Characterization of histidine residues essential for receptor binding and activity of nerve growth factor.

The role of the four histidine residues in receptor binding and activity of mouse nerve growth factor (NGF) was investigated using both site-directed mutagenesis and chemical modification with diethyl pyrocarbonate. Replacement of His-75 or His-84 with alanine resulted in decreased biological activity and decreased affinity for p140(trkA); however, with H75A only, a 5-fold increased affinity toward p75(LANR) was observed. The effect of simultaneous replacement of both His-75 and His-84 was neither additive nor synergistic. Slight perturbations in circular dichroism spectra and weakened self-association of the mutants indicated that His-75 and His-84 may be involved in stability, dimerization, and/or folding of NGF. Diethyl pyrocarbonate modification of His-4 and His-8 in the H75A/H84Q double mutant abolished neuritogenesis, binding to both receptors, and phosphorylation of p140(trkA) in PC12 cells. These chemical and mutational results confirm and clarify previous evidence for the involvement of His-75 and His-84 (Dunbar, J. C., Tregear, G. W., and Bradshaw, R. A. (1984) J. Protein Chem. 3, 349-356) or His-4 and His-8 (Shih, A., Laramee, G. R., Schmelzer, C. H., Burton, L. E., and Winslow, J. W. (1994) J. Biol. Chem. 269, 27679-27686) in receptor binding of NGF. At least three and possibly all four histidines, which are located in three spatially distinct regions, contribute to maintenance of functional sites that are essential for receptor binding and activity of NGF.

Signal transduction by these neurotrophins is initiated by the selective binding of each neurotrophin to its specific receptor(s) on the surface of distinct but overlapping populations of neurons. The trophic effects are then transmitted through a kinase signaling cascade to control the fate of the responsive cells (for review, see Ref. 15). The neurotrophins bind to two distinct classes of receptors, protein tyrosine kinase receptors encoded by the trk gene family (16 -18) and p75 LANR (19,20) of the tumor necrosis factor receptor superfamily (21). High affinity binding of NGF to TrkA induces the autophosphorylation and oligomerization of the receptor, which are thought to be the crucial initial steps for the subsequent neurotrophin signaling (22,23). In contrast, all neurotrophins bind to p75 LANR with similar low affinity, and the role of this low affinity neurotrophin receptor in neurotrophic signaling has been disputed (for review, see Refs. 24 and 25).
The x-ray crystallographic structure of NGF (26,27) has been determined. The NGF protomer has an elongated threedimensional structure with an unusual cysteine knot motif (28). Three pairs of antiparallel ␤-strands are connected by ␤-hairpin loop structures. Most of variable amino acid residues are concentrated in these loops. The carboxyl terminus (residues 113-118) and two loop regions (residues 43-48 and 76 -79) were poorly defined in the initial crystal structure determined (26) but were better defined in a later analysis containing five NGF monomer forms in two crystal forms (27). The amino terminus (residues 1-10) was ill defined in both determinations because of partial proteolysis and low electron density.
Recent studies on the structure-function relationships of NGF suggest that NGF has a distinct binding site for p75 LANR and for the TrkA receptor (for review, see Ref. 29). p75 LANR directly interacts with a group of positively charged residues (Lys-32, Lys-34, and Lys-95) that are located on the "north" end of the NGF molecule. Disruption of the p75 LANR receptor binding capability by site-directed mutagenesis of residues Lys-32, Lys-34, and Lys-95 did not affect those functions of NGF mediated by TrkA (30). On the other hand, NGF binds to the TrkA receptor through residues which are diffusely scattered in different variable regions. However, these TrkA binding residues may form a continuous extended binding surface across the north end approximately parallel to the 2-fold symmetry axis of the NGF molecule (31).
Besides these two distinct binding sites, there is an overlapping group of residues which affect the low affinity binding as well as the high affinity binding either simultaneously or differentially (29,31). Deletion of residues 9 -14 in the NH 2 terminus or residues 112-118 in the COOH terminus abolished binding to the NGF receptors on PC12 cells as well as the biological activity of NGF (32). Similar studies on the NH 2terminal region confirmed that deletion of residues 9 -13 greatly decreased binding to the TrkA receptor without significantly affecting binding to the LANR or the conformation of NGF (33). Proteolytic cleavage (34) or mutational truncation (35) of the NH 2 -terminal 1-9 residues affected the TrkA binding significantly and the p75 LANR binding to a lesser extent. Removal of the C-terminal dipeptide also affects biological activity in certain assays (36).
