SEARCH

SEARCH BY CITATION

Keywords:

  • barnase;
  • barstar;
  • ligand-receptor interactions;
  • force spectroscopy

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES

Results of the single molecule force spectroscopy study of specific interactions between ribonuclease barnase and its inhibitor barstar are presented. Experimental data obtained for the force loading rate ranging 2–70 nN/s are well approximated by a single straight line, from which the dissociation barrier of the width of 0.12 nm and height of 0.75–0.85×10−19 J can be inferred. The measured value of specific interaction does not depend on the NaCl concentration. This apparently contradicts the well-known dependence of the binding energy of this pair on the salt concentration, but such a “contradiction” is explained by the insensitivity of the force spectroscopy data to the relatively long-range electrostatic interaction. The latter essentially contributes to the value of barnase–barstar binding energy revealed by biochemical measurements, and it is exactly this electrostatic interaction which is influenced by the salt concentration. Copyright © 2010 John Wiley & Sons, Ltd.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES

Nowadays, the Atomic Force Microscopy (AFM)-based single molecule force spectroscopy became a versatile research tool enabling to obtain, among others, important and often unique in comparison with other methods data concerning the energy landscape of ligand–receptor pairs. The most efficiently interacting avidin–biotin and streptavidin–biotin pairs naturally appeared as the first objects having been investigated by the single molecule force spectroscopy (Lee et al., 1994; Florin et al., 1994), and numerous other pairs have been studied later. Analysis of the literature (see, among others, recent reviews Zlatanova et al., 2000; Schwesinger et al., 2000; Weisel et al., 2003; Lee et al., 2007) reveals that the corresponding data is now available for many tens, if not hundreds, ligand–receptor pairs.

By these reasons the lack of the force spectroscopy data concerning the ribonuclease barnase (110 amino acids) and its inhibitor barstar (89 amino acids; see Hartley, 2001 for review) seems somewhat surprising. This protein pair is not only “the second most effectively interacting one” just after avidin/streptavidin–biotin (dissociation constant for the former is around 10−14 M (Mariani et al., 1992; Schreiber and Fersht, 1993; Hartley, 2001) while it is equation image for the latter (Green, 1990), but it is much exploited in practice and is nowadays considered as a very prospective “constructor” to create multivalent complexes used as innovative protein therapeutics (Deyev et al., 2003). Both these proteins can be efficiently expressed, are stable and highly resistive to the harsh environmental conditions, and they are capable to reversible restoration of the structure and function after denaturation when an action of denaturating agents is removed. Their 3D structure is well known, and their association/dissociation reactions were much studied and modeled. It is also worthwhile to mention that barnase has been long considered as a kind of model system when the protein folding/defolding problem is concerned (Fersht, 1993).

All aforementioned circumstances make single molecule force spectroscopy study of barnase–barstar pair quite timely. Evidently, corresponding data is important for the better understanding of this system and hence for its biotechnological and medical applications. Another motivation for this study consists in the following. Experimental (Schreiber and Fersht, 1993; Schreiber and Fersht, 1995, 1996) and theoretical (Chong et al., 1998; Lee and Tidor, 2001a, 2001b; Schneierman and Honig, 2002; Dong et al., 2003; Wang et al., 2004; Spaar et al., 2006; Tang et al., 2006; Ababobu et al., 2007) data attest an important contribution of an electrostatic interaction to the barnase–barstar energy landscape. However, despite a lot of efforts spent for the corresponding modeling and calculations, even the sign of the effect still cannot be unambiguously predicted. Biochemical experiments clearly show a favorable contribution of an electrostatic interaction: the binding free energy in barnase–barstar complex reduces on the value of 3.1 kcal/mol for 500 mM NaCl concentration as compared with the pure water case (Schreiber and Fersht, 1993). At the same time, the theoretical predictions concerning electrostatic interactions in the barnase–barstar pair are inconsistent: while a number of papers predict favorable electrostatic contributions to binding affinity (Lee and Tidor, 2001a; Dong et al., 2003; Wang et al., 2004; Spaar et al., 2006), unfovarable (Lee and Tidor, 2001b) or net zero contributions (Schneierman and Honig, 2002) were also calculated. Furthermore, an analysis of the published computational results reveals their strong dependence on the details of the model exploited hence new researches in the field are needed.

