Stereoselectivity of Isoflurane in Adhesion Molecule Leukocyte Function-Associated Antigen-1

Background Isoflurane in clinical use is a racemate of S- and R-isoflurane. Previous studies have demonstrated that the effects of S-isoflurane on relevant anesthetic targets might be modestly stronger (less than 2-fold) than R-isoflurane. The X-ray crystallographic structure of the immunological target, leukocyte function-associated antigen-1 (LFA-1) with racemic isoflurane suggested that only S-isoflurane bound specifically to this protein. If so, the use of specific isoflurane enantiomers may have advantage in the surgical settings where a wide range of inflammatory responses is expected to occur. Here, we have further tested the hypothesis that isoflurane enantioselectivity is apparent in solution binding and functional studies. Methods First, binding of isoflurane enantiomers to LFA-1 was studied using 1-aminoanthracene (1-AMA) displacement assays. The binding site of each enantiomer on LFA-1 was studied using the docking program GLIDE. Functional studies employed the flow-cytometry based ICAM binding assay. Results Both enantiomers decreased 1-AMA fluorescence signal (at 520 nm), indicating that both competed with 1-AMA and bound to the αL I domain. The docking simulation demonstrated that both enantiomers bound to the LFA-1 “lovastatin site.” ICAM binding assays showed that S-isoflurane inhibited more potently than R-isoflurane, consistent with the result of 1-AMA competition assay. Conclusions In contrast with the x-ray crystallography, both enantiomers bound to and inhibited LFA-1. S-isoflurane showed slight preference over R-isoflurane.


Introduction
Many biologically active molecules, including drugs, exist in different chiral forms, and it is possible that each specific enantiomer can interact with their targets differently, thereby potentially eliciting different biological responses. Thus, the potential exists for a racemic mixture of a compound to cause diverse, and perhaps unwanted, side effects. Regardless, racemic drugs are used commonly in clinical medicine, as the purification or synthesis of one specific enantiomer can be costly. Novel drug discovery can be even more costly, so there has been recent interest in exploiting enantioselectivity in the last two decades. While the majority of drugs on the market were sold as racemic mixtures in early 1990s, about 40% of drugs were marketed as single enantiomers by 2002 [1]. Some of these single enantiomers were newly developed, while some were isolated from the previously marketed racemic mixtures [2]. The advantages of using specific enantiomers over the racemic mixture might include their less complex and more selective pharmacodynamic profiles, and the potential for improved therapeutic index and the reduction of side effects. In extreme cases, one enantiomer may act as an antagonist (such as R(-)-PN 202-791; calcium channel antagonist), while the other may be an agonist (such as S(+)-PN 202-791; calcium channel agonist) [3].
The interest in single enantiomers has also reached the anesthesia field, as demonstrated by the introduction of ropivacaine and levo-bupivacaine [3]. Further, even the inhalational general anesthetics have chiral centers. For example, isoflurane has been successfully separated into its enantiomers by Huang et al. and others in 1990s [4], allowing investigators to test the potency of each separately. The S-enantiomer of isoflurane was shown to be approximately 2-times more potent in vitro and in vivo than R-enantiomer isoflurane ( Table 1). However, the structural basis for the differential potency of isoflurane enantiomers has yet to emerge, but it has been reported that S-isoflurane binds with higher affinity to model proteins, like human serum albumin [5].
Previously we demonstrated that isoflurane interacted with the adhesion molecule leukocyte adhesion-associated antigen-1 (LFA-1). This protein is expressed ubiquitously on leukocytes and is involved in various immunological actions such as leukocyte arrest on the endothelium [6] and immunological synapse formation [7]. This heterodimeric molecule consisting of aand -b subunits is activated through the intracellular activation signals called the ''inside-out'' signal and undergoes dynamic conformational changes [8,9]. The a7 helix of the ligand binding domain (the aL I domain) of the a subunit is extremely flexible [10]. In a resting (inactive) state, the a7 helix remains helical, forming a pocket called the ''lovastatin site.'' In a high-affinity (active) state, however, an unwinding, downward movement of the a7 helix occurs causing a loss of the ''lovastatin site'', which allows the aL I domain to bind ligands at the metal-ion dependent adhesion site (MIDAS) (Figure 1). The X-ray crystallization of racemic isoflurane complexed with the aL I domain demonstrated that isoflurane bound to this ''lovastatin site'' [11], and the functional relevance of this binding in LFA-1 blockade by isoflurane was shown by the mutagenesis experiment [12]. Interestingly, further refinement of this complex suggested that only S-isoflurane, fit into the electron density map of this region; the R-isomer produced a steric clash with lining residues. This seemingly extreme example of isoflurane stereoselectivity on LFA-1 suggested by the X-ray crystallographic structure is striking as compared to the modest, functional stereoselectivity suggested by its central nervous system effects [13,14,15,16,17]. Our previous study demonsrated that the interaction of racemic isoflurane with LFA-1 had functional relevance in vivo [18]. Thus, if the suggested selectivity on LFA-1 can be demonstrated in solution, it may open the door to more selective application of these drugs in the perioperative setting.
In this study, we tested the hypothesis that the enantioselectivity of isoflurane suggested by the crystal structures of LFA-1isoflurane complex, is also observed in solution binding and functional studies.

