Nitrogen dioxide solubility and permeation in lipid membranes

Archives of Biochemistry and Biophysics 512 (2011) 190–196

Contents lists available at ScienceDirect

Archives of Biochemistry and Biophysics journal homepage: www.elsevier.com/locate/yabbi

Nitrogen dioxide solubility and permeation in lipid membranes Santiago Signorelli a,1,2, Matías N. Möller b,1,3, E. Laura Coitiño a, Ana Denicola b,⇑ a b

Laboratorio de Química Teórica y Computacional, Facultad de Ciencias, Universidad de la República, Iguá 4225, CP 11400 Montevideo, Uruguay Laboratorio de Fisicoquímica Biológica, Facultad de Ciencias, Center for Free Radical and Biomedical Research, Universidad de la República, Iguá 4225, CP 11400 Montevideo, Uruguay

a r t i c l e

i n f o

Article history: Received 14 April 2011 and in revised form 31 May 2011 Available online 15 June 2011 Keywords: Nitrogen dioxide Partition coefficient Permeability Lipid membrane

a b s t r a c t Nitrogen dioxide (NO2) is an important oxidant molecule in biology that is produced by several biological processes, and it is also an important air pollutant. It can oxidize proteins and lipids with important consequences on their biological functions. Despite its relevance, the interaction of NO2 with the cell barrier, the lipid membrane, is poorly understood. For instance, can lipid membranes limit NO2 diffusion? To estimate the permeability of lipid membranes to NO2 it is necessary to learn more about its solubility in the lipid phase. However, experimental data on NO2 solubility is very limited. To improve our knowledge on this matter, we used a mixed approach consisting in calculating the solubility of NO2 and related diatomic and triatomic gases (NO, O2, CO2, etc.) in different solvents using quantum calculations and Tomasi’s Polarizable Continuum Model and validating and correcting these results using experimental data available for the related gases. This approach led to an estimated partition coefficient for NO2 of 2.7 between n-octanol and water, and 1.5 between lipid membranes and water, meaning that NO2 is a moderately hydrophobic molecule (less than NO, more than CO2). Based on the solubility-diffusion permeability theory, the permeability coefficient was estimated to be 5 cm s1, up to 4000 times higher than that of peroxynitrous acid. It is concluded that lipid membranes are not significant barriers to NO2 transport. Ó 2011 Elsevier Inc. All rights reserved.

Introduction Nitrogen dioxide (NO2) is a free radical of increasing interest in biology [1]. It has long been recognized as an air pollutant and been studied extensively by environmental chemists, but it has just recently brought the attention of biologists. Exposure of animals to  NO2 results in several toxic effects, but most importantly lung injury caused by peroxidation of lipids, nitration of surfactant protein A and formation of carcinogenic nitrosamines [1]. At the cellular level, NO2 can be formed by several mechanisms, outlined in Scheme 1: the Lewis acid-promoted decomposition of NOderived peroxynitrite (I), in particular by metals and CO2 (present at high levels in mammalian cells) [2,3]; the autoxidation of nitric oxide (NO) (II), a reaction that is accelerated in lipid environments

⇑ Corresponding author. Fax: +598 2 525 0749. E-mail addresses: [email protected] (S. Signorelli), [email protected] (M.N. Möller), [email protected] (E.L. Coitiño), [email protected] (A. Denicola). 1 These authors contributed equally. 2 Present address: Laboratorio de Bioquímica, Departamento de Biología Vegetal, Facultad de Agronomía, Universidad de la República, Av. E. Garzón 780, CP 12900 Montevideo, Uruguay. 3 Present address: Department of Chemistry, Vanderbilt University, 7330 Stevenson Center, Station B 351822, Nashville, TN 37235, USA. 0003-9861/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2011.06.003

[4–6]; and peroxidase-catalyzed oxidation of nitrite (III), mainly myeloperoxidase, (MPO) [7]. In aqueous solution, NO2 is in equilibrium with its dimer N2O4 (K = 7  104 M1) that rapidly hydrolyzes to nitrite and nitrate (k = 1  103 s1) [8]. However, at the low concentrations expected in vivo (<0.1 lM), NO2 will be present mostly as a monomer (>99.7%).  NO2 is an oxidizing reactive nitrogen species and mainly reacts with low molecular weight antioxidants, proteins and unsaturated lipids. Electron transfer reactions are fast, as can be exemplified by the reaction of NO2 with cysteine thiolate (k = 2.4  108 M1 s1), tyrosinate (k = 2.9  107 M1 s1), or ascorbate (k = 6.4  107 M1 s1) [8–9]. However, hydrogen-atom abstraction and addition to double bonds are slower reactions. The reaction with tyrosine to form tyrosyl radical has a rate constant k = 3.2  105 M1 s1, while the addition to the double bonds of arachidonate or linoleate occur with k = 1  106 M1 s1 and k = 2  105 M1 s1, respectively [9]. In contrast, NO2 can react with other radicals at diffusion-controlled rates: NO2 reacts with tyrosyl radical to form 3-nitrotyrosine with k = 3  109 M1 s1 [9]. Modification of biomolecules by  NO2 leads to important biological consequences. Nitration of tyrosine residues results in a change of protein function, either inactivation (i.e., MnSOD, glutathione reductase) [10] or gain of function (such as activation of the peroxidatic activity of cytochrome c [11], or acceleration of the aggregation of Parkinson’s disease related

