Báo cáo khoa học: Thermodynamic characterization of interleukin-8 monomer binding to CXCR1 receptor N-terminal domain

Thermodynamic characterization of interleukin-8 monomer binding to CXCR1 receptor N-terminal domain Harshica Fernando1, Gregg T. Nagle2 and Krishna Rajarathnam1

1 Department of Biochemistry and Molecular Biology, University of Texas Medical Branch, Galveston, TX, USA 2 Department of Neuroscience and Cell Biology, University of Texas Medical Branch, Galveston, TX, USA

Keywords interleukin-8; isothermal titration calorimetry; monomer; N-terminal domain; thermodynamics

Correspondence K. Rajarathnam, 5.144 MRB, UTMB, Galveston, TX 77555-1055, USA Fax: +1 409 772 1790 Tel: +1 409 772 2238 E-mail: krrajara@utmb.edu

(Received 25 August 2006, revised 2 November 2006, accepted 7 November 2006)

doi:10.1111/j.1742-4658.2006.05579.x

Chemokines elicit their function by binding receptors of the G-protein-cou- pled receptor class, and the N-terminal domain (N-domain) of the receptor is one of the two critical ligand-binding sites. In this study, the thermo- dynamic basis for binding of the chemokine interleukin-8 (IL-8) to the N-domain of its receptor CXCR1 was characterized using isothermal titra- tion calorimetry. We have shown previously that only the monomer of IL-8, and not the dimer, functions as a high-affinity ligand, so in this study we used the IL-8(1–66) deletion mutant which exists as a monomer. Calori- metry data indicate that the binding is enthalpically favored and entropical- ly disfavored, and a negative heat capacity change indicates burial of hydrophobic residues in the complex. A characteristic feature of chemokine receptor N-domains is the large number of acidic residues, and experiments indicating that the using different buffers show no net proton transfer, CXCR1 N-domain acidic residues are not protonated in the binding pro- cess. CXCR1 N-domain peptide is unstructured in the free form but adopts a more defined structure in the bound form, and so binding is coupled to induction of the structure of the N-domain. Measurements in the presence of the osmolyte, trimethylamine N-oxide, which induces the structure of unfolded proteins, show that formation of the coupled N-domain structure involves only small DH and DS changes. These results together indicate that the binding is driven by packing interactions in the complex that are enthalpically favored, and are consistent with the observation that the N-domain binds in an extended form and interacts with multiple IL-8 N-loop residues over a large surface area.

mechanism, which involves interaction between the lig- and N-loop and the receptor N-terminal domain (N-domain) residues and between ligand N-terminal and receptor extracellular loop residues [5]. The largest sequence difference among chemokines and their re- ceptors is found in the N-loop and N-domain, respect- ively, suggesting that these residues encode both the specificity and promiscuity of interactions.

Abbreviations ASA, accessible surface area; CXCR1, CXC chemokine receptor 1; GPCR, G-protein-coupled receptor; IL-8, interleukin-8; ITC, isothermal titration calorimetry; N-domain, N-terminal domain; TMAO, trimethylamine N-oxide.

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

241

Chemokines constitute the largest family of proteins that mediate leukocyte recruitment and trafficking [1,2]. They show a remarkable range of receptor selec- tion and function, with some binding a single receptor, some binding multiple receptors, and some binding one receptor with high affinity and others with low affinity [3–7]. Chemokine receptors belong to the superclass of G-protein-coupled receptors (GPCRs), and structure–function studies show that all chemo- kines bind their receptors using the same two-site Interleukin-8 (IL-8, also known as CXCL8) and (such as related neutrophil-activating chemokines

H. Fernando et al.

Chemokine ligand–receptor interaction

amine N-oxide (TMAO), that promotes the folded state of the protein, and that the binding affinity of IL-8 for the N-domain is higher in osmolytes [24]. In this study, we have characterized the thermodynamic basis of IL-8 binding to the CXCR1 N-domain peptide using isothermal titration calorimetry (ITC).

MGSA ⁄ CXCL1 and NAP-2 ⁄ CXCL2) all have the characteristic N-terminal ELR residues, and bind and activate CXCR1 and CXCR2 receptors. IL-8 binds both receptors with high affinity, whereas all other lig- ands bind CXCR2 with high affinity and CXCR1 with low affinity [6,7]. Sequence analysis shows that the N-terminal ELR residues are conserved, whereas the N-loop residues are not, suggesting that the differences in binding may be due to binding of N-loop residues to the receptor N-domain.

Interestingly, sequence lowest the

that

We have shown previously that only the monomer of IL-8, and not the dimer, functions as a high-affinity ligand for receptor binding [22,25], so in this study we used the IL-8(1–66) deletion mutant which exists as a monomer. ITC measures the heat released or absorbed during a binding event, from which the free energy of binding (DG), enthalpy (DH), entropy (DS), and stoi- chiometry (n) are obtained in a straightforward man- ner, and also provides DCp by measuring heat released as a function of temperature [26]. To dissect coupling between structure induction and binding, we also measured binding in the presence of TMAO. As chemo- kine receptor N-domains are acidic in nature, binding experiments were also carried out using buffers with different heats of ionization to determine whether bind- ing is coupled to proton transfer. The data show that the binding is enthalpically favored and entropically disfavored, that coupled structure formation involves only small enthalpy and entropy changes, and that there is no net proton transfer. On the basis of the structure of the complex and structure–function studies, we propose that the favorable enthalpic contribution arises from optimal packing interactions of apolar resi- dues in the complex, and further propose that the ther- modynamic basis of the binding of all chemokine ligands to their receptor N-domains is similar to that observed for the IL-8 ⁄ CXCR1 system, and the ability to fine-tune the enthalpic and entropic components of the binding to the N-domain plays a key role in modu- lating affinity and ligand ⁄ receptor selectivity.