We believe that the methods of single molecule force spectroscopy are quite appropriate just for the case: electrostatic interaction is characterized by a relatively long spatial scale very different from that of the short range “contact” hydrogen bonds-mediated interprotein interactions. Force spectroscopy, which is known to be sensitive to the width of the energy barrier, looks like a natural tool to distinguish between them. Experimental results obtained confirm our anticipations and this, we believe, opens new perspectives for a systematic use of the force spectroscopy methods to clarify the role of different contributions to the total complex energy.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES

Barnase and barstar

Barnase and barstar were prepared as follows. Barnase was produced in TG-1 E. coli cells freshly transformed by the pPBn plasmid (Schulga et al., 1998). Starter culture was grown overnight in YTPSG medium (1% yeast extract, 1% tryptone, 100 mM NaCl, 45 mM K2HPO4, 5 mM KH2PO4, 0.5% glucose, pH 7.4), 100 mg/L ampicillin at 28°C. The same medium (1 L) was inoculated with 5 ml of the overnight culture. The bacteria were grown at 28°C until the cell culture reaches OD550 = 0.05 and then during additional 12–15 h at 37°C. All the following operations were performed at 4°C. Glacial acetic acid was added to the bacterial culture at 50 ml/L. The suspension was stirred for 30 min. The cells were removed by centrifugation at 7000g for 10 min. Phosphocellulose P11 in 20 mM Na-acetate buffer, pH 4.5 was added to the supernatant and the suspension was stirred for 3–4 h. Phosphocellulose was transferred into a column washed with the same buffer, and barnase was eluted with a linear gradient of 20–800 mM NH4-acetate buffer, pH8.0. Final purification of the barnase was performed on the column HiTrap 1 ml SP FF, equilibrated with 20 mM Na-acetate buffer, pH 5.5. Barnase was eluted with a linear salt gradient of 0–400 mM NaCl in 20 mM Na-acetate buffer, pH 5.5.

Barstar (C40,82A) was produced in HB101 E. coli cells freshly transformed the pMT643 plasmid (Jucovic and Hartley, 1996) and isolated according to modified procedure (Schreiber and Fersht, 1993). Starter culture was grown overnight in YTPSG medium (1% yeast extract, 1% tryptone, 100 mM NaCl, 45 mM K2HPO4, 5 mM KH2PO4, 0.5% glucose, pH 7.4), 100 mg/L ampicillin at 37°C. The overnight culture (5 ml) was added to 1 L YTPSM medium with the antibiotic and after 12 h cultivation at 37°C the bacterial cells were pelleted by centrifugation. All the following operations were performed at 4°C. Resuspended cell pellet in 30 ml ice-cold 50 mM Tris-HCl buffer, 100 mM NaCl, 10 mM EDTA, 10 mM DTT, pH8.0 was sonicated for 3–4 min on ice. The pellet was removed by centrifugation at 30 000g for 30 min. The supernatant was saturated with ammonium sulfate to 40% and centrifugated (30 000g, 30 min). The second supernatant was saturated with ammonium sulfate to 70–75% and again centrifuged for 30 min. The pellet was dissolved in 3–5 ml of 30 mM Tris-HCl buffer, 50 mM NaCl, 2 mM EDTA, 2 mM DTT, pH8.0. Proteins were fractionated by gel-filtration on Sephadex G100 SuperFine (16/100) column. Fractions containing barstar were collected and final purification was done on column HiTrap1ml Q Seph FF. Barstar was eluted with a linear salt gradient of 50–500 mM NaCl in Tris-HCl buffer, 2 mM EDTA, 2 mM DTT, pH8.0.

Proteins purity was verified by SDS-PAGE analysis.

AFM instrumentation

The same experimental setup and the procedure of tip and sample functionalization, which have been used for the single molecule force spectroscopy earlier and described in details in our papers (Chtcheglova et al., 2004; Favre et al., 2007), were exploited for this study which enables us to give here only a rather brief account. The data were obtained in Lausanne using the AFM Nanoscope IV “Picoforce”, Veeco Instruments, Santa Barbara, CA, USA. All experiments were performed using V-shaped Si3N4 (silicon nitride) cantilevers (Veeco) having a length of 200 µm and a nominal spring constant of 0.06 N/m. The spring constant of each cantilever was calibrated just prior to the measurements using the built-in calibration procedure of Nanoscope IV Picoforce AFM.

