lipid dependence of the Class I HAS family [11]. A possible reason for the cooperative activation of enzyme could be that Mg²⁺ ions are required either for the interactions between SeHAS and the multiple CL molecules in an active enzyme-lipid complex or for the ability to tightly pack CL molecules around SeHAS in order to activate it. In in vitro liposome systems Mg²⁺ has a strong interaction potential with CL [44] and can induce lamellar-to-inverted phase transitions and membrane fusion [45]. In intact E. coli membranes CL microdomains localize to regions of negative curvature [46]. Since CL has two head groups with four fatty acyl chains, its shape is conical and it naturally tends to create negative (concave) membrane curvature. In fact, HAS might be activated by CL because it decreases lateral pressure on the protein [47] and thus opens up the cytoplasmic active sites and intraHAS pore.

The close packing of negatively charged dual CL head groups (that can bind two Mg²⁺ ions), in regions of negative curvature might require a higher Mg²⁺ concentration to saturate formation of Mg²⁺-CL complexes. Previous radiation inactivation studies revealed that about 16 CL molecules are required for maximal activity of native SpHAS and SeHAS [22]. Thus, fully active HAS species are actually HAS-CL₁₆ complexes. Mg²⁺ binding to CL might facilitate formation of successive intermediate HAS•CL_x species. Intermediates approaching a fully activated HAS•CL₁₆ complex could become more and more stable, thus lowering the binding energy for the addition of each successive (Mg²⁺)₂-CL to the complex and creating a cooperative activation process.

For cells such as S. equisimilis or S. pyogenes the metabolic burden of continuously making an HA capsule and secreting large amounts of HA into the medium could be a factor in the evolution of structure-function relationships within their HA synthases. For example, the K_m of HAS for UDP-GlcNAc is relatively high so it does not interfere with the utilization of this import precursor by other glycosytransferases, e.g. for cell wall biosynthesis [37]. Since all cells require substantial Mg²⁺ concentrations for DNA, RNA, phospholipid and nucleotide metabolism, the cooperative activation of streptococcal HASs (thus requiring a higher Mg²⁺ concentration for HAS activity) might ensure that HA synthesis does not occur at the expense of cell survival by depleting too much Mg²⁺, energy, or sugar units.

This latter effect might have created continued selective pressure to retain the four Cys residues in mammalian HASs as well, since naturally occurring HAS mutants that lose the cooperative activation response to Mg²⁺, like the mutants used in this study, might impair cell survival or function. Whatever the reason, our results show that the cluster of four highly conserved Cys residues in SeHAS is required for the cooperative response to increasing Mg²⁺ concentration and for the severe inhibition of HAS function by divalent cations such as Cu²⁺.

Abbreviations

CL

Cardiolipin

HA

Hyaluronic Acid, Hyaluronate, Hyaluronan

HAS

HA Synthase

SeHAS

Streptococcus equisimilis HAS

SpHAS

Streptococcus pyogenes HAS