The role of electrostatics
Under previously reported reaction conditions8
(50-70 mM monovalent salt), restrictocin cleaves the SRL with a second-order rate constant kcat
of (1±0.3) × 10 M-1
(, closed circles) and a Km
of 76 ± 10 μM (Supplementary Fig. 1 online), similar to the published values for α-sarcin13,14
. The relatively slow reaction and weak binding of the SRL complicated characterization of the enzymatic reaction, prompting us to examine the effects of variations in pH, temperature and buffer composition (Supplementary Fig. 1 and Supplementary Table 1 online). The initial reaction pH (7.4) and temperature (37 °C) were already optimal for restrictocin. However, the reaction rate was much greater upon dilution of the monovalent salt concentration. The logarithm of kcat
decreased linearly with a slope (n) of -4.2 ± 0.3 as the logarithm of [KCl] increased (,closed circles). Experiments with other salts gave analogous results (Supplementary Fig. 2 online). At low salt concentration ([KCl] < 5mM), the rate of restrictocin-induced cleavage reached an apparent maximum and became salt independent. Under kcat
conditions ([SRL] » Km
), KCl concentration had no effect on the observed rate (, open circles), suggesting that the strong salt sensitivity of kcat
arises from changes in Km
values. We confirmed this by direct measurement of Km
at different salt concentrations (Supplementary Fig. 1). Under optimal conditions (10 mM Tris-HCl, 0-5 mM KCl, pH 7.0-7.4, 37 °C) restrictocin cleaves the SRL with a kcat
value of 108
, which exceeds published kcat
values for any ribotoxin by 104
To explore the origin of inhibition by KCl, we first tested whether KCl affects the active site of restrictocin. We used the cleavage rate of the dinucleotide GA as a measure of the ‘intrinsic activity’ of this endonuclease. The dinucleotide GA mimics the sequence surrounding the scissile bond of the SRL but lacks the flanking nucleotides. The concentration of KCl had no effect on the cleavage rate of the GA dinucleotide (, squares), suggesting that the catalytic properties of the restrictocin active site remain uniform over the entire range of KCl concentrations. To test whether the inhibition arises from saltinduced rearrangement of the SRL to a weaker binding structure, we determined the salt-rate profile for restrictocin cleavage of an unstructured single-stranded RNA (ssRNA) (Supplementary Fig. 2). KCl inhibited this reaction with the same concentration dependence as for the SRL, precluding a salt-induced change in the SRL structure as the reason for the observed inhibition.
Alternatively, the steep salt dependence of the SRL reaction may reflect competition between the ribotoxin and the salt cations for electrostatic interactions with the RNA, as established for numerous other protein-nucleic acid binding reactions15-18
. The surface of restrictocin is positively charged (pIcalc
9.13) and could interact electrostatically with the negatively charged RNA. These electrostatic interactions are expected to redistribute the ion atmosphere surrounding the SRL and restrictocin upon complex formation, rendering binding dependent on the salt concentration. The salt dependence,
, reflects the number of ion pairs released (a negative slope) from the RNA and protein upon complex formation18,19
. For restrictocin-catalyzed cleavage of the SRL, changes in kcat
reflect changes in Km
() and Km
is equal to the equilibrium dissociation constant Kd
(Supplementary Fig. 3 online). Therefore, we substituted
to obtain a value of n = -4.2 ± 0.3.
To test further the model wherein restrictocin interacts with the SRL electrostatically, we measured the single-turnover kinetic parameters k2 (the maximum rate constant under single-turnover conditions with saturating restrictocin) and K½ (restrictocin concentration required to achieve half-maximal velocity under single-turnover conditions) for a series of RNA oligonucleotides of increasing length (Supplementary Table 2 online). These substrates contained at least one embedded GA site and were cleaved by restrictocin predominantly after purines. At 5 mM KCl, all non-SRL substrates reacted with similarly slow rates (k2 values of (6 ± 4) × 10-4 s-1), but showed improvement in K½ values as substrate length increased (). Beginning with K½ = 0.5 ± 0.2 mM for the pGpA dinucleotide, each additional nucleotidyl group provided 2.3 ± 0.1-fold improvement to K½, or 0.51 ± 0.03 kcal mol-1 to the binding energy. Beyond ~20 nucleotides, this effect attenuated until the binding affinity reached a maximum of ~10 nM for oligonucleotides containing 25 residues or more. Such substrates bind restrictocin with the same affinity as the SRL, indicating that substrate length, not sequence or structure, dictates the stability of the binary complex and suggesting that binding occurs via nonspecific electrostatic interactions.