Results and Discussion

The structure of IL-8 is known, and the structural basis for its function has been well studied [8–17]. The receptor structures are not known and are difficult to obtain because of their membrane-embedded state. IL-8 and all other chemokine receptors share some unique properties compared with other members of GPCR class A receptors. Chemokines (molecular mass (cid:2) 8 kDa) are unusually large for a GPCR class A lig- and, as most are small molecules (< 1 kDa) with a rigid scaffold. In general, the sequence length of the GPCR N-domain correlates with ligand size [18], and chemokine receptors are a notable exception, as their N-domains are unusually short ((cid:2) 40 residues) com- pared with the size of their ligands ((cid:2) 70 residues). Further, in contrast with most GPCR class A recep- tors, chemokine receptor N-domains are also highly acidic. identity between CXCR1 and CXCR2 lies in the N-domain ((cid:2) 50%), and the N-domains are also of different length. Results from mutagenesis studies on both IL-8 and the CXCR1 receptor suggest the binding interactions between the IL-8 N-loop and receptor N-domain residues cover an extended interface, and can be described either by a model that involves mul- tiple weak interactions or by a ‘hot spots’ model which involves few strong interactions [12,19,20]. In principle, both models allow the chemokine ⁄ receptor to fine-tune and regulate binding affinity and ⁄ or ligand selectivity. Currently, little is known about the relative enthalpic (van der Waals, hydrogen-bonding, and electrostatic interactions) and entropic (solvation ⁄ desolvation, loss of conformational flexibility and dynamics) contribu- tions to binding, and such knowledge is essential for understanding the relationship between structure and the thermodynamics of binding.

studied, and for instance,

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

242

On the basis of structure–function data, a general two- site mechanism of ligand–receptor interaction has been proposed for all chemokines. Binding involves interac- tions between the chemokine ligand N-loop and the receptor N-domain, and ligand N-terminal and recep- tor extracellular loop residues. N-domain peptides for various chemokine receptors including CXCR1 have been shown to bind to their cognate ligands, indicating that studying isolated domains may give considerable insight into the molecular basis of binding and func- tion [11,21,27–31]. The IL-8 ⁄ CXCR1 pair is one of studies using the best CXCR1 N-domain peptide have shown that IL-8 dimer dissociation is essential for high-affinity binding, IL-8 binds the isolated N-domain with an affinity similar to that for the N-domain in the intact receptor, and so can be studied outside the context of the intact receptor [21]. Such studies have already provided valu- able insights into the molecular basis of ligand selectiv- ligand dimerization and binding affinity [22–24]. ity, We have recently shown that the receptor N-domain adopts a definite structure in the osmolyte, trimethyl-

H. Fernando et al.

Chemokine ligand–receptor interaction

and that the CXCR1 N-domain plays a major role in determining binding affinity and not in ligand selectiv- ity [22–24].

Design and characterization of IL-8(1–66) monomer

synthesis

Previous studies using a ‘trapped’ monomer and native protein that exists as both monomers and dimers have shown that dimer dissociation is essential for high- affinity binding to the receptor [22]. The trapped L25NMe monomer contains a non-natural NMe- amino acid as a dimer interface residue, and was synthesized by solid-phase chemical [32]. Comparison of the trapped monomer and native dimer structures shows that the last six C-terminal residues (67–72) are unstructured in the monomer and struc- tured in the dimer [8,10]. Therefore, we suspected that deleting these residues would result in a monomer. in residues Trp57–Ser72 being structured and helical the dimer, whereas the monomer will have lower helical content, as it is missing residues 67–72. An HSQC spec- trum of the IL-8(1–66) monomer shows the characteris- tic upfield (Phe17 and Val58) and downfield (Gln8 and Lys20) shifted peaks previously observed in the native dimer and the trapped monomer (Fig. 1). Chemical shift and NOESY data analyses indicate that IL-8(1– 66) adopts a structure similar to that of the trapped L25NMe monomer. We have also characterized the dynamics of IL-8(1–66) from 15N-T1, T2, and 1H-15N NOE relaxation measurements (unpublished observa- tions). The correlation time (sc) of 5.2 ns calculated from relaxation data is consistent with that expected for a 7.7-kDa protein. Previous studies have shown that the activity of IL-8(1–66) is similar to that of native IL-8 [34]. We also observed that our recombinant IL-8(1–66) is as active as native IL-8, and show below that it binds with the same affinity as the trapped L25NMe monomer to the CXCR1 N-domain peptide.

indicate that We used ITC to determine the enthalpy (DH), entropy (DS), and the free energy (DG) of binding of monomeric IL-8 to the receptor CXCR1 N-domain. The binding isotherm of IL-8(1–66) and the trapped L25NMe monomers to the CXCR1 N-domain are shown in Fig. 2. The upper panels show the thermo- grams, and the lower panels show the integrated heat fitted to a standard binding isotherm. The negative the interaction is exothermic peaks (DH < 0); to noise the data show excellent signal We had previously observed from ultracentrifugation studies [33] that the IL-8(1–66) deletion mutant is a monomer at micromolar concentrations, and we now observe from NMR and ITC studies that it is a mono- mer up to millimolar concentrations. The circular dichroism (CD) spectrum of the IL-8(1–66) monomer indicates that it is folded and shows a profile similar to that observed for the native IL-8(1–72), both showing characteristic minima at (cid:2) 222 nm (Fig. 1). Higher ellipticities and a pronounced minimum at (cid:2) 208 nm for the native protein are consistent with C-terminal

A

B

Fig. 1. Characterization of the IL-8(1–66) monomer. (A) CD spectra of a 25-lM solution of the IL-8(1–66) monomer (solid line) and the native IL-8(1–72) dimer(dash line) in 50 mM sodium phosphate ⁄ 50 mM NaCl, pH 8.0 buffer. (B) 15N-1H HSQC NMR spectrum of the IL-8(1–66) monomer. The observed chemical shifts are similar to that observed for the trapped monomer, and some of the upfield and downfield shifted peaks characteristic of a folded protein are labeled. The spectrum was acquired on a Varian Unity 750-MHz spectrometer in 50 mM acetate buffer, pH 5.5 at 25 (cid:2)C.