S. Signorelli et al. / Archives of Biochemistry and Biophysics 512 (2011) 190–196

191

it would be very difficult to obtain the experimental solubility of  NO2 in n-octanol, since N2O4 reacts with alcohols to form organic nitrates and nitrites [29]. Our approach consisted first in collecting available experimental solubility data for diatomic and triatomic gases (NO, O2, N2, CO2, N2O and O3) in different solvents. Then, performed quantum calculations combining Density Functional Theory (DFT) with Tomasi’s Polarizable Continuum Model (PCM) to calculate the solubility from the free energy of NO2 solvation in those solvents [30–32]. The same calculations were made for the whole set of aforementioned gases, and used as control comparing theoretical and experimental data. This procedure allowed us to estimate the solubility of NO2 in n-octanol and lipid membranes, and then calculate its permeability coefficient using the solubility-diffusion permeability theory. Scheme 1. Biological sources of NO2.

protein a-synuclein [10]). The reaction of NO2 with polyunsaturated lipids can lead not only to lipid oxidation but also to the formation of nitro-lipids [12,13], which have been shown to have potent anti-inflammatory and other cell signaling properties [14– 17]. Much of the free radical oxidizing damage is subject to compartmentalization by cell membranes, which offer significant resistance to the passage of most reactive species [18]. It is therefore relevant to know whether lipid membranes are permeable to NO2, to define if  NO2 formed in one subcellular compartment can reach and react with targets in another compartment or another cell. Lipid membranes represent no barrier to the diffusion of NO or O2 [2], but they do represent a significant barrier for peroxynitrous acid and hydrogen peroxide and are practically impermeable to peroxynitrite anion, superoxide and carbonate radical [18]. The only quantitative data on lipid membrane permeability to NO2 comes from Khairutdinov et al. studies on the permeability of lipid membranes to peroxynitrite [19]. They used a mathematical model that included the permeability of NO2 and estimated that the permeability coefficient for NO2 would be between 4  104 and 10 cm s1, a range too wide to make useful predictions, ranging from minimal to significant membrane resistance to NO2 permeation. When membrane permeability is not known, it is usual to use Overton’s rule, which dictates that membrane permeability to a given molecule is proportional to the solubility of that molecule in an organic solvent, such as n-octanol [20]. This relation is widely used to predict pharmacokinetics and biodistribution of drugs in the organism [21–23]. As will be mentioned in Section Permeability of lipid membranes to NO2, a mechanistic and quantitative version of this relation is the solubility-diffusion theory of permeability [20,24,25] that states that the permeability coefficient in the membrane (Pm) is directly proportional to the solubility in an organic solvent – or in the membrane – (see Eq. (7) below). Whereas the available data shows that NO2 is 10 times more soluble in water than O2, experimental data on NO2 solubility in organic solvents is fairly scarce, which limits the use of the solubility-diffusion permeability theory to calculate membrane permeability. Only solubility of NO2 in decane, chloroform, carbon tetrachloride, tetrachloroethane, and nitrate esters have been determined [26– 28]. Such scarcity in empirical data motivated us to use an in silico approach involving quantum calculations to get theoretical solubilities for NO2 in different solvents, with an especial interest in n-octanol. Apart from being the ‘‘classical’’ solvent used in partitioning experiments, the solubility of some molecules of interest has been determined both in n-octanol and lipid membranes, so that determining the solubility of NO2 in n-octanol will improve our estimation of its solubility in lipid membranes. Furthermore,

Methods Molecular systems and experimental data relationships The molecules chosen were diatomic and triatomic gases containing oxygen and/or nitrogen atoms: x = nitrogen, nitric oxide, oxygen, carbon dioxide, nitrous oxide, ozone, and nitrogen dioxide (N2, NO, O2, CO2, N2O, O3, NO2, respectively). The solubility data was expressed as the partition coefficients (K Py=w ) between a given organic solvent y and water w, using the reported solubilities at 1 atm and 25 °C: y=w