Covalent attachment of proteins onto the tip and sample surfaces without special linkers was used. (Currently we are repeating this and some other force spectroscopy experiments using benzaldehyde-PEG6-NHS linkers (Ebner et al., 2008), these results will be published elsewhere). Very briefly, the procedure of tip/sample functionalization was as follows. One percentage solution of glutaraldehyde (Sigma) was used as a coupling agent when functionalizing tips with barnase with the concentration of 250 mg/L after their initial intensive cleansing. The loosely attached proteins were then removed by extensive washing with PBS buffer. The protein-functionalized tips were used immediately for making measurements. Substrates (freshly cleaved muscovite mica) were functionalized for 5 min, followed by processing in a 1% v/v 3-aminopropyltriethoxysilane (APTES, Sigma) solution in water. This procedure was followed by processing the substrate in a 1% v/v glutaraldehyde solution in water for 15 min. After rinsing with deionized ultra high quality (UHQ) (resistivity 18 MΩ·cm) water, the samples were immersed into a solution of barstar with the concentration of 250 mg/L for 15 min. The non-reacted and loosely bonded proteins were subsequently removed by extensive washing with PBS buffer.

AFM images of mica surface functionalized in the aforementioned way demonstrate the formation of smooth protein monolayers (single molecules are seen as globules) with an average thickness of a few nanometers; for reference see e.g., Figure 2 from the paper (Chtcheglova et al., 2004). Our previous single molecule force spectroscopy studies as well as literary data attest that for the surface functionalization method and protein concentrations used, mostly single- and double-molecule interactions are observed. In particular, and this was checked up again in these series of experiments, the lowering of concentration does not lead to the appearance of peaks with lower specific force in force histograms but strongly decreases the percentage of the successful approach—contract cycles, that is such cycles where specific interaction events are recorded. For the concentrations used, the percentage of successful cycles is 20–25% for the 1000–1500 total number of cycles performed with the same cantilever/sample pair.

Typical pulling off (contraction) curve characteristic for the single molecule barnase–barstar interaction (“specific interaction event”) is presented in Figure 1. Fuzzy logic-based software by Kasas et al., 2000, was used to process such a data semiautomatically. This software assigns a grade ranging from 0 to 1 to each measurement, which could be interpreted as a “specific interaction event”. The main criteria to give such a grade are the following. One criterion is the “quasiverticality” of the signal jump associated with the bond rupture (which takes place at Z = 120 nm in Figure 1). Relaxation of the cantilever position after such a rupture is a very fast process, which, for the determined by the given force loading rate scale of Figure 1 and similar graphs, necessitates such a quasiverticality. After the jump, force curve is to be seen as a horizontal straight line and thus should constitute approximately a right angle with the quasivertical “jumping part” of the curve. The position of this same straight line determines the zero force level, and the value of the signal change at the moment of the rupture is taken as a value of the specific interaction force. Further, the part of the force curve just before the rupture (which corresponds to Z position between 110 and 120 nm in Figure 1) should be seen as an approximately straight line without peculiarities. For each measurement, the software determines an average slope of this part of the curve (initially in Newton par nanometer), which then can be easily recalculated into Newton per second using the known value of the pre-established Z-scanning rate (in nm/s) amply determined by the full amplitude of the Z-displacement of the piezo and the force ramp repetition rate. This same slope is then taken as a physically sound force loading rate, which should not be mixed up with still often used nominal loading rate formally determined from the Z-scanning rate and cantilever spring constant.

thumbnail image

Figure 1. Typical force curve of the specific interaction between barnase and barstar. The first minimum of the force curve (around Z = 100 nm) corresponds to the non-specific adhesion and hence it is disregarded by our data processing software (Kasas et al., 2000); the second minimum (Z = 120 nm) is a specific interaction. The approximation of the “close to the bond rupture” part of the force curve by a straight line whose slope is used to determine an actual value of the force loading rate is shown.

Download figure to PowerPoint

An example of the force histogram is presented in Figure 2. To be sure that we measured specific interactions, different control experiments have been performed at the same conditions. As usual, first of all this includes the study of the “partial system”, i.e., the study of an interaction between non-functionalized mica and tip, functionalized mica and non-functionalized tip, non-functionalized mica and functionalized tip. Second, we have searched for specific interactions in the protein pairs barnase–barnase and barnase–avidin. Not less than 500 force curves were analyzed for any case. All of those yielded negative results: only rare (overall less than for the 1% of all approach-contraction cycles) events mimicking the specific interaction was observed for any control experiment. This is an unavoidable noise, and such a situation is quite common for the field. Similar decrease of the number of observations of specific interaction events was observed when free barnase or barstar with the concentration of 250 mg/L was added to solution to block the possibility of receptor-ligand observation during the approach-contraction cycles.

thumbnail image

Figure 2. Force histogram corresponding to the force loading rate of equation image.

