Nobody’s Perfect: Choice of the Buffer and the Rate of Cu2+ Ion–Peptide Interaction

The choice of correct pH buffer is crucial in chemical studies modeling biological processes involving Cu2+ ions. Popular buffers for physiological pH are known to form Cu(II) complexes, but their impact on kinetics of Cu(II) complexation has not been considered. We performed a stopped-flow kinetic study of Cu2+ ion interactions with four popular buffers (phosphate, Tris, HEPES, and MOPS) and two buffers considered as nonbinding (MES and PIPPS). Next, we studied their effects on the rate of Cu2+ reaction with Gly-Gly-His (GGH), a tripeptide modeling physiological Cu(II) sites, which we studied previously at conditions presumably excluding the buffer interference [KotuniakR.; Angew. Chem., Int. Ed.2020, 59, 11234–11239]. We observed that (i) all tested pH 7.4 buffers formed Cu(II) complexes within the stopped-flow instrument dead time; (ii) Cu(II)-peptide complexes were formed via ternary complexes with the buffers; (iii) nevertheless, Good buffers affected the observed rate of Cu(II)-GGH complex formation only slightly; (iv) Tris was a competitive inhibitor of Cu(II)-GGH complexation; while (v) phosphate was a reaction catalyst. This is particularly important as phosphate is a biological buffer.


