(Received for publication, May 9, 1996, and in revised form, October 9, 1996)
From the Protein Engineering Network of Centres of
Excellence (PENCE) and the Department of Biochemistry, University
of Alberta, Edmonton, Alberta T6G 2S2, Canada, the
¶ Theodor-Kocher Institute, University of Bern, P.O. Box 99, CH-3000 Bern, Switzerland, and
PENCE, the Biomedical Research
Centre and the Department of Biochemistry, University of British
Columbia, Vancouver, British Columbia V6T 1Z3, Canada
Neutrophil-activating peptide-2 (NAP-2) and
melanoma growth-stimulatory activity (MGSA) are members of the
chemokine family of inflammatory proteins. The structures of NAP-2,
determined by x-ray crystallography, and MGSA, elucidated by NMR
spectroscopy, revealed a tetramer and dimer, respectively. In order to
address the relevance of multimeric species to their activities on
neutrophils, analogs of NAP-2 and MGSA were synthesized in which the
backbone amide proton of Leu-22 in NAP-2, and Val-26 in MGSA, was
substituted with the bulky methyl group (NH NCH3).
These analogs were shown to be monomeric by sedimentation equilibrium
ultracentrifugation studies and were similar to the corresponding
native protein in assays for neutrophil elastase release and
Ca2+ mobilization from IL-8R1 and IL-8R2 transformed cells.
Sedimentation equilibrium studies of the native NAP-2 and MGSA were
also carried out to address the association behavior. For NAP-2, there
was no evidence for the tetramer, but an equilibrium between monomers and dimers and the dissociation constant was calculated to be 50-100
µM. Similarly, MGSA showed a monomer-dimer equilibrium with a Kd of ~5 µM. The data from
the monomeric analogs and also the calculation of dissociation
constants indicate that NAP-2 and MGSA have a tendency to associate
above the concentrations required for maximal activity or for receptor
activation, but at functional concentrations they are predominantly
monomers.
Recruitment and accumulation of circulating leukocytes at the site of inflammation are mediated in part by a family of small molecular weight proteins called chemokines (1-3). They have four conserved cysteines and are called either CXC or CC chemokines, depending on whether the first two cysteines are separated by one amino acid (CXC) or are adjacent (CC). Functionally, the CXC chemokines predominantly attract neutrophils, whereas CC chemokines attract monocytes, lymphocytes, and eosinophils.
Three well studied proteins of the CXC chemokine family are
interleukin-8 (IL-8)1, melanoma
growth-stimulatory activity (MGSA), and neutrophil-activating peptide-2
(NAP-2). The structure of IL-8 has been solved by both NMR and x-ray
methods, and both approaches indicated it to be a dimer (4, 5). NMR
studies of MGSA (6, 7) and x-ray studies of a variant of NAP-2 (M6L)
(8) indicated them to be a dimer and tetramer, respectively. The
tertiary and quaternary structures of IL-8 and MGSA are similar, and
NAP-2 adopts a similar structure while showing an additional level of
complexity by forming a tetramer. The tertiary structure consists of a
series of turns and loops in the N-terminal region followed by three
-strands and a C-terminal
-helix. In the dimeric structure, the
three
-strands from each monomer form a six-stranded
-sheet with
the helices transversing the
-sheet. On the basis of the dimeric structure of IL-8, it was previously thought that dimerization is
essential for function. However, we subsequently demonstrated that an
analog in which Leu-25 was substituted with N-methylleucine (L25NMe) was monomeric and had similar functional properties as the
native IL-8 (19). The structure of this synthetic IL-8 monomer was
solved by 1H NMR spectroscopy and was shown to be largely
similar to that of the monomeric unit in the NMR and x-ray structures
of the native dimer (16). Taken together, these observations suggested
that dimerization of IL-8 is essential neither for structural integrity nor for functional activation.
