Understanding Ligninase-Mediated Reactions of Endocrine Disrupting

Jun 24, 2011 - Griffin, Georgia 30223, United States. §. Department of Civil and Environmental Engineering, University of Virginia, Charlottesville, ...
0 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/est

Understanding Ligninase-Mediated Reactions of Endocrine Disrupting Chemicals in Water: Reaction Rates and Quantitative StructureActivity Relationships Liang Mao,†,‡ Lisa M. Colosi,§ Shixiang Gao,†,* and Qingguo Huang‡,* †

State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing 210093, P. R. China ‡ College of Agricultural and Environmental Sciences, Department of Crop and Soil Sciences, University of Georgia, Griffin, Georgia 30223, United States § Department of Civil and Environmental Engineering, University of Virginia, Charlottesville, Virginia 22904, United States

bS Supporting Information ABSTRACT: We have verified in our previous work that lignin peroxidase (LiP) mediates effective removal of selected natural and synthetic estrogens. The efficiency of these reactions was greatly enhanced in the presence of veratryl alcohol (VA), a chemical that is produced along with LiP by certain white rot fungi, for example, Phanerochaete chrysosporium. In this study, we systematically evaluated the kinetic behaviors of LiP-mediated reactions for six endocrine disrupting compounds (EDCs), that is, steroid estrogens and their structural analogs, in both the presence and absence of VA. Resulting kinetic parameters were then correlated with structural features of LiP/ substrate binding complexes, as quantified using molecular simulation, to create quantitative structureactivity relationship (QSAR) equations. These equations suggest that binding distance between a substrate’s phenolic proton and δN of HIS470 s imidazole ring plays an important role in modulating substrate reactivity toward LiP in both the presence and absence of VA. This information provides insight into an important enzymatic reaction process that occurs in the natural environment affecting EDC transformation, a process that may be used in engineered systems to achieve EDC removal from water.

’ INTRODUCTION Endocrine disrupting chemicals (EDCs) exhibit a range of toxicity effects and represent one prominent contemporary challenge in water quality management and engineering.19 EDCs tend to be relatively bioaccumulative, persistent, and capable of eliciting physiological responses at extremely low concentrations.1,2,4 These features create unique challenges with respect to EDCs removal in conventional water treatment procedures.10,11 Enzyme-mediated oxidative coupling processes, as a class of highly efficient catalytic reactions, have been shown potentially promising in dealing with EDCs.3,7,8,12,13 A range of naturally occurring extracellular enzymes are able to catalyze reactions leading to removal of EDCs from water.8,14,15 Our earlier studies revealed that lignin peroxidase (LiP) mediates effective reactions of three steroid estrogens, estrone (E1), 17β-estradiol (E2), and estriol (E3) and one synthetic steroid 17R-ethinylestradiol (EE2), to form oligomeric products via radical coupling.3,12 Our results also showed that the efficiency of these LiP-mediated estrogen reactions is enhanced in the presence of VA, a secondary metabolite that Phanerochaete chrysosporium produces along with LiP in nature.12 r 2011 American Chemical Society

LiP contains at its catalytic center an iron porphyrin group, the reactivity of which is modulated by surrounding residues comprising a catalytic cavity on the distal side of the heme. ARG43, PHE46 and HIS47 (shown in Figure S1 in the Supporting Information (SI)) are said to be particularly important in modulating the reactivity of the LiP catalytic center.16 It is known that LiP mediates substrate oxidation via a three step catalytic cycle (shown at left in SI Figure S2). First, the ferric-porphyrin center of resting phase LiP is oxidized by H2O2 to yield an intermediate called LiP compound I (LiPI). One electron oxidation of a single molecule of phenolic substrate leads to formation of a second intermediate called LiP compound II (LiPII), and subsequent one electron oxidation of a second substrate molecule restores the enzyme to its neutral resting state.17 The third reaction step, oxidation of the second substrate molecule, is the rate-limiting step of the catalytic cycle.18,19 It is also known that LiPII can react with excess H2O2 to form a Received: December 21, 2010 Accepted: June 2, 2011 Published: June 24, 2011 5966