■ INTRODUCTION
Biological processes, including those involving Cu 2+ ions, take place in pH-buffered environments necessary for the smooth running of (bio)chemical reactions.Unfortunately, physiological buffers cannot be fully employed in quantitative laboratory studies aimed to reproduce and understand such reactions because they introduce too much physical and chemical interference to the studied systems.For example, the blood serum buffer is composed of bicarbonate/carbonic acid, phosphates, and serum albumin protein. 1 The bicarbonate buffer is instable in vitro, and serum albumin forms very strong Cu(II) complexes. 2Therefore, it is necessary to use an alternative "chemical" buffer.This, however, is not an easy task.Known buffers with a satisfactory capacity in the physiological pH range, roughly 7.0−7.7,−7 According to a comprehensive review by Ferreira et al., 8 most of the buffers listed as "noncoordinating" in 1966 by Good and co-workers, 9 and updated by Rorabacher and co-workers in 1997, 10 actually form Cu(II) complexes.Summarizing the data reviewed by Ferreira et al., only MES (2-(N-morpholino)ethanesulfonic acid) is noncoordinating in its buffering range around pH 6.A few buffers cited by Ferreira et al. have not yet been rigorously studied regarding Cu(II) complexation.For example, PIPPS ((1,4-piperazinedipropanesulfonic acid)), which has a pK a of 7.97 and provides effective buffering in the pH range of 7−9, 11 was used in a study of dissolution of cuprous minerals, with no coordination-based effects reported. 12arallel interactions of buffers with Cu 2+ ions are known to affect the primary reaction equilibria with the ligands of interest.If not taken into account, these interactions may result in significant errors of stability constant determinations 13 and can even change the course of reaction by forming ternary complexes. 14Very little is known, however, about the effects that buffers may exert on the kinetics of such reactions.
The ATCUN/NTS motifs consist of N-terminal tripeptide sequences, characterized by the free N-terminus, a His residue in position three and any intervening amino acid, except Pro. 15,16These motifs bind Cu(II) ions spontaneously in a broad pH range, yielding square-planar four-nitrogen (4N) complexes characterized with high stability (log C K = 12−15 at pH 7.4).They are also inert in reactions of Cu(II) ion dissociation/exchange. 17The ATCUN/NTS motifs are present in many extra-and intracellular human proteins, but in most cases, their biological function remains to be established. 18Notably, these motifs are present in human serum albumin (Asp-Ala-His N-terminal sequence) and human Ctr1 cell membrane copper transporter (Met-Asp-His Nterminal sequence), two proteins involved in copper transport. 13ATCUN/NTS peptides have been massively used in chemical studies modeling the chemical reactivity of their parent proteins with respect to Cu(II) ions. 16ecently, we and our collaborators used fast kinetic methods (stopped-flow and freeze-quench) coupled with EPR and ultraviolet−visible (UV−vis) spectroscopies to elucidate the molecular mechanism of Cu 2+ complexation by Gly-Gly-His (GGH), the simplest ATCUN/NTS peptide model. 19,20In a separate study, this Cu(II) binding kinetics was also studied for the N-terminal peptide of Alzheimer's Disease-associated peptide Aβ 4−16 . 21The reaction consists of three major steps, presented in Scheme 1, which are characterized by distinct coordination environments for the Cu(II) ion.In the initial one, completed within 100 μs, the Cu 2+ aqua ion forms a single bond with the peptide nitrogen, with a preference for imidazole ring atoms (1N complex, Early Complex (EC) in Scheme 1).The EC was documented by EPR spectra of freezequenched samples.The next species (2N complex, IC in Scheme 1) is formed within 1 ms and contains a macrochelate loop complemented by the other terminal nitrogen donor (Gly amine if the initial anchoring was at the imidazole).This species is more stable and was directly detected in stoppedflow experiments.It eventually undergoes a rearrangement into a thermodynamically stable 4N complex (t 1/2 ∼ 100 ms) in which the equatorial Cu(II) coordination sphere is fully saturated in a square-planar structure.Additional interactions in the first and second coordination sphere of the IC enabled by amino acid substitutions in positions upstream or directly downstream of the His-3 residue can increase the lifetime of the IC substantially (up to several seconds) but do not affect the overall course of the reaction. 22We recently explored conceptually the impact of this reaction mechanism on the biological copper transport processes, noting the correspondence of lifetimes of the intermediate species with physiological cycles. 23he EC and IC contain the Cu(II) ion bound to just one or two donor atoms (1N and 2N, respectively; Scheme 1).This makes them prone to ternary interactions by exchanging the labile water molecules present at other positions around the Cu(II) ion in D 4h symmetry.Such interaction was observed for HEPES buffer.It affected the rate of Cu 2+ ion binding to the Aβ 4−16 peptide. 21Inspired by this observation, here we applied stopped-flow kinetics to investigate how buffers suitable for pH 7.4 affect the Cu 2+ reaction with GGH.We considered HEPES, Tris, MOPS, and phosphate, which are commonly used in biochemical studies, and PIPPS as a candidate weakly interacting buffer.For the sake of consistency, we also performed experiments with MES, as in the previous GGH study. 19The MES pK a is 6.15, so it could not be used to buffer the pH 7.4. 24tructures and systematic names of the studied buffers are listed in Scheme 2. From the structural point of view, MES and MOPS are morpholine derivatives while HEPES and PIPPS contain the piperazine ring.Tris is the main representative of its own buffer family containing the characteristic primary amine, and phosphate is the main component of physiological buffer, most commonly used as phosphate-buffered saline (PBS), the common buffering medium in biochemistry and cell biology.Reactions of Cu 2+ ions with buffers were performed using 3.2 mM CuCl 2 solution, acidified with HCl at pH 4 to prevent the Cu(OH) 2 precipitation prior to the reaction, and 400 mM solutions of phosphate, HEPES, PIPPS, MOPS, and Tris, pH 7.4, and MES, pH 6.0.The choice of a 125-fold excess of buffer over the Cu 2+ ions was dictated by test experiments, where a significant pH drift of the reaction mixture was noticed at a lower buffer excess.The same buffer excess was used in our previous studies. 19,22The CuCl 2 solutions were mixed with buffer solutions at 1:1 ratios in an SFM-300 diode-array stopped-flow apparatus (BioLogic, Seyssinet-Pariset, France) with the 400−900 nm detection range in a 1 cm quartz cuvette.The dead time of the stopped-flow instrument was 2 ms.The experiments were carried out at 25 °C, with a typical flow rate of 15 mL/min.After each series of measurements, the system was rinsed 3 times with 0.2% HCl and 3 times with water after each run in order to remove traces of Cu(OH) 2 precipitates.The course of reaction was monitored in 2 time windows: 1.5 and 300 s.In both cases, the number of collected spectra was 1000.In short runs, the time interval between the spectra was 1.5 ms (thus providing the first spectrum at 3.5 ms).In long runs, the integration time was increased to 32 ms (first spectrum at 34 ms); the first 500 spectra were recorded with 32 ms intervals and the remaining 500 spectra with 576 ms intervals.

