Structural insights into the activation mechanism of melibiose permease by sodium binding Meritxell Granella, Xavier Leóna,1, Gérard Leblancb, Esteve Padrósa,2, and Víctor A. Lórenz-Fonfríaa,2 a Unitat de Biofísica, Departament de Bioquímica i de Biologia Molecular, Facultat de Medicina, and Centre d’Estudis en Biofísica, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain; and bInstitut de Biologie et Technologies-Saclay, Service de Bioenergétique, Biologie Structurale et Mécanismes, Commissariat à l’Energie Atomique-Saclay, F-91191 Gif sur Yvette, France
Edited by H. Ronald Kaback, University of California, Los Angeles, CA, and approved November 2, 2010 (received for review June 18, 2010)
The melibiose carrier from Escherichia coli (MelB) couples the accumulation of the disaccharide melibiose to the downhill entry of Hþ , Naþ , or Liþ. In this work, substrate-induced FTIR difference spectroscopy was used in combination with fluorescence spectroscopy to quantitatively compare the conformational properties of MelB mutants, implicated previously in sodium binding, with those of a fully functional Cys-less MelB permease. The results first suggest that Asp55 and Asp59 are essential ligands for Naþ binding. Secondly, though Asp124 is not essential for Naþ binding, this acidic residue may play a critical role, possibly by its interaction with the bound cation, in the full Naþ -induced conformational changes required for efficient coupling between the ion- and sugar-binding sites; this residue may also be a sugar ligand. Thirdly, Asp19 does not participate in Naþ binding but it is a melibiose ligand. The location of these residues in two independent threading models of MelB is consistent with their proposed role. infrared spectroscopy ∣ ligand binding ∣ membrane proteins ∣ sugar/cation symporter
ccording to the chemiosmotic principles, secondary active transporters comprise membrane proteins that couple in an obligatory fashion the discharge of an ionic gradient (or that of a solute gradient) to the “uphill” translocation of different solutes in the same direction (symporters or cotransporters) or in the opposite one (antiporters or exchangers) (1). Thermodynamic considerations and a wealth of kinetic, biochemical, and biophysical studies have led to the consensual view that substrate translocation relies on the alternating-access concept (2), stating that at any moment a single binding site in a polar cavity is accessible to only one side of the membrane (see for example, recent reviews and references therein in refs. 3–8). The recent elucidation of the atomic structure of almost a dozen of transporters provides strong support to the validity of the alternatingaccess concept (see reviews cited above). Finally, the diversity of conformation(s) adopted in the different transporters crystals has yielded insights into the structural basis of the various steps of the symporter cycle. Still, many issues regarding the conformational changes, especially those involved in ligand binding and in the coupling of the ligand binding sites, remain largely unanswered. In this context, the melibiose permease (MelB) of Escherichia coli, which belongs to the Glycoside-Pentoside-Hexuronide: Cation symporter family (9) (a submember of the major facilitator superfamily, MFS), is a convenient Naþ symporter to analyze the molecular and structural basis of the interaction of the coupling ion with the transporter. MelB efficiently couples the uphill transport of α- or β-galactosides to the favorable entry of Naþ , Liþ , or Hþ (H3 Oþ ) (10,11). In the past, this property has been extensively exploited to investigate the molecular and structural basis of the ion–MelB interaction and implications in the coupling properties inherent to the symport mechanism (9, 12–16). This strategy led to show, for example, that the three cations compete for the same binding site, that the sugar/ion stoichiometry is 1∕1 and that either cosubstrate enhances the affinity for its 22078–22083 ∣ PNAS ∣ December 21, 2010 ∣ vol. 107 ∣ no. 51
companion substrate as a manifestation of their coupling (9, 17). MelB models strongly suggest that it is a 12 transmembrane (TM) helices transporter (18–23). Its structural fold (24) (see Fig. 1) is expected to follow that of crystallized transporters of the MFS superfamily (25–27) rather than the one observed in all the Naþ symporters so far crystallized, which display a structurally conserved inverted repeat of five TM helices (see reviews cited above). Accordingly, the MelB fold most likely consists in two distinct N-terminal and a C-terminal six-helix bundles, pseudosymmetrically related by an axis running through a central cavity nearly perpendicular to the membrane (24), as other members of the MFS superfamily show (28). As for the above cited symporters, the MelB transport cycling model includes several intermediate steps (12, 30) (Fig. S1). Evidence that conformational changes occur at different stages of the MelB transport cycle has been obtained using biochemical and electrophysiological methods, intrinsic and fluorescence energy transfer spectroscopy, and substrate-induced infrared difference (IRdiff ) spectroscopy by attenuated total reflection (ATR) (13, 15, 31–34). In the past, several mutagenesis studies have shown that four Asp residues (Asp19, Asp55, Asp59, Asp124), located in the putative N-terminal six-helices bundle of MelB, are crucial for Naþ -dependent affinity increase and transport of melibiose, justifying their assignment as Naþ ligands (35–38). Nevertheless, direct evidence showing that mutation of these residues directly interferes with Naþ binding is still lacking and the assigned role of these residues remains still tentative. The aim of this work was to reevaluate the role of these Asp residues in MelB, with special emphasis in their involvement in the Naþ and melibiose binding and in their coupling process. To do so we have mostly relied on substrate-induced IRdiff spectroscopy by ATR. In this technique, substrate-induced IRdiff spectra are measured in response to the interaction of the coupling ion with MelB or to the interaction of the sugar with the ion–MelB binary complex (32, 33, 39). The structural information contained in the amide I and amide II regions of the IRdiff spectra provides insight into the conformational changes occurring at these early stages of the symporter reaction, including those related to the coupling process. To aid in the characterization and quantitative interpretation of IRdiff spectra we have applied a spectral correlation analysis. The results from IRdiff spectroscopy have been complemented by intrinsic fluorescence spectroscopy. Our results point to which residues are involved in Naþ binding and suggest a Author contributions: E.P. designed research; M.G. performed research; G.L. and V.A.L.-F. contributed new reagents/analytic tools; M.G., X.L., and V.A.L.-F. analyzed data; and G.L., E.P., and V.A.L.-F. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1
Present address: Centre de Biotecnologia Animal i Teràpia Gènica, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain.