KP

¼

solubility of x in solvent y ðmoles L1 Þ solubility of x in water ðmoles L1 Þ

ð1Þ

The solubility of NO2 in water has been indirectly determined in several occasions [33–35], while the experimental data on  NO2 solubility in organic solvents was obtained from [26–28]. Computational modeling Quantum DFT modeling at the (U)B3LYP/6-311G⁄ level [36–38] combined to the Integral Equation Formalism implementation of the Tomasi´s Polarizable Continuum Model (IEF-PCM) [39,40] performed over molecular shaped cavities generated using Bondi’s radii [41] and complemented by non electrostatic contributions was performed using Gaussian09 [42]. Representative structures were determined both in gas phase and in each of the 15 considered solvents for NO2 and the six reference gases. All the species were checked to be stable minima by inspection of the Hessian eigenvalues in each of the considered isotropic media, and free energies at 298 K and 1 atm were calculated. The 15 isotropic solvents used in the calculations were: water, ethanol, methanol, n-octanol, methylene chloride, acetonitrile, dimethylsulfoxide, acetone, dichloroethane, heptane, cyclohexane, chlorobenzene, toluene, chloroform and benzene, described using the set of parameters included in the Gaussian09’s database. IEF-PCM allows to determine the electrostatic component of the free energy for a molecular species x in a given solvent (DGelec-sol). Complemented by classical descriptions of the non-electrostatic contributions to the solvation energy, including cavitation (DGcav) and repulsion–dispersion (DGrd) terms, respectively calculated according to Claverie–Pierotti’s scale particle theory [43,44] and Floris and Tomasi’s expressions [45,46] the method provides a very good cost-exactness trade-off for a wide range of molecular properties, including partition coefficients [30–32,47]. The free energy of solvation DGsolv for each molecule in each different solvent is calculated by the following equation:

    DGsolv ¼ Gsol  Ggas ¼ Gelec-sol  Ggas þ DGcav þ DGrd

ð2Þ

192

S. Signorelli et al. / Archives of Biochemistry and Biophysics 512 (2011) 190–196

where

Gsol ¼ Gelec-sol þ DGcav þ DGrd

ð3Þ

The difference in free energy of solvation between a given organic solvent and water (DGsolv.y/w) can then be used to calculate the corresponding partition coefficient K Py=w as shown in Eqs. (4)– (6).

DGsolvy=w ¼ DGsolvy þ DGsolvw

Table 1 Experimental solubility of different gases in water and organic solvents.

Solvents Water

ð4Þ Decane

DGsolvy=w ¼ RT ln

y=w KP

ð5Þ Chloroform

y=w KP



¼e

DGsolvy=w RT

ð6Þ Carbon Tetrachloride

Results and discussion 1,1,2,2-Tetrachloroethane

Experimental data correlations When comparing the experimental solubilities of the diatomic gases O2, N2 and NO in different solvents, a fairly good correlation was observed, r2 = 0.952 between O2 and N2, and r2 = 0.863 between O2 and NO (Fig. 1A). If triatomic gases behaved in a similar fashion, then, we may be able to use a similar correlation to extrapolate missing data, and estimate the solubility of NO2 in n-octanol. Hence, we collected experimental solubility data for molecules similar to NO2 in different solvents and analyzed it in search of correlations (Table 1 and Table S1). Unfortunately, as shown in Fig. 1B–D, no correlation was found between NO2 and any other diatomic or triatomic gas molecule. Among the triatomic gases, only a weak correlation between N2O and CO2 was observed (r2 = 0.705, Fig. 1C). Given that no correlation was found for NO2

Solubility (M/atm  103) KP Ref.   NO2 O2 NO N2 12 1 [35] 60 5 [28] 21 1.8 [27] 17 1.4 [27] 73 6.1 [28]

N2O

CO2

O3

14 7.3 [49]

0.63 1 [35] 6.5 10 [50] 5 7.9 [54] 6.7 11 [56]

25 1 [35] 91 3.6 [51] 150 6 [55] 188 7.5 [57]

34 1 [35] 64 1.9 [52] 114 3.3 [54] 111 3.3 [56]

NF

NF

NF

NF

12 1 [35] 51 4.3 [53] 87 7.2 [53] 80 6.7 [53] 72 6.0 [53]

1.27 1 [48] 11.1 8.7 [48] 9.2 7.2 [54] 12 9.8 [56]

1.95 1 [49]

NF

NF

NF

NF: not found.

that could let us calculate its K Poctanol=w and the difficulty in measuring it experimentally, we decided to use a theoretical approach to determine the solubility of NO2 in solvents through quantum calculations. Theoretical calculations and relations with experimental data Unlike the empirical data, the computed results showed a regular behavior both for the diatomic and the triatomic gases (Figure S1). This is an indication that there are solvent–solute

Fig. 1. Correlation between the experimental partition coefficients of different gases. (A) O2 and diatomic gases NO and N2; (B) O2 and triatomic gases; (C) CO2 and diatomic and triatomic gases; (D) NO2 and diatomic and triatomic gases. Good correlations were observed between O2 and N2 (r2 = 0.952) or NO (r2 = 0.863), but not with the triatomic gases. No correlations were observed between NO2 and any of the other gases, probably due to the limited amount of data. In the triatomic gases, the only correlation observed was between CO2 and N2O (r2 = 0.705).