GGH, CuCl
Reactions of Cu 2+ ions with the GGH peptide in the presence of buffers were performed similarly, with 4 mM GGH dissolved in respective 400 mM buffers mixed at a 1:1 volume ratio, using initial Cu 2+ concentrations between 1.4 and 3.6 mM (final concentrations: 2 mM GGH in 200 mM buffer, 0.7−1.8mM Cu 2+ ).The reactions were monitored for 1.5 s with a time resolution of 1.5 ms or for 10 s with a time resolution of 10 ms.
The phosphate concentration dependence experiments were performed using initial concentrations of 1.6 mM Cu 2+ and 2.0 mM GGH in phosphate buffers, pH 7.4, of various concentrations between 50 mM and 1 M (final concentrations: 0.8 mM Cu 2+ and 1 mM GGH in 25−500 mM phosphate buffer).
The pH dependence of reaction was studied for MES, phosphate, HEPES, and MOPS buffers in the pH ranges of 5.5−6.5, 6.5−7.7,6.4−7.7, and 6.4−7.8,respectively, using initial concentrations of 3.2 mM Cu 2+ and 4.0 mM GGH in 400 mM buffer (final concentrations: 1.6 mM Cu 2+ and 2.0 mM GGH in 200 mM buffer).The reactions were monitored for 1.5 s with a time resolution of 1.5 ms.
All reactions were repeated at least 5 times under each experimental condition.The kinetic data processing and the analyses of concentration and pH dependences of rate constants were performed by using Origin2024.The observed rate constants (k obs ) were determined by fitting the absorption values at 525 nm to the monoexponential function, unless indicated otherwise.The UV−vis spectra were smoothed, when necessary, using a Fast Fourier Transform (FFT) smoothing procedure implemented in Origin2024, using a 45-point filter window.