To whom correpondence may be addressed. E-mail: [email protected]
or [email protected]
This article contains supporting information online at www.pnas.org/lookup/suppl/ doi:10.1073/pnas.1008649107/-/DCSupplemental.
sound mechanism for the coupling of the Naþ and sugar binding sites. Finally, we used experimentally plausible homology models of MelB for a more detailed interpretation of our results. Results In the IRdiff technique the concentrations of substrates in contact with the protein are modulated by a continuous flow of buffers, probing only the sample nearby (∼0.7 μm) to the ATR crystal surface (Fig. 2A and Fig. S2). Given the large and widely distributed number of IR active vibrations in a protein, almost any interaction with the substrate will induce an IRdiff spectrum (40), in contrast to other spectroscopic methods relying on localized probes. Information about protein structural changes is mainly constrained to the 1;700–1;500 cm−1 region (amide I and II), whereas changes in amino acid side chains are more broadly distributed in frequencies (Fig. 2B). We have demonstrated previously that IRdiff is able to disclose the conformational changes induced by substrates binding (and translocation) in WT MelB (32, 33). Additionally, a mutant without cysteines with a WT phenotype (hereafter C-less) and a mutant with similar substratebinding properties to WT but defective in sugar translocation (R141C) have been characterized as well by this technique (39). Here, the same technique was used for detecting and characterizing substrate-induced conformational changes on mutants of the above-mentioned amino acids, derived from C-less MelB. The C-less and mutants were purified, reconstituted in their native E. coli lipids, deposited over an ATR crystal, and placed in contact with a buffer solution (see Materials and Methods and Fig. 2A). From a quantitative comparison of the shape of the structure-sensitive amide I and amide II protein bands of the absorbance spectra, we confirmed that the introduced mutations had a small (for D55C, D59C, and D124C) or insignificant (for D19C and C-less) structural effect on MelB (Fig. S3). This observation reasonably discards that the observed defective substrate-binding phenotypes (see below) could be due to loss of the protein native structure. Effect of Naþ Binding on the Protein Structure. In the control C-less permease addition of 10 mM Naþ to the medium (threefold Granell et al.
Fig. 2. Revealing MelB residues participating to Naþ binding. (A) MelB reconstituted in E. coli lipids in contact with Naþ -free buffer (100 mM KCl, 20 mM MES, pH 6.6) is probed by the IR evanescent electromagnetic field generated in the ATR crystal–sample interface. (B) Naþ -induced IR difference spectra at 4 cm−1 resolution of MelB C-less and C-less mutants of candidate residues. Difference spectra were obtained by replacing Naþ -free buffer by a buffer media containing Naþ at the concentration indicated on the left-hand side. The buffer exchange protocol and data acquisition scheme are shown in Fig. S2 A and B. All the difference spectra (here and in Fig. 3) were normalized to the amount of probed protein (Fig. S3). The average difference spectrum from three or more (D19C, D124C, and D59C) or two (C-less and D55C) independent experiments is shown, with each experiment representing here and in Fig. 3 the average from at least 25 repetitions conducted on each sample. A sample containing only E. coli lipids was used as negative control. For visualization purposes, the D124C difference spectrum is also shown after multiplication by 3.5. (C) Spectral similarity and intensity of the Naþ -induced IR difference spectra in the mutants compared to the C-less (see Materials and Methods and Fig. S4). The error bar corresponds to one standard error of the mean. (D) Dendrogram clustering samples according to their spectral similitude in response to Naþ (see Materials and Methods and Fig. S4C).
above the affinity constant) (30) generates a reproducible difference spectrum, formally originated from the substitution of a proton (H3 Oþ ) for Naþ in the cation-binding site. This difference spectrum includes many discrete spectral bands well above the noise level (Fig. 2B). Peaks in the difference spectrum reflect not only interaction(s) of Naþ with the cationic-binding site ligands and associated local structural adjustments, but also induced protein structural changes responsible for the well-established increase in sugar affinity following Naþ binding. Even if only few of the arising peaks have been tentatively assigned to given amino acids or to specific secondary structure components (32, 33), the difference spectrum in the protein amide I and II regions offers a powerful fingerprint to globally and quantitatively characterize the structural response of the different mutants to Naþ with respect to C-less permease. To make the quantitative comparison between any mutant and C-less as unbiased as possible, a linear regression analysis encompassing the structure-sensitive 1;710–1;500 cm−1 region from their difference spectra was applied (see Materials and Methods and Fig. S4 for further details). This global analysis provides two outputs (Fig. 2C). First, the linear correlation parameter, R2 , quantifies the spectral similarity of a mutant response relative to the C-less, i.e., the percentage of spectral features in common PNAS ∣ December 21, 2010 ∣
vol. 107 ∣
no. 51 ∣
Fig. 1. Yousef-Guan’s model of the MelB structure obtained by threading through a crystal structure of the lactose permease of E. coli (24). (A) Two views of the solvent excluded surface of the model including helices I, II, and IV as a reference. (B) A cartoon representation of the protein backbone, with helices I, II, and IV highlighted. The putative Naþ ligands as proposed in Poolman et al. (9) are shown in stick representation (Asp19, Asp55, Asp59, and Asp124). The figure was produced from the coordinates provided in ref. 24.