To determine whether the role of electrostatics reflects a conserved feature of ribotoxins, we examined the salt-rate dependence of the SRL cleavage by the restrictocin homolog α-sarcin. α-sarcin cleaved the SRL with a similarly steep salt dependence (n = -4.8 ± 0.4; Supplementary Fig. 2) and showed the same considerable improvement in kinetic parameters at low salt concentrations (kcat/Km =(1 ± 0.4) ×107 M-1 s-1 and Km = 8 ± 2 nM at 5 mM KCl), due to better binding as described above for restrictocin. All sequenced members of the α-sarcin family have calculated isoelectric points in the basic range, suggesting that our observations have functional implications for this entire family.
Ribosomes as targets
As described above for the SRL, the ribotoxin loses 104-to 105-fold in catalytic power as the salt concentration increases from 5 mM to physiological levels of 100-150 mM (, closed circles). To test whether placing the SRL within the natural cellular target, the ribosome, restores this lost catalytic potential, we conducted experiments using ribosomes from rat liver as a substrate. These ribosomes sedimented as a mixture of 80S particles and polysomes over a range of MgCl2 (0.1-2 mM) and KCl (10 mM-150 mM) concentrations, according to analytical ultracentrifugation (AUC) analysis (data not shown; see Methods). At 50 mM KCl and 1 mM MgCl2, restrictocin cleaves the SRL within ribosomes with a kcat/Km of (3 ± 1)× 107 M-1 s-1, more than 1,000-fold faster than it cleaves the SRL oligonucleotide under the same conditions (). The increase in rate arises from a corresponding improvement in ribotoxin binding: ribosomes react with kcat = 1.1 ± 0.5 s-1 (, plateau) and Km = 30 ± 10 nM (data not shown), whereas the SRL reacts with kcat = 1.4 ± 0.4 s-1 () and Km ~100 μM (data not shown). Disruption of the ribosomes by phenol extraction of ribosomal proteins decreased the rate by 104-fold (Supplementary Fig. 4 online and data not shown), supporting the argument that perturbing ribosomal proteins, ribosomal structure or both contributes to this loss. These results indicate that the ribosomal context strongly enhances binding of the ribotoxin.
Figure 2 Ribosomes as substrates for restrictocin. (a) Salt-rate profiles for restrictocin cleavage of ribosomes and the SRL in the presence of 1 mM MgCl2. Solid lines were obtained by fitting the data to (more ...)
To assess whether electrostatic interactions impart the additional energy for binding between restrictocin and the ribosome, we determined the salt dependence for restrictocin-mediated ribosome cleavage under subsaturating conditions. Because the ribosome-cleavage buffer contained 1 mM MgCl2
to stabilize 80S particles, we also determined the KCl profile for the SRL reaction in the presence of 1 mM MgCl2
to allow direct comparison between the ribosome and the SRL oligonucleotide. In the presence of 1 mM MgCl2
, KCl still strongly inhibited the SRL cleavage reaction (, open circles), but as expected for mixed-salt buffers20,21
, the maximum cleavage rate (kobs
= (5 ± 2) × 105
, where kobs
is the observed rate constant and E0
is the total enzyme concentration) and slope (n = -3.0 ± 0.2) decreased relative to the profile obtained in the absence of magnesium (kobs
= (1 ± 0.2) × 108
; n =-4.2 ± 0.3).
The ribosome reacts with a steeper dependence on KCl concentration than does the SRL oligonucleotide (n = -6.0 ± 0.4; , colored circles). A sharp transition occurs near 130 mM KCl, possibly reflecting dissociation of the 80S species to give 60S and 40S subunits, as suggested by AUC (data not shown). The slope of -6 for the KCl dependence suggests that more electrostatic contacts form upon restrictocin binding to the ribosome than upon binding to the SRL RNA. These results support the argument that additional electrostatic interactions stabilize the restrictocin-ribosome complex relative to the restrictocin-SRL complex. To illustrate the apparent catalytic advantage for ribosome cleavage versus SRL oligonucleotide cleavage, we plotted kobs/E0 values for cleavage of ribosomes relative to SRL over a range of KCl concentrations (). Between 50 and 140 mM KCl, restrictocin cleaves ribosomes ~103-fold faster than SRL oligonucleotides, and this effect is entirely due to tighter binding, as the ribosome and SRL react with the same kcat values (1 s-1).