■ RESULTS AND DISCUSSION
Interactions of Cu 2+ Ions with Buffers in the Absence of GGH.Stopped-flow experiments, aiming at investigating early stages of Cu 2+ ions interactions with buffers, were preceded by benchtop tests consisting of manual mixing of appropriate reactants according to the protocol further implemented in stopped-flow studies: 3.2 mM CuCl 2 + 400 mM buffer, equal volumes, pH 6.0 for MES, pH 7.4 for other buffers.For the sake of comparison, two additional reactions of Cu 2+ ions were performed, one with an equal volume of 50 mM NaOH and another with an equal volume of distilled water.The samples were left for 60 min on the bench and checked visually.Tris buffer and pure water were the only ones that did not exhibit formation of Cu(II) precipitates.
Next, these reactions were studied on a stopped-flow instrument equipped with a diode-array detector.The data were collected with two different time intervals, enabling monitoring the reaction in two distinct time windows: from 3.5 ms to 1.5 s and from 34 ms to 5 min.The complete sets of data are provided in Figures S1 and S2, whereas comparisons of initial (at 3.5 ms) and final (separately at 1.5 s and 5 min) absorption spectra are presented in Figure 1.The shorter of these observation periods corresponded to the typical time window of GGH reactions in MES buffer, studied previously. 19he longer one was considered as sufficiently corresponding to macroscopic observation times and still assuring the stability of the instrument baseline.Figure S3 presents the kinetic traces recorded at the d−d band maximum over the 1.5 s observation period.These traces clearly confirm that the binding of buffer molecules to the Cu 2+ ion occurred within the dead time of the measurements.
At 3.5 ms, the spectrum of Cu 2+ /MES, centered at 812 nm, was identical to that of CuCl 2 proving no interaction with buffer molecules under examined conditions (Figure 1A, black and light blue lines).Unlike MES, for all other buffers, a significant increase of the d−d band intensity, accompanied by a blue shift, was observed.Such a clear spectral change vs that of the Cu(H 2 O) 6 2+ ion indicates the formation of Cu(II)/ buffer complexes.The fast rates of these reactions are not surprising.In our previous study of Cu(H 2 O) 6 2+ ion reaction with GGH, we observed the completion of formation of the first peptide complex (EC, Scheme 1) within 100 μs. 19With buffer concentrations 2 orders of magnitude higher than GGH, we can estimate the timespan of this second-order reaction to be within single microseconds.
The parameters of these initial spectra are presented in Table 1 and compared with the available literature values.Those for Cu(H 2 O) 6 2+ (in MES, pH 6.0), the transiently soluble Cu(OH) 2 , the CuHPO 4 species, and the steady state Tris complexes are in agreement with previous publications. 3,6,25,26The HEPES complex is not stable at pH 7.4, slowly decomposing into a hydroxide species, but its coordination mode was extrapolated from pH and spectroscopic titrations. 6The d−d band shapes and absorption maxima for MOPS and PIPPS are very similar to each other and red-shifted by just 10 nm compared to HEPES.These spectra were not published before.They indicate a coordination mode like in HEPES, with one nitrogen atom in the Cu(II) coordination sphere.
Quite remarkably, not only Tris but also several other studied systems remained well-soluble within 1.5 s after mixing the reagents.The solutions of MOPS and MES exhibited no trace of turbidity, HEPES exhibited it just a little, while the hydroxide solution, PIPPS, and phosphate demonstrated a higher absorption at 470 nm (in the ascending order).This wavelength was selected to monitor the formation of insoluble Cu(II) aggregates (Figure 2 aggregates was so strong that the residual absorption band of the initial complex became barely discernible and could not be quantitated.In all other systems, the original absorption bands were clearly seen despite the baseline elevation due to light scattering on these aggregates.Comparing the d−d band parameters in the presence of phosphate with those of the Cu(II) aqua ion, the partially soluble Cu(OH) 2 one can see that the Cu(II) ion is coordinated by phosphate rather than hydroxide ions.HEPES and MOPS systems did not evolve into pure Cu(OH) 2 either but retained the coordinated buffer molecule.The formation of Cu(OH) 2 particles at pH 6.0 in the MES system was only partial, leaving a significant portion of the Cu(H 2 O) 6 2+ ions in solution.The probably most unexpected result of this part of research is the highly differentiated evolution of systems containing HEPES, MOPS, and PIPPS buffers, despite their chemical similarity (Figure 1C).It shows the complicated nature of the observed precipitation processes.For four organic buffers, including MES, the effects of organic molecules on precipitate formation were strikingly differentiated.The PIPPS system evolved similarly to those of hydroxide and phosphate for about 10 s, whereas MES, HEPES, and MOPS systems underwent a rapid rearrangement process at ca. 60−70 s, followed by a more or less steady buildup of turbidity (Figure 2).They, however, evolved differently for the first minute of the process.For example, the Cu(II) ions in MES remained remarkably stable as apparent Cu(H 2 O) 6 2+ species for about a minute before the sudden increase of turbidity.The behavior of the MOPS system was the most peculiar one.Before the event at ca. 70 s,  its turbidity evolved sigmoidally with a significant lag period, of ca. 2 s, during which the solution remained clear.
While the role of subtle differences in buffer molecule structures in Cu(OH) 2 precipitation may pose an interesting research topic, deeper analysis of these phenomena is beyond the scope of this study.However, several important observations could be made as below.
1.All tested buffers suitable for pH 7.4 formed Cu(II) complexes within the dead time of the stopped-flow instrument.2. The presence of phosphate, HEPES, or PIPPS buffers results in the formation of Cu(II) aggregates with no lag period, but with different aggregation rates.The aggregation in HEPES was significantly slower than in the other two buffers.3. The MOPS buffer provided a remarkable time window (ca. 2 s under the tested conditions) in which no aggregates were visible.Therefore, it provides an option for studying fast reaction kinetics of Cu(II) ions at pH 7.4.One ought to remember, however, that the Cu(II) ion remains coordinated to the MOPS in this time window (see Table 1).4. The analogous time window is as long as 60 s for MES at pH 6.0, with Cu(II) ions present as Cu(H 2 O) 6 2+ .Hence, if the reaction can be studied at pH 6.0, then MES should be the buffer of choice.Effects of Buffers on the Interactions of Cu 2+ Ions with GGH.We next studied how the buffers affect the Cu 2+ reaction with GGH.All data are presented in Figures S4 and Figure 3 presents the initial spectra recorded at 3.5 ms of the reactions, while Figure S5 provides pairwise comparisons of the spectra recorded at 3.5 ms in the absence and presence of GGH.As shown in Figure 3, all buffers at pH 7.4 yielded spectra which differed from that of IC (the initial Cu(II)/ GGH spectrum recorded in MES at pH 6.0; 19 see Table 2 for parameters).The reaction in Tris started from the pure Tris complex, while in HEPES, MOPS, PIPPS, and phosphate, the initial spectra were different from both that of IC and those obtained in the absence of GGH.This demonstrates the initial formation of ternary complexes in all cases except Tris.
The initial d−d bands recorded in HEPES, MOPS, and PIPPS were blue-shifted by ca. 25 nm, and their intensities increased by ca.50%, compared to MES (Table 2).This can be interpreted in terms of the reaction mechanism delineated previously. 19Due to the dead time of the stopped-flow system, the first species detected is IC (Scheme 1), a two-coordinate complex that contains a macrochelate loop between the amine and imidazole nitrogen atoms coordinated to Cu(II).This leaves two equatorial Cu(II) coordination sites available for the tertiary amine nitrogen of HEPES, MOPS, or PIPPS.As argued above, buffer molecules probably bind to Cu(II) much faster, in the single microsecond time scale, so the GGH rather binds to preformed Cu(II)-buffer complexes.