193

S. Signorelli et al. / Archives of Biochemistry and Biophysics 512 (2011) 190–196

Fig. 2. Empirical scaling factors obtained for KP in each solvent. The ratio between experimental and theoretical KP shows that calculated values were generally moderately precise with variable accuracy. The values calculated for ethanol, methanol and DMSO were very close to the experimental ones, while they were significantly underestimated in n-octanol. In this case, the underestimation factor calculated with the available experimental data can be used to scale and correct the theoretical results.

interaction factors that are not being taken into account by the theoretical calculations at the level of description employed. A practical way to enhance the exactness of the predictions provided by a given theoretical level without enlarging the associated computational cost consists in introducing empirical ‘‘scaling factors’’ determined by fitting theoretical predictions against their known experimental counterparts. This strategy has been thoroughly used for obtaining at low cost very accurate molecular properties such as quantum-mechanics force fields for vibrational spectroscopy, redox potentials for radicals in solution, molecular polarizabilities, and others [58–61]. Scaling factors were determined here for each solvent taking into account the ratio between experimental and theoretical KP. Fig. 2 shows that ratio to be between 1.5 and 27, indicating a low to moderate underestimation of KP with a moderately precise prediction. The accuracy was fairly good for ethanol, methanol and DMSO (<1.6 times underestimation), but not so good for other solvents. The precision, assessed by the standard deviation and the relative errors for each solvent (Table S3) indicated that the predicted values of solubility are on average affected by a 40% relative error (relative errors ranging from 20% to 110%). For n-octanol, the ratio was 27, with a relative error of 34%. So that even though the theoretical approach significantly underestimated the solubility of the different gases in n-octanol, it did so in a consistent manner. Therefore, we can use this ratio for n-octanol to correct (scale) the purely theoretical KP value (K octanol=w ¼ 0:10), P octanol=w and get (K P ¼ 2:7), expected to be closer to the experimental value. These values, together with those of other alcohols are shown in Table 2. When comparing KP for NO2 and O2 in decane/ water (4.97 vs. 8.74, Table 1) or in octanol/water (2.7 vs. 5.65, Tables 2 and 3), it can be appreciated that KP for NO2 is roughly

Table 2 Theoretical and corrected theoretical alcohol/water partition coefficients for NO2.

n-Octanol Ethanol Methanol

Theoretical KP

Scaling factor

Corrected KP

0.1 4.8 5.1

27 1.5 1.6

2.7 7.4 8.1

50% that of O2, indicating that NO2 may be considered slightly less hydrophobic than O2. Solubility of NO2 in lipid membranes The octanol/water partition coefficient is widely used in medicinal chemistry to predict the biodistribution of drugs and their ability to cross biological membranes [21–23]. n-Octanol is often argued to represent the bulk solvent properties of the membrane: the alkyl part accounting for the lipid alkyl chains and the polar alcohol group accounting for the polar headgroups in the membrane surface. However, the lipid membrane is intrinsically different from a bulk solvent. Lipids in the membrane form an interfacial highly-ordered, packed, and heterogeneous phase [25,67]. For instance, molecular dynamics simulations of phospholipid membranes studying the solubility of NO or O2 show that the distribution is heterogeneous, preferring the middle of the bilayer

Table 3 Relationship between membrane and octanol partition coefficients. Lipid system

T (°C)

KP

KP ratio Octanol/lipid

Ref.

EYPC DLPC LDL DMPC n-Octanol

O2 25 25 25 30 25

3.2 3.2 2.6 3.1 5.6

1.8 1.8 2.2 1.8

[4,62] [4] [4,62] [63] [48]



EYPC DLPC LDL n-Octanol

NO 25 25 25 25

3.6 3.6 3 6.5

1.8 1.8 2.2

[4,62] [4] [4,62] [64]

EYPC EYPC EYPC/Chol EYPC/Chol n-Octanol n-Octanol

CO2 25 37 25 37 25 37

0.95 0.99 0.74 0.71 1.3 1.5

1.4 1.5 1.8 2.1

[65] [66] [66] [66] [66] [66]

Average

1.8 ± 0.2

194

S. Signorelli et al. / Archives of Biochemistry and Biophysics 512 (2011) 190–196

to the interfacial carbonyl/headgroup region [18,68–70]. Furthermore, it has been observed that the physical state of the membrane can modify the solubility of O2 (and other molecules) in the membrane at least ten times [63,71], which would not be expected by a ‘‘bulk solvent’’ model. When we compare the values of KP between lipid membranes (in fluid/liquid–crystalline state) and water determined for O2,  NO and CO2 (Table 3) with the corresponding K octanol=w , we find P that solubility in the membrane is in all cases lower than in octanol, by a relatively constant factor: 1.8 ± 0.2 (Table 3). The K octanol=w for NO2 was calculated to be 2.7 from quantum calculaP tions and the empirical scaling factor (Table 2). Considering the 1.8 octanol/membrane correction factor, we estimate that the partition coefficient between lipid membranes and water (K membrane=w ) P should be 1.5. Note that this value is half the value previously used membrane=w by Squadrito and Postlethwait (K P ¼ 3:0Þ [26] and five times higher than that used by Lim et al. (K Poctanol=w ¼ 0:3Þ [3]. Based membrane=w on K P ¼ 1:5 only a modest increase in reaction rates (50%) relative to the aqueous phase would be expected in the membrane as a result of the higher local concentration. Lim et al. have recently modeled the reactivity of NO2 in a cell environment and concluded that only 4% of the NO2 would react with polyunsaturated fatty acids (PUFAs) in the membrane (considering K NO2 = 0.3), and most P of it would react with glutathione and ascorbate in the cytosol [3]. Given the complexity of the kinetic model we cannot recalculate how much NO2 would react with PUFAs in the membrane considering a fivefold increase in KP, but it will likely be more than 4%. Permeability of lipid membranes to NO2 A good solubility in lipids is associated with a good membrane permeabilty, and this is expressed in the solubility-diffusion theory [24], where the permeability coefficient of the membrane (Pm) to a given molecule depends directly on the solubility of this molecule in the membrane (KP), its diffusion coefficient in the membrane (Dm), and inversely on the membrane thickness (dm) [24]:

Pm ¼

Dm K P dm

ð7Þ

So far we have a good estimation of the solubility of NO2 in the membrane (K membrane=w ¼ 1:5), but the diffusion coefficient of NO2 P in the lipid membrane is not known. In this case, we can consider a similar sized molecule such as CO2 (molecular weight 44.0 vs. 46.0 g/mol for NO2) which should have a similar hydrodynamic radius and diffusion coefficient through the membrane. The permeability coefficient for CO2 in planar lipid bilayers has been recently determined by Missner et al. [72]. Using planar membranes of different composition, including diphytanoyl phosphatidylcholine, lipid raft-like diphytanoyl phosphatidylcholine, cholesterol and sphingomyelin, or red blood cell membrane-like containing egg phosphatidylcholine, egg phosphatidylethanolamine, brain phosphatidylserine, cholesterol, and egg sphingomyelin, the authors found PCO2 to be equal or greater than 3.2 cm s1 [72]. The lack of change of PCO2 after altering the composition of the membrane indicates that unstirred layer effects were still important and that the determined PCO2 is very likely an underestimation. Bearing this in mind, we can use this value as a working lower limit of P CO2 . Taking Eq. (7) and considering that Dm and dm should be equal for both molecules, we find that:

PNO2 ¼ PCO2

2 K NO P 2 K CO P

It may be argued that CO2 and NO2 have structural differences that prevent this comparison, such as the linear geometry of CO2 compared to the angular one exhibited by NO2, or that whereas the latter has a permanent dipole moment (0.316 D) the former does not. However, the low dipole moment is more likely to cause a difference in partition rather than in diffusion within the membrane, and similar molecules with different dipole moments – such as NO (0.159 D) and O2 (0 D) – show very similar diffusion and permeability coefficients in phospholipid membranes [18,62,73]. The difference in shape (linear vs. angular) is also expected to have little impact on its permeability coefficient, since the relevant dimension is the minimum cross-sectional area, which will be very similar for both molecules [25]. Considering a lipid membrane 4 nm thick, and DNO2(water) = 2  105 cm2 s1 [33], the permeability of an equally thick layer of water would be Pw = 50 cm s1, only 10 times larger than Pm, indicating a low membrane resistance to NO2 transport. For comparison, the Pm for NO is 93 cm s1 [73], whereas Pm for ONOOH is 1.3  103 cm s1 [19] (Scheme 2). This calculated low resistance agrees with experimental results obtained by Connor et al. while studying the effect of phospholipid monolayers on NO2 reactive absorption into a glutathione solution [74]. These authors found that phospholipids containing unsaturated fatty acids offered no resistance to NO2 reactive absorption, while dipalmitoyl phosphatidylcholine reduced the reactive absorption by approximately 50% when compressed (lower area by phospholipid) [74]. This raises the possibility that highly packed membranes, such as sphingomyelin–cholesterol rich lipid rafts, may offer higher resistance to NO2 permeation than the bulk plasma membrane, and this may be reflected by a lower level of protein and lipid oxidation in lipid rafts. Nitration of tyrosine in lipid membranes by peroxynitrite Biological tyrosine nitration is a radical process, where first tyrosine is oxidized to tyrosyl radical and then NO2 reacts in a termination reaction to generate nitrotyrosine. Protein tyrosine nitration has been associated with both biological loss and gain of function [10,11], and often correlates with pathological states such as cardiovascular disease [75]. It has recently been found that the nitration of tyrosine in lipid membranes by biologically relevant nitrating systems is more efficient than nitration of tyrosine in solution [76–80]. Most of these studies have been done using peroxynitrite as the nitrating agent. Peroxynitrite is a biologically relevant reactive species that is formed by the diffusion-limited reaction between superoxide and NO. It is a strong oxidant participating in direct reactions with biomolecules but it can also generate NO2 and hydroxyl radical (OH) upon protonation. It is not clear whether ONOOH diffuses into the membrane and then generates the radicals in situ, or if the radicals are generated in the aqueous phase and then diffuse into the membrane. Comparing the corresponding permeability coefficients for ONOOH (Pm = 1.3  103

ð8Þ

Using the recently reported value for PCO2, we get PNO2 = 5 cm s1, a value that is in the upper range earlier estimated by Khairutdinov et al. [19].

Scheme 2. Membrane permeability and solubility of reactive nitrogen species.