dx.doi.org/10.1021/es201460s | Environ. Sci. Technol. 2011, 45, 5966–5972

Environmental Science & Technology temporarily inactivated form LiPIII (step 4 in SI Figure S2), which can lead to permanent inactivation. VA is an active substrate of LiP 16,20 that can be oxidized to a cation radical (VA•+) during steps two and three of the LiP catalytic cycle.2123 This, in effect, helps the reduction of LiPII to complete the catalytic cycle and, thus, suppresses the side reaction leading to inactivated LiPIII. Some studies have also suggested that VA can bind directly to LiPIII and return it to the native LiP state (step 5 in SI Figure S2), thus further mitigating LiP inactivation.24 Recent studies indicates that the VA•+ cation radical can, in turn, bind with the oxoferryl moiety of LiPII to form a LiPII-VA•+ complex.24,25 Formation of this complex is believed to enhance LiP catalytic performance; however, such enhancement has not been verified by direct experimental data, and the molecular basis of such enhancement is unclear. In the present study, we have examined initial reaction rates for apparent removal of six EDCs, as mediated by LiP in the absence or presence of VA. Our results indicate that enzyme catalytic rate constant (kCAT) and enzyme affinity (KM) vary markedly among studied chemicals and also in the absence or presence of VA. It should be noted that our study focused on initial reaction rates such that enzyme inactivation is presumed to have no significant effect on measured kinetics parameters. Marked differences among kCAT and KM values for systems with or without VA are thus directly attributable to VA impacts on LiP catalytic performance. A novel quantitative structureactivity relationship (QSAR) approach, which considers enzymesubstrate binding conformation rather than just substrate characteristics, was used to explore observed variability in LiP catalytic behaviors on a molecular level. This approach was validated in our previous work with horseradish peroxidase (HRP)-mediated reactions of phenolic substrates.26 Using this approach, we were able to establish significant QSAR equations that provide insight about the mechanism of electron transfer during LiP reactions with EDCs; help elucidate the molecular basis for differences in LiP catalytic activity toward different substrates; and clarify the role of VA during LiP reactions. This understanding is important because LiP-mediated reactions of EDCs occur in the natural environment and thus influence their environmental fate and risks.12 Our earlier study also suggests that this enzymatic process may be used in engineered systems to achieve EDC removal from water.3

’ MATERIALS AND METHODS Materials. All reagents were ACS grade or better. Purity of all estrogens was >98%. EDCs E1, E2, E3, EE2, nonylphenol (NP), and bisphenol A (BPA), and veratryl alcohol (VA; 3, 4-Dimethoxybenzyl alcohol, 96%), lignin peroxidase (LiP), and hydrogen peroxide (30 wt %) were purchased from Sigma-Aldrich (St. Louis, MO). Molecular structures of E1, E2, E3, EE2, NP and BPA are provided in SI Figure S3. Acetonitrile (ACN) and methanol were HPLC grade, and obtained from Fisher Scientific (Pittsburgh, PA). Citric acid and dibasic sodium phosphate were purchased from J.T. Baker Chemical Co. (Phillipsburg, NJ). Assessment of MichaelisMenten Reaction Kinetics. Experiments were carried out to assess the initial reaction rates of each studied substrate at various initial concentrations, and the data were used to construct MichaelisMenten curves and conduct associated kinetic analyses. Reaction rates were also evaluated similarly for systems containing VA. Reactions were

ARTICLE

conducted at room temperature using glass test tubes as reactors and following a scheme that we have used in our earlier studies.12 Each reactor contained 2 mL reaction medium prepared in citrate-phosphate buffer solution (CPBS, 10 mM, pH 4.6), comprising various initial concentrations of a substrate (the range varied for different substrates but was generally between 0 and 32 μM, and at least seven different concentrations were examined), 0.005 U mL1 LiP, 10 μM H2O2, and with or without the presence of 1 mM VA. The working solution with the highest EDC concentration, that is, 32 μM was prepared in buffer from concentrated stock solutions. The final methanol concentration in this working solution was less than 5% and we have tested in our previous work that such a methanol concentration did not interfere with LiP reactions.3 Lower concentrations were prepared via serial dilution in buffer from this working solution. Reactions were initiated by addition of H2O2 and then handshaken for 40 s prior to termination of reaction by addition of 2 mL methanol. The mixture of reaction solution and methanol was then sampled for HPLC analysis as described in the SI part II. Three replicate experimental reactors and one blank were prepared and tested for each reaction condition. Blank reactors were identical to experimental reactors except that they did not contain LiP. Only one blank reactor was used for each reaction condition because preliminary tests showed excellent agreement among sets of three replicate blanks. Following HPLC measurement of substrate concentration in both the blank (S0) and experimental tubes (St), initial reaction rates (r0) were calculated using the formulation r0 = (S0  St)/ Δt. The reaction time Δt was 40 s for all experiments, based on preliminary tests in which it was determined that this time interval captures the pseudofirst-order rate behavior of the enzymatic reactions. This interval also ensured reproducible handling of the reactors. Initial rate data were then fit to MichaelisMenten equation, r0 = rmax  S0/(S0 + KM), to obtain each substrate’s maximum reaction rate (rmax) and Michaelis constant (KM). We assumed the 40 s reaction time was sufficiently short such that the active enzyme concentration remained unchanged from the initial enzyme concentration [E0], and thus calculated a normalized maximum initial reaction rate according to kCAT = rmax/[E0]. Computational Models for LiPII and LiPII-VA•+ Complex. Molecular simulation was performed using molecular and quantum mechanics algorithms available as part of HyperChem Molecular Modeling System, release 8.0 (Hypercube, Inc.: Gainesville, Florida). For substrate chemicals, preliminary geometry optimization was achieved using the OPLS molecular mechanics force field and the Polak-Ribiere optimization algorithm. Subsequent quantum optimization for electronic structure was achieved using the ZINDO/1 semiempirical method and Polake-Ribiere optimization. Following determination of optimal quantum structures, the following properties were computed for each substrate: molecular volume (VM); energy of the highestoccupied molecular orbital (EHOMO); and energy of the lowestunoccupied molecular orbital (ELUMO). The structural coordinates of a model LiP (entry 1B82) were downloaded from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB). To obtain the LiPII model from this resting enzyme model, we added an oxygen to the heme iron and created a double bond between the oxygen and the iron atom. Geometry optimization for heme iron and its double-bonded oxygen atom was achieved using the OPLS and the Polak-Ribiere algorithms, with all residues fixed except the heme iron and its double-bonded oxygen atom. 5967