Mouse NGF has four histidine residues at positions 4, 8, 75, and 84 that are strongly but not completely conserved among NGF species and the other neurotrophins (37). Chemical modification of the histidine residues of des(1-9)NGF with diethyl pyrocarbonate (DEP) greatly reduced the receptor binding affinity with rabbit superior cervical ganglia, suggesting that His-75 and/or His-84 are responsible for the receptor binding (38). Interestingly, the absence of the two histidines when the NH 2 -terminal region was removed did not affect the sensitivity to DEP modification as measured by binding affinity in these studies. In accordance with these results, mutational substitution of residues including His-84 in the variable region IV (79 -88) affected the specificity of NGF binding to TrkA (31) and point mutation of His-4 or His-8 to alanine markedly reduced biological activity and TrkA binding (35). Because of the uncertainties and discrepancies in these studies of histidine residues, we felt it important to compare and contrast mutation and modification of histidine residues in purified recombinant NGF.
In this study we have used site-directed mutagenesis as well as chemical modification techniques to elucidate more thoroughly the role of histidine residues in structure and function of NGF. Recombinant mouse NGF, expressed in Sf21 cells, has previously been shown to be produced as the intact polypeptide chain that is correctly processed on the N terminus (36). Histidine mutants of NGF were purified to electrophoretic homogeneity and characterized by bioactivity, receptor binding, and biophysical properties. We have demonstrated that histidine residues 75 and 84 and histidine residues 4 and/or 8 play an important role in receptor binding as well as in biological activity of NGF. Negligible conformational changes in the secondary structure were produced either by mutagenesis or by chemical modification, but some loosening of the dimeric structure was noted after mutation of His-75 or His-84. EXPERIMENTAL PROCEDURES mNGF-Mouse NGF was isolated from the submaxillary glands of adult male Swiss Webster mice and further purified by fast protein liquid chromatography as described previously (36,39). This preparation is about 95% intact chains as determined by SDS-PAGE and nonequilibrium isoelectric focusing (40). The integrity of the N terminus of wild type recombinant NGF was also verified by amino acid composition, indicating 4.4 His residues per molecule of subunit.
Antibodies-Polyclonal anti-TrkA antibody was purchased from Santa Cruz Biotechnology, Inc. Anti-NGF antibody was from CRC, Inc. Preparation of monoclonal anti-NGF antibody N60 has been previously described (41). Horseradish peroxidase-conjugated secondary antibodies to rabbit or mouse IgG were from Organon Pharmaceuticals and Bio-Rad, respectively.
PC12 Bioassay-PC12 cell culture and defined medium bioassay were performed as described (42). Cells bearing neurites of more than one cell body length were counted 48 h after the addition of NGF, and the experiment was repeated at least three times.
Preparation of Histidine Mutants of Mouse NGF-Site-directed mutagenesis was performed in the pTZ19U vector by polymerase chain reaction-based gene splicing by overlap extension (43). All mutant vectors were sequenced over 200 base pairs upstream and downstream of the mutated site to ensure fidelity. Subcloning to the pBlueBac III shuttle vector and further manipulation in the baculovirus were performed as described previously (33,36). The pBlueBac III vector was used to facilitate screening of positive recombinant baculovirus plaques and to maximize the secretion of the expressed proteins. Three percent fetal bovine serum was added to Excell 401 medium to support expression in 200 -400 ml of suspension cultures. From 2 to 5 liters of culture, mutant proteins were purified to homogeneity on SDS-PAGE using CM52, N60 monoclonal antibody immunoaffinity chromatography, and MonoS FPLC as described previously (33,36). Since the N60 antibody only recognizes native NGF (44), this purification assures that the preparation of mutant has folded more or less correctly. The proteins were further concentrated without loss of activity by lyophilization after a reversed phase fast protein liquid chromatography chromatography (33). The mutant proteins were stored at Ϫ80°C until use. Protein concentration of purified recombinant NGF was determined by densitometric scanning of Western blots and by absorption using A 280 nm 1 mg/ml ϭ 1.6 (45).