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

243

H. Fernando et al.

Chemokine ligand–receptor interaction

A

B

Fig. 2. Representative isothermal titration calorimetric profiles of IL-8 (1–66) and L25NMe IL-8 monomers binding to the CXCR1 N-domain. The titrations were car- ried out at 25 (cid:2)C in 50 mM Hepes ⁄ 50 mM NaCl, pH 8.0 buffer, and are shown in (A) and (B), respectively. The upper panels rep- resent the ITC thermograms, and the lower panels represent the fitted binding iso- therms.

Fig. 3. Sequence of the CXCR1 N-domains.

the trapped L25NMe monomer

is 8.6 lm, binding is

ratio, and could be adequately fitted to a single-site binding model. Control titration of IL-8(1–66) in to a buffer showed a weak exothermic peak, which further confirms that IL-8(1–66) is a monomer (not shown), as dimer dissociation is endothermic. The thermodynamic parameters for the binding of the IL-8(1–66) monomer to the N-domain peptide were observed to be similar [22]. to those of For the IL-8(1–66) monomer, the binding constant (KD) enthalpically favored (DH )11.8 kcalÆmol)1) and entropically disfavored (TDS )4.8 kcalÆmol)1). For the trapped monomer, the thermodynamic parameters are DH )10.5 kcalÆmol)1, TDS )3.4 kcalÆmol)1, and KD 6.0 lm.

Enthalpy of binding

residues are shaded grey and underlined, respectively. The structure of the complex reveals that binding is dominated by burial of apolar residues, involving van der Waals interactions between IL-8 Tyr13, Phe17, Phe21, Leu43 and receptor Pro21, Pro22, Tyr27, and Pro29 residues (numbering corresponds to the human sequence; Fig. 3). The structure also shows evidence of less well-defined electrostatic interactions between IL-8 Lys15, Arg47, Lys11 and receptor N-domain Asp24, Glu25, Asp26 residues. These observations show that residues that mediate binding in the complex are quite conserved between the human and rabbit sequences.

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

244

In addition to the structure, knowledge of how spe- cific residues contribute to binding affinity is essential. Proximity of residues in the structure does not always mean that they are involved in favorable interactions, and even if involved in favorable interactions, struc- tures cannot provide the relative strengths of the indi- vidual interactions; such information can be inferred only from mutagenesis studies. Mutagenesis studies in IL-8 have shown that both apolar (Ile10, Tyr13, Phe17, Phe21) and charged residues (Lys11, Lys15 and Lys20) are important [12,16,17,35,36]. However, inter- pretation of the receptor mutagenesis studies has been Enthalpic factors typically include van der Waals, hydrogen-bonding, and electrostatic interactions. The structure of a ligand–receptor N-domain complex is essential to identify the pairwise interactions and to describe how different interactions contribute to the observed enthalpy. The only structure available is that of IL-8 complexed to a chemically synthesized human CXCR1 N-domain peptidomimetic [11]. The sequence of the peptidomimetic is shown in Fig. 3 (labeled as p1). It corresponds to residues 9–29 and contains a sin- gle six-amino hexanoic acid linker (shown as lin) for residues 15–19. Sequences of our rabbit CXCR1 34- mer (residues 11–44) and the corresponding human CXCR1(9–39) are also shown. Identical and conserved

H. Fernando et al.

Chemokine ligand–receptor interaction

Table 1. Thermodynamic parameters for binding of IL-8(1–66) to the CXCR1 N-domain in different buffers. Measurements were carried out at 25 (cid:2)C, and the reported values are the mean of two experiments. All buffers contained 50 mM NaCl. DHITC is the experimentally mea- sured binding enthalpy, and DHion is the ionization enthalpy of the buffer.

Buffer, pH 8.0

n

KD (lM)

DHITC (kcalÆmol)1)

TDS (kcalÆmol)1)

DHion (kcalÆmol)1)

50 mM phosphate 50 mM Hepes 50 mM Tris

1.1 1.05 1.1

8.7 ± 0.3 8.6 ± 1 7.5 ± 1.5

)11.6 ± 0.1 )11.8 ± 0.1 )12.3 ± 0.1

)4.8 ± 0.1 )4.8 ± 0.1 )5.2 ± 0.1

1.0 5.0 11.5

N-domain, ITC measurements were carried out using single-component buffers with different ionization en- thalpies ranging from 1.0 to 11.5 kcalÆmol)1. Table 1 lists the values of DH, TDS, and KD in three different buffers. In all buffers, DHITC values were similar, within experimental error. The independence of meas- ured DHITC from DHionization indicates that binding is not accompanied by net protonation ⁄ deprotonation events.

receptors

Additional experiments such as mutagenesis and calorimetry measurements on the CXCR1 N-domain are necessary to provide a more definitive answer on the role of negative charges in binding. It is also poss- ible that the N-terminal acidic residues are necessary for interactions with extracellular matrix constituents and integrins, and that such interactions play a more vital role in the leukocyte recruitment process and ⁄ or in angiogenesis [39]. less straightforward. A characteristic feature of the N-domain is the preponderance of Asp ⁄ Glu residues, so it is reasonable to assume that some of these are involved in binding to the positively charged IL-8 Lys11, Lys15 and Lys20 residues. However, mutating Asp ⁄ Glu residues in the CXCR1 receptor N-domain in fact, mutagenesis did not seem to affect binding; studies identified only four residues (Thr18, Pro21, Pro22 and Tyr27) as being important [19,37]. All of these residues except Thr18 were also identified as from the NMR structural studies. being important On the other hand, mutational studies using a CXCR1 N-domain peptide showed that some of the acidic residues, in addition to the apolar residues, are [20]. involved in binding to neutrophil Although the results from the mutational studies of the receptor N-domain are inconclusive, they do indi- cate that the binding involves interactions with mul- tiple IL-8 N-loop residues over a large surface area. We determined the heat capacity (DCp)

for IL- 8(1–66) monomer binding to the CXCR1 N-domain by measuring enthalpy (DH) at several temperatures ranging from 20 to 35 (cid:2)C. Table 2 lists the thermo- dynamic parameters, and the data show that at all temperatures, the measured enthalpies are exothermic, and that the interaction is energetically less favorable at higher temperatures, as evidenced by the increased values of the dissociation constants. Provided that the temperature dependence of DH is linear over the tem- perature range studied, DCp is obtained as the slope of DH versus temperature. Figure 4 shows a plot of

Table 2. Thermodynamic parameters for binding of IL-8(1–66) to the CXCR1 N-domain as a function of temperature. All measure- ments were carried out in 50 mM Hepes ⁄ 50 mM NaCl, pH 8.0 buf- fer, and the reported values are the mean of two experiments.