dx.doi.org/10.1021/es201460s |Environ. Sci. Technol. 2011, 45, 5966–5972

Environmental Science & Technology Following optimization, the ironoxygen bond length was 1.73 Å, a reasonable value comparable to ironoxygen bonds in compound II forms of other peroxidases. For example, the ironoxygen bond in HRPII (C1A, entry 1h55) is 1.83 Å and in CCPII (Cytochrome C peroxidase) is 1.65 Å.27 Prior to using the LiPII model in molecular simulations, we computationally reassigned an appropriate charge distribution for a subset structure comprising the porphyrin ring, the heme iron, its associated oxygen, and proximal residue HIS176.28 The ZINDO/1 UHF method was used to assign charges on these elements, and all other residues were constrained in space using the “Mechanical Atoms” function in HyperChem. The total charge of the subset structure was set to zero, and the multiplicity (2S+1) was set to triplet. This reflects the two unpaired electrons on the FedO center.29,30 The charges on the heme iron, heme-bound oxygen, and N on HIS1760 s imidazole ring were (+0.05), (0.34), and (0.13), respectively, as shown in SI Figure S4 (left). We used LiPII in subsequent enzymesubstrate binding simulations because it is known that the third step in LiP catalytic cycle (i.e., the LiPII-mediated step in SI Figure S2), is rate limiting. To achieve a suitable computational model of the LiPII-VA•+ complex, the LiPII model referenced above was merged with that of VA•+. It was known that the heme in LiP is buried inside the protein with only a small opening connecting the active site to the outside of the protein (SI Figure S5). It has also been shown that this small opening allows for diffusion of VA•+ into the heme site.31 As such, VA•+ was placed at a random location within this small opening and then subjected to a series of optimizations (as described in SI part II) to achieve a LiP II-VA•+ complex model (as shown in SI Figure S4 right). The distance between VA•+ and heme was around 7.2 Å. The same process described above was used to reassign charge distribution for the subset of LiP II comprising VA•+, the heme iron and its associated oxygen in LiP II-VA•+ model. The multiplicity was set to quartet to reflect the two unpaired electrons of the FedO and one unpaired electron of the VA•+.25 The calculated net charges on the heme iron (+0.01) and the heme-bound oxygen (0.16) are shown in SI Figure S4 (right). Simulation of EnzymeSubstrate Binding and Calculation of Average Binding Distances. To simulate binding between a substrate and LiPII, the computational model of LiPII was merged with that of a substrate. The substrate was first placed at a random position within LiP II0 s distal region, and then subjected to a series of optimizations to achieve a LiP II-substrate complex model. The OPLS algorithm was then utilized in a 1000-step Monte Carlo (MC) simulation, during which all model components were held fixed in place except for the substrate and the three distal residues known to be critical during substrate docking: ARG43, PHE46 and HIS47. Following completion of the first 1000-step MC simulation, the lowest energy conformation of substrate was recalled and its geometry was optimized using OPLS and Polak-Ribiere. The resultant optimized geometry was then used as starting configuration for the next 1000-step MC simulation in which the substrate, ARG43, PHE46, and HIS47 were once again allowed to move freely. This process was repeated for the same substrate at 810 randomly selected starting positions. Each starting position usually required 38 sets of 1000-step MC simulations to achieve optimization. For each step of the MC simulation, we recorded six pieces of information. These included the potential energy corresponding to each step’s conformation and the distances between the substrate’s phenolic proton and five benchmark locations: (1) the heme-bound oxygen (H-HemeO); (2) the heme iron (H-HemeFe); (3) the imidazole δN on HIS47