Protein expression and purification
The aL I domain wild type (WT, residues 128-307 of the integrin aL subunit) (low affinity conformer) of LFA-1 were expressed in BL21 (DE3) cells. The aI domains were expressed as inclusion bodies, which were solubilized and refolded as previously described [19]. Soluble full ectodomain LFA-1 protein was expressed from CHO Lec 3.2.8.1 cells stably transfected with LFA-1 plasmids and purified as previously described [20].

Type of experiment
Study results Reference

In vivo
The lipid emulsion of isoflurane injection to rats. S(+) was 40+/28% more potent than R(2) at producing loss of righting reflex.
[ 35] In vivo The inhalation of isoflurane enantiomers to rats. Minimum alveolar concentMAC was S(+); 1.06%, R(2); 1.62%, suggesting S(+) was about 50% more potent. [36] In vivo The determination of MAC using rats. Inhalational route. S(+): 0.0144 +/2 0.0012 atm, R(2); 0.0169 +/2 0.0020 atm. Not statistically significant. [37] In vivo Intraperitoneal injection of isoflurane enantiomers to mice. Sleep time was longer with S(+) over R (2). [38] In vitro S(+) was more potent and efficacious than R(2) in enhancing [ 3 H] flunitrazepam binding to GABA A receptor complex. [13] In vitro S(+) had greater increase in GABA receptor mediated IPSCs than R(2) [14] In vitro S(+) was 1.6 times as potent as R(2) in augmenting GABA gated Cl flux. [15] In vitro Cl flux induced by GABA was potentiated by isoflurane. The maximum stereoselectivity occurred at 0.3 mM isoflurane (S . R), about 2 times. [16] In vitro No stereospecific effects of isoflurane in isolated guinea pig hearts (the effect on LVP, AV conduction, coronary flow) [48] In vitro No stereoselectivity of isoflurane seen to inhibit isradipine binding to L-type calcium channel [49] In vitro S(+) bound bovine serum albumin with slower association and dissociation rates than R (2). No difference in static condition. [50] In vitro S(+) was twofold more effective than R(2) both in eliciting the anesthetic-activated potassium current and in inhibiting a current mediated by neuronal nicotinic acetylcholine receptors [17] The reported studies of isoflurane enantiomers are summarized in the Fluorescence microscopy 293 T cells were transiently transfected with LFA-1 on glass bottom dishes (MatTek Corporation; Ashland, MA, USA). LFA-1 on the cells was stained with TS1/12 antibody (Immune Disease Institute, Boston, MA, USA) followed by anti-mouse Cy5 antibody (Life Technologies; Grand Island, NY, USA). 1-AMA was added to medium, incubated for 10 minutes, and then the cells were examined with fluorescence microscopy.

Chiral gas chromatography
Isoflurane enantiomers (S-isoflurane and R-isoflurane) were kindly provided by Baxter (Cambridge, MA, USA). Racemic isoflurane was from Abbott Laboratories (Abbott Park, IL, USA). We first used chiral gas chromatography with mass spectrometry detection (GC/MS) to evaluate the purity of each optical enantiomer. Samples of isoflurane were diluted in heptane (Sigma) and injected into the gas chromatography (Model G1540N-6410N, Agilent Technologies; Santa Clara, CA, USA). We used a 30-m, 0.25-mm, 0.25 mm 20% permethylated b-cyclodextrin chiral 20B column (Agilent Technologies). The parameters were a split ratio of 1: 20, injector temperature of 70muC, oven temperature of 60uC, and carrier gas hydrogen at 1 mL/min. Mass spectrometry quantification was conducted in single ion model at m/z 51, 117 and 141.