S. Signorelli et al. / Archives of Biochemistry and Biophysics 512 (2011) 190–196

cm s1, [19]) and NO2 (Pm = 5 cm s1, this work) it is evident that ONOOH is a more hydrophilic molecule than NO2. Thus, we expect that the majority of ONOOH would homolyze in the aqueous phase, generating NO2 and OH that can then diffuse into the membrane. Nevertheless, we have to point that OH reacts with tyrosine to generate tyrosyl radical 4.4  104 times faster than NO2 (kOH = 1.4  1010 M1 s1 [81], kNO2 = 3.2  105 M1 s1 [9]). This big difference in rate constants indicates that although the fraction of ONOOH located in the membrane is low, the locally produced  OH could be an important source of tyrosyl radical. In addition, OH generated in the aqueous phase has been found to also react with molecules located inside lipid membranes [82]. Thus, both aqueous and membrane generated OH will be important oxidants of tyrosine residues in lipid membranes. The implication of OH in membrane tyrosine nitration is further demonstrated by experimental observation of tyrosine hydroxylation in membranes and by the decrease in tyrosine nitration when using OH scavengers such as mannitol [83]. Furthermore, it could have an additional role by promoting lipid peroxidation-mediated tyrosine nitration [79]. In an elegant work, Zhang et al. used transmembrane peptides containing tyrosine residues at different depths into the lipid bilayer to study its nitration [76]. When using peroxynitrite, they observed that tyrosine residues deeper in the bilayer were nitrated more efficiently [76]. Based on our results on membrane permeability towards nitrating species, these results could be interpreted as ONOOH-generated OH producing tyrosyl radicals at all depths in the membrane, while NO2 remaining more concentrated in the midbilayer, favoring nitrotyrosine formation there. In conclusion, although most of the homolysis of ONOOH will occur in the aqueous phase, nitration of tyrosines in membranes by peroxynitrite will not be due exclusively to reactions of hydrophobic NO2 diffusing into the membrane, but will be promoted by  OH produced in or close to the membrane. Conclusions Nitrogen dioxide is slightly less hydrophobic than NO and O2, but more hydrophobic than CO2. Our results indicate that NO2 is 1.5 times more soluble in a lipid membrane than in water. Its permeability coefficient through lipid membranes was estimated to be 5 cm s1, almost twice as much that of CO2 and about 10 times less than that of NO or O2. This also means that permeation of NO2 through a membrane should occur up to 4000 times faster than for ONOOH, something that may be amenable to in vitro experimentation. As a general conclusion, it can be said that membranes represent a low barrier to NO2 transport, thus biochemical NO2 reactions will not be significantly limited by the presence of a lipid membrane. Acknowledgments We thank Dr. Rafael Radi for critical reading of the manuscript. This work was supported by CSIC (Comisión Sectorial de Investigación Científica), Universidad de la República, and ANII (Agencia Nacional para la Innovación e Investigación), Uruguay. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.abb.2011.06.003. References [1] O. Augusto, M.G. Bonini, A.M. Amanso, E. Linares, C.C. Santos, S.L. De Menezes, Free Radic. Biol. Med. 32 (2002) 841–859.

