Lucille
Babel
a,
Minh-Ha
Nguyen
b,
Cédric
Mittelheisser
a,
Marie
Martin
b,
Katharina M.
Fromm
a,
Olivier
Walker
b and
Maggy
Hologne
*b
aUniversity of Fribourg, Department of Chemistry, chemin du musée 9, Fribourg 1700, Switzerland
bUniversité de Lyon, CNRS, UCB Lyon1, Institut des Sciences Analytiques, UMR 5280, 5 rue de la Doua, Villeurbanne 69100, France. E-mail: maggy.hologne@univ-lyon1.fr
First published on 9th August 2021
SilE and SilB are both proteins involved in the silver efflux pump found in Gram-negative bacteria such as S. typhimurium. Using model peptides along with NMR and CD experiments, we show how SilE may store silver ions prior to delivery and we hypothesize for the first time the interplay between SilB and SilE.
Fig. 1 The sequence of SilB-p is extracted from the complete sequence of SilB Salmonella typhimurium (Uniprot Q9ZHD0, see ESI†). The sequence of SilE-p is extracted from the complete sequence of SilE Salmonella typhimurium (Uniprot Q9Z4N3, see ESI†). |
We report qualitative and quantitative aspects of the interplay between SilE-p and SilB-p with respect to silver binding and transfer. We have first investigated the ability of SilE-p to interact with SilB-p by means of NMR. To this end, we have assigned the corresponding NMR chemical shifts of SilE-p and SilB-p by a combination of 1H–1H TOCSY and 1H–1H NOESY experiments (see ESI†). The corresponding 1H,15N-HSQC spectra for SilE-p and SilB-p exhibit an ensemble of chemical shifts distributed on a narrow region in the 1H frequency dimension (Fig. S3A and B and ESI† for assignments). This hallmark, added to the fact that no NOE between amide protons of residues i, i + 3 could be detected, is a clear indication that SilE-p and SilB-p are disordered in the free state.17 To detect a potential interaction between SilE-p and SilB-p, we monitored signal shifts in a 1H,15N-HSQC spectrum of a SilE-p/SilB-p mixture at a 1:1 molar ratio and compared it with their isolated 1H,15N-HSQC spectra in the free state. As shown on Fig. S3C and D (ESI†), the spectra of the isolated SilE-p and SilB-p nicely overlap with the spectrum corresponding to the SilE-p/SilB-p mixture and therefore provide a clear evidence that SilE-p and SilB-p do not interact with each other. To map the interaction of silver with either SilE-p or SilB-p, we monitored the chemical shift perturbations (CSPs) in a series of 1H,15N-HSQC spectra upon addition of silver ions. As illustrated on Fig. 2, several residues on both SilE-p or SilB-p present significant CSPs while some showed a strong decrease of their signal intensities, indicative of intermediate exchange. It has to be recalled that CSPs include either, even light, structural rearrangement and interaction with silver ions. On SilE-p, the most affected residues cluster around two binding sites that comprise K82, M83 and R89, M90 (see Fig. 2A). Additionally, M83 and M90 exhibit a strong signal broadening indicative of intermediate-to-slow exchange and hence a tight binding with silver ions. The magnitude of the CSPs of residues participating in interaction saturate at approximately 2:1 molar ratio of silver ions to SilE-p and therefore indicates a 2:1 stoichiometry, in good agreement with the proposed presence of two binding sites (see Fig. 3A). To quantify the binding interactions, we have used a 2:1 binding model (see ESI†) and assumed two binding sites with identical affinity. We have derived a dissociation constant of 4.5 ± 1.5 μM from the NMR titration data in good agreement with the Kd reported previously via a different method.6 By means of the same methodology, we have investigated the interaction of SilB-p with silver ions. As can be seen on Fig. 2B, the most significant CSPs are located around R407 flanked by M406 and H408 while a second binding site is identified around A418, involving M416 and M419. Surprisingly, M426 does not show any significant CSPS, thus does not participate in silver binding. A closer inspection of the corresponding CSPs of the two binding sites reveals a distinct behaviour upon silver titration. The first site that induces chemical shifts of M416, A418 and M419 exhibits a hyperbolic titration curve that reaches a plateau after 2 equivalents of silver ions (see Fig. 3B). In the case of the second binding site resulting in CSPs of R405, M406, R407 and H408, the titration curve shows a sigmoidal shape and suggests a cooperative silver binding. Moreover, the Ag+ concentration needed to reach the bond state is twice compared to the previous binding site. This effect