ARTICLE

(H-HIS47); 4) the oxygen on MET172 (H-MetO); and (5) the nitrogen on MET172 (H-MetN). MET172 was selected as a benchmark residue because it is known to help transfer electrons from VA0 s binding site TRP171 to the heme.20 Substrate/LiPIIVA•+ binding was also quantified using the same approach for each substrate. Over the course of a docking simulation, each distance was measured at each MC step, thus a range of binding distances (2.716.5 Å) was collected for each substrate. To facilitate QSAR analysis, the range of binding distances collected during analysis of each substrate was synthesized into an energy-weighted average distance. In particular, binding distances for each sampled enzymesubstrate ensemble were weighted according to the inverse of their measured potential energy value, as fit to the exponential distribution.26 This distribution takes the form f(x) =λeλx, where x is a random variable equal to the potential energy of a particular configuration, f(x) is the probability of x, and λ is a characteristic parameter of each substrate’s conformational energy distribution equal to the inverse arithmetic average of all energies sampled for that substrate. If d is the characteristic binding distance associated with the potential energy (x) of a sampled configuration, the average characteristic binding distance μ was calculated according to μ =∑(d  f(x))/∑ f(x). Use of the f(x) as a weighting factor allowed lower-energy conformations, which are more favorable and thus more probable, to be weighted more heavily in the computation of average distance than less favorable, higher-energy conformations.

’ RESULTS AND DISCUSSIONS Reaction Kinetics. The MichaelisMenten curve of each substrate is shown in Figure 1 by plotting the initial reaction rate versus the initial concentration of each substrate. These data were then fit to the MichaelisMenten equation to obtain KM and kCAT. Resulting values are presented alongside corresponding standard errors of regression in Table 1. From this data, it can be seen that both constants vary greatly among the six selected substrates in the system without VA, spanning nearly 2 orders of magnitude. In the presence of VA, kCAT increased for E2 and NP but decreased for E3, EE2, and BPA (E1 could not be evaluated because an impurity in VA interfered with HPLC detection of E1). All but one substrate (EE2) exhibit a statistically significant difference in kCAT for systems with versus without VA at R = 0.05. This is despite the fact that some substrates exhibit increased kCAT and others exhibit decreased kCAT in the presence of VA. Although they were not the focus of this study, we did note that KM values decreased for E2, E3, EE2 and BPA but increased for NP. Relationship between Reaction Rate Constant and Substrate Molecular Structure. In order to understand what factors affect LiP-mediated EDC reaction rates, QSAR was first attempted using the traditional approach, whereby reactivity (as parametrized using lnkCAT) is correlated with molecular descriptors for selected substrates. In particular, we evaluated QSAR models for three parameters related to electron transfer because LiP mediates oxidation of its substrate. These three parameters included EHOMO; ELUMO; and the difference between these two, ELUMO  EHOMO, which is sometimes referred to as “gap energy”. These parameters have been examined in previous QSAR studies of peroxidase reactions,26,32 and values for each of our selected substrates are shown in SI Table S1. As shown in Figure 2, no apparent linear correlation was found between lnkCAT and EHOMO, ELUMO, or ELUMO  EHOMO. Because we previously found that the LiP estrogen binding site is accessible only via a 5968

dx.doi.org/10.1021/es201460s |Environ. Sci. Technol. 2011, 45, 5966–5972

Environmental Science & Technology

ARTICLE

Figure 1. MichaelisMenten curves for LiP-catalyzed estrogen reactions with and without the presence of VA. Reaction rates were determined at the following conditions: [LiP] = 0.005 U/mL, [H2O2] = 0.01 mM, [VA] = 1 mM, pH 4.6 (10-mM CPBS). Reaction time was 40 s. Three replicate experiments were performed for each reaction condition. Error bars are (1 standard deviation. The reaction of E1 in the presence of VA was not measured because of an experimental artifact.

Table 1. Measured Ln(kCAT) and KM Values for Estrogens and Simulation-Estimated Binding Distances simulation-estimated distance (Å)a

measured parameters compound

KM (μM)

ln(kCAT) (s1)

ln(SE)c (s1)

H-HIS47

H-HemeO

H-HemeFe

H-MetO

H-MetN 11.91

El

2.56

1.45

0.44

6.5

6.17

7.44

9.87

E2

0.9

0.21

2.41

7.87

7.41

7.95

10.42

12.41

E3

4.93

1.73

1.07

6.26

5.96

6.57

9.93

12.12 11.29

EE2

14.39

1.86

0.39

4.94

3.99

4.73

8.57

NP

4.54

2.04

0.49

5.7

5.73

6.29

9.7

12.55

BPA

28.51

5.35

0.94

3.16

4.12

4.91

8.29

10.6

E2+VAb E3+VAb

0.06 0.51

0.8 0.76

1.78 2.71

7.74 7.57

7.39 7.48

7.6 7.41

11.23 10.12

13.47 12.7

EE2+VAb

13.49

1.69

2.01

6.36

7.64

7.32

9.69

11.89

NP+VAb

16.87

2.94

0.74

3.77

2.9

3.87

12.21

9.68

BPA+VAb

27.77

4.64

0.05

3.57

3.11

3.84

8.34

11.14

a Distances between the substrate phenolic proton and the heme-bound oxygen (H-HemeO), heme iron (H-HemeFe), the δN of HIS470 s imidazole ring (H-HIS47), the oxygen of MET172(H-MetO) and the nitrogen of MET172 (H-MetN) are obtained through molecular simulations of enzyme/ substrate complexes comprising either LiP II (rows 1 to 6) or LiPII-VA•+(rows 7 to 11). b Denotes molecular simulations of enzyme/substrate complexes comprising LiPII-VA•+. c ln(SE) is the natural logarithm of the standard error for the estimate of k CAT.