with the control C-less. A high spectral similarity for a mutant is expected to correlate with structural changes in response to the substrate highly similar to those of the C-less. Second, the slope of the linear correlation gives the relative intensity of the spectral features in common with the C-less intensity. A relative intensity lower than 100% for any given mutant implies either a reduced affinity for the added substrate or smaller structural changes in response to substrate binding than for the C-less. As seen in Fig. 2B, D55C and D59C did not show any clear peak assignable to the protein in their difference spectra (flat signal below ∼1;720 cm−1 ). Even after rising the Naþ concentration up to 50 mM the spectral response was less than 4% of that observed in the C-less at 10 mM (Fig. 2C), suggesting a major reduction in the affinity constant for Naþ . The inability of these mutants to bind Naþ is further confirmed by a similar response of a negative control containing E. coli lipids without MelB (Fig. 2b), and also by their insignificant spectral similarity with respect the C-less (Fig. 2C). These results confirm that Asp55 and Asp59 are essential side chains for Naþ binding. In contrast, D19C displayed an almost WT- or C-less-type signal in terms of intensity and similarity (Fig. 2 B and C), meaning that the previous suggestion that Asp19 is involved in Naþ binding (9) is not correct. The behavior of D124C deserves some special comments. This mutant displays a Naþ -induced IRdiff spectrum, indicating that it retains the ability to bind Naþ (Fig. 2B), implying that Asp124 is not essential for Naþ binding to MelB. However, the resulting difference spectrum shows a moderate similarity (∼50%) and a reduced intensity (∼15%) with respect to the C-less difference spectrum (Fig. 2C). The moderate similarity suggests that the Naþ -induced structural changes of D124C may be less complete that those occurring in C-less. This conclusion is supported by the fact that D124C lacks some intense peaks present in C-less (e.g., at 1,640 and at 1;575 cm−1 , Fig. 2B), whereas all peaks in D124C IRdiff spectrum are also seen in C-less. On the other hand, the markedly smaller intensity of the difference spectrum of D124C appears not to be completely due to a reduced affinity for Naþ . Increasing the Naþ concentration from 10 to 50 mM increased the D124C signal 1.9-fold, suggesting an increase of the Naþ -
affinity constant from 3 to ∼15 mM as a consequence of Asp124 mutation. Such a reduction of affinity could only account for a twofold reduction in the intensity of the IRdiff spectrum with respect the C-less, whereas a sevenfold reduction was observed instead. Consequently, not only the structural changes induced by Naþ in D124C are less complete than in C-less, but are also of smaller amplitude, i.e., Asp124 is required for full and nativelike structural protein changes in response to Naþ binding. To further complete the characterization of the studied mutants, we performed a comparison of the similitude of the Naþ -induced IRdiff spectra across all mutants, collected in a correlation matrix (Fig. S4C) used to construct a dendrogram (conceptually similar to a phylogenetic tree) (Fig. 2D). The responses of the D55C and D59C mutants and of the lipid sample appear unclustered, as expected for samples without any specific interaction with Naþ . In contrast, D19C and D124C are clustered together with the C-less, reflecting that they bind Naþ . However D124C is the most distant among them, highlighting the abovecommented notion that in D124C the interaction with Naþ triggers structural changes that differ from C-less and D19C. Binding of Melibiose in the Absence of Naþ. We next explored how each mutation affects the MelB ability to bind melibiose in the absence of Naþ , i.e., when the only possible coupling ion is Hþ . In these experiments the sugar was added at a concentration of 50 mM (Fig. S2), a value close to the half-saturating concentration of C-less for melibiose (30). Because protons are present in the medium, substrate translocation becomes possible and it may contribute also to the difference spectra. We should note, however, that this contribution is not the dominant one, because R141C that binds but does not translocate the cosubstrates (41) shows only a modest modification of the sugar-induced IRdiff signal compared to that of the functional C-less (39). Fig. 3A shows that mutants D55C and D59C gave rise to a melibiose-induced IRdiff spectra reasonably similar to that of the C-less (∼70%), even if the decreased intensity in the case of D59C (∼35%) may suggest a ∼4-5-fold reduction of the sugar affinity (Fig. 3B). After a complete spectral similarity comparison
lipids 1800 1700 1600 1500 1400 1300
wavenumber / cm
D124C D55C D59C
1800 1700 1600 1500 1400 1300
wavenumber / cm-1
without Na + with 10 mM Na
C -le ss D 19 C D 12 4C D 55 C D 59 C
10 mM Na+
melibiose relative effect in the spectra intensity (%)
Relative to C-less (in %)
mel + 10 mM Na+
60 40 20 0
C -le s D s 19 D C 12 4C D 55 C D 59 C lip id s
D 10 mM mel
C -le ss D 55 C D 59 C D 19 D C 12 4C lip id s
50 mM melibiose
50 mM mel + H+
C -le s D s 19 D C 12 4C D 55 C D 59 C lip id s
Relative to C-less (in %)
Fig. 3. Involvement of the studied MelB residues in the melibiose binding and the Naþ -melibiose coupling. (A) IR difference spectra at 4 cm−1 resolution of MelB C-less and site-point mutants of the C-less induced by 50 mM melibiose in Naþ free media. The buffer exchange protocol and data acquisition scheme are shown in Fig. S2C. The average of three or more (C-less, D19C, D124C, and D59C) or two (D55C) independent experiments is shown. An experiment with only E. coli lipids (without protein) is included as negative control. (B) Comparison of the spectral similarity and intensity of the melibiose-induced IR difference spectra of the C-less mutants with that of the C-less (see Materials and Methods and Fig. S5). Error bars correspond to one standard error of the mean. (C) Dendrogram clustering samples according to their spectral similitude in response to melibiose (see Materials and Methods and Fig. S5C). (D) Melibiose-induced IR difference spectra at 4 cm−1 resolution of MelB C-less (10 mM melibiose) and several C-less mutants (50 mM melibiose) in the presence of 10 mM Naþ (see Fig. S2 D and E). The average difference spectrum from three (D19C) or two (C-less, D124C, D55C, and D59C) independent experiments is shown. An experiment without protein (only E. coli lipids) is included as a negative control. (E) Effect of the presence of Naþ in the intensity of the difference spectra induced by melibiose. An increase on the intensity induced by Naþ is a signature of a preserved Naþ -melibiose coupling. (F) Comparison of the spectral similarity and intensity of the melibioseinduced IR difference spectra in the presence of Naþ with respect to the C-less (see Materials and Methods and Fig. S6). 22080 ∣
Granell et al.
Binding of Melibiose in the Presence of Naþ. The shape of the melibiose-induced IRdiff spectra of mutants (Fig. 3 D and F) were not affected by the presence of 50 mM Naþ in the medium, in contrast to the C-less at 10 mM Naþ (compare Fig. 3A and D; note the different scale in these figures). Moreover, the presence of Naþ did not enhance the intensity of the IRdiff spectrum induced by melibiose, as observed in the C-less (Fig. 3E). Therefore, the coupling between the cation- and sugar-binding sites, reflected in the synergistic enhancement of the melibiose-induced IRdiff spectrum by Naþ , is not functional in any of the mutants. Although this behavior can be (partially) accounted for their impaired Naþ (D55C, D59C) and melibiose (D19C, D124C) binding phenotypes, it is possible that at least some of these residues are also required for a functional coupling mechanism. The correlation matrix and the constructed dendrogram (Fig. S6 C and D) confirm that none of the mutants display a native-like response to melibiose in the presence of Naþ . Both the spectral responses of mutants with retained melibiose binding (D55C and D59C) and sodium binding (D19C and D124C) were closer among them than to the C-less, in agreement with the notion of an impaired substrate coupling. Fluorescence Spectroscopy of Mutants. As a result of the substrate coupling, Trp fluorescence intensity increases upon binding of Naþ to MelB WT or C-less in the presence of melibiose, reflecting conformational changes responsible for the Naþ -induced increase in the affinity for the melibiose (13). None of the mutants studied here showed any significant intensity variation upon incubation with melibiose or with Naþ (in the presence of melibiose) (Fig. S7). As indicated previously, comparison of the mutant’s FTIR absorbance spectra discarded overall structural alteration in the protein caused by the mutations. Therefore, we can conclude that the structural changes responsible for the increase in the melibiose affinity upon Naþ binding and affecting Trp residues environment, are lacking in the mutants. In agreement with the IRdiff results, the fluorescence results confirm that none of the studied mutants preserves a coupling mechanism between binding of Naþ and the sugar. Depending on the particular mutant, this behavior may be caused by different defects (lack of sodium or of sugar binding, or impairment of coupling). Granell et al.