1-AMA competition assay
The competition of 1-AMA binding with the isoflurane enantiomers was performed using the aL I domain. The aL I domain (0.2 mM) was first pre-equilibrated with 10 mM 1-AMA (Sigma; St. Louis, MO, USA). After addition of 0.1 mM to 4 mM of S-, or R-isoflurane, samples were excited at 380 nm and emission spectra were collected from the range of 400 nm to 700 nm. The reduction of the fluorescence signal at 540 nm was corrected by subtracting the baseline fluorescence curves of 1-AMA and the aL I domain, and plotted against isoflurane enantiomer concentration. Both competition curves were fitted to variable slope Hill curves. Analysis was performed using PRISM 5 software (Graph Pad Software, La Jolla, CA, USA).

Docking simulation of S-and R-isoflurane
The structure of the aL I domain was obtained from the Protein Data Bank (PDB) 1ZOO [22]. The program GLIDE (Schrodinger; Cambridge, MA, USA) was used to perform rigid molecular docking of isoflurane enantiomers with the aL I domain, as previously described [20]. The structures of the isoflurane enantiomers were obtained through PubChem (http://pubchem. ncbi.nlm.nih.gov/). Isoflurane binding position was sought with the grid size of 25 6 25 6 25 Å 3 and the centroid grid residue of Tyr-257. No positional constraint was applied. GLIDE has a scoring system (''glidescore'') that ranks docked pairs based on the predicted interaction free-energy. The pair with the most negative glidescore is considered to have the highest affinity, thus we selected the docked pair with the most negative glidescore.
Cell-based Intercellular adhesion molecule-1 (ICAM-1) binding assay ICAM-1 is the major endogenous ligand for LFA-1. A cellbased LFA-1: ICAM-1 binding assay was performed using flow cytometry as previously described [23]. Briefly, 293 T cells transiently transfected with LFA-1 were harvested in HBS containing 10 mM EDTA 48 hours after transfection. Cells were washed with HBS, and then resuspended in HBS. Cells were aliquoted to polymerase chain reaction tubes (Axygen; Union City, CA, USA) and then centrifuged. Cell pellets were given150 ml aliquot of HBS, 2 mM MnCl 2 containing isoflurane at 2 6 final concentrations, and another 150 ml aliquot of HBS containing 10 mg/mL ICAM-1-Fca fusion protein. Tubes were immediately capped, mixed, and incubated for 30 minutes at room temperature. Following wash, cells were incubated with anti-human IgA-FITC (Invitrogen) as a secondary antibody for 30 minutes. The cells were then washed and subjected to flow cytometry analysis using a FACScan instrument (BD Bioscience; San Jose, CA, USA). ICAM-1 binding % was defined as [mean fluorescence intensity (MFI) of samples at various concentrations of isoflurane divided by mean fluorescence intensity of mock-treated sample] 6 100%.

Statistical analysis
All the statistical analyses were performed using PRISM 5 software. The details of statistical analysis were described in the corresponding figure legends. P,0.05 was considered statistically significant.

Results
The interaction of 1-AMA with LFA-1 1-AMA is a small molecule with environment-dependent fluorescence and general anesthetic properties as demonstrated by the potentiation of gamma-aminobutyric acid (GABA)-ergic transmission [24,25,26]. It has been used to explore protein cavities, and binds to a well-characterized ''general anesthetic site'' in horse spleen apoferritin (HSAF) [26,27]. We previously  demonstrated that 1-AMA exhibited fluorescence shift in the presence of aL I domain, suggesting that 1-AMA interacted with LFA-1 [21]. Our previous direct photolabeling work showed that propofol bound the ''lovastatin site.'' In addition propofol displaced 1-AMA from LFA-1. We explored the size of available cavities on the surface of aL I domain and found that only the ''lovastatin site'' was a larger cavity than 1-AMA. Thus we concluded that 1-AMA bound the ''lovastatin site.'' This site is also the binding site of LFA-1 allosteric antagonists, and therefore we hypothesized that 1-AMA would block LFA-1. As shown in

The binding of 1-AMA to LFA-1 on the cell surface
The interaction of 1-AMA with LFA-1 was studied in a cellbased system as well. Using microscopy, we asked if 1-AMA binds to LFA-1 on the surface of transfected cells. 1-AMA fluorescence was clearly co-localized with LFA-1 ( Figure 2B), however, it was also seen in non-transfected cells, indicating that it has other binding targets. Because there are many proteins with hydrophobic cavities that can accommodate 1-AMA (,500 s 3 ) [28], this was predicted. Our previous work showed that isoflurane bound LFA-1 at the ''lovastatin site'' [12,20], so we hypothesized that isoflurane and 1-AMA would compete at this site, with the Senantiomer being more potent than R-isoflurane. Because 1-AMA displacement requires a small amount of reagents and the available amount of isoflurane enantiomers is limited, we opted to use this method in a cell-free system.