Analysis of NH 2 -terminal Sequence and Amino Acid Composition-Protein sequencing was performed on an ABI 470A protein sequencer with online identification of the phenylthiohydantoin-derivatives using an ABI 120A HPLC and a PTHC18 narrow bore column according to the manufacturers' protocols. Amino acid composition was analyzed by hydrolyzing the samples in 6 N HCl with 0.1% phenol at 150°C for 1 h and derivatizing the hydrolysate with phenyl isothiocyanate using the PICOTAG TM system with Waters' protocols.
Receptor Binding Assay-Iodination of NGF and suspension receptor binding assays, either with PC12 cells or Sf21 cells containing ectopically overexpressed TrkA, were carried out as described previously (33,39). The cold chase experiment (39,46) was performed to differentiate the stable binding sites from the labile, fast binding sites on PC12 cells. In order to obtain the true inhibition constant (K i ), the specific binding data from the competition curves with unlabeled mNGF or mutant NGF were fit by a nonlinear least square Marquardt algorithm (BASICFIT) to the one class competition isotherm of Equation 1 where b ϭ pM ligand bound; B max ϭ maximal ligand bound in pM/10 6 cells; [L] ϭ concentration of ligand, i.e. radiolabeled NGF; [I] ϭ concentration of inhibitor, i.e. cold or mutant NGF; K d ϭ dissociation constant in picomolar concentration for labeled NGF determined from consideration of the competition data as a cold dilution of the labeled NGF; and K i ϭ dissociation constant in picomolar concentration for cold or mutant NGF in the competition experiment. The lines in each figure are drawn to this equation with the best fit parameters.
TrkA Phosphorylation Assays-TrkA phosphorylation assays were performed with PC12 cells in 10-cm culture plates (Falcon) by Western blotting of cell lysates with anti-Trk antibody (Santa Cruz) and the ECL chemiluminescence system (Amersham Corp.) as previously described (44).
Circular Dichroism (CD)-CD spectra were recorded by a computerdirected Jasco 710 spectropolarimeter at room temperature. Protein samples including appropriate controls were measured in a quartz cuvette with a 2-mm path length at 50 g/ml protein concentration in 10 or 30 mM potassium phosphate, pH 6.1. Spectra were typically recorded as an average of 6 scans from 250 to 185 nm.
Histidine Modification-The lyophilized mNGF and His mutants were dissolved in 50 mM potassium phosphate, pH 6.1, at a concentration of approximately 5 M. The samples were centrifuged at 13,000 ϫ g for 10 min at 4°C and the NGF concentration was estimated from the absorption at 280 nm (47). NGF was modified as described previously (38,48) with a 10-fold molar excess of DEP to histidine, and the reaction was continued for 40 min at 25°C. The extent of the reaction was monitored by the increased absorbance at 240 nm. The volume of ethanol used for DEP solution did not exceed 2% of the final volume of the reaction mixture.

Expression and Purification of Recombinant NGF Histidine
Mutants-The recombinant His mutants were designed to be expressed and secreted as the mature protein in insect Sf21 cell culture. H75A, H84Q, and H75A/H84Q were expressed at the level of wild type recombinant NGF. However, the expression of H84A and H84I was reduced 2-5-fold below that of wild type recombinant NGF. Thus, replacement of His-84 with a different amino acid significantly affected the expression level either in serum-supplemented or serum-free media (Table I). His-75 is fully conserved throughout the neurotrophin family, while His-84 is replaced with Gln in both mouse BDNF and NT-3. Expression of H84Q and the double mutant H75A/H84Q was as good as wild type recombinant NGF (Table I and Ref. 36), while replacement of the hydrophilic His-84 with the hydrophobic and aliphatic residues, Ala and Ile, severely reduced the expression of the corresponding mutants. Whether the lower expression was caused by decreased transcription or occurred post-translationally was not investigated.
The NH 2 -terminal sequence and amino acid composition of H75A/H84Q were analyzed to verify the authenticity of the engineered protein product. The double mutant protein had the same NH 2 -terminal sequence as mature NGF for 10 residues (SSTHPVFHMG . . . ), indicating that the prepro region of the protein was fully processed on secretion. No secondary sequence was detected, confirming the lack of proteolysis to des (1)(2)(3)(4)(5)(6)(7)(8) forms. Analysis of the total amino acid composition showed that the double mutant protein had only 2 histidine residues (2.0 mol/mol), representing those at positions 4 and 8, with the number of the other amino acid residues being the same as that of wild type recombinant or authentic mouse NGF (data not shown).