Discussion On the MelB Naþ Ligands. Taking into account the high sensitivity
of infrared spectroscopy, any change in the protein structure or in the protonation state of side chains should be detected in the substrate-induced IRdiff spectra. Therefore, we can conclude that (i) Asp55 and Asp59 are essential ligands to Naþ because neither D55C nor D59C exhibit any structural variations upon incubation with the Naþ coupling ion. This conclusion is consistent with previous functional results. D59C does not show any transport coupled to Hþ or Naþ , implicating this side chain in cation and/or melibiose binding (35, 42, 43). D55C, which still shows measurable transport with Hþ , has selectively lost the capacity to cotransport the sugar with Naþ , suggesting an implication of Asp55 in Naþ binding (37, 38). (ii) Asp19 does not interact with Naþ , because D19C displays a Naþ -induced difference spectrum nearly identical to that of C-less. Asp19 was formerly suggested as a ligand to Naþ due to its impaired melibiose transport (9), which we can now explain from its involvement as a melibiose ligand. (iii) Most probably, Asp124 interacts with Naþ because D124C gives rise to a difference spectrum of only 15% intensity and ∼50% similarity when compared to C-less. An attractive interpretation is that Asp124 is a conditional ligand, establishing an interaction with Naþ only when the ion is already bound to Asp55 and Asp59. Although not essential for Naþ binding, this interaction would be favorable, increasing the affinity for Naþ . But more importantly, it would presumably lead Asp124 to approach the bound Naþ , driving by these means part of the observed Naþ -induced conformational changes, the ones lacking in D124C. Such hypothesis is further developed below in view of molecular models for MelB. Side Chains Interacting with Melibiose. Our study suggests which side chains may serve as melibiose ligands. D19C and D124C showed only small and rather featureless difference spectra upon melibiose incubation in the presence of Hþ or Naþ . Therefore, Asp19 and Asp124 are most likely ligands to melibiose, which is in agreement with the absence of transport for D19C and D124C (36, 42, 44). But in spite of their low intensity, the melibioseinduced difference spectra of these two mutants showed a reproducible and similar shape (Fig. 3), implying some interaction of melibiose with MelB. It may be possible that in D19C and D124C the affinity of the active binding site for melibiose drops to the point that melibiose binding can only induce marginal structural changes. Alternatively, the observed faint spectroscopic changes could originate from the interaction of melibiose with a secondary low-affinity binding site as described for the LeuTAa transporter (45). Concerning Asp55 and Asp59, the melibiose-induced IRdiff spectra of their mutants are of significant intensity. Therefore, we can conclude that Asp55 and Asp59 are not essential residues for melibiose binding. When the coupling ion is Hþ , the decreased signal of D55C and D59C as compared to C-less may be due either to the contribution of these residues to conform the right binding environment for the sugar, to defective Hþ binding in these mutants, or both. In the presence of Naþ , the spectra are similar to those obtained in the presence of Hþ , in shape and intensity, showing that there is absence of Naþ binding in these two mutants even after melibiose is bound. The lack of Naþ binding makes it unfeasible to obtain information about the role of Asp55 or Asp59 in melibiose interaction when Naþ is the coupling cation. Our conclusions agree well with functional properties previously described. An example is provided by binding studies of D55C and D59C. Using the high affinity melibiose analog p-nitrophenyl-alpha-D-galactopyranoside, an affinity around 70% of C-less in the presence of Hþ and between 15% and 25% in the presence of Naþ where obtained (36). On the other hand, substitution of Asp55 with Cys does not drastically modify the Hþ -driven sugar translocation properties of MelB (37). These PNAS ∣ December 21, 2010 ∣
vol. 107 ∣
no. 51 ∣
(Fig. S5C) both the IRdiff spectra of D55C and D59C mutants appear clustered with that of the C-less (Fig. 3C), reflecting a similar melibiose binding phenotype in terms of protein structural changes. This result provides additional evidence that melibiose binds to these two mutants in a native-like way. In contrast, D19C and D124C gave rise to small and rather featureless melibiose-induced difference spectra (Fig. 3A), with nearly negligible intensity (∼5%) and low similarity (25;000 scans for every difference spectrum. Fluorescence measurements were performed using a UV-visible QuantaMaster™ spectrofluorimeter. MelB modeling was done with the I-Tasser server (46). ACKNOWLEDGMENTS. We thank Dr. Lan Guan for a critical reading of the manuscript, and Elodia Serrano and Neus Ontiveros for skillful technical assistance. This work was supported a postdoctoral fellowship (40607) from the Universitat Autònoma de Barcelona and Marie Curie Reintegration Grant PIRG03-6A-2008-231063 (to V.L.-F.), by Ministerio de Ciencia e Innovación Grants BFU2006-04656/BMC, BFU2009-08758/BMC, and in part by a grant from the Commissariat à l’Energie Atomique.
Granell et al.
Granell et al.