The purity of S-and R-isoflurane enantiomers
We validated that the S-and R-enantiomers as supplied by the company were greater than 99% pure, and that the racemic mixture consisted of 50% (62%) of each. (Figure S1A-C).

Interactions of the aL I domain with two isoflurane enantiomers
Our hypothesis was that only S-isoflurane binds to the ''lovastatin site''. In contrast, however, both enantiomers reduced fluorescence intensity, suggesting that they both interact with the lovastatin site on the aL I domain ( Figure 3A). The dissociation constants (Kd) for S-isoflurane and R-isoflurane (374.8 mM and 460.2 mM, resp.) were not statistically different ( Figure 3B).

Docking of S-and R-isoflurane into the aL I domain
Both enantiomers were successfully docked to the ''lovastatin site'' using GLIDE ( Figure 4B and 4C). The neighboring amino acid residues are listed in Table 2. In both S-and R-isoflurane docking models, the trifluoromethyl heads were in the same orientation and formed hydrophobic interactions with Ile-235 and Leu-302. However, the difluoromethyl groups assumed different orientations. The difluoromethyl group in R-isoflurane had hydrophobic interactions with Leu-298 and Leu-302, while in Sisoflurane it interacted with Tyr-257 and Leu-302. Consistent with the 1-AMA experiments, the comparison of the docking scores indicated an inability to detect a significant difference of affinity to the aL I domain between S-and R-isoflurane ( Table 3).

Inhibition of LFA-1 by S-and R-isoflurane
The above results indicate that both isoflurane enantiomers bind to the ''lovastatin site,'' although subtle differences may exist which might be detectable using functional assays. Thus, we tested if both enantiomers could inhibit ligand (ICAM) binding. We used Mn 2+ to activate LFA-1, which mimics inside-out signaling as described above. Also we confirmed that concentrations of S-, Rand racemic isoflurane remained steady during the incubation for 30 minutes. Both S-and R-isoflurane inhibited the binding of ICAM-1 to LFA-1 ( Figure 5). Interestingly, S-isoflurane inhibited ICAM-1 binding ,50% more potently than R-isoflurane, indicating the presence of functional enantioselectivity of isoflurane enantiomers to LFA-1.

Discussion
Here we have demonstrated that both S-and R-isoflurane interact with LFA-1 at the lovastatin site, and impair its ligand binding. 1-AMA displacement assay and docking simulations were unable to detect a statistically significant difference in the interaction of the aL I domain with isoflurane enantiomers. Both enantiomers inhibited the binding of LFA-1 to ICAM-1 in cell- Table 2. The distance of near-by amino acid residues from isoflurane (, 4 Å ). based assays, with the S-enantiomer being more potent. The enantioselectivity of isoflurane on anesthetic targets in the CNS has been documented as in Table 1, and in general indicate a very modest effect. This study is the first to investigate the enantioselectivity of targets relevant to immune function.