On the basis of structure–function studies, residues that could be involved in protonation ⁄ deprotonation- coupled binding are the IL-8 N-loop residue His18 and any of the Asp ⁄ Glu residues in the receptor N-domain. The possibility that the Asp ⁄ Glu could be protonated on binding was especially intriguing, considering that only chemokine receptor sequences show the prepon- derance of negatively charged residues. Thermody- namic parameters measured by ITC can be influenced by the choice of buffer if proton transfer accompanies the binding process. In this case, the measured enthalpy of binding is linearly related to the intrinsic ionization enthalpy of the buffer. The relationship between the ionization enthalpy of the buffer and the measured enthalpy is given by the following equation [38]:

n

KD (lM)

DH (kcalÆmol)1)

TDS (kcalÆmol)1)

Temperature ((cid:2)C)

20 25 30 35

1.0 1.05 1.05 1.1

5.2 ± 0.3 8.6 ± 1.0 12.9 ± 4.0 12.3 ± 0.4

) 10.6 ± 0.4 ) 11.8 ± 0.1 ) 13.2 ± 0.2 ) 14.2 ± 0.1

) 3.5 ± 0.5 ) 4.8 ± 0.1 ) 6.5 ± 0.1 ) 7.2 ± 0.2

DHITC ¼ DHbinding þ nDHionization

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

245

where DHITC is the experimentally observed binding enthalpy, DHbinding is the buffer-independent binding enthalpy, DHionization is the ionization enthalpy of the buffer, and n is the net number of protons transferred during binding. To investigate whether there is a net proton transfer on IL-8 binding to the receptor

H. Fernando et al.

Chemokine ligand–receptor interaction

for residues 15–19,

Fig. 4. Temperature dependence of the enthalpy of binding of IL- 8(1–66) monomer to CXCR1 N-domain in 50 mM Hepes ⁄ 50 mM NaCl, pH 8.0 buffer. The solid line represents the least-squares fit of the experimental data.

ted and experimental DCp values have the same sign but differ by (cid:2) 170 calÆmol)1ÆK)1. Despite the limita- tions of the structure such as the N-domain peptidomi- metic containing a linker the agreement between calculated and experimental DCp is quite good, suggesting that burial and packing of apo- lar residues are the predominant determinants for binding. The mutagenesis studies do indicate a role for electrostatic interactions, and the observation that the experimental DCp is smaller than the structure-based calculated DCp also suggests that the structure may be missing some of these native interactions. The struc- ture of the complex was calculated on the basis of intermolecular NOEs; NOEs between charged residues are the most difficult to assign, especially if one of the interacting partners (receptor N-domain) is not isotopi- cally labeled.

Entropy of binding

experimental thermodynamic parameters

DH versus temperature, and the data indicate a slope [DCp ¼ d(DH) ⁄ dT] of )238 calÆmol)1ÆK)1. Correlating experimentally measured DCp and structural changes on binding and ⁄ or folding has shown that positive DCp involves burial of polar residues, and negative DCp involves burial of apolar residues, respectively [40]. Thermodynamic parameters for ligand–protein interactions can be related to changes in solvent- accessible surface area (ASA) upon binding. Empirical equations have been derived that allow comparison of the and structure-based calculated parameters based on sur- face area parameterization. This approach is clearly an approximation; according to this parameterization, in heat capacity arise from binding- the changes induced changes in the solvent polar and apolar ASA. To evaluate the energetic contributions of the binding interface, we made use of the NMR solution structure of IL-8 complexed to the human CXCR1 receptor N-domain [11]. The thermodynamic parame- ters were calculated for the complex with and without the receptor N-domain using the vadar program [41]. The DCp was calculated using the following equa- tion:

DCp ¼ 0:45DASAapolar (cid:3) 0:26 ASApolar Discussion of entropic factors in terms of structure is less straightforward. As motional properties correlate with entropy, a detailed knowledge of the conforma- tional flexibility and dynamic motions before and after binding of both the partners is essential to quantita- tively discuss entropic changes in structural terms. It has now become increasingly clear that the experiment- ally determined structures are a snapshot of one of many conformations that a protein can adopt, and fast and slow that proteins undergo a variety of motions [42]. For instance, NMR relaxation measure- ments show that protein backbone atoms undergo fast dynamics (nanosecond–picosecond time scale) about the average structure, and, further, such dynamics con- tribute significantly to the entropy of the protein [43]. It is generally thought that the conformational flexibil- ity and dynamic motions are reduced on binding, and so would be entropically disfavored. In contrast with conventional thinking, NMR relaxation studies have shown that the backbone dynamics in the bound form are not always quenched and may remain the same or actually increase [44]. Further, release of water on binding, which is entropically favored, should also be considered. If the binding interface were predomin- antly hydrophobic, the release of ordered water from interacting partners upon association could dominate the binding process [45].

changes on binding

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

246

where DASAapolar and DASApolar are the changes in ASA of the apolar and polar residues, respectively [39]. The structure-based calculations provide a DCp of )407 calÆmol)1ÆK)1 and DASAapolar and DASApolar of )1354 and )777 A˚ 2, indicating that more of the apolar residues are buried on complex formation. The calcula- IL-8 is highly structured in the free form, and NMR studies of the complex suggest that IL-8 does not undergo structural to the N-domain [11]. On the other hand, our CD data show that the CXCR1 N-domain is unstructured in the free form, and relatively more structured in the bound

H. Fernando et al.

Chemokine ligand–receptor interaction

cription for the role of entropy in binding. Future structural and dynamic studies of the complex should provide such an answer.