29. Wilson TH, Ding PZ (2001) Sodium-substrate cotransport in bacteria. Biochim Biophys Acta 1505(1):121–130. 30. Meyer-Lipp K, et al. (2006) The inner interhelix loop 4-5 of the melibiose permease from Escherichia coli takes part in conformational changes after sugar binding. J Biol Chem 281:25882–25892. 31. Meyer-Lipp K, Ganea C, Pourcher T, Leblanc G, Fendler K (2004) Sugar binding induced charge translocation in the melibiose permease from Escherichia coli. Biochemistry 43:12606–12613. 32. León X, Lórenz-Fonfría VA, Lemonnier R, Leblanc G, Padrós E (2005) Substrate-induced conformational changes of melibiose permease from Escherichia coli studied by infrared difference spectroscopy. Biochemistry 44:3506–3514. 33. León X, Lemonnier R, Leblanc G, Padrós E (2006) Changes in secondary structures and acidic side chains of melibiose permease upon cosubstrates binding. Biophys J 91:4440–4449. 34. Ganea C, Pourcher T, Leblanc G, Fendler K (2001) Evidence for intraprotein charge transfer during the transport activity of the melibiose permease from Escherichia coli. Biochemistry 40:13744–13752. 35. Pourcher T, Deckert M, Bassilana M, Leblanc G (1991) Melibiose permease of Escherichia coli: Mutation of aspartic acid 55 in putative helix II abolishes activation of sugar binding by Naþ ions. Biochem Biophys Res Commun 178:1176–1181. 36. Pourcher T, Zani ML, Leblanc G (1993) Mutagenesis of acidic residues in putative membrane-spanning segments of the melibiose permease of Escherichia coli. I. Effect on Naþ -dependent transport and binding properties. J Biol Chem 268:3209–3215. 37. Zani ML, Pourcher T, Leblanc G (1993) Mutagenesis of acidic residues in putative membrane-spanning segments of the melibiose permease of Escherichia coli. II. Effect on cationic selectivity and coupling properties. J Biol Chem 268:3216–3221. 38. Wilson DM, Wilson TH (1992) Asp-51 and Asp-120 are important for the transport function of the Escherichia coli melibiose carrier. J Bacteriol 174:3083–3086. 39. León X, Leblanc G, Padrós E (2009) Alteration of sugar-induced conformational changes of the melibiose permease by mutating Arg141 in loop 4-5. Biophys J 96:4877–4886. 40. Kotting C, Gerwert K (2005) Proteins in action monitored by time-resolved FTIR spectroscopy. ChemPhysChem 6:881–888. 41. Abdel-Dayem M, Basquin C, Pourcher T, Cordat E, Leblanc G (2003) Cytoplasmic loop connecting helices IV and V of the melibiose permease from Escherichia coli is involved in the process of Naþ -coupled sugar translocation. J Biol Chem 278:1518–1524. 42. Ding PZ, Wilson TH (2001) The effect of modifications of the charged residues in the transmembrane helices on the transport activity of the melibiose carrier of Escherichia coli. Biochem Biophys Res Commun 285:348–354. 43. Matsuzaki S, Weissborn AC, Tamai E, Tsuchiya T, Wilson TH (1999) Melibiose carrier of Escherichia coli: Use of cysteine mutagenesis to identify the amino acids on the hydrophilic face of transmembrane helix 2. Biochim Biophys Acta 1420:63–72. 44. Ding PZ, Wilson TH (2001) Cysteine mutagenesis of the amino acid residues of transmembrane helix I in the melibiose carrier of Escherichia coli. Biochemistry 40:5506–5510. 45. Singh SK, Piscitelli CL, Yamashita A, Gouaux E (2008) A competitive inhibitor traps LeuT in an open-to-out conformation. Science 322:1655–1661. 46. Roy A, Kucukural A, Zhang Y (2010) I-TASSER: A unified platform for automated protein structure and function prediction. Nature Protoc 5:725–738. 47. Dave N, Lórenz-Fonfría VA, Leblanc G, Padrós E (2008) FTIR spectroscopy of secondarystructure reorientation of melibiose permease modulated by substrate binding. Biophys J 94:3659–3670. 48. Lórenz-Fonfría VA, Granell M, León X, Leblanc G, Padrós E (2009) In-plane and outof-plane infrared difference spectroscopy unravels tilting of helices and structural changes in a membrane protein upon substrate binding. J Am Chem Soc 131:15094–15095. 49. Cordat E, Leblanc G, Mus-Veteau I (2000) Evidence for a role of helix IV in connecting cation- and sugar-binding sites of Escherichia coli melibiose permease. Biochemistry 39:4493–4499. 50. Vardy E, Arkin IT, Gottschalk KE, Kaback HR, Schuldiner S (2004) Structural conservation in the major facilitator superfamily as revealed by comparative modeling. Protein Sci 13:1832–1840. 51. Lemieux MJ (2007) Eukaryotic major facilitator superfamily transporter modeling based on the prokaryotic GlpT crystal structure. Mol Membr Biol 24:333–341.
PNAS ∣ December 21, 2010 ∣
vol. 107 ∣
no. 51 ∣
1. Mitchell P (1972) Chemiosmotic coupling in energy transduction: a logical development of biochemical knowledge. J Bioenerg 3:5–24. 2. Jardetsky O (1966) Simple allosteric model for membrane pumps. Nature 211:969–970. 3. Guan L, Kaback HR (2006) Lessons from lactose permease. Annu Rev Biophys Biomol Struct 35:67–91. 4. Krishnamurthy H, Piscitelli CL, Gouaux E (2009) Unlocking the molecular secrets of sodium-coupled transporters. Nature 459:347–355. 5. Abramson J, Wright EM (2009) Structure and function of Naþ -symporters with inverted repeats. Curr Opin Struct Biol 19:425–432. 6. Padan E (2008) The enlightening encounter between structure and function in the NhaA Naþ -Hþ antiporter. Trends Biochem Sci 33:435–443. 7. Law CJ, Maloney PC, Wang DN (2008) Ins and outs of major facilitator superfamily antiporters. Annu Rev Microbiol 62:289–305. 8. Forrest LR, Rudnick G (2009) The rocking bundle: A mechanism for ion-coupled solute flux by symmetrical transporters. Physiology 24:377–386. 9. Poolman B, et al. (1996) Cation and sugar selectivity determinants in a novel family of transport proteins. Mol Microbiol 19:911–922. 10. Tsuchiya T, Wilson DM, Wilson TH (1985) Melibiose-cation cotransport system of Escherichia coli. Ann NY Acad Sci 456:342–349. 11. Pourcher T, Leclercq S, Brandolin G, Leblanc G (1995) Melibiose permease of Escherichia coli: Large scale purification and evidence that Hþ , Naþ , and Liþ sugar symport is catalyzed by a single polypeptide. Biochemistry 34:4412–4420. 12. Pourcher T, Bassilana M, Sarkar HK, Kaback HR, Leblanc G (1990) The melibiose∕Naþ symporter of Escherichia coli: Kinetic and molecular properties. Philos Trans R Soc Lond B Biol Sci 326:411–423. 13. Mus-Veteau I, Pourcher T, Leblanc G (1995) Melibiose permease of Escherichia coli: Substrate-induced conformational changes monitored by tryptophan fluorescence spectroscopy. Biochemistry 34:6775–6783. 14. Mus-Veteau I, Leblanc G (1996) Melibiose permease of Escherichia coli: Structural organization of cosubstrate binding sites as deduced from tryptophan fluorescence analyses. Biochemistry 35:12053–12060. 15. Maehrel C, Cordat E, Mus-Veteau I, Leblanc G (1998) Structural studies of the melibiose permease of Escherichia coli by fluorescence resonance energy transfer. I. Evidence for ion-induced conformational change. J Biol Chem 273:33192–33197. 16. Franco PJ, Wilson TH (1996) Alteration of Naþ -coupled transport in site-directed mutants of the melibiose carrier of Escherichia coli. Biochim Biophys Acta 1282:240–248. 17. Damiano-Forano E, Bassilana M, Leblanc G (1986) Sugar binding properties of the melibiose permease in Escherichia coli membrane vesicles. Effects of Naþ and Hþ concentrations. J Biol Chem 261:6893–6899. 18. Pourcher T, Bibi E, Kaback HR, Leblanc G (1996) Membrane topology of the melibiose permease of Escherichia coli studied by melB-phoA fusion analysis. Biochemistry 35:4161–4168. 19. Hacksell I, et al. (2002) Projection structure at 8 Å resolution of the melibiose permease, an Na-sugar co-transporter from Escherichia coli. EMBO J 21:3569–3574. 20. Purhonen P, Lundback AK, Lemonnier R, Leblanc G, Hebert H (2005) Three-dimensional structure of the sugar symporter melibiose permease from cryo-electron microscopy. J Struct Biol 152:76–83. 21. Dave N, et al. (2000) Secondary structure components and properties of the melibiose permease from Escherichia coli: A fourier transform infrared spectroscopy analysis. Biophys J 79:747–755. 22. Gwizdek C, Leblanc G, Bassilana M (1997) Proteolytic mapping and substrate protection of the Escherichia coli melibiose permease. Biochemistry 36:8522–8529. 23. Botfield MC, Naguchi K, Tsuchiya T, Wilson TH (1992) Membrane topology of the melibiose carrier of Escherichia coli. J Biol Chem 267:1818–1822. 24. Yousef MS, Guan L (2009) A 3D structure model of the melibiose permease of Escherichia coli represents a distinctive fold for Naþ symporters. Proc Natl Acad Sci USA 106:15291–15296. 25. Abramson J, et al. (2003) Structure and mechanism of the lactose permease of Escherichia coli. Science 301:610–615. 26. Huang Y, Lemieux MJ, Song J, Auer M, Wang DN (2003) Structure and mechanism of the glycerol-3-phosphate transporter from Escherichia coli. Science 301:616–620. 27. Hirai T, et al. (2002) Three-dimensional structure of a bacterial oxalate transporter. Nat Struct Biol 9:597–600. 28. Abramson J, Kaback HR, Iwata S (2004) Structural comparison of lactose permease and the glycerol-3-phosphate antiporter: Members of the major facilitator superfamily. Curr Opin Struct Biol 14:413–419.