The d−d band blue shift observed for HEPES, MOPPS, and PIPPS is in accord with that, however, its extent is less than expected for a peptidic square-planar complex.It should be 50−70 nm rather than the observed 25 nm when quantified using the published formula of Sigel and Martin (see Table S1 in ref 19 for examples of such calculations). 27The less than expected blue shift could be caused by incomplete coordination, with the observed spectrum being a sum of contributions of 2N (IC) and 2N + N (ternary complex) chromophores.This is, however, less likely for 200 mM buffer concentrations and the estimated log K of 3−4 based on data for Cu(II) complexes of HEPES 4 and MOPS. 7Alternatively, the diminished blue shift may suggest a lower than D 4h symmetry of the ternary complex, 28 resulting from the huge sterical hindrance exerted by the bulk of tertiary amine (see Scheme 2).For the phosphate buffer, the initial d−d band was redshifted by 14 nm vs IC.This effect is consistent with a replacement of a water molecule by phosphate in the IC coordination sphere. 29able 3 provides rate constants calculated by monoexponential fitting of the reaction traces at 525 nm, as provided in Figure 4, with the monoexponential function.The experimental and fitted curves shown in Figure S6 were reasonably accurate for all buffers, except Tris, which could be more reliably fitted with a biexponential function.This deviation may stem from the presence of two Cu(II)/Tris complexes, CuTris and CuTris 2 (see Table 2 and the related discussion).Nevertheless, all reactions exhibited isosbestic points, as shown  in Figure S4.This means a single-step buffer molecule replacement by peptide nitrogens in the course of IC-to-4N conversion.Our previous studies performed in the absence of significant buffer interference indicated that the rate-limiting step of this conversion is the acquisition of a rare conformation by the GGH peptide chain.This conformation is suitable for simultaneous insertion of peptide nitrogens into the first coordination sphere of Cu(II). 19,22Reaction rates in HEPES, MOPS, and PIPPS were similar to each other, and somewhat faster than that in MES.This effect should be considered as apparent because the 4N complex formation is altogether strongly pH-dependent due to high intrinsic pK a of peptide nitrogens, 15,27 and the MES reaction was performed at pH 6.0 rather than 7.4.This issue is treated in more detail below.The presence of phosphate accelerated the 4N complex formation 5-fold, compared to HEPES, MOPS, and PIPPS, with ca. 5% of reaction product being formed already within the instrument dead time (a shoulder at 525 nm in the spectrum in Figure 3).In contrast, the reaction in Tris was more than 10-fold slower compared with other nitrogen buffers.At pH 7.4, Tris forms strong bis-complexes with Cu(II) ions, saturating its coordination sphere. 6Hence, one can propose that only the mono-complex, a minor species under the present conditions, is mechanistically susceptible to the GGH assault.This creates a bottleneck in the reaction of Tris with the Cu/GGH complex.
Therefore, the order of events in buffered GGH reactions under conditions applied in this work is as follows: the Cu(II) ion is first intercepted by a buffer molecule, forming a 1:1 complex (except for Tris), which then reacts with GGH to form a ternary complex involving a peptidic macrochelate.Both these events take part within the dead time of the instrument but can be reconstructed on the basis of the published freeze-quench data for the Cu(II)/GGH system in MES. 19This ternary intermediate complex (TIC) undergoes a rate-limiting intramolecular reaction to yield the 4N species and simultaneous dissociation of the buffer molecule.The latter step is assured by the fact that the end point of all GGH reactions was identical with the 4N Cu(II)-GGH spectrum. 19he buffer competition step is not rate-limiting except for the Tris system, as discussed above.The k obs obtained directly from the experimental data in this work corresponds to the ratio k on3 /k off3 in Scheme 3.
In order to gain deeper insight into these phenomena, the effects of 200 mM buffers on the Cu(II)/GGH reaction were studied over a broader range of Cu 2+ to GGH ratios and pH values.The Cu(II) concentration effect was studied first, with GGH concentration fixed at 2 mM.The MES system in which there was little interference by the buffer in the Cu(II) reaction with GGH was also studied as a reference.In all instances, the Cu(II) concentration increase slowed the reaction, without affecting its one-step character.The linear dependence of k obs on the Cu 2+ to GGH ratio is presented in Figure 5, and parameters of the fit are given in Table S1.
We analyzed these data using the value of relative slope of straight line approximation of the reaction rate dependence on the GGH concentration (−B/A in Table S1).For MES at pH 6.0, the buffer molecule does not bind the Cu(II) ion effectively, and the reaction rate depends solely on intramolecular rearrangement of the coordinated GGH molecule, as presented in Scheme 1 and discussed in detail in ref. 22.The relative slope for this case was 0.22, which means that k obs decreased by ca.10% in the tested range of Cu(II) concentrations.The scale of this effect is thus not large.Remembering that the observed reaction step is intramolecular, and the formation of IC, its substrate, occurred before the beginning of the observation, one can consider free GGH to be a weak reaction catalyst.The contribution to this effect of the reaction product (the 4N complex) can be excluded because the exchange of Cu 2+ ions between the 4N ATCUN/NTS complexes was studied previously and was found to occur on the time scale of days. 17,30−32 Hence, we can narrow our considerations to the interactions of IC, which is labile. 19As demonstrated recently for a series of GGH analogues with individual nitrogen atoms modified by methylation or acetylation, 22 the 4N complex formation, which involves the formation of two Cu−N bonds, is very highly cooperative.It requires a specific prearrangement of the  Taking this into account, we can propose that a larger excess of GGH molecules over IC allows for more efficient scanning for reactive peptide chain conformation by way of Cu 2+ exchange among the GGH pool.
Quite interestingly, the apparent GGH catalysis also acts for Tris.The binding of Tris to the Cu(II) ion is so strong that it prevents the Cu(OH) 2 precipitation in the absence of GGH (Figure 1).Accordingly, Tris slowed the 4N complex formation by an order of magnitude, but the GGH catalytic effect remained and was actually even slightly larger than for MES.A contribution of the GGH/Tris competition for Cu(II) binding could be responsible for this enhancement.Interestingly, the extent of the GGH catalytic effect is similar in the phosphate buffer, despite its opposite overall effect on the 4N formation rate (Table S1).The reasons are probably similar as well and involve GGH/phosphate competition.
The catalytic GGH effect was less pronounced for MOPS, HEPES, and PIPPS, with deviations from linearity.Molecules of these three buffers are structurally similar and are very bulky.Perhaps the coordinated buffer molecule slows down the postulated Cu 2+ exchange between the GGH molecules, but, as clearly seen in Figure 5, this difference is small to negligible.
Phosphate buffer is a uniquely strong catalyst for the 4N complex formation.This effect is particularly relevant due to the role of phosphate as a component of physiological and laboratory buffers.Its ca.10-fold acceleration of the studied reaction appears to reflect another phenomenon observed previously, namely, a nearly 4-fold reaction rate decrease by the amidation of the C-terminal carboxylate of GGH. 22The GGH carboxylate is not involved in Cu(II) binding in the 4N complex, as it points away from the coordinated Cu(II) ion. 33 steric hindrance is absent from the IC.The Cu(II)/ carboxylate interaction can be further enabled in IC by the 2+ charge of the Cu(II) ion, but nevertheless, it rather has an indirect character.The reaction acceleration was observed in reactions of substituted GGH analogues containing the Glu residue in positions 2 or 4 (GEHG-amide and GGHE-amide, respectively), but the IC absorption spectra confirmed the absence of carboxylate coordination.In contrast, the carboxylate binding to the Cu(II) in the EGHG-amide IC complex slowed the reaction, rather than accelerating it. 22herefore, an anionic group near the Cu(II) ion can accelerate the 4N complex formation if it does not impose limitations on the conformation of the Cu(II)-bound peptide.
We tend to speculate that in the TIC, the bound phosphate can help withdraw the peptide bond proton upon the Cu(II) ion assault, thus accelerating the reaction (see Scheme 4 for a cartoon representation of this concept).All of these processes are completed on the microsecond time scale under the given experimental conditions.In the TIC containing a single phosphate molecule, the GGH conformation productive in terms of the 4N complex formation is made available, and the 4N complex formation is enhanced by phosphate-assisted proton transfer, as indicated by blue arrows.In the TIC containing two phosphate molecules, such rearrangement is not possible until one coordinated phosphate ion dissociates from the Cu(II) ion.Charges and water molecules are omitted for clarity.