Fig. 5. CD spectra of free (solid line) and bound (dash line) CXCR1 receptor N-domain. The spectrum of the bound form was obtained by subtracting the spectrum of a 25-lM solution of free IL-8(1–66) monomer from the spectrum of an equimolar mixture (25 lM each) of IL-8(1–66) and receptor N-domain in 50 mM sodium phos- phate ⁄ 50 mM NaCl, pH 8.0 buffer.

Binding and folding

three different

form (Fig. 5). The N-domain peptide in the free form shows a minimum at 199 nm, which is characteristic of random coil structure, and on binding, the minimum is shifted to (cid:2) 203 nm and also shows a new peak at (cid:2) 220 nm. We measured the CD spectra of the bound N-domain at stoichiometries, and observed the spectrum of the bound N-domain to be essentially the same (data not shown). The CD spectra rule out helical structure (absence of characteristic double-well minima at 208 and 222 nm), suggesting that the N-domain binds in an extended fashion. These observations are also consistent with the previous NMR structural studies, which show that the receptor N-domain binds in an extended form to a cleft formed by the IL-8 N-loop residues [11].

We have discussed the calorimetry data so far simply in terms of binding, and have not explicitly considered contribution of enthalpy and entropy from folding of the N-domain. Mechanistically, binding could be des- cribed by a model in which the N-domain adopts a structure only on binding, or by an ensemble model in which the free N-domain exists in multiple freely inter- converting substates, one of which corresponds to a folded state that is binding-competent. In the former model, binding precedes folding, and in the latter, fold- ing precedes binding. Although these two models are mechanistically not equivalent, they are thermodynam- ically equivalent. Therefore, it is possible, in principle, to dissect the thermodynamics of folding and binding. We have shown previously that the CXCR1 N-domain is structured in the osmolyte, TMAO, and that IL-8 binds to the N-domain with higher affinity in TMAO [22]. In that study, a CXCR1 N-domain modified with a fluorescent tag was used; fluorescence spectroscopy was used to show that the N-domain becomes struc- tured in the presence of TMAO [DGfolding 1.7 kcalÆmol)1 (C1 ⁄ 2) 1.6 m TMAO]. and a transition mid-point Therefore, to dissect the contribution between binding and folding of the N-domain, we carried out binding experiments in the presence of TMAO. Our rationale was that the N-domain is structured in TMAO, so the measured thermodynamics should be predominantly due to binding. It is now well established that organic osmolytes such as TMAO promote structure of parti- ally and natively unfolded proteins and impart biologi- cal activity to these proteins, and so serve as excellent tools for studying the thermodynamic basis of protein folding [46].

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

247

The ITC data show that the binding is entropically disfavored, and also the relatively smaller change in entropy could be interpreted as entropic factors play- ing only a marginal role in complex formation. How- ever, this is not necessarily true, as the change in individual entropic factors, such as release of water or change in backbone dynamics, may be significant, and the overall change cancels out the individual con- tributions. Folding of N-domain on binding to IL-8 would be entropically disfavored, as the N-domain is unstructured in the free form and structured in the bound form. Knowledge of the dynamic characteris- tics of both IL-8 and N-domain before and after binding and whether binding is accompanied by release or retention of water molecules is lacking, and is also essential to provide a more quantitative des- The ITC thermograms for 1 m and 2 m TMAO are shown in Fig. 6, and the thermodynamic parameters are listed in Table 3. We could not carry out the bind- ing at higher TMAO concentrations because of limited solubility of TMAO and the 34-mer. Our previous studies have shown that a significant fraction of the N-domain peptide should be folded in 2 m TMAO [24]. The data indicate that in 2 m TMAO, binding is tighter (lower KD values), and that both the enthalpy and entropy values are higher. The approximately threefold increase in binding affinity is comparable to the approximately fivefold increase observed in our previous fluorescence studies, which is quite good, considering the intrinsic differences between the fluor-

H. Fernando et al.

Chemokine ligand–receptor interaction

C

A

B

Fig. 6. Representative isothermal titration calorimetric profiles of IL-8 (1–66) monomer binding to CXCR1 N-domain in TMAO. The titrations were carried out at 25 (cid:2)C in 50 mM Hepes, 50 mM NaCl, pH 8.0 buffer, and the data for 0, 1, and 2 M TMAO are shown in panels A, B, and C, respectively. The upper panels represent the ITC thermograms, and the lower panels represent the fitted binding isotherms.

Table 3. Thermodynamic parameters for binding of IL-8(1–66) to the CXCR1 N-domain in TMAO. All measurements were carried out in 50 mM Hepes ⁄ 50 mM NaCl, pH 8.0 buffer at 25 (cid:2)C, and the reported values are the mean of two experiments.

[TMAO] (M)

n

KD (lM)

DH (kcalÆmol)1)

TDS (kcalÆmol)1)

intermolecular interactions (binding) and not because of intramolecular interactions (folding). Small chan- ges in the thermodynamic parameters on folding also suggest that the folded form is only slightly more stable.

Conclusion

0 1 2

1.05 1.1 1.0

8.6 ± 1 3.4 ± 0.6 2.7 ± 0.3

)11.8 ± 0.1 )12.6 ± 0.1 )15.5 ± 0.7

)4.8 ± 0.1 )5.1 ± 0.1 )7.7 ± 0.1

escent-tagged and unlabeled N-domain, and that the binding was measured using two different techniques. that The ITC measurements suggest that study is