Supporting Information Granell et al. 10.1073/pnas.1008649107 SI Materials and Methods Expression, Purification, and Reconstitution of Melibiose Permease (MelB) Mutants. A recombinant pK95ΔAHB plasmid with a cas-
sette containing the melB gene encoding a permease with a 6-Histag at its C-terminal end (1) and devoid of its four native cysteines (C-less MelB) (2) is used as a background for further permease engineering and as a control. Escherichia coli DW2-R (ΔmelB, ΔlacZY) (3) was transformed with the plasmid harboring the mutated MelB (C-less, D19C, D55C, D59C, or D124C, respectively). Freshly transformed cells were grown at 30 °C in M9 medium supplemented with 54 mM glycerol, 0.2% (wt∕vol) casamino acids, 1 mM thiamine, and 0.3 mM ampicillin, until an OD600 of 1.0–1.2 was reached. The cells were washed and resuspended in 50 mM Tris • HCl, 50 mM NaCl, and 5 mM 2-mercaptoethanol at pH 8. Inverted membrane vesicles were prepared by means of a Microfluidizer 110S (20,000 psi) and washed with 50 mM Tris • HCl, 50 mM NaCl, and 5 mM 2-mercaptoethanol at pH 8. Mutated His-tagged MelB was purified from inverted membrane vesicles. The protein was solubilized by using 1% (wt∕vol) 3-(laurylamido)-N,N′-dimethylaminopropylamine oxide. Chromatography procedures were used to prepare MelB protein solubilized in 0.1 % (wt∕vol) dodecylmaltoside ðDDMÞ (1). MelB reconstitution into liposomes (protein/lipid ratio 1∕2 wt∕wt) was performed by removing the detergent with Bio-Beads SM-2. MelB content was assayed by a Lowry procedure including 0.2% (wt∕vol) sodium dodecyl sulfate and using bovine serum albumin as standard (4). Sample Preparation and FTIR Data Acquisition. A sample of 20 μL of a proteoliposomes suspension (about 150 μg of protein) was spread homogeneously on a germanium attenuated total reflection (ATR) crystal (50 × 10 × 2 mm, Harrick Scientific Products, yielding 12 internal reflections at the sample side) and dried under a stream of nitrogen. The substrate-containing buffer and the reference buffer (containing no MelB substrates) were alternatively perfused over the proteoliposomes film at a rate of ∼1.5 mL∕ min. The film was exposed to the substrate-containing buffer for 4 min and washed with the reference buffer for 10 min (30 min when the substrate-containing buffer had 50 mM melibiose or 50 mM NaCl). The switch of buffers was carried out by a computer-controlled electrovalve. For each cycle, 1,000 scans at a resolution of 4 cm−1 were recorded and a minimum of 25 spectra were taken and averaged in order to increase the signal-to-noise ratio (i.e., a total of ≥25;000 scans for every difference spectrum). Spectra were recorded with a FTS6000 Bio-Rad spectrometer, equipped with a mercury-cadmium-telluride detector. Sample temperature was adjusted to 20 °C using a cover jack placed over the ATR crystal and connected to a circulating thermostatic bath. The cover jack temperature was controlled with a fitted external probe. Data Corrections for FTIR Difference Spectra. The experimental difference spectrum contains four possible contributions: (i) sample (protein and lipid) difference spectrum induced by the substrate (s); (ii) water difference spectrum induced by the substrate(s) solvation; (iii) absorbance of the substrate(s) (in our case melibiose, because the cations do not absorb IR light); (iv) change in the swelling of the film, with an apparent gain or loss of water with a concomitant apparent loss or gain of sample (protein and lipid). The latter contribution was corrected by subtracting, from the experimental difference spectrum, an absorption spectrum of MelB proteoliposomes in the substrate-containing buffer. The Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
subtraction factor used was that able to remove the lipid CH2 bands at 2,920 and 2;850 cm−1 . Contributions (ii) and (iii) were corrected subtracting from the experimental difference spectrum a difference between the substrate-containing buffer and the reference buffer. The subtraction factor was adjusted to flatten the water band between 3,700 and 2;800 cm−1 and to remove bands coming from the substrates (melibiose give an intense band at 1;080 cm−1 , whereas cations give no bands). Quantitative Comparison of Intensity and Similarity of FTIR Difference Spectra. All the difference spectra presented were normalized for
the amount of protein contributing to the IR signal, as described below. Quantitative comparison of two normalized difference spectra (an input and a reference spectrum) was performed by a linear regression in the 1;710–1;500 cm−1 interval. This region includes the protein conformation-sensible amide I and amide II bands from the peptide bond. The correlation analysis provides two relevant parameters, the correlation coefficient and the slope. The correlation coefficient, R2 , measures the spectral similarity of the input with respect to the reference spectrum in response to the added substrate. The slope quantifies the relative intensity of common features in the input with respect to the reference spectrum arising in response to the added substrates. The correlation analysis provides an estimate of the uncertainty in the slope, but not for the uncertainty in R2 . Final uncertainties for R2 were obtained from replicate experiments and for the slope by combining uncertainty estimates for individual experiments and replicate experiments. For the absorbance spectra, we used a slightly different region in the correlation analysis (1;700– 1;500 cm−1 ) to minimize contributions from the C═O lipid absorbance. Here, R2 quantifies spectral differences with respect to the reference, whereas the slope only reflects differences in the amount of protein sensed by the evanescent wave with respect to the reference. This slope was used to normalize the difference spectra of MelB mutants for the actual amount of protein sensed by the evanescent wave, and so contributing to the IR difference spectra. In order to enhance its sensitivity and to minimize the potentially pernicious contribution of instrumental baseline features, the correlation analysis was performed on the first derivative of the difference spectra and on the second derivative of the absorbance spectra. Analysis was performed in Matlab v7 (Mathworks). First and second derivatives were performed in the Fourier domain with phase correction, with a cut point of 0.167 cm (6 cm−1 ) and 0.125 cm (8 cm−1 resolution) for the first and second derivatives, respectively, using a Sinc2 filter in both cases (5). Linear correlation was performed using the Matlab built-in function “regress.” A spectral similarity (correlation) matrix was computed from a correlation analysis for all possible combinations of input and reference spectra. This matrix collects the spectral similarity between different MelB mutants, computed as R2 × 100. From the correlation matrices we constructed dendrograms. A dendrogram groups spectral elements (i.e., mutants or type of proteins) accordingly to their interspectral distance. We used the Matlab built-in function “dendrogram” using an average linkage, with a measure of distance between two elements given by ð1 − R2 Þ × 100. Fluorescence Spectroscopy. Fluorescence measurements were performed using a UV-visible QuantaMaster™ spectrofluorometer and processed with the Felix 32 software (Photon Technology International). For the Trp fluorescence spectra, the excitation wavelength was set at 290 nm, with a bandwidth of 5 nm. The 1 of 9