Inorganic Chemistry
In order to better understand this phenomenon, we followed the effect of the phosphate concentration on the 4N complex formation.At all tested phosphate concentrations, the spectra evolved analogously to that presented in Figure S4, with clearly visible isosbestic points.The initial spectra of the TIC at 3.5 ms are presented in Figure S7. Figure 6 presents kinetic traces at 525 nm for these reactions, while monoexponential fits to these traces are presented in Figure S8.The obtained k obs values are plotted in Figure 7 vs the phosphate concentration, and compared to the values of λ max of TIC d−d bands.A very accurate correspondence between these parameters was found depending on the concentration range: In the range of 25−200 mM, the linear increase of k obs was accompanied by a constant λ max value of 726 ± 2 nm.Starting from 200 mM, the k obs decreased linearly, which corresponded to a linear red shift of λ max , up to 753 ± 2 nm at 500 mM.
The excellent correspondence of these two parameters prompted us to provide the following explanation for the role of phosphate in the studied reaction.At concentrations up to 200 mM phosphate, the linear rate constant increase is due to catalysis by free, rather than coordinated phosphate, which can interact with both coordinated and free GGH molecules, enhancing the productive GGH conformation in both cases.The intercept of the line approximating this interaction, which extrapolates k obs to zero phosphate, 27 ± 1 s −1 , describes the contribution of the Cu(II)-bound phosphate ion to the reaction rate.It cannot be compared directly with the value for MES at pH 6.0 because the 4N formation process is pHdependent on its own. 17The deceleration of the reaction above 200 mM phosphate can be then readily explained by the d−d band red shift, which must be caused by the binding of another phosphate ion to TIC.The weak binding constant about 1 M −1 , expected for this interaction, explains the linearity of the effect in the studied pH range.The resulting reaction inhibition mechanism is analogous to the case of Tris because electrostatics forces the two phosphate ions to trans positions around the Cu(II) ion.This would preclude productive GGH coordination.The structural aspects of these interactions are also illustrated in Scheme 4.
Further insight into the role of buffers in the Cu(II)/GGH reaction was provided by experiments at various pH values.These experiments were performed for 200 mM MES, phosphate, HEPES, and MOPS buffers in the pH ranges of 5.5−6.5, 6.5−7.7,6.4−7.7, and 6.4−7.8,respectively.The results are summarized in Figure 8.The pH dependence of k obs in phosphate buffer had a sigmoidal character and could be fitted with the Hill equation. 34The obtained pK value was 7.22 ± 0.09, with the Hill (cooperativity) coefficient of n = 1.4 ± 0.3.This apparent pK is higher than that of the H 2 PO 4 − ion dissociation by 0.5 pH units, and the value of n for the latter process should be 1.Therefore, the observed effect must originate from a more complicated interaction, which corroborates the structural effect of phosphate ions on GGH, as postulated in the preceding section.One should note, however, that the acceleration of the studied reaction with increasing pH is primarily due to the enhancement of Cu(II)assisted deprotonation of GGH peptide nitrogen atoms, which