has already been underlined in several other cases where intrinsic disorder might optimize allosteric coupling in proteins.18 To quantify the binding affinity of silver with SilB-p, we have used a sequential model that takes into account the cooperative binding of the silver ions. This model is based on two binding events and thus relies on two dissociation constants associated with the two binding sites (see Appendix 1, ESI†). Overall, we derived two Kds of 4 ± 2 and 571 ± 80 μM respectively, showing a clear difference between the two SilB-p binding sites and a negative cooperativity of the second binding site. The lowest Kd is associated with the site that primarily saturates and encompasses M416–M419 while the highest Kd is committed to the binding site that saturates at higher [Ag+] concentration and encompasses R405-H408. To rule out the possibility that the much higher Kd of the second binding site could result from its own affinity, we have engineered two other shorter peptides SilB-p1 and SilB-p2 that comprise the two individual binding sites (see ESI†). Mass spectrometry reveals that each shorter peptide is prone to bind one silver ion (see Fig. S4, S5 and Table S1, ESI†). Additionally, as can be seen on Table S2 and Fig. S6, S7 (ESI†), the derived Kds for SilB-p1 and -p2 have a similar affinity with a Kd of 8 ± 2 and 2 ± 1 μM respectively. It therefore supports a negative cooperative binding effect solely detected in the case of SilB-p that comprises the two binding sites.
Fig. 3 Representative titration curves for SilE-p starting from a free state of SilE-p or a mixture of SilE-p/SilB-p as a function of the molar ratio of Ag+ (A) and SilB-p starting from a free state of SilB-p or a mixture of SilB-p/SilE-p as a function of the molar ratio of Ag+ (B). Symbols represent experimental data while plain lines represent fitted data according to the models presented in ESI.† |
A possible explanation for the decrease in affinity may be due to an increase of energy required for the second binding site to reach the bond state. According to these observations, it is likely that silver ions first bind to the site that embraces M416 and M419 and then to the second site including M406. This clearly pinpoints a new binding site for SilB that has neither been detected in its CusB13 counterpart nor in SilB from Cupriavidus metallidurans.14 Conversely to SilE-p that undergoes structural folding upon silver binding, SilB-p remains fully unstructured in its silver-bound state. This finding is supported by several observations. First and foremost, we did not detect any NOE contact between amide protons of residues i, i + 3 in the silver-bound SilB-p. Second, we have observed low ellipticity above 210 nm and negative bands near 195 nm during a silver titration in circular dichroism experiments (see Fig. S9, ESI†).
That SilE-p and SilB-p both bind silver ions has encouraged us to investigate their interplay. Accordingly, we used an equimolar mixing of SilE-p and SilB-p to which we added a graduated amount of silver ions from [Ag+]:[peptide] ratios of 1:1 to 8:1 with respect to SilE-p or SilB-p. From a first glimpse at Fig. 2C and D, SilE-p and SilB-p present similar binding sites for silver whether they are taken individually or together. Nevertheless, their CSP intensities or titration curves present several discrepancies.
From the SilB-p side, the measured CSPs show a similar pattern compared to those recorded for an independent SilB-p alone in the presence of silver (compare with Fig. 2B). Moreover, the titration curves corresponding to SilB-p or the SilB-p/SilE-p mixture nicely overlap upon addition of silver (Fig. 3B) for the first binding site that encompasses M416 and M419. As a result, our data support the fact that the presence of SilE-p does not affect the binding of silver on the first site of SilB-p. For the second binding site that clusters around M406, only a slight difference could be detected in the titration curves with a slower increase of the saturation rate. The picture is drastically different for the SilE-p in such a mixture as (i) the recorded CSPs display weaker values at saturation for both binding sites (Fig. 2D) and (ii) the titration curves present different shapes that become sigmoidal. Additionally, a larger amount of silver is necessary for SilE-p to reach a plateau (see Fig. 3A). These experimental results strongly support the fact that when SilE-p and SilB-p are both present, SilB-p first binds silver before SilE-p could accommodate the remaining silver ions.