emission spectrum was collected with a bandwidth of 8 nm. The concentration of reconstituted MelB used in the fluorescence experiments was 23 ng∕mL in 100 mM KPi buffer, pH 7. Prior taking the spectra, the sample was sonicated for 30 s at 4 °C in an ultrasonic bath. MelB Modeling with the I-Tasser Server. The I-Tasser server performs automated protein structure predictions from the amino acid sequence (http://zhanglab.ccmb.med.umich.edu/I-TASSER/). Models are built based on multiple-threading alignments by LOMETS and iterative TASSER simulations (6, 7). After submission of the full-length sequence, the server identified the glycerol-31. Pourcher T, Leclerc S, Brandoli G, Leblanc G (1995) Melibiose permease of Escherichia coli: Large scale purification and evidence that Hþ , Naþ , and Liþ sugar symport is catalyzed by a single polypeptide. Biochemistry 34:4412–4420. 2. Weissborn AC, Botfield MC, Kuroda M, Tsuchiya T, Wilson TH (1997) The construction of a cysteine-less melibiose carrier from E. coli. Biochim Biophys Acta 1329:237–244. 3. Botfield MC, Wilson TH (1988) Mutations that simultaneously alter both sugar and cation specificity in the melibiose carrier of Escherichia coli. J Biol Chem 263:12909–12915.
phosphate transporter (GlpT, 1pw4_A) as the preferred template, but used also the crystal structure of the lactose permease (LacY, 1pv6_A). The server delivered five atomic models, with varying confidence score. Model 2 was selected based on a favorable location of known side chains important for MelB function. The 3D structure retrieved from the server was used without further modifications, except from the optimization of side-chain conformation of Asp19, Asp55, Asp59, and Asp124 to reduce to the minimum their distance to the putative bound Naþ . Physically allowed rotamers for these side chains were identified using the software Chimera (http://www.cgl.ucsf.edu/chimera/) 4. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ (1951) Protein measurement with the Folin phenol reagent. J Biol Chem 193:265–275. 5. Lórenz-Fonfría VA, Padrós E (2004) Curve-fitting of Fourier manipulated spectra comprising apodization, smoothing, derivation and deconvolution. Spectrochim Acta, A 60:2703–2710. 6. Zhang Y (2008) I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9:40. 7. Roy A, Kucukural A, Zhang Y (2010) I-TASSER: A unified platform for automated protein structure and function prediction. Nat Protoc 5:725–738.
Fig. S1. Extended model for Naþ -driven transport of melibiose by MelB. States 1–3 represent outward-facing MelB conformations (without and with the stepwise binding of substrates), whereas states 4–6 are the equivalent inward-facing conformations. The transition between outward- and inward-facing conformations is expected to occur through an occluded state (state 3′), where the bound substrates are neither accessible to the periplasm not to the cytosol.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
2 of 9
Fig. S2. Substrate-induced difference buffer exchange and IR spectra data acquisition scheme for (A) 10 mM Naþ -induced IR difference spectra; (B) 50 mM Naþ -induced IR difference spectra; (C) 50 mM melibiose-induced IR difference spectra; (D) 10 mM melibiose-induced IR difference spectra in the presence of 10 mM Naþ ; and (E) 50 mM melibiose-induced IR difference spectra in the presence of 10 mM Naþ . A buffer without substrates, buffer A, and a buffer with substrate(s), buffer B [buffer A plus substrate(s)] were alternatively and repetitively flowed over the sample, modulating the substrates concentration in the sample compartment (light blue for Naþ, and purple for melibiose). Both buffers A and B contained 100 mM KCl and 20 mM MES at pH 6.6, and buffer A extra KCl until compensating for any ionic strength differences with buffer B. The estimated substrate concentration profile during the buffers exchange is indicated and reflects the approximated substrate concentration in the buffer in contact with the sample (the estimated time constant of buffer exchange in the sample compartment is ∼2.4 min). Data acquisition, Abs (marked in red), was performed in the last 5 min of buffer A flow (after wash), and buffer B flow, averaging 1,000 spectra at 4 cm−1 resolution in each. The average expected substrate concentration during data acquisition is indicated. The difference between the absorbance spectra in buffer B minus A provides the substrate-induced IR difference spectrum (ΔAbs). Only one cycle is shown for clarity, lasting between 22 and 44 min, but typically ∼25 of such cycles were performed and averaged for one experiment.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
3 of 9
Fig. S3. Structural comparison of MelB mutants using IR spectroscopy of hydrated samples. (A) Absorbance IR spectrum at 4 cm−1 resolution of the Cys-less (C-less) MelB mutant in a hydrated film with the buffer contribution subtracted (black). The spectrum includes the structure-sensitive amide I and II regions. The second derivative at 8 cm−1 resolution (orange) provides more resolved details, robust to small differences in the buffer subtraction, and therefore more suitable for the structural comparison performed below. (B) Second derivative in the amide I/II region for the MelB mutants considered in this study (blue), plus the R141C mutant and the WT MelB, and two unrelated α-helical membrane proteins: the ATP/ADP mitochondrial transporter (Anc2), a secondary transporter as MelB; and the proton pump bacteriorhodopsin (bR). As a visual reference, the spectrum of C-less MelB is shown in red. Vertical blue bars (barely visible) corresponds to 1 SE (n ¼ 4–8). All spectra intensities were normalized to the C-less. (C) Illustration of the method used for structural comparison between two spectra, based in measuring their linear correlation in the amide I/II regions. The R2 gives the similarity, while the slope gives their relative intensity, used to normalize the spectra in B, and the difference spectra of C-less mutants to the C-less in Figs. 2 and 3 of the manuscript. (D) Representation of all possible comparisons between pair of protein, with a color code indicating their degree of similarity, measured as R2 × 100. The color map has been shifted to increase the contrast. (E) Dendrogram representing the spectral distance (related to the structural distance) between different proteins forms, measured as ð1 − R2 Þ × 100. Note how all MelB variants are clustered together, but with two resolved subclasses: one for D124C, D55C, and D59C forms, and another for C-less, WT, D19C, and R141C forms.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