Download figure to PowerPoint

We do not see the necessity to present all this huge and uninformative data massive in the paper, hence for an illustration, in Figure 3 we reproduce only the results of the testing of the following partial system (more than one thousand of approach-contraction curves were recorded): AFM tips functionalized with barstar versus samples without any protein deposited but treated with APTES exactly in the same fashion as this is done for other experiments.

thumbnail image

Figure 3. Force histogram of the control experiment performed with the following partial system: AFM tips functionalized with barnase versus samples without any protein deposited but treated with APTES exactly in the same fashion as this is done for other experiments. Force loading rate ranges equation image.

Download figure to PowerPoint

EXPERIMENTAL RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES

The main results of our study are presented in Figures 2–4. In Figure 4 we present the dependence of the specific interaction force F* on the force-loading rate equation image (in N/s) obtained for experiments performed in a PBS buffer solution (50 mM phosphate, 150 mM NaCl, pH 7.4 at 25°C). Each point on this graph is an average of no less than 250 specific interaction events. As an example, in Figure 2 we present a force histogram corresponding to the force loading rate ranging equation image.

thumbnail image

Figure 4. Dependence of the specific interaction between barnase and barstar on the force loading range. Black quadrants: experiments performed in PBS buffer solution (50 mM phosphate, 150 mM NaCl, pH 7.4 at 25°C), red circles: experiments performed in PBS buffer solution with additional NaCl (50 mM phosphate, 650 mM NaCl). Straight line shows the linear fit to the 150 mM NaCl data. Typical experimental errors of the specific force for both experiments are shown. The indicated value of the loading rate is just the central value of the interval of the loading rates used for the averaging which in typical case is equal to 2 nN/s.

Download figure to PowerPoint

Experimental data is well approximated by a single straight line equation image with equation image, equation image. According to the Bell theory of the dependence of the specific interaction force (i.e., the most probable single bond rupture force) on the force-loading rate, such a dependence is given by the relation (Bell, 1978; Grubmueller et al., 1996; Evans and Ritchie, 1997; Schwesinger et al., 2000; Zlatanova et al., 2000; Friedsam et al., 2003; Weisel et al., 2003; Lee et al., 2007)

  • equation image(1)

Here kB is a Boltzmann constant, T is a temperature, equation image is a barrier thickness and equation image is the natural off-rate, i.e., the dissociation rate where no external force is applied on the single molecule complex.

When applying this relation, one should have in mind that, first, it is derived for the condition that the loading rate equation image is independent on the force F, which is not exactly true (the part of the force curve just before the rupture, as for example the part corresponding to Z position between 40 and 70 nm in Figure 1, is only an approximately straight line). This relation also should be modified if more than one potential well (barrier) contribute to the force-induced complex dissociation (see e.g., (Dettmann et al., 2000; Strunz et al., 2000; Derenyi et al., 2004) for some generalizations in this direction).

The values equation image and equation image follow from the approximation of the data given before. To analyze the meaning of this data, first it looks instructive to compare them with those obtained for other receptor–ligand pairs. Although the most single molecule force spectroscopy experiments reveal the value of the barrier thickness equation image ranging 0.2–1 nm, cf. (Zlatanova et al., 2000; Schwesinger et al., 2000; Weisel et al., 2003; Lee et al., 2007), an observation of such narrow barriers as 0.05–0.15 nm is not uncommon in the field. In particular, they are characteristic for the most efficiently interacting streptavidin–biotin and avidin–biotin pairs, as this was reported, for different loading rates including the loading rate equation image (single complexes were kept stretched by a constant force (Favre et al., 2007)), in (Yuan et al., 2000; Lo et al., 2001; Favre et al., 2007). For example, as narrow barrier value as 0.024 nm was measured for avidin–biotin interaction for the force loading rate ranging 1.7–9.6 nN/s in de Odrowaz Piramowicz et al., 2006. Similar barrier thicknesses were reported for some other ligand–receptor pairs as well, see e.g., Table 2 in the review Lee et al., 2007.