5969

dx.doi.org/10.1021/es201460s |Environ. Sci. Technol. 2011, 45, 5966–5972

Environmental Science & Technology

ARTICLE

Figure 3. Correlations between ln(kCAT) and the simulation-measured distance of (H-HIS47). (a) Six ln(kCAT) values obtained for the systems without VA; (b) Eleven ln(kCAT) values obtained for the systems with (red dots) and without (black dots) VA.

Figure 2. Correlation between the ln(kCAT) values and selected substrate parameters for reactions systems that do not contain VA. Molecular descriptors are abbreviated as follows: EHOMO is energy of the highest molecular orbital; ELUMO is energy of the lowest unoccupied molecular orbital; ELUMO  EHOMO is the arithmetic difference between ELUMO and EHOMO (the so-called “gap energy”); and Volume is molecular volume.

small channel from the protein surface (SI Figure S5),12 we also hypothesized that the molecular volume (VM) might be related to lnkCAT. In particular, we expected that larger molecules might exhibit slower LiP reactions because they may become sterically hindered during interaction with the enzyme’s active site. The lack of linear correlation between lnkCAT and VM in Figure 2 discredits this hypothesis. Relationship between Reaction Rate Constant and Enzyme Substrate Binding Conformation. Once it was determined that traditional QSAR could not be utilized to evaluate LiP reactivity toward estrogen substrates, we turned our attention toward QSAR molecular descriptors that better encapsulate characteristics of the enzymesubstrate complex, rather than just the substrate molecule itself. This is consistent with our approach from previous work.26 Several strategically selected distances were recorded from molecular simulation of the reaction between LiP and several estrogens, specifically the distances between the substrate proton and five landmark locations on the enzyme catalytic pocket: (H-HemeO); (H-HemeFe); (H-HIS47); (H-MetO); and (H-MetN). Energy-weighted averages were then calculated for each of these distances as exhibited by each of six estrogenic substrates. The results are summarized in Table 1. Linear regression yielded no apparent correlations between ln(kCAT) and the distances (H-Heme O), (H-Heme Fe), (H-Met O), or (H-Met N) from Table 1. Correlation coefficients for these relationships were quite low (R2 = 0.520.65) (as shown in SI Figure S6). In contrast, ln(kCAT) is strongly correlated to (H-HIS47), the distance between the substrate’s phenolic proton and the δN of HIS470 s imidazole ring (R2 = 0.89), as shown in Figure 3a. This observation is consistent with current understanding of the electron transfer mechanism during LiP reactions, since it has been suggested that HIS47, in conjunction with ARG43, plays a key role during substrate oxidation by functioning as proton acceptor for the bound phenolic substrate.16

Once we had identified the intermolecular binding distance as an important predictor of LiP reactivity during interaction with EDCs, we used the average H-HIS47 binding distances (μHHIS47) to predict lnkCAT via one-parameter linear regression equation. The resulting equation (eq 1) exhibits very good correlation. The negative coefficient for μH-HIS47 indicates that increased binding distance between the proton and HIS47 retards the reaction rate by hindering electron transfer. Interestingly, eq 1 is qualitatively consistent with the QSAR outcome for prediction of lnkCAT during HRP reactions with estrogenic phenols even though it does not require use of an electrochemical molecular descriptor (e.g., EHOMO). lnðkCAT Þ ¼ 7:94  1:02  μH-HIS47 ðR 2 ¼ 0:89Þ

ð1Þ

To better understand the reaction between LiP and estrogens under certain conditions, we also measured initial reaction rates in systems containing VA. We found that LiP reaction behaviors are significantly modified in the presence of VA. This is made evident by the differences in MichaelisMenten plots for systems with and without VA in Figure 1. As described earlier, VA reacts with LiP to form a radical cation VA•+, which can bind to iron porphyrin to form a LiPII-VA•+ complex.25 Since this mechanism involves modification of the LiP catalytic center, changes in kCAT compared to the system without VA are perhaps not unexpected. It is however interesting that EDCs in Figure 1 exhibit different kCAT trends. E2 and NP exhibit a significant increase in the presence of VA, E3 and BPA exhibit a significant decrease, and EE2 exhibits a slight decrease. We further examined the correlation between lnkCAT values and the (H-HIS47) distance for the reactions in the presence of VA. The resulting correlation, eq 2, is quite similar to eq 1, which embodies kCAT as a function of binding distance for the same reactions in the absence of VA (Figure 3b). Linear regression for all eleven lnkCAT values (five substrate reactions with VA plus six without VA) yields eq 2. lnðkCAT Þ ¼ 7:18  0:88  μH-HIS47 ðR 2 ¼ 0:87Þ 0