4 of 9
Fig. S4. Illustration of how the quantitative comparison of the intensity and similarity of the Naþ -induced IR difference spectra of MelB mutants with respect to the C-less form was performed. For the C-less, D19C, D124C, and the lipids 10 mM Naþ was used to induce the difference spectra, and 50 mM Naþ for the rest. (A) Naþ -induced IR difference spectrum of C-less MelB at 4 cm−1 resolution (black), and its phase-corrected first derivative at 6 cm−1 resolution (orange). The phase-corrected first derivative was used in the posterior analysis, because it shows more details and is more robust to baseline errors. (B) The phase-corrected first derivative of the Naþ -induced IR difference spectra of MelB mutants were plotted with respect to the C-less spectrum. Here only the plots of D19C and D59C versus the C-less are shown for clarity (gray circles). The linear regression (in red) provides the correlation coefficient, R2 , that quantifies how well the mutant difference spectrum resembles that of C-less (i.e., spectral similarity relative to C-less) and the slope, m, that quantifies the relative intensity of similar features between mutants and C-less (i.e., spectral intensity relative to C-less). As a negative control we used a Naþ -induced IR difference spectrum of E. coli lipids. Note the differences in intensity scale between C-less, D19C, and D124C on one side, and D55C, D59C, and the lipids on the other side. (C) Correlation matrix collecting all possible comparisons between pair of Naþ -induced difference spectra, with a color code indicating their degree of similarity, measured as R2 × 100.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
5 of 9
Fig. S5. Illustration of the quantitative comparison of the intensity and similarity of 50 mM melibiose-induced IR difference spectra of MelB mutants with respect to the C-less form, in the absence of Naþ . (A) Melibiose-induced IR difference spectrum of C-less MelB at 4 cm−1 resolution (black) and its phase-corrected first derivative (orange). (B) The phase-corrected first derivative of the melibiose-induced IR difference spectra of MelB mutants were plotted with respect to the C-less spectrum, although here for clarity only the plots of D19C and D59C versus the C-less are shown for clarity (gray circles). The linear regression (in red) provides the correlation coefficient, R2 , and the slope, m, which measure the spectral similarity and intensity of the mutants with respect to the C-less. As a negative control we used a melibiose-induced IR difference spectrum of E. coli lipids alone. Note the differences in scale among spectra. (C) Correlation matrix collecting all possible comparisons between pair of melibiose-induced difference spectra, with a color code indicating their degree of similarity, measured as R2 × 100.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
6 of 9
Fig. S6. Illustration of the quantitative comparison of the intensity and similarity of melibiose-induced IR difference spectra of MelB mutants with respect to the C-less form in the presence of 10 mM Naþ . Except for the C-less (10 mM melibiose), 50 mM melibiose was used to induce the difference spectra. (A) Melibiose-induced IR difference spectrum of C-less at 4 cm−1 resolution (black) and its phase-corrected first derivative (orange). (B) The phase-corrected first derivative of the melibiose-induced IR difference spectra of MelB mutants were plotted with respect to the C-less spectrum (here only D19C and D59C versus C-less are shown for clarity). The linear regression (in red) provides the correlation coefficient, R2 , and the slope, m, which measure the spectral similarity and intensity of the mutants with respect to the C-less. As a negative control we used a melibiose-induced IR difference spectrum of E. coli lipids alone in the presence of 10 mM Naþ . (C) Correlation matrix collecting all possible comparisons between pair of melibiose-induced difference spectra in the presence of Naþ , with a color code indicating their degree of similarity, measured as R2 × 100. (D) Dendrogram constructed from the elements of the correlation matrix, classifying samples accordingly to their spectral distance, measured as ð1 − R2 Þ × 100.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
7 of 9
Fig. S7. Effect of adding melibiose and NaCl on the fluorescence of MelB permease mutants. Representative plot of fluorescence variation induced by the addition of melibiose (10 mM final concentration) in a nominally Naþ -free medium, followed by addition of NaCl (10 mM final concentration). Arrows indicate addition times. Samples of 1 mL containing proteoliposomes at 23 ng∕mL of protein in 100 mM potassium phosphate buffer at pH 7.0 and 20 °C were illuminated at 290 nm with a bandwidth of 5 nm, and the relative variation in fluorescent light emitted at 335 nm (ΔF∕F 0 ) was plotted over time.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
8 of 9
Fig. S8. Model of the MelB structure obtained by multiple-threading alignments and iterative simulations using the I-TASSER server (1, 2). The model represents an inward-facing conformation. The possible ligands to Naþ (Asp19, Asp55, Asp59, and Asp124) as proposed in Poolman et al. (3) are shown in stick representation. To make the model comparable to that of Fig. 1, residues 1–5 and 449–473 were removed. 1. Zhang Y (2008) I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9:40. 2. Roy A, Kucukural A, Zhang Y (2010) I-TASSER: A unified platform for automated protein structure and function prediction. Nat Protoc 5:725–738. 3. Poolman B, et al. (1996) Cation and sugar selectivity determinants in a novel family of transport proteins. Mol Microbiol 19:911–922.
Granell et al. www.pnas.org/cgi/doi/10.1073/pnas.1008649107
9 of 9