At the first glance, the measured off rate, equation image, appears to be too rapid for such efficiently interacting system as barnase–barstar: remind, that biochemical data attests for the same temperature and similar pH and ionic strength the off rate as slow as equation image (Schreiber and Fersht, 1993). By applying the Kramers theory of chemical bond dissociation in liquids equation image with the typical pre-exponential factor equation image (Bell, 1978; Grubmueller et al., 1996; Evans and Ritchie, 1997; Izrailev et al., 1997), the latter value corresponds to the value of the dissociation barrier equation image equal to equation image. In the same approximation our data gives only equation image as the barrier value. Similar discrepancies with the biochemical data are not surprising and actually they are quite common for the single molecule force spectroscopy method: numerous examples can be found in reviews (Zlatanova et al., 2000; Schwesinger et al., 2000; Weisel et al., 2003; Lee et al., 2007) and other papers. Note, for instance, that comparably large off rates, ranging 1–30 s−1, have been earlier measured (again, for different force loading rates) by the force spectroscopy method even for the streptavidin/avidin–biotin pairs (de Odrowaz Piramowicz et al., 2006). The reason of this discrepancy is, of course, the aforementioned insufficiency of the oversimplified single well (single barrier) model of the interprotein interactions which constitutes the basis for the derivation of Equation (1). This situation is illustrated by the model two-barrier energy landscape of Figure 5: if the whole landscape is composed by narrower internal and wider external wells, as it seems is often the case, already for rather small stretching forces the whole barrier ceases to contribute to the measured values, only the narrower internal barrier persists thus providing the experimental force spectroscopy data. Indeed, really slow force loading rates of the order of a few (tens) picoNewton per second are required to measure the whole “wide” barrier by the single molecule force spectroscopy method, cf. (Dettmann et al., 2000; Strunz et al., 2000; Derenyi et al., 2004).

thumbnail image

Figure 5. A graph illustrating the plausible energy landscape for the barnase–barstar interaction in absence (A) and presence of the applied force F = 35 pN (B). Experimental values for the internal barrier width and height taken from the present study, total barrier height from Schreiber and Fersht (1993) and electrostatic barrier width of 3 nm from Spaar et al. (2006) are used when preparing this graph. It is seen that already under the action of such a rather small force, the broad electrostatic interaction-related external well is not felt by a force spectroscopy method while the narrow internal barrier is only slightly modified.

Download figure to PowerPoint

In the Introduction we have already discuss the inconsistency of the data pertaining to the contribution of the electrostatic interaction to the energy landscape of the barnase–barstar pair. This inconsistency makes difficult the detailed comparison of our experimental data with the theory, and for the current analysis we are obliged to limit ourselves with the conclusions that the total contribution of an electrostatic interaction is favorable and equal to a few kcal/mol (experimental data), and that the corresponding characteristic barrier width is a few nanometers large (for the latter the value of 3 nm has been derived in (Spaar et al., 2006), see especially Figure 6 and Table 1 there). This means that as small stretching force as ∼ 30 pN already smears out an electrostatic interaction-related broad external barrier (cf. Figure 5). Not surprisingly that the latter cannot be seen in the force spectroscopy experiments performed with the force loading range lying in the nN/s range.

To check up this conclusion we performed numerous force spectroscopy experiments in such a manner that all experimental conditions apart from the salt concentration are the same. The salt concentration changes from zero (deionized ultra high quality (resistivity 18 MΩ·cm) water) to 700 mM (necessary quantity of salt was added to our PBS buffer solution). These experiments gave negative results: no statistically sound dependence of the force data on the salt concentration was observed. For example, for the 700 mM salt concentration the processing of the experimental data result in the values which are only ca. 10% different (this is less than an experimental error which is estimated as approximately 20%) from those reported above for the 150 mM salt concentration: equation image and equation image.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES

Single force spectroscopy data presented in this paper attest an efficient and highly specific interaction of the ribonuclease barnase with its inhibitor barstar and reveals an existence of narrow (with an effective width of 0.12 nm) and sufficiently high (equation image) “internal” well/barrier in their interaction energy landscape. Together with the “external” potential well given by the relatively long range electrostatic interaction, which can be estimated as ∼equation image, these observations quite reasonably explain the overall dissociation barrier of the order of equation image inferred from experimental biochemical data.

No dependence of the specific interaction on the salt concentration was observed in our experiments. This is consistent with the relatively long range nature of the electrostatic interaction (it is only the latter which is influenced by the salt concentration via the changes of the charge screening range): such an interaction simply does not affect the data obtained in the force spectroscopy experiment performed at the typical conditions. Hence our results clearly demonstrate the possibility of the use of the force spectroscopy study to separate the short range “contact” and long range electrostatic interprotein interactions in a systematic way.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES

The authors thank G. A. Nalgranyan for the help during the beginning of the experiments. The financial support of Swiss National Science Foundation (grant no. 200020-119993), Russian Foundation for Basic Research (no. 09-04-01201-a) and Russian Federal Agency for Science and Innovation are gratefully acknowledged.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. EXPERIMENTAL RESULTS AND DISCUSSION
  6. CONCLUSION
  7. Acknowledgements
  8. REFERENCES