His Mutants Show Reduced Biological Activity-The biological activity assayed by PC12 neurite outgrowth in defined medium showed that His mutants retain biological activity but that it is reduced 4 -10-fold compared to mNGF, depending on the substitution (Table I and Fig. 1). Comparison of H75A/ H84Q (27% activity) to H75A (23%) or to H84Q (20%) showed that the effects of the mutation at both His-75 and His-84 were neither additive nor synergistic. This reduced biological activity could be explained in several ways: 1) His-75 and His-84 may play an independent role in different subsites for the high affinity receptor binding; 2) each mutation may result in a conformational change that subsequently alters the receptor binding activity; 3) each His residue may be important for the stability of the NGF homodimer so that mutation of His residues significantly contribute to dimer instability.
Receptor Binding Assays with His Mutants-To determine if any of these mutants has a significantly altered binding affinity for either the low affinity p75 LANR receptor or for the high affinity TrkA receptor, the binding affinity of each purified mutant for PC12, A875, and Sf21-TrkA cells was determined by equilibrium displacement binding studies with 125 I-NGF. H84A, H84Q, and H75A/H84Q bound to PC12 cells with a  Table I. binding affinity similar to that of mNGF (Table I and Fig. 2A). Binding to PC12 cells under these conditions is largely determined by low affinity p75 LANR receptor sites, since the number of these receptors is about 10 -20 times larger than TrkA (39,49). However, binding of H75A to p75 LANR showed approximately a 10-fold higher affinity than mNGF ( Fig. 2A), suggesting that the mutation of His-75 significantly increased the binding affinity toward p75 LANR . In contrast, limited studies suggested that H84I appeared to bind to p75 LANR with an affinity severalfold lower than the other His mutants. The relative affinities of several mutants for binding to p75 LANR were confirmed using A875 cell lines that only express p75 LANR and not TrkA. The EC 50 for H75A was about 10-fold lower than H84Q, H84A, H75A/H84Q, or wild type NGF (data not shown).
On the other hand, binding of His mutants to the TrkA receptor ectopically expressed in Sf21 cells correlated with the biological activity of each mutant shown by neurite extension promoted in PC12 cells (Fig. 2B and Table I). Binding of both H75A and H84Q was reduced by about 5-10-fold. As with the bioactivity, the double mutation at both His-75 and His-84 reduced the binding to TrkA receptor approximately 10-fold lower than mNGF, indicating no cooperative effect between these two residues. Thus from these studies, His-75 and His-84 are not essential for biological activity; however, each of these 2 His residues makes some contribution, either directly or indirectly, to interaction with TrkA.
DEP Modification Disrupts Biological Activity and Receptor Binding-Earlier studies of DEP-modified 2.5S NGF and des(1-9)NGF concluded that His-4 and His-8 would be dispensable for receptor binding (38). We exploited the advantage of the chemical modification of His and availability of the H75A/ H84Q mutant lacking two His residues to directly modify the His-4 and His-8 residues in order to help elucidate the role of these two residues in NGF-receptor interaction. The quantitative DEP modification of His residues in the ␤-NGF preparation was repeated essentially the same as the earlier modification of the 2.5 S NGF (38). The H75A, H84A, and H75A/H84Q mutants were also modified with a 10-fold molar excess of DEP with about 90 -100% of the remaining His residues reacting within 40 min at 25°C as shown by the increased absorbance at 240 nm (data not shown). Because of the slow but significant lability of the DEP modification, the modified proteins were immediately and simultaneously used for PC12 bioassay, receptor binding, TrkA phosphorylation, and circular dichroism spectroscopic measurement; cold chase-stable receptor binding experiments were done later. Modification of His residues of mNGF by DEP led to a marked loss in bioactivity in the PC12 bioassay. DEP-NGF and DEP-H75A/H84Q had about 60 and 45% of the activity of their corresponding controls when the percentages of cells with extended neurites were counted after treatment at 1 nM concentration for 48 h; however, this number is difficult to quantitate because of the reversibility of the DEP modification and the length of time required for bioassay. Since receptor binding and TrkA phosphorylation studies require only about 2 h, we focused attention on these shorter assays.