of both IL-8

small are

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

248

the folding energetically and enthalpically disfavored (DG is 0.8 kcalÆmol)1, DH 3.7 kcalÆmol)1) and entropically favored (TDS 2.9 kcalÆmol)1). Essentially the same enthalpic and entropic factors (such as van der Waals interactions and loss of conformational flexi- bility) that govern binding also govern folding and important induction of structure [26,47]. The most observation is that these thermodynamic changes on folding compared with binding. The measured heat capacity also has contributions from folding and binding. Most protein folding and struc- ture-induction events are accompanied by a negative DCp, which is to be expected because of burial of apolar residues [45]. Our experimentally determined DCp is negative, which, however, most likely reflects the binding process, as the apolar residues of the N-domain are buried predominantly because of IL-8 binding to its The thermodynamic basis of receptor CXCR1 N-domain has been characterized using ITC. This report describes how different enthalpic and entropic factors could mediate chemo- kine binding to its receptor N-domain, and is one of the few calorimetric studies that describes thermo- dynamics of a GPCR class receptor. The major con- clusion from this the binding is enthalpically favored and is mediated by optimal packing interactions of apolar residues and to a les- ser extent also by electrostatic and hydrogen-bonding interactions. Future high-resolution structure deter- mination of the complex and thermodynamic meas- and CXCR1 N-domain urements mutants should provide a more quantitative relation- ship between enthalpy and entropy and different binding interactions, and be able to distinguish between a model that involves multiple weak interac- tions and a hot-spots model that involves a few key interactions providing most of the binding energy. We propose that all chemokine receptor N-domains interact with their ligands using principles observed for IL-8 ⁄ CXCR1 system, and that binding affinity and receptor selectivity are mediated by modulating

H. Fernando et al.

Chemokine ligand–receptor interaction

CD studies the enthalpic and entropic components of the bind- ing to the receptor N-domain.

Experimental procedures

All CD spectra were collected on a Jasco J-720 spectropola- rimeter at 25 (cid:2)C in a 50 mm sodium phosphate ⁄ 50 mm NaCl, pH 8.0 buffer. Samples of the receptor N-domain, IL-8(1–66) monomer, and native IL-8 dimer were exten- sively dialyzed against the buffer and then filtered before determination of the protein concentration. Spectra were recorded from 260 to 195 nm with a scan rate of 10 nmÆ min)1 in a 0.1-cm-path-length cuvette. Scans of the buffer alone were averaged and subtracted from the averaged spectrum of each sample. Spectra of the bound receptor N-domain were obtained by measuring the spectra of the equimolar mixture of the receptor N-domain and IL-8(1– 66), and subtracting the spectra of the free IL-8(1–66) at the same concentration. Raw ellipticities were plotted, because molar ellipticities cannot be accurately determined in the case of protein complexes.

Cloning, expression, and purification of IL-8(1–66) monomer

peptide

concentrations were measured

and

0.04–0.07 mm for

The IL-8(1–66) construct was generated by introducing a stop codon after residue 66 in the wild-type IL-8(1–72) cloned in the pet32Xa vector at the LIC site. PCR amplifica- tion was carried out using the upstream primer 5¢-GAGA AGTTTTTGTAAGGCTCTAACTCTCCTCTG-3¢ and the downstream primer 5¢-AGAGTTAGAGCCTTACAAAAA CTTCTCCACAAC-3¢ with the QuickChange Site-Directed Mutagenesis kit (Stratagene Inc., La Jolla, CA, USA). The IL-8(1–66) monomer was expressed and purified using a pro- tocol similar to that used for wild-type human IL-8 [23]. Briefly, transformed Escherichia coli BL21DE3pLysS cells were grown in Luria–Bertani medium in the presence of ampicillin to a A600 of 0.5, and induced with 1 mm isopropyl b-d-thiogalactopyranoside for 4 h at 37 (cid:2)C. The pelleted cells were solubilized in lysis buffer (500 mm NaCl, 20 mm Tris ⁄ HCl, 5 mm benzamidine, 5 mm imidazole, pH 8.0), and then subjected to four freeze–thaw cycles and sonication. The protein-containing supernatant was loaded on to a Ni ⁄ nitril- otriacetate column and eluted with the same buffer as above, except containing 250 mm imidazole. Fractions containing the protein were pooled and dialyzed against the cleavage (20 mm Tris ⁄ HCl, 50 mm NaCl, 2 mm CaCl2, buffer pH 7.4). The dialyzed protein was cleaved with Factor Xa, and then purified by RP-HPLC using a gradient of acetonit- rile in 0.1% heptafluorobutyric acid. The fractions contain- ing protein were pooled, lyophilized, and stored at )20 (cid:2)C until further use. Both MS and analytical HPLC show that the recombinant IL-8(1–66) is pure with no evidence of impurities. The mass was verified using MALDI TOF MS.

ITC

The rabbit CXCR1 34-mer (LWTWFEDEFANATGMPP VEKDYSPSLVVTQTLNK) used in this study is the same as that was used in all of our previous studies [22–25], and was synthesized at the Biomedical Research Center, Van- couver, Canada. The peptides were purified by RP-HPLC and eluted with a gradient of acetonitrile in 0.1% trifluoro- acetic acid, and the mass was confirmed MALDI TOF MS. The sequence corresponds to residues 11–44, and is missing the first 10 residues, which have been shown not to be essential for binding [20]. At the time we initiated our calorimetric and biophysical studies, we synthesized both human and rabbit peptides, and observed that the rabbit CXCR1 34-mer was easy to synthesize, better behaved, and showed good heat signature.

The ITC experiments were performed using the VP-ITC system at 298 K as described previously [48]. The proteins and the CXCR1 34-mer peptide were extensively dialyzed against the appropriate buffer, centrifuged, filtered and degassed just before the start of the experiment. Protein using and UV absorbance spectroscopy, and the absorption coeffi- cients were determined by amino-acid analysis: for IL-8(1– 66), 7044 m)1Æcm)1, and for the N-domain peptide, 14962 m)1Æcm)1. Protein concentrations used for the titra- tion ranged from 0.5 to 0.8 mm for the IL-8(1–66) mono- mer, the CXCR1 N-domain peptide. For ITC experiments carried out in the presence of TMAO, the samples were dialyzed in the appropriate buffer, and aliquots of TMAO were added from a 4-m stock solution. The 1.42-mL sample cell and the injector were first washed with the dialysis buffer before the CXCR1 34-mer and the IL-8(1–66) monomer were intro- duced into the sample cell and injector, respectively. One to five injections of 3 lL followed by 20–25 of 9 lL were made, with a 6-min equilibration period between injec- tions. The reference cell was filled with distilled water. Control experiments such as protein and TMAO titration into buffer alone were performed to evaluate the heats of dilution, and subtracted from the experimental titration results. The heat of dilution of IL-8(1–66) was small and exothermic, providing further evidence that it is a mono- mer, as dimer dissociation is endothermic. The heats of dilution of the peptide and the buffer were small com- pared with the heat of reaction. Data were fitted using a nonlinear least squares routine using a single-site binding in Origin for ITC version 5.0 (Microcal), varying model the stoichiometry (n), binding constant (Kb), and binding enthalpy (DH(cid:2)).