Two receptors (IL-8R1 and IL-8R2) that bind CXC chemokines
have been identified and characterized in neutrophils, and they belong
to the superfamily of seven transmembrane domain-containing proteins
that bind to G-proteins. IL-8R2 binds IL-8, MGSA, and NAP-2 with equal
affinity. In contrast, IL-8R1 binds only IL-8 with high affinity, but
it does bind MGSA and NAP-2 with low affinity (9-11). The N-terminal
residues Glu-4, Leu-5, and Arg-62 (the
"ELR" motif) (Fig. 1) are conserved and are
absolutely essential for receptor binding in MGSA, NAP-2, and IL-8
(12-15). Differential binding to the two receptors has been attributed
to a region of about 10 residues immediately preceding the first
-strand, and these residues are similarly located away from the
interface (17, 18). Mutation of charged residues at the dimer interface
in IL-8 had no affect on activity (18). Thus, functional data indicate that the residues at the dimer interface are not directly involved in
receptor binding.
In order to address the relevance of higher order states observed in structures of NAP-2 and MGSA, monomeric analogs of NAP-2 and MGSA were synthesized using a similar strategy employed in the synthesis of the monomeric IL-8 (19). For MGSA, Val-26 was substituted with N-methylvaline, and for NAP-2, Leu-22 was substituted with N-methylleucine. In addition to disrupting hydrogen bonding across the dimer interface, the modification also introduces a bulky methyl group, which prevents monomers from coming close to each other. We show here that the monomeric analogs of NAP-2 and MGSA have an activity similar to that of the corresponding native proteins for neutrophil elastase release and induce a similar rise in free Ca2+ in Jurkat cells transfected with either IL8-R1 or IL8-R2, indicating that dimerization is not essential for receptor binding and functional activation. Furthermore, calculations of Kd by sedimentation equilibrium for both NAP-2 and MGSA indicate that they are predominantly monomeric at functional nanomolar concentrations.
All proteins were chemically synthesized, purified, and characterized as discussed in detail previously (20). The monomeric MGSA and NAP-2 were synthesized using the same strategy as that used for IL-8 (19). Val-26 in MGSA and Leu-22 in NAP-2, the residues corresponding to Leu-25 in IL-8, were substituted with N-methylleucine and N-methylvaline, respectively. The synthesis was performed using t-butoxycarbonyl chemistry, and longer reaction times were required for the addition of the protected amino acid following the N-methyl-amino acid (NH2-terminal amino acid), apparently due to slow reaction kinetics. The analogs were purified by HPLC and characterized by ion spray mass spectroscopy (SCIEX AP III) as described previously. MGSA was synthesized without the last Asn residue, and this analog was shown to have the same activity as the full-length protein. Elastase release from freshly purified neutrophils (21) and Ca2+ induction from Jurkat cells transfected with IL-8R1 and IL-8R2 receptors were performed as described previously (22).
Ultracentrifugation StudiesSedimentation equilibrium studies were performed on a Beckman Spinco model E analytical ultracentrifuge using Raleigh interference optics. The samples were extensively dialyzed for a period of 48 h against the appropriate buffer. 100 µl of the samples was loaded into 12-mm double-sector, charcoal-filled Epson cells equipped with sapphire windows. Sedimentation equilibrium runs of all of the proteins were carried out at a concentration of ~1 mg/ml in 100 mM NaCl, 50 mM sodium phosphate, pH 7.0. Runs were carried out at rotor speeds between 24,000 and 32,000 rpm, depending on the sample, for a period of 48 h, and photographs were taken when fringes could be resolved across the boundary region between the protein solution and the solvent. For both native NAP-2 and MGSA, data were collected using at least two different rotor speeds, two different pH values, and/or two different starting concentrations.
The average molecular weights (Mav) from sedimentation equilibrium runs were calculated using the equation,
![]() |
(Eq. 1) |
The dissociation constant for the monomer-dimer equilibrium was calculated from the equation (25),
![]() |
(Eq. 2) |
The data were fitted using the curve-fitting routine in Table Curve (Jandel Scientific). The estimated Kd is given at a 95% confidence interval. For the monomeric proteins, the data could be fitted to a single species, and for the native proteins, the data could be fitted to a simple monomer-dimer equilibrium. There was no need to fit the data to a more complex monomer-dimer-tetramer equilibrium, and when this was tried there was no significant improvement of the residuals, indicating that the monomer-dimer equilibrium adequately fits the data.