195

[2] A. Denicola, B.A. Freeman, M. Trujillo, R. Radi, Arch. Biochem. Biophys. 333 (1996) 49–58. [3] C.H. Lim, P.C. Dedon, W.M. Deen, Chem. Res. Toxicol. 21 (2008) 2134–2147. [4] M.N. Möller, Q. Li, D.A. Vitturi, J.M. Robinson, J.R.J. Lancaster, A. Denicola, Chem. Res. Toxicol. 20 (2007) 709–714. [5] M.N. Möller, Q. Li, J.R. Lancaster Jr., IUBMB Life 59 (2007) 243–248. [6] X. Liu, M.J. Miller, M.S. Joshi, D.D. Thomas, J.R. Lancaster Jr., Proc. Natl Acad. Sci. USA 95 (1998) 2175–2179. [7] J. Byun, D.M. Mueller, J.S. Fabjan, J.W. Heinecke, FEBS Lett. 455 (1999) 243–246. [8] R.E. Huie, Toxicology 89 (1994) 193–216. [9] W.A. Prutz, H. Monig, J. Butler, E.J. Land, Arch. Biochem. Biophys. 243 (1985) 125–134. [10] J.M. Souza, G. Peluffo, R. Radi, Free Radic. Biol. Med. 45 (2008) 357–366. [11] C. Batthyany, J.M. Souza, R. Duran, A. Cassina, C. Cervenansky, R. Radi, Biochemistry 44 (2005) 8038–8046. [12] H. Rubbo, R. Radi, M. Trujillo, R. Telleri, B. Kalyanaraman, S. Barnes, M. Kirk, B.A. Freeman, J. Biol. Chem. 269 (1994) 26066–26075. [13] V.B. O’Donnell, J.P. Eiserich, P.H. Chumley, M.J. Jablonsky, N.R. Krishna, M. Kirk, S. Barnes, V.M. Darley-Usmar, B.A. Freeman, Chem. Res. Toxicol. 12 (1999) 83– 92. [14] P.R. Baker, F.J. Schopfer, V.B. O’Donnell, B.A. Freeman, Free Radic. Biol. Med. 46 (2009) 989–1003. [15] A. Trostchansky, H. Rubbo, Free Radic. Biol. Med. 44 (2008) 1887–1896. [16] H. Rubbo, A. Trostchansky, V.B. O’Donnell, Arch. Biochem. Biophys. 484 (2009) 167–172. [17] F.J. Schopfer, Y. Lin, P.R. Baker, T. Cui, M. Garcia-Barrio, J. Zhang, K. Chen, Y.E. Chen, B.A. Freeman, Proc. Natl Acad. Sci. USA 102 (2005) 2340–2345. [18] M.N. Möller, J.R. Lancaster Jr., A. Denicola, The interaction of reactive oxygen and nitrogen species with membranes, in: S. Matalon (Ed.), Current Topics in Membranes 61, Free Radical Effects on Membranes, Academic Press Inc., 2008, pp. 23–43. [19] R.F. Khairutdinov, J.W. Coddington, J.K. Hurst, Biochemistry 39 (2000) 14238– 14249. [20] A. Missner, P. Pohl, ChemPhysChem 10 (2009) 1405–1414. [21] P. Artursson, K. Palm, K. Luthman, Adv. Drug Deliv. Rev. 46 (2001) 27–43. [22] C.A. Lipinski, F. Lombardo, B.W. Dominy, P.J. Feeney, Adv. Drug Deliv. Rev. 46 (2001) 3–26. [23] E.H. Kerns, J. Pharm. Sci. 90 (2001) 1838–1858. [24] A. Finkelstein, J. Gen. Physiol. 68 (1976) 127–135. [25] T.X. Xiang, B.D. Anderson, Biophys. J. 75 (1998) 2658–2671. [26] G.L. Squadrito, E.M. Postlethwait, Nitric Oxide 21 (2009) 104–109. [27] S.N. Mendiara, L.J. Perissinotti, Appl. Magn. Reson. 25 (2003) 323–346. [28] B.A. Lur’e, I.V. Arkhipov, V.N. Apal’kova, Russ. J. Phys. Chem. 60 (1986) 1144– 1147. [29] E.H. White, W.R. Feldman, J. Am. Chem. Soc. 79 (1957) 5832–5833. [30] A. Bidon-Chanal, O. Huertas, M. Orozco, F.J. Luque, Theor. Chem. Acc. 123 (2009) 11–20. [31] A. Klamt, B. Mennucci, J. Tomasi, V. Barone, C. Curutchet, M. Orozco, F.J. Luque, Acc. Chem. Res. 2009 (2009) 489–492. [32] I. Soteras, F. Forti, M. Orozco, F.J. Luque, J. Phys. Chem. B 113 (2009) 9330–9334. [33] Y.N. Lee, S.E. Schwartz, J. Phys. Chem. 85 (1981) 840–848. [34] J.L. Cheung, Y.Q. Li, J. Boniface, Q. Shi, P. Davidovits, D.R. Worsnop, J.T. Jayne, C.E. Kolb, J. Phys. Chem. A 104 (2000) 2655–2662. [35] R. Sander, Compilation of Henry’s Law Constants for Inorganic and Organic Species of Potential Importance in Environmental Chemistry, 1999. Available from: . [36] A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652. [37] R. Krishnan, J.S. Binkley, R. Seeger, J.A. Pople, J. Chem. Phys. 72 (1980) 650–654. [38] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B: Condens. Matter 37 (1988) 785–789. [39] S. Miertuš, E. Scrocco, J. Tomasi, Chem. Phys. 55 (1981) 117–129. [40] E. Cances, B. Menucci, J. Tomasi, J. Chem. Phys. 107 (1997) 3032–3041. [41] J.L. Pascual-Ahuir, E. Silla, I. Tuñón, J. Comput. Chem. 15 (1994) 1127–1138. [42] M.J. Frisch et al., Gaussian 09, Gaussian Inc., Wallingford, CT, 2009. [43] R. Pierotti, Chem. Rev. 76 (1976) 717–726. [44] P. Claverie, J.P. Daudey, J. Langlet, B. Pullman, D. Piazzola, M.J. Huron, J. Phys. Chem. 82 (1978) 405–418. [45] F. Floris, J. Tomasi, J. Comput. Chem. 10 (1989) 616. [46] F. Floris, J. Tomasi, J.L. Pascual-Ahuir, J. Comput. Chem. 12 (1991) 784. [47] J. Tomasi, B. Mennucci, R. Cammi, Chem. Rev. 105 (2005) 2999–3093. [48] R. Battino, T. Rettich, T. Tominaga, J. Phys. Chem. Ref. Data 12 (1983) 163–178. [49] A.W. Shaw, A.J. Vosper, J. Chem. Soc. Faraday Trans. 1 (8) (1976) 1239–1244. [50] P.J. Hesse, J. Chem. Eng. Data 41 (1996) 195–201. [51] C.L. Young, Oxides of nitrogen, in: C.L. Young (Ed.), IUPAC Solubility Data Series, Pergamon Press, Exeter, GB, 1981, p. 172. [52] R.J. Wilcock, R. Battino, W.F. Danforth, E. Wilhelm, J. Chem. Thermodyn. 10 (1978) 817–822. [53] A. Bin, Sci. Eng. 28 (2006) 67–75. [54] K. Shirono, T. Morimatsu, F. Takemura, J. Chem. Eng. Data 53 (2008) 1867– 1871. [55] W. Kunerth, Solubility of carbon dioxide and nitrous oxide in certain solvents, Phys. Rev. 19 (1922) 512–524. [56] P. Lühring, A. Schumpe, J. Chem. Eng. Data 34 (1989) 252–257. [57] L. Yen, J. McKetta, J. Chem. Eng. Data 7 (1962) 288–289. [58] A.P. Scott, L. Radom, J. Phys. Chem. 100 (1996) 16502–16513. [59] Y.N. Panchenko, C.W. Bock, G.R. De Mare, J. Struct. Chem. 46 (2005) 49–58. [60] H. Wang, C. Yu, Res. Chem. Intermed. 36 (2010) 1003–1019.