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

249

Synthesis of the CXCR1 N-domain peptide

H. Fernando et al.

Chemokine ligand–receptor interaction

Acknowledgements

function identified by design of analogs and CXC che- mokine hybrids. J Biol Chem 269, 16075–16081. 13 Rajarathnam K, Clark-Lewis I, Dewald B, Baggiolini M & Sykes BD (1996) 1H NMR evidence that Glu-38 interacts with the N-terminal domain in interleukin-8. FEBS Lett 399, 43–46.

14 Rajarathnam K, Dewald B, Baggiolini M, Sykes BD &

We thank Dr Bolen for access to instrumentation, Drs Bolen, Ro¨ sgen, and Konkel for critical reading of the manuscript, and Ms Eapen for technical assistance. This work was supported by grants from the National Institutes of Health and the American Heart Associ- ation (to KR).

References

Clark-Lewis I (1999) Disulfide bridges in interleukin-8 probed using non-natural disulfide analogs: dissociation of roles in structure and function. Biochemistry 38, 7653–7658.

1 Luster AD (2002) The role of chemokines in linking

innate and adaptive immunity. Curr Opin Immunol 14, 129–135.

15 Rajarathnam K, Clark-Lewis I & Sykes BD (1994) 1H NMR studies of interleukin-8 analogs: characterization of the domains essential for function. Biochemistry 33, 6623–6630.

2 Moser B, Wolf M, Walz A & Loetscher P (2004) Che- mokines: multiple levels of leukocyte migration control. Trends Immunol 25, 75–84.

3 Loetscher P & Clark-Lewis I (2001) Agonistic and

16 Lowman HB, Fairbrother WJ, Slagle PH, Kabakoff R, Liu J, Shire S & Hebert CA (1997) Monomeric variants of IL-8: effects of side chain substitutions and solution conditions upon dimer formation. Protein Sci 6, 598– 608.

antagonistic activities of chemokines. J Leukoc Biol 69, 881–884.

17 Suetomi K, Lu Z, Heck T, Wood TG, Prusak DJ,

4 Fernandez EJ & Lolis E (2002) Structure, function, and inhibition of chemokines. Annu Rev Pharmacol Toxicol 42, 469–499.

Dunn KJ & Navarro J (1999) Differential mechanisms of recognition and activation of interleukin-8 receptor subtypes. J Biol Chem 274, 11768–11772.

18 Ji TH, Grossmann M & Ji I (1998) G protein-coupled receptors. I. Diversity of receptor–ligand interactions. J Biol Chem 273, 17299–17302.

5 Crump MP, Gong JH, Loetscher P, Rajarathnam K, Arenzana-Seisdedos F, Virelizier J-L, Baggiolini M, Sykes BD & Clark-Lewis I (1997) Solution structure and basis for functional activity of stromal derived cell- derived factor-1: dissociation of CXCR4 activation from binding and inhibition of HIV-1. EMBO J 16, 6996– 7007.

6 Murphy PM (1997) Neutrophil receptors for interleu-

19 Leong SR, Kabakoff RC & Hebert CA (1994) Complete mutagenesis of the extracellular domain of interleukin-8 (IL-8) type A receptor identifies charged residues med- iating IL-8 binding and signal transduction. J Biol Chem 269, 19343–19348.

kin-8 and related CXC chemokines. Semin Hematol 34, 311–318.

20 Attwood MR, Borkakoti N, Bottomley GA, Conway EA, Cowan I, Fallowfield AG, Handa BK, Jones PS, Keech E, Kirtland SJ, et al. (1996) Identification and characterization of an inhibitor of interleukin-8: a recep- tor based approach. Bioorg Med Chem Lett 6, 1869– 1974.

21 Gayle RB, Sleath PR, Srinivason S, Birks CW, Wee-

7 Ahuja SK & Murphy PM (1996) The CXC chemokines growth-regulated oncogene (GRO) a, GROb, GROc, neutrophil-activating peptide-2, and epithelial cell- derived neutrophil-activating peptide-78 are potent agonists for the type B, but not the type A, human interleukin-8 receptor. J Biol Chem 271, 20545–20550. 8 Clore GM, Appella E, Yamada M, Matsushima K &

rawarna KS, Cerretti DP, Kozlosky CJ, Nelson N, Bos TV & Beckmann MP (1993) Importance of the amino terminus of the interleukin-8 receptor in ligand interac- tions. J Biol Chem 268, 7283–7289.

22 Fernando H, Chin C, Ro¨ sgen J & Rajarathnam K

(2004) Dimer dissociation is essential for interleukin-8 (IL-8) binding to CXCR1 receptor. J Biol Chem 279, 36175–36178.

Gronenborn AM (1990) Three-dimensional structure of interleukin 8 in solution. Biochemistry 29, 1689–1696. 9 Baldwin ET, Weber IT, St Charles R, Xuan JC, Appella E, Yamada M, Matsushima K, Edwards BF, Clore GM & Gronenborn AM (1991) Crystal structure of interleu- kin 8: symbiosis of NMR and crystallography. Proc Natl Acad Sci USA 88, 502–506.

10 Rajarathnam K, Clark-Lewis I & Sykes BD (1995) 1H NMR solution structure of an active interleukin-8 monomer. Biochemistry 34, 12983–12990.