ð2Þ

2

Equation 2 s significant R and its close similarity to eq 1 suggest that the presence of VA in LiP reaction systems mediates modification of the enzyme’s catalytic center via VA•+ binding but that it does not change the principal electron transfer pathway. HIS47 still serves as the primary point of contact for electron shuttle. Changes in kCAT in the presence of VA are thus primarily attributable to changes in μH-HIS47, a result of steric or physical interactions when VA•+ is present in the catalytic pocket. 5970

dx.doi.org/10.1021/es201460s |Environ. Sci. Technol. 2011, 45, 5966–5972

Environmental Science & Technology

ARTICLE

the importance of the distance between a substrate’s phenolic proton and the imidazole δN on HIS47. Our simulation protocol also captures LiP conformational changes resulting from VA•+ binding to the catalytic center and their associated impacts on LiP reactivity. Such simulation-based QSAR models can serve as screening tools to predict the reactivity of other substrates, and they may also provide guidance during design of recombinant enzymes that have enhanced reactivity toward specific chemicals of interest.37,38

’ ASSOCIATED CONTENT Figure 4. Stereo view of the catalytic cavity contained in LiP (left) and HRP (right).12,26 These enzyme structures were downloaded from RCSB protein data bank (www.rcsb.org/pdb) and viewed in HyperChem Molecular Modeling System.

Proposed Electron Transfer Mechanism. Figure 4 depicts the active sites of both LiP and HRP, showing that each contains an iron porphyrin surrounded by PHE, HIS, and ARG residues comprising a catalytic pocket.19,26 Interestingly, HIS47 in LiP is positioned similarly to HIS42 in HRP, relative to the iron porphyrin. One widely accepted reaction mechanism for HRP suggests that electron transfer from the substrate to the heme takes place not at the iron atom itself but at the δN of HIS420 s imidazole ring.26,33 Our observation of strong correlation between ln(kCAT) and μH-HIS47, as summarized by eqs 1 and 2, may suggest that the δN on LiP0 s HIS470 s imidazole ring plays an analogous role to that of the δN on HRP0 s HIS42 imidazole ring. That is, both HIS residues may serve as the point of contact for electron transfer or help to deprotonate the substrate, thus facilitating the electron transfer from the substrate to the heme. The electron transfer mechanism has been relatively well studied for HRP, and a so-called “push-pull” mechanism has been proposed.3336 This concept suggests that distal histidine (HIS42) acts initially as a proton acceptor and then as a donor working together with a charged arginine (ARG38) to “pushpull” apart the bound phenolic substrate.34 As shown in SI Figure S7, the postulated process includes formation of ROH-heme complex, deprotonation of ROH by the distal histidine to form RO—O (heme oxygen) bond, and RO—O bond cleavage with assistance from the protonated imidazole and the arginine residue within the active site. For LiP, the short distance between the δN of HIS470 s imidazole ring and the heme O (2.88 Å), in conjunction with the close proximity of ARG43, may enable HIS47 to mediate a “push-pull” deprotonation mechanism similar to that of HRP. In reaction systems containing VA, the formation of LiPIIVA•+ does not seem to alter the principal electron transfer mechanism. To reiterate from above, this is evidenced by the close similarity between eqs 1 and 2. Changes in ln(kCAT) for VAcontaining systems can seemingly be justified solely by changes in μH-HIS47, which reflect modification of the LiP catalytic pocket arising from LiPII-VA•+ formation. These changes are captured by our molecular simulations. For example, the μH-HIS47 for E2 and NP decreased from 7.87 and 5.7 to 7.74 and 3.77 Å, respectively, and their respective ln(kCAT) values increased from 0.21 and 2.04 to 0.80 and 2.94 in the presence of VA. Similarly, μH-HIS47 values for E3, EE2 and BPA increased in the presence of VA, mediating decreases in ln(kCAT). Our QSAR study provides insight into the impact of enzyme/substrate binding on LiP reactivity, calling attention to

bS

Supporting Information. (I): (i) Table S1 Substrate parameters used in QSAR relationships for LiP reactivity; (ii) Stereo view of LiP’s catalytic cavity; (iii) Schematic depiction of the three-step LiP catalytic cycle; (iv) Molecular structures of selected natural and synthetic estrogens; (v) Charge distribution for selected subsets of enzyme intermediates; (vi) Stereo view of the open channel that connects heme pocket to LiP surface; vii) Correlation between ln(kCAT) values and the simulationestimated average binding distances (Å); (viii) Schematic representation of the HRP-catalyzed EDC reaction cycle. (II) Additional description of certain experimental procedures. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Phone: 86-25-89680359 (S. G.); 770-229-3302 (Q. H.). Fax: 86-25-89680360(S. G.); 770-412-4734 (Q. H.). E-mail: ecsxg@ nju.edu.cn (S. G); [email protected] (Q. H.).