Sedimentation
equilibrium ultracentrifugation data of V26NMe MGSA (Fig.
2A) indicated that this analog is a monomer,
since the calculated molecular weight (7,880) (Equation 1) was very similar to the experimental molecular weight (7,640). The data could be
fitted to a single species with no evidence of a dimer form. The
Mav of native MGSA at pH 7.0 was calculated from
Equation 1 to be 13.7 kDa (monomer molecular weight, 7,865) with a mass distribution between 11 and 15 kDa, indicating that it was in equilibrium between monomers and dimers (Fig. 2A). The data
could be fitted to a monomer-dimer equilibrium with a dissociation
constant (Kd) of 4 ± 3 × 106 M. The Kd at pH 5.0 was calculated to be 4.3 ± 1.4 × 10
5
M and indicated a weaker dimer interface at pH 5.0 than at
pH 7.0. Consistent with these results, NMR at pH 5.5 (7) and at pH 5.1 (6) indicated that the last four residues are unstructured. The
dependence of dimer association as a function of pH indicates that
electrostatic interactions play a role in stabilizing the dimer
structure.
Ultracentrifugation data of L22NMe NAP-2 (Fig.
2B) indicated that this analog is a monomer (calculated
molecular weight, 7,646; experimental molecular weight, 7,775) with no
evidence for dimeric species. The average molecular mass of native
NAP-2 at pH 7.0 (Fig. 2B) was calculated to be 11.6 ± 0.4 kDa (monomer molecular weight, 7,632), with a mass distribution
from 10 to 14 kDa, indicating that it exists in an equilibrium between
monomers and dimers. The data could be fitted adequately to a
monomer-dimer equilibrium with a Kd value of
5.3 ± 2.0 × 105 M (Fig.
3). Unlike the case for MGSA, there was no appreciable difference in the Kd values between pH 5.0 and 7.0 (~100 and ~50 µM, respectively).
These values indicate a relatively weak dimer interface compared with MGSA and IL-8. Lack of a pH effect is consistent with the NMR studies of NAP-2, which showed no differences in the Kd value for pH values between 5 and 7 (26). At 250 mM NaCl and pH 7.0, the authors calculated a dissociation constant for the monomer-dimer equilibrium as ~300 µM and for a dimer-tetramer equilibrium as ~900 µM (Table I). However, from chemical cross-linking studies, a dissociation constant of ~0.3 µM was reported for NAP-2 (27). Ultracentrifugation data are largely consistent with the NMR studies, and the ~6-fold difference is within the experimental error of the two techniques. It is likely that tetramers were not detected at the concentrations used in the sedimentation equilibrium experiment due to the large Kd value for the dimer-tetramer equilibrium. Overall, data from this study and those reported by NMR studies (26) are inconsistent with the low Kd values measured by cross-linking studies.
|
X-ray studies of a NAP-2 variant, having the substitution Met-6 to Leu
(M6L), showed the structure to be a tetramer (8). In addition to having
a dimer interface like that for IL-8 (designated as A/B interface), the
second interface involves residues in the N-terminal region (designated
as A/C interface). In solution structures of macrophage inflammatory
protein-1 (28) and RANTES (29, 30), which are CC chemokines,
N-terminal residues constitute the dimer interface. The crystal
structure of platelet factor-4, a CXC chemokine, showed it
to be a tetramer, and it has two dimer interfaces similar to NAP-2, an
A/B interface like IL-8 and an A/C interface like RANTES (31, 32). By
necessity, the crystals of NAP-2 were grown at high concentrations of
the protein (17 mg/ml in 100 mM sodium acetate, 200 mM ammonium acetate, pH 4.6), and this could be the reason
for the formation of the tetramer. It has been suggested by the authors
that the mutation M6L may have favored interaction at the A/C interface
compared with the native NAP-2, since platelet factor-4 has a Leu at
the corresponding position. Since L22NMe NAP-2 is a monomer, the
sedimentation equilibrium data suggest that the A/B interface is
stronger than the A/C interface for NAP-2.