As expected, the receptor binding of the more intact ␤-NGF preparation was disrupted (NGF control data, Figs. 3 and 4) similarly to that previously described for 2.5 S NGF (Dunbar et al., 1984), except that a clear demonstration of the effect both on TrkA (Fig. 3B) and on p75 LANR (Fig. 3A) was now attainable. With the H75A/H84Q mutant, equilibrium binding studies showed that the capability to bind either the low affinity p75 LANR of PC12 cells (Fig. 3A) or the high affinity TrkA receptor of Sf21-TrkA cells (Fig. 3B) was also destroyed by chemical modification with DEP. The effect of the DEP modification on high affinity receptor binding of H75A/H84Q was also confirmed by the cold chase-stable binding experiment performed with PC12 cells (Fig. 4). In the cold chase-stable binding protocol, the fast dissociation component is rapidly dissociated within seconds, and the slow component, correlated with high affinity binding, remains stably bound during the chase (33,36,39,46,49). The DEP-modified H75A/H84Q at 1 nM concentration, i.e. a 10-fold molar excess over 125 I-NGF, did not compete with NGF for cold chase-stable sites (Fig. 4). Parallel experiments with 1 nM DEP-modified NGF, H75A, and H84A exhibited a similar lack of competition in cold chasestable receptor binding assays (Fig. 4), confirming that the major effect of DEP modification on binding was on His residues other than 75 or 84. This trend in lack of competition after modification was also apparent at 100 pM concentration of competitor (data not shown). These results are consistent with His-4 and/or His-8 being important for both high and low affinity receptor binding, in agreement with Shih et al. (35). However, our results might also be due to a general disruption  Table I. of protein conformation upon DEP modification of His-4 and His-8.
TrkA Phosphorylation Assays-The dose-response effect of mutants on TrkA phosphorylation was studied at two NGF mutant concentrations. TrkA phosphorylation can be detected with 5 ng of mNGF/ml of medium (50). With this subsaturating NGF concentration (5 ng/ml, 190 pM), H75A (Fig. 5A, lane 3 compared to lane 1), H84Q (data not shown), and H75A/H84Q (Fig. 5B, lane 1 compared to lane 5) stimulated TrkA phosphorylation nearly as intensely as mNGF, consistent with their somewhat lower TrkA binding affinity (Fig. 2B). However, significantly reduced TrkA phosphorylation was observed with H84A (B, lane 3 compared to lane 5), since its binding affinity is only about 2% of wild type NGF. At saturating ligand concentrations of 50 ng/ml (1.9 nM), all NGF mutants induced TrkA phosphorylation to a similar extent (data not shown), indicating that the reduced autophosphorylation at low mutant concentration was a simple reflection of the dose response of NGF mutant binding to TrkA.
DEP modification of the remaining intact His residues at positions 4 and 8 in H75A/H84Q reduced the TrkA phosphorylation severalfold (Fig. 5B, lane 2 compared to lane 1 and Table I). However, DEP modification of three or all four resi- dues, including the two NH 2 -terminal His residues plus either H75A or H84Q, resulted in a greater reduction in TrkA phosphorylation as shown with the DEP-modified mNGF, H75A, and H84A. These results (summarized in Table I) suggest that the two NH 2 -terminal His residues, 4 and 8, are important, but that His-75 and His-84 also help maintain a structural region important for TrkA phosphorylation.
Circular Dichroism of Histidine Mutants With and Without DEP Modification-For all unmodified His mutants the maximum negative molar ellipticity was shifted slightly toward the lower wavelength (Fig. 6A), suggesting that some degree of conformational change might be induced by the site-directed mutagenesis. For H75A and H75A/H84Q only a small change in magnitude of the trough at 205 nm was observed. A major effect was found for H84A, with the most negative value at 205 nm, suggesting that the 11% activity (Table I) could be at least partially attributed to an unfavorable conformational change. The less drastic substitution of Gln for His, rather than Ala for His, in the double mutant H75A/H84Q did not have this effect on CD spectra or conformation and was used for most subsequent studies.