’ ACKNOWLEDGMENT S.G. thanks the support from National Basic Research Program (2009CB421604) and Key Projects of the National Natural Science Foundation (20737001 and 20677024) of China. L.M. acknowledges the supported by Shanghai Tongji Gao Tingyao Environmental Science & Technology Development Foundation. The study was support in part by U.S. EPA STAR grant G6M10518 and HATCH fund. The content of the paper does not necessarily represent the views of the funding agencies. ’ REFERENCES (1) Welshons, W. V.; Thayer, K. A.; Judy, B. M.; Taylor, J. A.; Curran, E. M.; vom Saal, F. S. Large effects from small exposures. I. mechanisms for endocrine-disrupting chemicals with estrogenic activity. Environ. Health Prospect. 2003, 111, 994–1006. (2) Kidd, K. A.; Blanchfield, P. J.; Mills, K. H.; Palace, V. P.; Evans, R. E.; Lazorchak, J. M.; Flick, R. W. Collapse of fish population after exposure to a synthetic estrogen. Proc. Natl. Acad. Sci. 2007, 104, 8897–8901. (3) Mao, L.; Lu, J.; Habteselassie, M.; Luo, Q.; Gao, S.; Cabrera, M.; Huang, Q. Ligninase-mediated removal of natural and synthetic estrogens from water: II. reactions of 17β-estradiol. Environ. Sci. Technol. 2010, 44, 2599–2604. (4) Jobling, S.; Nolan, M.; Tyler, C. R.; Brighty, G.; Sumpter, J. P. Widespread sexual disruption in wild fish. Environ. Sci. Technol. 1998, 32, 2498–2506. (5) Ying, G.; Kookana, R. S. Degradation of five selected endocrinedisrupting chemicals in seawater and marine sediment. Environ. Sci. Technol. 2003, 37, 1256–1260. 5971

dx.doi.org/10.1021/es201460s |Environ. Sci. Technol. 2011, 45, 5966–5972

Environmental Science & Technology (6) Fan, Z. S.; Casey, F. X. M.; Hakk, H.; Larsen, G. L. Persistence and fate of 17β-estradiol and testosterone in agricultural soils. Chemosphere 2007, 67, 886–895. (7) Suzuki, K.; Hirai, H.; Murata, H.; Nishida, T. Removal of estrogenic activities of 17 β -estradiol and ethinylestradiol by ligninolytic enzymes from white rot fungi. Water Res. 2003, 37, 1972–1975. (8) Auriol, M.; Filali-Meknassi, Y.; Tyagi, R. D.; Adams, C. D. Oxidation of natural and synthetic hormones by the horseradish peroxidase enzyme in wastewater. Chemosphere 2007, 68, 1830–1837. (9) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Pharmaceuticals, hormones, and other organic wastewater contaminants in U.S. streams, 1999 2000: a national reconnaissance. Environ. Sci. Technol. 2002, 36, 1202– 1211. (10) Westerhoff, P.; Yoon, Y.; Snyder, S.; Wert, E. Fate of endocrine disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 2005, 39, 6649–6663. (11) Johnson, A. C.; Sumpter, J. P. Removal of endocrine-disrupting chemicals in activated sludge treatment works. Environ. Sci. Technol. 2001, 35, 4697–4703. (12) Mao, L.; Huang, Q.; Lu, J.; Gao, S. Ligninase-mediated removal of natural and synthetic estrogens from water: I. Reaction behaviors. Environ. Sci. Technol. 2009, 43, 374–379. (13) Lu, J.; Huang, Q.; Mao, L. Removal of acetaminophen using enzyme-mediated oxidative coupling process: I. reaction rates and pathways. Environ. Sci. Technol. 2009, 43, 7062–7067. (14) Auriol, M.; Filali-Meknassi, Y.; Adams, C. D.; Tyagi, R. D. Natural and synthetic hormones removal using the horseradish peroxidase enzyme: temperature and pH effects. Water Res. 2006, 40, 2847–2856. (15) Yu, J.; Taylor, K. E.; Zou, H.; Biswas, N.; Bewtra, J. K. Phenol conversion and dimeric intermediates in horseradish peroxidasecatalyzed phenol removal from water. Environ. Sci. Technol. 1994, 28, 2154–2160. (16) Choinowski, T.; Blodig, W.; Winterhalter, K. H.; Piontek, K. The crystal structure of lignin peroxidase at 1.70 Å resolution reveals a hydroxyl group on the Cβ of Tryptophan 171: a novel radical site formed during the redox cycle. J. Mol. Biol. 1999, 286, 809–827. (17) Blodig, W.; Smith, A. T.; Winterhalter, K.; Piontek, K. Evidence from spin-trapping for a transient radical on tryptophan residue 171 of lignin peroxidase. Arch. Boichem. Biophys. 1999, 370, 86–92. (18) Koduri, R. S.; Tien, M. Kinetic analysis of lignin peroxidase: Explanation for the mediation phenomenon by veratryl alcohol. Biochemistry 1994, 33, 4225–4230. (19) Banci, L.; Ciofi-Baffoni, S.; Tien, M. Lignin and Mn peroxidasecatalyzed oxidation of phenolic lignin oligomers. Biochemistry 1999, 38, 3205–3210. (20) Gelpke, M. D. S.; Lee, J.; Gold, M. H. Lignin peroxidase oxidation of veratryl alcohol: effects of the mutants H82A, Q222A, W171A, and F267L. Biochemistry 2002, 41, 3498–3506. (21) Harvey, P. J.; Schoemaker, H. E.; Palmer, J. M. Veratryl alcohol as a mediator and the role of radical cations in lignin biodegradation by Phanerochaete chrysosporium. FEBS Lett. 1986, 195, 242–246. (22) Gilardi, G.; Harvey, P. J.; Cass, A. E. G.; Palmer, J. M. Radical intermediates in veratryl alcohol oxidation by ligninase NMR evidence. Biochim. Biophys. Acta 1990, 1041, 129–132. (23) Chuang, N.; Aust, S. D. Veratryl alcohol-mediated indirect oxidation of pentachlorophenol by lignin peroxidase. Arch. Biochem. Biophys. 1995, 322, 143–148. (24) Mao, L.; Lu, J.; Gao, S.; Huang, Q. Transformation of 17βestradiol mediated by Lignin peroxidase: The role of veratryl alcohol. Arch. Environ. Contam. Toxicol. 2010, 59, 13–19. (25) Khindaria, A.; Nie, G.; Aust, S. D. Detection and characterization of the lignin peroxidase compound II-veratryl alcohol cation radical complex. Biochemistry 1997, 36, 14181–14185. (26) Colosi, L. M.; Huang, Q.; Weber, W. J., Jr. Quantitative structure activity relationship based quantification of the impacts of