11 Skelton NJ, Quan C, Reilly D & Lowman H (1999) Structure of a CXC chemokine-receptor fragment in complex with interleukin-8. Structure 7, 157–168. 12 Clark-Lewis I, Dewald B, Loetscher M, Moser B &

23 Rajagopalan L & Rajarathnam K (2004) Ligand selec- tivity and affinity of chemokine receptor CXCR1: role of N-terminal domain. J Biol Chem 279, 30000–30008. 24 Rajagopalan L, Ro¨ sgen J, Bolen DW & Rajarathnam K (2005) Novel use of an osmolyte to dissect thermody- namic linkages between receptor N-domain folding, ligand binding, and ligand dimerization in a chemokine- receptor system. Biochemistry 44, 12932–12939.

Baggiolini M (1994) Structural requirements for IL-8

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

250

H. Fernando et al.

Chemokine ligand–receptor interaction

interleukin-8 in the high affinity interaction with the interleukin-8 receptor type A. J Biol Chem 270, 10428– 10431.

25 Rajarathnam K, Prado GN, Fernando H, Clark-Lewis I & Navarro J (2006) Probing receptor binding activity of interleukin-8 dimer using a disulfide trap. Biochemistry 45, 7882–7888.

26 Leavitt S & Freire E (2001) Direct measurement of pro- tein binding energetics by isothermal titration calorime- try. Curr Opin Struct Biol 11, 560–566.

27 Ye J, Kohli LL & Stone MJ (2000) Characterization of binding between the chemokine eotaxin and peptides derived from the chemokine receptor CCR3. J Biol Chem 275, 27250–27257.

37 He´ bert CA, Chuntharapai A, Smith M, Colby T, Kim J & Horuk R (1993) Partial functional mapping of the human interleukin-8 type A receptor. Identification of a major ligand binding domain. J Biol Chem 268, 18549– 18553 (erratum appears in J Biol Chem 269, 16520). 38 Baker BM & Murphy KP (1996) Evaluation of linked protonation effects in protein binding reactions using isothermal titration calorimetry. Biophys J 71, 2049– 2055.

28 Mayer KL & Stone MJ (2000) NMR solution structure and receptor peptide binding of the CC chemokine eotaxin-2. Biochemistry 39, 8382–8395.

39 Parker SL, Parker MS, Sah R & Sallee F (2005) Angio- genesis and rhodopsin-like receptors: a role for N-term- inal acidic residues? Biochem Biophys Res Commun 335, 983–992.

40 Prabhu NV & Sharp KA (2005) Heat capacity in pro-

29 Booth V, Keizer DW, Kamphuis MB, Clark-Lewis I & Sykes BD (2002) The CXCR3 binding chemokine IP- 10 ⁄ CXCL10: structure and receptor interactions. Bio- chemistry 41, 10418–10425.

teins. Annu Rev Phys Chem 56, 521–548.

41 Willard L, Ranjan A, Zhang H, Monzavi H, Boyko

30 Farzan M, Chung S, Li W, Vasilieva N, Wright PL,

RF, Sykes BD & Wishart DS (2003) VADAR: a web server for quantitative evaluation of protein structure quality. Nucleic Acids Res 31, 3316–3319.

Schnitzler CE, Marchione RJ, Gerard C, Gerard NP, Sodroski J & Choe H (2002) Tyrosine-sulfated peptides functionally reconstitute a CCR5 variant lacking a criti- cal amino-terminal region. J Biol Chem 277, 40397– 40402.

31 Mizoue LS, Bazan JF, Johnson EC & Handel TM

42 Wand AJ (2001) Dynamic activation of protein func- tion: a view emerging from NMR spectroscopy. Nat Struct Biol 8, 926–931.

43 Jarymowycz VA & Stone MJ (2006) Fast time scale

dynamics of protein backbones: NMR relaxation meth- ods, applications, and functional consequences. Chem Rev 106, 1624–1671.

(1999) Solution structure and dynamics of the CX3C chemokine domain of fractalkine and its interaction with an N-terminal fragment of CX3CR1. Biochemistry 38, 1402–1414.

32 Rajarathnam K, Sykes BD, Kay CM, Dewald B, Geiser T, Baggiolini M & Clark-Lewis I (1994) Neutrophil acti- vation by monomeric interleukin-8. Science 264, 90–92. 33 Rajarathnam K, Kay CM, Clark-Lewis I & Sykes BD

44 Zidek L, Novotony MV & Stone MJ (1999) Increased protein backbone conformational entropy upon hydro- phobic ligand binding. Nat Struct Biol 6, 1118–1121. 45 Murphy KP, Xie D, Garcia KC, Amzel LM & Freire E (1993) Structural energetics of peptide recognition: angiotensin II ⁄ antibody binding. Proteins 15, 113–120.

(1997) Characterization of quaternary structure of inter- leukin-8 and functional implications. Methods Enzymol 287, 89–105.

46 Bolen DW & Baskakov IV (2001) The osmophobic effect: natural selection of a thermodynamic force in protein folding. J Mol Biol 310, 955–963.

47 Murphy KP & Freire E (1992) Thermodynamics of

34 Clark-Lewis I, Schumacher C, Baggiolini M & Moser B (1991) Structure-activity relationship of interleukin-8 determined using chemically synthesized analogs. J Biol Chem 266, 23128–23134.

structural stability and cooperative folding behavior in proteins. Adv Protein Chem 43, 313–361.

48 Wiseman T, Williston S, Brandts JF & Lin LN (1989) Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal Biochem 179, 131–137.

35 Williams G, Borkakoti N, Bottomley GA, Cowan I, Fallowfield AG, Jones PS, Kirtland SJ, Price GJ & Price L (1996) Mutagenesis studies of interleukin-8. Identification of a second epitope involved in receptor binding. J Biol Chem 271, 9579–9586.

36 Schraufstatter IU, Ma M, Oades ZG, Barritt DS &

Cochrane CG (1995) The role of Tyr13 and Lys15 of

FEBS Journal 274 (2007) 241–251 ª 2006 The Authors Journal compilation ª 2006 FEBS

251