The CD spectra of DEP-modified NGF was also recorded immediately after the modification (Fig. 6B). This DEP modification of NGF did not cause a significant change in the ellipticity of the CD spectrum between 190 and 260 nm. Pairs of similar CD curves were obtained for each DEP-modified and unmodified histidine mutant (data not shown). This result agrees with the earlier observations in which DEP-modified 2.5 S NGF retained the fluorescence spectra of the unmodified NGF (38).
His Mutation Perturbed the Stability of NGF Dimer-Considering that His-75 lies adjacent to a Trp residue involved in the dimeric interface in the south loop region and His-84 is between two Thr residues also involved in the dimeric interface in the ␤ sheet backbone region (26), replacement of either residue could influence the stability of the dimeric NGF molecule. One convenient way to check the dimer stability is to examine the relative distribution of monomeric and dimeric forms of NGF on a nonreducing SDS-PAGE followed by Western blotting (51). This method has been shown in our laboratory to be a reliable, albeit qualitative, method to assess neurotrophin dimer stability. 2 Both single His mutants showed increased dissociation relative to wild type NGF (Fig. 7), with the H84A mutant being more dissociated than H75A. Moreover, the effect of the double mutation was additive, suggesting that both His-75 and His-84 are important in subunit-subunit interactions in NGF. Their contribution to the dimer stability is reasonable because crystal structure analyses show that both His-75 and His-84 are near the dimer interface (26,27). This enhanced dissociation of the NGF dimer in the His mutants must be taken into account in interpreting both the decreased binding to TrkA and the enhanced binding of H75A to p75 LANR . The dimer stability was not much affected by DEP modification (Fig. 7), but the intensity of the dimeric band was slightly increased in DEP-modified samples, possibly because of the more hydrophobic nature of the NH 2 -terminal region of the DEP-modified NGF, which might moderately contribute additional interactions to the dimer stability of NGF.

DISCUSSION
Previous structure-function studies of NGF performed either by chemical modification (38) or by homolog scanning, point, or deletion mutagenesis (31,32,35,52) did not provide complete information on the role of each of the four histidine residues underlying the molecular mechanism of neurotrophin signaling. Disruption of receptor binding was previously reported for DEP-modified 2.5 S NGF (38). These authors suggested that 2 S. Swiatkowski and K. E. Neet, unpublished data.
FIG. 6. CD spectra of native and DEP-modified histidine mutants. The CD spectra were measured at a concentration of 50 g/ml protein in a cylindrical quartz cell of 2.0-mm light path length at room temperature. The circular dichroism of each spectrum was averaged over 6 scans and then converted to mean residue ellipticity ((degrees⅐cm 2 /dmol) ϫ 10 Ϫ3 ). A, mNGF, H75A, H84A, and H75A/H84Q in 10 mM potassium phosphate buffer, pH 6.1. B, mNGF and DEP-mNGF in 30 mM potassium phosphate buffer, pH 6.1. the observed lower receptor binding was caused exclusively by the chemical modification of His-75 and/or His-84, since cleavage of the NH 2 -terminal 9 residues to form des(1-9)NGF did not appear to affect the receptor binding or to alter DEP sensitivity. However, these experiments were conducted with a more heterogeneous preparation of NGF than used in this report, making interpretation difficult. Moreover, subsequent experiments have converged on the conclusion that the N terminus is critical for functional p140 trkA receptor binding (33)(34)(35). For example, truncation of the NH 2 -terminal 1-9 residues has now been shown to significantly reduce binding to and phosphorylation of TrkA (34). Moreover, these earlier experiments of DEP modification of NGF (38) measured binding to a microsomal membrane protein preparation from superior cervical ganglia that contains both high and low affinity binding sites (53); therefore, the interpretation of the results is less clear than in cell systems that contain only a single receptor type. In the current study we exploited both site-directed mutagenesis and chemical modification to demonstrate that at least three, and perhaps four, histidine residues play an important role in the structure, receptor binding, and activity of NGF. Our results suggest that His-75 and His-84 make modest contributions to high affinity binding, in basic agreement with Dunbar et al. (38). However, our results also indicate that His-4 and/or His-8 (or at least the N terminus) play a critical role in TrkA binding and biological activity, in agreement with the mutagenesis studies of Shih et al. (35).