196

S. Signorelli et al. / Archives of Biochemistry and Biophysics 512 (2011) 190–196

[61] C.M. Baker, A.D.J. MacKerell, J. Mol. Model. 16 (2010) 567–576. [62] M. Möller, H. Botti, C. Batthyany, H. Rubbo, R. Radi, A. Denicola, J. Biol. Chem. 280 (2005) 8850–8854. [63] W.K. Subczynski, J.S. Hyde, Biophys. J. 41 (1983) 283–286. [64] T. Malinski, Z. Taha, S. Grunfeld, S. Patton, M. Kapturczak, P. Tomboulian, Biochem. Biophys. Res. Commun. 193 (1993) 1076–1082. [65] G.G. Power, H. Stegall, J. Appl. Physiol. 29 (1970) 145–149. [66] S.A. Simon, J. Gutknecht, Biochim. Biophys. Acta 596 (1980) 352–358. [67] L.R. De Young, K.A. Dill, Biochemistry 27 (1988) 5281–5289. [68] T. Sugii, S. Takagi, Y. Matsumoto, J. Chem. Phys. 123 (2005) 184714. [69] P. Jedlovszky, M. Mezei, J. Phys. Chem. B 107 (2003) 5322–5332. [70] Y. Wang, J. Cohen, W.F. Boron, K. Schulten, E. Tajkhorshid, J. Struct. Biol. 157 (2007) 534–544. [71] E.S. Smotkin, F.T. Moy, W.Z. Plachy, Biochim. Biophys. Acta 1061 (1991) 33–38. [72] A. Missner, P. Kugler, S.M. Saparov, K. Sommer, J.C. Mathai, M.L. Zeidel, P. Pohl, J. Biol. Chem. 283 (2008) 25340–25347. [73] W.K. Subczynski, M. Lomnicka, J.S. Hyde, Free Radic. Res. 24 (1996) 343–349. [74] L.M. Connor, A. Bidani, J. Goerke, J.A. Clements, E.M. Postlethwait, J. Appl. Physiol. 91 (2001) 2024–2034.

[75] S. Pennathur, C. Bergt, B. Shao, J. Byun, S.Y. Kassim, P. Singh, P.S. Green, T.O. McDonald, J. Brunzell, A. Chait, J.F. Oram, K. O’Brien, R.L. Geary, J.W. Heinecke, J. Biol. Chem. 279 (2004) 42977–42983. [76] H. Zhang, K. Bhargava, A. Keszler, J. Feix, N. Hogg, J. Joseph, B. Kalyanaraman, J. Biol. Chem. 278 (2003) 8969–8978. [77] H. Zhang, J. Joseph, J. Feix, N. Hogg, B. Kalyanaraman, Biochemistry 40 (2001) 7675–7686. [78] S. Bartesaghi, G. Ferrer-Sueta, G. Peluffo, V. Valez, H. Zhang, B. Kalyanaraman, R. Radi, Amino Acids 32 (2007) 501–515. [79] S. Bartesaghi, J. Wenzel, M. Trujillo, M. Lopez, J. Joseph, B. Kalyanaraman, R. Radi, Chem. Res. Toxicol. 23 (2010) 821–835. [80] N. Romero, G. Peluffo, S. Bartesaghi, H. Zhang, J. Joseph, B. Kalyanaraman, R. Radi, Chem. Res. Toxicol. 20 (2007) 1638–1648. [81] S. Solar, W. Solar, N. Getoff, J. Phys. Chem. 88 (1984) 2091–2095. [82] D. Barber, J. Thomas, Radiat. Res. 74 (1978) 51–65. [83] S. Bartesaghi, V. Valez, M. Trujillo, G. Peluffo, N. Romero, H. Zhang, B. Kalyanaraman, R. Radi, Biochemistry 45 (2006) 6813–6825.