Amino acid R-isoflurane (Å ) S-isoflurane (Å )
Our experimental results were inconsistent with structural analysis of previous x-ray crystallography experiments, which indicated that R-isoflurane could not bind this pocket due to steric clash with Tyr-257 [11]. While X ray crystallography is a very useful tool to provide detailed structural information, it may not reflect subtle changes in protein dynamics that allow accommodation of ligands (''induced fit''). Also, numerous lattice contacts at the protein-protein interfaces can subtly alter the crystal structure from the solution structure. Examples are alpha-lactalbumin [29], myoglobin [30] and aspartate transcarbamylase [31]. Finally, it is known that crystallization may not demonstrate structural variability as in the case of intestinal fatty-acid-binding apoprotein where X-ray crystallization did not demonstrate the variability of backbone structure [32]. Both 1-AMA displacement and cellbased ICAM-1 binding assays, were performed in a solution setting where dynamics should be retained. The ''lovastatin site'' underneath the C-terminal a7-helix was demonstrated to be the most variable region in reported I-domain X-ray crystallographic structures [33], as demonstrated by the difference of cavity size depending on the crystallization conditions [21]. The region was quite flexible in solution nuclear magnetic resonance structures [10]. Therefore, the probable explanation of the discrepancy between the structural and our data is a loss of normal protein flexibility or dynamics [10] that might normally result in ''induced fit'', and which are reduced in the crystal due to multiple lattice contacts, leading into the altered features of small molecule binding sites. Our study further stresses the importance of solutionbased assay for the validation of crystallographic studies.
Pfeiffer's rule states that the stereoselectivity of a drug is a function of the mass of drug necessary to produce its pharmacological effects [34]. Given that isoflurane's plasma concentration that produces general anesthetic effects is relatively high (, mM range), it is expected that isoflurane would demonstrate only modest stereoselectivity. Accordingly, previous in vivo studies of isoflurane enantiomers did not demonstrate more than a 20-50% difference in anesthetic effect [35,36,37,38]. Our cell-based functional results showed that a similar magnitude of functional stereoselectivity was the case for LFA-1 as well. Previously, we demonstrated that isoflurane also bound the top domain of the b2 subunit called the bI domain, which is also an important regulator of ICAM-1 binding [20]. There could also be a difference of affinity against this domain between S-and R-isoflurane, which may explain the small difference of the results between 1-AMA displacement assay (or docking simulation) using only the aL I domain and cell-based ligand binding assay. Unfortunately, the fact that the isolated b2 I domain protein has not been successfully expressed [39] prevented us from testing this hypothesis.
LFA-1 antagonism by isoflurane has a few intriguing clinical implications. For example, LFA-1 knockout mice had a higher mortality than wild-type mice after intraperitoneal S. pneumoniae infection [40]. However, Emoto et al. demonstrated that LFA-1 knockout mice were resistant to lipopolysaccharide-induced liver injury [41]. Further, Miyamoto et al. showed that LFA-1 knockout mice had resistance to Listeria monocytogenes infection [42]. Blockade of LFA-1 has been shown to be beneficial in graft survival in various transplantation models [43,44,45]. A clinical pilot study of allo-islet transplantation recipients treated with the humanized anti-LFA-1 antibody efalizumab showed a favorable outcome [46]. In addition, efalizumab has been used to treat psoriasis [47]. Unfortunately, this drug has been withdrawn from the market due to a risk of fatal brain infections. These studies demonstrate that LFA-1 blockade could be beneficial or detrimental depending on circumstances. Also it has to be noted that many of these studies have been done with permanent LFA-1 blockade (knockout mice), or antibody blockade which usually has a prolonged effect. In our previous studies, we demonstrated that a short exposure of isoflurane (2-4 hours) attenuated neutrophil recruitment in the setting of skin inflammation, a process in which LFA-1 plays a large role. In the future, the impact of transient LFA-1 blockade by anesthetics needs to be examined. The answer to this question will allow us to consider to redesigning, or reformulating our anesthetic drugs to mitigate this LFA-1 functional alteration.
Additional finding in this study is that 1-AMA, which is a general anesthetic binding site probe, possessing general anesthetic activity, is an LFA-1 antagonist and bound LFA-1 on the cell Figure 5. The ligand binding assay of leukocyte function-associated antigen-1 (LFA-1) using isoflurane enantiomers. The binding of LFA-1 to ICAM-1 in the presence of isoflurane (racemate, enantiomers) was tested using 293T cells overexpressing LFA-1 using flow cytometry as described in the method. Data represent mean +/-S.D. of triplicates. To compare any difference in ICAM-1 binding % in the presence of racemic, Sand R-isoflurane, statistical analysis was performed using two-way ANOVA with Bonferroni post hoc analysis. * denotes p,0.05. doi:10.1371/journal.pone.0096649.g005 surface. However, the existence of other targets is likely, which clouds interpretation.
In conclusion, we have demonstrated that both isoflurane enantiomers interact with LFA-1, an important component of immunologic cascades, with evidence for subtle enantioselectivity. The clinical significance of this interaction remains to be determined in the future.

Supporting Information
Figure S1 Chiral gas chromatography of isoflurane. The gas chromatographic traces were shown. (A) racemic isoflurane, (B) S-or (C) R-isoflurane. The area under the curve was used to calculate the composition of S-and R-isoflurane in each solution.