Chemical Modification of Histidine-Although modification of histidine is selectively achieved with DEP, the possibility of nonspecific side reactions involving other residues such as lysine, tyrosine, and tryptophan (48) cannot be entirely excluded. All three types of residues were shown by site-directed mutagenesis to be more or less involved in the binding of NGF to p75 LANR without concomitant loss of activity (30,54,55). In the modification of NGF with DEP (38), however, no loss of absorbance at 280 nm or changes in the fluorescent properties that would indicate modification of aromatic amino acid residues, such as tryptophan or tyrosine, was observed. Furthermore, after prolonged incubation in buffer the DEP-modified NGF was fully active and bound to the receptor with full capacity (data not shown), arguing against the possible modification of lysine or tryptophan residues, which are irreversible. A similar reversal of binding inhibition was seen after treatment with hydroxylamine (38). Similarly, the excessive modification of histidines, which would result in a ring opening reaction of the imidazole side chain (56), makes the reaction irreversible. Thus, the effects of DEP most probably can be attributed to modification only of His residues.
Site-directed Mutagenesis of His-75 and His-84 -Single or double replacement of His-75 and His-84 gave a different effect on several activities of NGF. Increased hydrophobicity at either His-75 or His-84 showed no significant effect on the binding affinity for p75 LANR . Introduction of a bulkier side chain at His-84, i.e. H84I, reduced the p75 LANR binding to a significant extent. Two factors may affect the binding capability of NGF, proximity of the residue to the binding site for p75 LANR and the charge of the residue. Several other positively charged residues, Lys-32, Lys-34, Lys-95, and Glu-35, were shown to be involved in the binding of NGF to p75 LANR , and all are located near the same site on the surface of NGF (31). His-84 is about 16 -25 Å from this charged cluster and His-75 is much farther away. Replacement of His-75 with Ala destroys a hydrogen bond between Asp-72 and His-75 (26), and NGF gains two net negative charges per dimer. The H75A mutant has a significantly increased binding affinity for p75 LANR , perhaps due to loss of this hydrogen bond (Fig. 8A), loosening of the dimer interface, and subtle effects transmitted to the p75 LANR binding site.
In contrast, the replacement of His-75 or His-84 with Ala reduced the bioactivity by 4-and 10-fold and the binding affinity for the TrkA receptor by 10-and 50-fold, respectively. The decrease in binding affinity correlates well with the lowered stimulatory effect on the neurite extension in PC12 cells. In further agreement with the altered binding, activation of TrkA, as shown by autophosphorylation, was also distinctly different between mNGF and the histidine mutants at subsaturating concentrations (5 ng/ml), but not when saturating concentrations of the mutants (50 ng/ml) were used. The differential effect on the binding affinity for two classes of the receptors suggests that His-75 and His-84 are more critical for the TrkA receptor binding and only indirectly involved in p75 LANR binding. This result is consistent with conclusions from DEP mod- ification of des(1-9)NGF (38) and homolog mutational analysis of the variable region IV, containing His-84 (31). However, the His-84 side chain probably does not provide a specific "fit" to the TrkA binding site, since the substitution of His-84 by Ala has a greater effect than substitution by Gln in both neuritogenic and TrkA assays.
Interestingly, the H75A mutant had a binding affinity toward p75 LANR that was increased by 5-fold in PC12 cells and to 10-fold in A875 cells. This enhanced binding occurred despite the lack of change in the CD spectra and the slightly increased ease of dissociation to monomer. Occasional reports of increased affinity have appeared, e.g. 30 -40% increase of His-4 mutants for p75 binding (35), but not of the magnitude observed here. The exact cause of this enhanced affinity is not known, but may suggest that mutants with increased binding, as well as selectivity, can be constructed and utilized. Even though H75A or H84Q alone had a 4.8-and 1.1-fold enhanced binding affinity, respectively, the double mutant H75A/H84Q had a binding affinity for p75 LANR that was decreased to 48% compared to mNGF (Table I). Thus, an interaction appears to occur between these two His sites, resulting in a decreased p75 LANR affinity in the double mutant. However, biological activity, TrkA binding, and cold chase-stable binding for H75A/ H84Q were similar to that for H75A and H84Q (Table I), indicating the lack of a required interaction between these two residues in Trk binding.