Neutrophil elastase release activity of
native MGSA and NAP-2 and the monomeric L22NMe MGSA and V26NMe NAP-2
analogs are shown in Fig. 4. The data indicate that the
activities of the monomeric analogs are indistinguishable from those of
the native proteins. Neutrophils have two receptors that bind ELR
motif-containing chemokines, and both receptors elicit chemotaxis
exocytosis and a rise in Ca2+ level response. It has been
shown that IL-8, NAP-2, and MGSA elicit the same transient induction of
Ca2+ from Jurkat cells transfected with IL-8R2, whereas up
to 100 nM, only IL-8 elicits Ca2+ release
response from cells transfected with IL-8R1 (22). Similar experiments
were performed with the monomeric analogs (Fig. 5). The
data clearly indicate that both the monomeric analogs elicit similar
functional response as the native protein. The relative distribution of
monomers and dimers for different total concentrations of NAP-2
calculated from the Kd at pH 7 is shown in Fig. 6. It is observed that at functional concentrations
(0.1-10 nM), NAP-2 exists predominantly as a monomer, and
at concentrations used for structural determination (~1
mM) it exists predominantly as a dimer. The distribution
profile for MGSA, calculated for a Kd of 4 µM, shows a similar trend and indicates that it is
predominantly monomeric at functional concentrations. The Kd for IL-8 has been determined by sedimentation
equilibrium and microcalorimetry studies as 10-30 µM
(33, 34) (Table I), which indicates that IL-8 also exists as a monomer
at physiological concentrations (35, 36). The data from this study for
MGSA and NAP-2 and previous studies for IL-8 provide experimental
evidence that these chemokines have a strong propensity to associate at high concentrations but bind as monomers to the neutrophil
receptors.
Implications for Receptor Binding
Various models for
interaction between chemokines and their receptors have been discussed
(19, 37). The data presented here are consistent with a model in which
a monomeric chemokine ligand binds to a monomeric receptor. Several
lines of evidence indicate that the active species is also the monomer
for some of the CC chemokines. At high concentrations, I-309 and
monocyte chemoattractant protein-3 exist as monomers with no evidence
of dimerization (34, 37). Equilibrium constants for monocyte chemoattractant protein-1 and -2 are in the µM region
(34, 37), and a variety of studies have shown that the functionally
active species for macrophage inflammatory protein-1 is a monomer
(38).
Our data do not rule out the possibility that dimerization of the native chemokine may occur during or after initial binding of a monomer to the receptor. This is not possible for the monomeric chemokines discussed here, since the same mechanisms that prevent dimerization in solution will prevent dimerization on the receptor. However, there is no reason to believe that dimerization of the ligand is necessary for G-protein-coupled seven-transmembrane receptors. Most of the ligands for this family of receptors tend to be small molecular weight compounds such as formylmethionylleucylphenylalanine and epinephrine. For the growth factor type receptors, which have a distinctly different topology, dimerization of the receptor, and not the ligand, has been shown to be essential (39). All of the data presented here are from in vitro studies, and it is possible that these proteins may dimerize in vivo due to local high concentrations in the microenvironment at the receptor binding site. The dimerization may regulate the active concentration available for receptor binding or may stabilize the protein from proteolytic cleavage. A number of in vivo and in vitro studies seem to indicate that heparin binding on the cell surface is critical to generate a concentration gradient for leukocyte attraction and, hence, receptor binding (40-43). Preliminary studies from our laboratory suggest that the interaction of monomeric and native dimeric IL-8 with heparin is highly differential.3 Presently, all of the available data from our study and other studies indicate that chemokines have a tendency to associate at high concentrations but that the monomeric unit is sufficient to bind the receptor for functional activation.