The ribosome concentration in vivo
exceeds the Km
for restrictocin-catalyzed cleavage by 30-to 300-fold (Km
~ 30 nM; data not shown). We therefore determined the salt-rate profiles at various ribosome concentrations, including those greatly exceeding the Km
values. As the ribosome concentration increases, the salt-rate profiles shift further along the KCl axis (). This shift occurs because reactions containing elevated ribosome concentrations require higher salt concentrations to induce subsaturating conditions. The high in vivo
concentration of ribosomes is therefore expected to offset salt inhibition at 100-150 mM KCl, allowing restrictocin to operate near its catalytic optimum inside cells (, yellow box).
Distal basic residues contribute to ribosome targeting
To test directly whether the stronger binding of restrictocin to the ribosome results from electrostatic interactions distinct from those at the SRL interface, we constructed charge reversal mutations (lysine and arginine to aspartate) of three basic residues (Arg21, Lys28 and Lys63) that lie outside the restrictocin-SRL interface in the crystallographically determined structure9
(). Residue 63 lies near this interface, whereas residues 21 and 28 reside on the opposite face. We investigated three restrictocin variants for comparison to the wild-type protein: the single mutant R63D, the double mutant R21D K28D and the triple mutant R21D K28D R63D (designated as 3/D). The three mutants cleave the SRL oligonucleotide with nearly the same rate as does wild-type restrictocin (), indicating that the mutated residues make little energetic contribution to SRL binding. The triple mutant 3/D gives the same salt-rate profile for SRL cleavage as the wild-type ribotoxin, within error (n = -3.9 ± 0.1; Supplementary Fig. 5 online). These findings suggest that the mutated residues do not contribute to electrostatic interactions between the SRL and restrictocin, in accord with the cocrystal structure, where none of the three mutated residues contact the SRL9
Figure 3 Mutation of cationic surface residues of restrictocin mitigates the ribosome advantage over SRL. (a) Residues targeted for mutation are highlighted on the model of restrictocin docked onto the ribosome (see (more ...)
In contrast to SRL cleavage, the mutant ribotoxins cleave ribosomes with markedly decreased rates compared to wild-type ribotoxin (). The single charge reversal on the front surface of the ribotoxin (K63D) reduced kcat/Km by a factor of 100 ± 30; the double charge reversal on the back surface (R21D K28D) reduced kcat/Km by a factor of 30 ± 3. The simultaneous reversal of the three charges (triple mutant 3/D) reduced kcat/Km by 2,000 ± 400-fold, such that ribosome cleavage and SRL oligonucleotide cleavage occurred with similar rates (). As expected, the 2,000-fold reduction in kcat/Km for the triple mutant 3/D results from weakened binding (Km for the 3/D mutant increased by greater than 700-fold; Supplementary Fig. 5). The triple mutation also attenuates the slope of the salt-rate profile for ribosome cleavage (n = -2.2 versus n = -6 for wild-type protein; Supplementary Fig. 5). These basic surface residues of restrictocin, which have little or no effect on cleavage of the SRL oligonucleotide, contribute to the ribotoxin’s ability to bind and cleave ribosomes, supporting a role for electrostatics in proteinribosome recognition.
Maximal rate of ribosome inactivation by restrictocin
At low salt concentration, the binding affinity of restrictocin for the ribosome approaches the concentration of ribosomes in the reaction (10 nM). Under these condition, the salt-independent parameter kcat (rather than kcat/Km) governs the reaction kinetics, masking the lowsalt behavior of kcat/Km. To explore kcat/Km under low-salt conditions, we used subsaturating ribosome concentrations ( « 10 nM) throughout. As the ethidium staining protocol cannot detect ribosomes at such low concentrations, we developed a more sensitive assay that detects formation of the α-fragment (the 3′ product of SRL cleavage in 28S rRNA) by hybridization of a radiolabeled [32P]DNA probe complementary to nucleotides 4371-4399 of rat 28S rRNA ().
Figure 4 Monitoring restrictocin activity at dilute ribosome concentrations. (a) Detection of α-fragment cleavage product by hybridization of rRNA with 32P-radiolabeled DNA probe (see (more ...)
Using this alternative assay (), we obtained the salt-rate profile for restrictocin reactions containing 10 pM ribosomes (). In the high salt regime (>40 mM KCl), the new profile overlaps with that obtained from reactions containing 10 nM ribosomes, confirming the validity of the assay. However, because reactions at 10 pM ribosomes remain under kcat
conditions throughout the titration, the profiles diverge as the salt concentration decreases, revealing a new apparent maximum for kcat
of (2 ± 1) × 109
(at 20 mM KCl and 1 mM MgCl2
). Further improvement in kcat
occurs upon dilution of MgCl2
(, inset). Analogous improvement happens in reactions with the SRL oligonucleotide: the salt-rate plateau in the absence of magnesium (, filled circles) increases by ~100-fold compared to that in the presence of 1 mM MgCl2
(, open circles). At 0.1 mM MgCl2
(10 mM KCl), restrictocin induces formation of the α-fragment with a second-order rate constant of (1.7 ± 0.2) × 1010
, one of largest kcat
values ever reported for an enzymatic reaction12
(; see Discussion). Independent AUC experiments confirmed that, under these ionic conditions, the ribosomes remain in the native 80S configuration (data not shown). No additional increase in kcat
occurs upon further dilution of MgCl2
or ribosomes, suggesting that the value of (1.7 ± 0.2)× 1010
probably represents the physical limit imposed on the reaction rate by diffusion.