Scheme 2 .
Scheme 2. Structures and Acronyms of Organic Buffers Used in This Study a

Figure 1 .
Figure 1.Absorption spectra resulting from mixing of 3.2 mM Cu(H 2 O) 6 2+ ions (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 400 mM buffers, pH 6.0 for MES and pH 7.4 for other buffers (final concentrations: 1.6 mM Cu(II) and 200 mM buffer).Additional reactions for 3.2 mM Cu(H 2 O) 6 2+ ions with water (final pH 4.3) and with 50 mM NaOH (final pH 12.4) were also performed.Panels A−C present absorption spectra at different time points of the reaction: (A) 3.5 ms (initial spectra), (B) 1.5 s, and (C) 5 min (end of observation period).The spectra are color-coded according to the labels in the graph throughout the article.
) due to the absence of d−d absorption of initial Cu(II) complexes.Eventually, only Tris fully stabilized Cu(II) ions in solution, as indicated by the lack of turbidity and only a slight shift of the d−d band position and intensity during the 5 min period of observation (Figure S2).In PIPPS, the scattering of light on the formed Cu(OH) 2

Figure 2 .
Figure 2. Absorbance changes at 470 nm corresponding to Cu(II) precipitates formation, resulting from mixing of 3.2 mM Cu(H 2 O) 6 2+ ions (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 400 mM buffers, pH 6.0 for MES and pH 7.4 for other buffers (final concentrations: 1.6 mM Cu(II) and 200 mM buffer).Additional reactions for 3.2 mM Cu(H 2 O) 6 2+ ions with water and 50 mM NaOH (final pH of 12.4) were performed.Panels (A−C) present kinetic traces in different reaction time windows: (A) 3.5 ms−1.5 s, (B) 34 ms−16 s, (C) 34 ms−300 s.In panel (C), 60−90 s time window was marked in gray.Kinetic traces are color-coded according to the labels in Figure 1: red, phosphate; orange, NaOH; black, MES; dark blue, HEPES; light blue, water; dark green, MOPS; light green, PIPPS; magenta, TRIS.