ARTICLE

enzyme-substrate binding on rates of peroxidase-mediated reactions of estrogenic phenolic chemicals. J. Am. Chem. Soc. 2006, 128, 4041–4047. (27) Gerini, M. F.; Roccatano, D.; Baciocchi, E.; Nola, A. D. Molecular dynamics simulations of lignin peroxidase in solution. Biophys. J. 2003, 84, 3883–3893. (28) HyperChem Reference Manual; Hypercube, Inc.: Gainesville, FL, 1996; p 172. (29) Khindaria, A.; Aust, S. D. EPR detection and characterization of lignin peroxidase porphyrin π-cation radical. Biochemistry 1996, 35, 13107–13111. (30) Kuramochi, H.; Noodleman, L.; Case, D. A. Density functional study on the electronic structures of model peroxidase compounds I and II. J. Am. Chem. Soc. 1997, 119, 11442–11451. (31) Poulos, T. L.; Edwards, S. L.; Wariishi, H.; Gold, M. H. Crystallographic refinement of lignin peroxidase at 2 Å. J. Mol. Biol. 1993, 268, 4429–4440. (32) Van Haandel, M. J. H.; Rietjens, I. M. C. M.; Soffers, A. E. M. F.; Veeger, C.; Vervoort, J.; Modi, S.; Mondal, M. S.; Patel, P. K.; Behere, D. V. Computer calculation-based quantitative structure-activity relationships for the oxidation of phenol derivatives by horseradish peroxidase compound II. Biol. Inorg. Chem. 1996, 1, 460–467. (33) Filizola, M.; Loew, G. H. Characterization of polyethylene glycolated horseradish peroxidase in organic solvents: Generation and stabilization of transient catalytic intermediates at low temperature. J. Am. Chem. Soc. 1998, 120, 8020–8025. (34) Poulos, T. L.; Kraut, J. The stereochemistry of peroxidase catalysis. J. Biol. Chem. 1980, 255, 8199–8205. (35) Rodriguez-L opez, J. N.; Lowe, D. J.; Hernandez-Ruiz, J.; Hiner, A. N. P.; Garcia-Canovas, F.; Thorneley, R. N. F. Mechanism of reaction of hydrogen peroxide with horseradish peroxidase: Identification of intermediates in the catalytic cycle. J. Am. Chem. Soc. 2001, 123, 11838–11847. (36) Dawson, J. H. Probing structure-function relations in Hemecontaining oxygenases and peroxidases. Science 1988, 240, 433–439. (37) Colosi, L. M.; Huang, Q.; Weber, W. J., Jr. Validation of a twoparameter quantitative structure activity relationship as a legitimate tool for rational re-design of horseradish peroxidase. Biotechnol. Bioeng. 2007, 98, 295–299. (38) Colosi, L. M.; Huang, Q.; Weber, W. J., Jr. QSAR-assisted design of an environmental catalyst for enhanced estrogen remediation. Chemosphere. 2010, 81, 897–903.

5972

dx.doi.org/10.1021/es201460s |Environ. Sci. Technol. 2011, 45, 5966–5972