Ribosomes bind many restrictocin molecules at once
Linear Poisson-Boltzmann calculations of the electrostatic surface potential, for both the ribosome and the ribosomal subunits from archaeal sources24,25
, assign negative potential to much of the ribosomal surface, including the region containing the SRL, with relatively few positively charged areas. In the crystal structure of the 50S ribosomal subunit, 23S rRNA and five proteins (L3, L6, L13, L14 and L24e) comprise the surface near the SRL (). Like most ribosomal proteins26
, these proteins have bipolar character, with the majority of basic residues buried inside the ribosome and acidic residues exposed to solvent. Many of these acidic residues are conserved among bacterial, archaeal and eukaryotic ribosomes (sequence alignments of ribosomal proteins across the species not shown; bacteria contain no counterpart to L24e), implying that the electrostatic properties of the ribosomal surface may be important for function.
Figure 5 Side view of the 2.4Å crystal structure of the large ribosomal subunit41,42. Ribosomal proteins L3, L6, L13, L14 and L24e are colored by electrostatic potential (see Methods: red, negative; blue, (more ...)
To test this view of the ribosomal surface potential, we assessed whether restrictocin binds multiple sites on the ribosome. We determined the stoichiometry of binding by monitoring restrictocincatalyzed cleavage of the radiolabeled SRL oligonucleotide in the presence of ribosomes. Reaction of the SRL oligonucleotide under subsaturating (kcat/Km) conditions provides a sensitive measure of the free restrictocin concentration (, open circles). Restrictocin binding by the ribosome decreases the concentration of free ribotoxin, thereby reducing the rate of SRL oligonucleotide cleavage. At low restrictocin concentrations ( <5 nM), 10 nM ribosomes inhibited the rate of SRL oligonucleotide cleavage by 100-to 200-fold, suggesting that the ribosomes bind greater than 99% of the restrictocin molecules. The reaction shows a modest dependence on restrictocin until the ribotoxin concentration approaches 400 nM, where the dependence acquires a strong linear sensitivity (, filled circles). We obtained an analogous profile with 5 nM ribosomes, except that reactions acquired strong linear sensitivity at lower restrictocin concentrations (about 200 nM; , diamonds). These observations suggest that ribosomes sequester free restrictocin until available binding sites become saturated.
Figure 6 Titration of the restrictocin-binding sites on the ribosomal surface. (a) Restrictocin cleavage of the 32P-radiolabled SRL RNA in the absence and in the presence of ribosomes (Rb). Reactions (more ...)
To verify that the titration profile reflects restrictocin binding to the ribosome, we performed the same experiment using the triple mutant of restrictocin, 3/D. As described above, this mutant cleaves the SRL oligonucleotide with nearly the same rate as wild-type restrictocin but binds ribosomes much more weakly. Ribosomes had essentially no effect on the rate of SRL oligonucleotide cleavage by the 3/D mutant (), demonstrating that the inhibitory effect in the reaction with wild-type restrictocin reflects restrictocin binding to the ribosome.
Binding of one restrictocin to one ribosome cannot account for the data in (dotted line), as the concentration of restrictocin required to restore SRL cleavage greatly exceeds the concentration of ribosomes. We analyzed the data according to a model in which the ribosome contains k
restrictocin-binding sites that competitively inhibit restrictocin-catalyzed cleavage of the SRL oligonucleotide. Fitting the titration profiles in to this model (see Methods) gives k = 49 ± 3. Data from analogous experiments containing different concentrations of ribosomes, KCl and magnesium (ribosome integrity under all conditions was checked by AUC) quantitatively supports this conclusion (see Methods). The large number of binding sites on the ribosome for restrictocin but not for the triple mutant 3/D suggests that restrictocin binds nonspecifically to much of the ribosomal surface via electrostatic interactions. These observations agree with the global view of the ribosomal surface potential derived from linear Poisson-Boltzmann calculations24,25
and have implications for ribotoxin function.