Figure 3 .
Figure 3. Initial spectra at 3.5 ms resulting from mixing of 3.2 mM Cu(H 2 O) 6 2+ ions (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 4 mM GGH dissolved in 400 mM buffers, pH 6.0 for MES and pH 7.4 for other buffers (final concentrations: 1.6 mM Cu(II) and 2 mM GGH in 200 mM buffer).The spectra are color-coded according to the labels in the graph.

Figure 4 .
Figure 4. Absorbance changes at 525 nm corresponding to 4N CuGGH complex formation resulting from mixing of 3.2 mM Cu(H 2 O) 6 2+ ions (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 4 mM GGH dissolved in different 400 mM buffers, pH 6.0 for MES and pH 7.4 for other buffers (final concentrations: 1.6 mM Cu(II) and 2 mM GGH in 200 mM buffer).Inset presents the full course of the reaction in TRIS.The kinetic traces are color-coded according to the labels in the graph.

Figure 5 .
Figure 5. Dependences of observed rate constants (k obs ) for the formation of 4N CuGGH complex on the c Cu /c GGH ratio in the presence of different buffers: red, phosphate; black, MES; blue, HEPES; dark green, MOPS; light green, PIPPS; magenta, TRIS.k obs values were calculated from the data obtained by mixing different copper concentrations (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 4 mM GGH dissolved in different 400 mM buffers (final concentrations: 0.7−1.8mM Cu(II) and 2 mM GGH in 200 mM buffer).Error bars for individual determinations are within the symbols in all cases.

Scheme 4 .
Scheme 4. Hypothetic Mechanism of Catalysis and Inhibition of 4N Complex Formation in the Presence of a Phosphate Buffer a

Figure 6 .
Figure 6.Absorbance changes at 525 nm corresponding to 4N CuGGH complex formation resulting from mixing of 1.6 mM Cu(H 2 O) 6 2+ ions (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 2 mM GGH in the presence of various phosphate concentrations, pH 7.4 (final concentrations: 0.8 mM Cu(II) and 1 mM GGH).The phosphate concentrations are colorcoded according to the labels in the graph.

Figure 7 .
Figure 7. Dependences of calculated rate constants (black circles) and position of the d−d band of TIC at 3.5 ms (blue squares) for the formation of the 4N CuGGH complex at different phosphate concentrations (25−500 mM).Data points were determined from the absorbance spectra resulting from mixing of 1.6 mM Cu(H 2 O) 6 2+ ions (delivered by dissolving CuCl 2 in distilled water) with the same volumes of 2 mM GGH in the presence of various phosphate concentrations, pH 7.4 (final concentrations: 0.8 mM Cu(II) and 1 mM GGH).Error bars for individual determinations are marked.

Figure 8 .
Figure 8. pH dependence of k obs determined for the reaction of 4N CuGGH complex formation in different buffers.The dotted line represents the sigmoidal fit (Hill equation, pK 7.22 ± 0.09, n = 1.4 ± 0.3, R 2 = 0.995) to phosphate data, and the dashed line represents the linear fit to MES, MOPS, and HEPES data taken together (y = A + B × x; A (intercept) = −26 ± 1 s −1 , B (slope) = 5.5 ± 0.2 s −1 /pH unit; R 2 = 0.97).Circles representing data points are color-coded according to the labels in the graph.Error bars for individual determinations are within the symbols in all cases. 2 •2H 2 O, HEPES, PIPPS, MOPS, MES, and Tris were purchased from Merck (Darmstadt, DE).HCl, NaOH, and disodium and monosodium phosphates were purchased from ChemPur (Piekary S ́laskie, PL).

Table 1 .
Parameters and Assignments of the d−d Bands Recorded at the Beginning (3.5 ms) and End (300 s) of Cu(H 2 O) 6 a The reaction with sodium hydroxide is included for comparison.b pH 7.4, Unless stated otherwise.c Coordination mode inferred by analogy with HEPES.d Values could not be established.

Table 2 .
Parameters and Assignments of d−d Bands Recorded at 3.5 ms for Reactions of Cu(H 2 O) 6 2+ Ions with GGH in Buffers at pH 7.4, Except for MES, pH 6.0.The relative error of λ max determination is ±2 nm

Table 3 .
Observed Rate Constants (k obs ) for the Formation of the 4N CuGGH Complex (Final Concentrations of 1.6 mM Cu(II), 2 mM GGH) in the Presence of 200 mM Buffers at pH 7.4 (Except for MES, pH 6.0) a a Statistical errors of determinations on the last significant digits are given in parentheses.