To lend credence to our hypothesis, we carried out competition assays. Starting from a preformed SilB-p/Ag+ complex at a [SilB-p]/[Ag+] ratio of 1:2, we added SilE-p at a [SilB-p]/[SilE-p] ratio of 1:1. According to Fig. 4B, the SilB-p binding site that encompasses M419 does not display any signal shift. Conversely, the addition of SilE-p caused the second binding site that clusters around M406 to shift toward the original position they occupied in the free state. We then increased the amount of Ag+ to reach a [SilB-p]/[Ag+] ratio of 1:4 and finally 1:8 (Fig. 4C and D). While the first binding site (M419) does not experience any further shift, the signals associated with the second binding site (R405) shift back to the position they occupied at a 1:2 ratio and reach a plateau at a 1:8 ratio. Our observations clearly demonstrate that SilE-p outcompetes the second binding site of SilB-p for silver binding and acts as a regulator when the silver concentration increases. This mechanism allows SilB-p to bind silver at a much higher concentration compared to its concentration if SilE-p would be absent.
We carried out a similar experiment from the SilE-p side with a preformed SilE-p/Ag+ complex at a [SilE-p]/[Ag+] ratio of 1:2 (Fig. 5A). We then added SilB-p at a [SilB-p]/[SilE-p] ratio of 1:1 that causes the signals associated with both binding sites to shift back to the position they occupied in the free state of SilE-p.
Nevertheless, one can notice that this effect is less pronounced than the one experienced by the second binding site of SilB-p when SilE-p is added. This is likely due to the presence of the first and strong binding site of SilB-p that outcompetes SilE-p. Adding a further amount of silver causes the signals to saturate at a 1:4 ratio with no further variation at 1:8 (Fig. 5C and D). These competition assays led us to hypothesize that SilE-p may bind the excess of silver that SilB-p would not accommodate when the silver concentration increases.
Based on this set of experimental observations, we may draw a conclusion related to the interplay mechanism between SilB-p and SilE-p. While the function or silver binding capability of the C-terminus of SilB from Salmonella typhimurium has never been reported, neither in the case of its counterpart in the Cus system nor in the case of SilB from Cupriavidus metallidurans, we have shown that either SilB-p or SilE-p possess two silver binding sites with different affinities. In the case of SilE-p, the two binding sites present a similar affinity in the micromolar range while SilB-p exhibits two binding sites that show a negative cooperativity upon silver binding. From our findings, we can hypothesize a synergetic mechanism between SilE and SilB. The negative binding cooperativity of SilB is likely to obey to a rapid remodelling of the system after a significant modification of the silver concentration. We can hypothesize that the C-terminus of SilB would slowly usher silver ions from M406 to M419 prior to a possible release to SilC via SilA. When the silver concentration significantly increases, SilE may play the role of a regulator that stores silver ions to avoid an overload of SilB. With such a mechanism, there is no disruption of the silver withdrawal from the cell and therefore maintenance of the silver resistance (see ESI† for a sketch of the possible interplay between SilB and SilE). Of course, this assumption has to be demonstrated in the context of the full native proteins SilE and SilB in a further study.
K. M. F. had the idea to study SilE and excerpts thereof and its possibility to act as chaperone to transfer silver ions to SilB (SNSF project 178827) and corrected the final manuscript. L. B. carried out mass spectrometry and 1D NMR studies, C. M. synthesized the SilB peptides, M. H. N. and M. M. carried out 2D NMR spectroscopy and CD experiments, M. H. carried out 2D NMR experiments and analysis. Finally, M. H. and O. W. wrote the manuscript. This project has also received financial support from the CNRS through the MITI interdisciplinary programs.
Footnote |
† Electronic supplementary information (ESI) available: Experimental methods, sequence alignment, binding constant determination, NMR data, CD spectra. See DOI: 10.1039/d1cc02597j |
This journal is © The Royal Society of Chemistry 2021 |