CD spectra of these mutants indicate that a slight conformational change took place upon mutagenesis because the maximum negative ellipticity of H75A and H84A slightly shifted toward the lower wavelength region. The binding of each His mutant to the N60 antibody during purification supports the lack of a major change in conformation, since this antibody only recognizes the native conformation (44). The most significant change in CD spectra was observed in H84A, in which the maximum negative ellipticity was the largest of all and the overall shape of the spectrum was altered from that of mNGF. Thus, these two histidine residues also moderately contribute to the global conformation of NGF. This conformational change might be induced by the removal of a hydrogen bond between His-75 and Asp-72 in the same protomer (Fig. 8A) or by the elimination of an ionic (H-bond) interaction between His-84 and Asp-105 (Fig. 8B), which have been shown to form a zinc binding site in the NGF crystals (27). In H75A/H84Q with a more normal CD spectrum, these interactions might compensate for each other or, alternatively, the substitution with Gln in H75A/H84Q may allow a polar interaction that H84A cannot make. The CD spectrum of H84Q by itself was not determined. CD spectral analysis showed that little or no change in secondary structure was caused by the DEP modification.
A significant decrease in dimer stability of the His mutants (Fig. 7) could contribute to the altered properties such as decreased biological activity, increased (or decreased) p75 LANR receptor binding, decreased TrkA receptor binding, decreased receptor phosphorylation, and altered CD spectra. Because the instability of the His mutants would be amplified at the extremely diluted conditions of the bioassay, the mutational contribution could be larger than generally thought at these positions. On the other hand, DEP modification of the NH 2terminal His residues has little influence on the dimer stability, although a large effect on function, suggesting a more direct role in binding receptor.
Functional Role of His-4 and His-8 -The NH 2 -terminal residues, containing His-4 and His-8, are thought to be flexible enough to be fully solvent-accessible (26). His-4 is highly conserved in sequence across species of NGF but is varied in other members of the neurotrophin family. His-8 is conserved in NT-3 but not in BDNF or NT4/5. H75A/H84Q was quantitatively modified with the specific reagent DEP, and the effect on activity, receptor binding, and structure of NGF was measured. Both histidine residues at positions 4 and 8 are indistinguishable in their reactivity toward DEP in the H75A/H84Q mutant. Simultaneous chemical modification of His-4 and His-8 completely disrupted the activity of H75A/H84Q that remained after the site-directed mutagenesis of the His-75 and His-84 residues. However, differences were apparent in the effects on p140 trkA and on p75 LANR interactions.
The reduction of binding to p140 trkA upon DEP modification of wild type or H75A/H84Q NGF (Ͻ1%, Table I) consistently correlated with loss of neuritogenic activity and receptor activation as shown by the ligand-stimulated Trk autophosphorylation. A decrease of 4 -200-fold in TrkA activity has also been reported either from simple replacement of residues 1-8 with the corresponding residues from BDNF (31,35) or after deletion of certain residues in the 10 N-terminal residues (32,34,35). Point substitution of His-4 by alanine or by aspartate or substitution of both His-4 and His-8 by alanine decreased TrkA binding and PC12 neuritogenic activity by greater than 40-fold; however, substitution of His-8 by alanine was stated to be only 1.5-fold lower than wild type (35). Our results agree with Shih et al. (35) that His-4 and/or His-8 are important, particularly since the DEP-modified H75A/H84Q mutant had similar properties to the 1-9 deletion reported by Shih et al. (35). Thus, an important role exists for His-4/His-8 as well as for His-75/ His-84 in NGF activity with TrkA.
Several earlier results are consistent with the concept that His-4 and His-8 play a minimal role in the p75 LANR binding. Deletion or replacement of the first eight amino acids in the NH 2 terminus resulted in a greater decrease in the binding to p140 trkA than p75 LANR (30,32,35). The H4A, H4D, and H4A/ H8A mutations had little or no effect on p75 LANR binding (35). Deletion of residues 9 -13, which necessarily shifts the position of His-4 and His-8 relative to the remainder of the protein, similarly affected p140 trkA binding more than p75 LANR binding (33). Nevertheless, the chemical modification of the double mutant with DEP reported here decreased the binding of NGF to p75 LANR . The DEP reaction may not have modified binding site residues for p75 LANR but rather indirectly affected binding to p75 LANR or p140 trkA . The N-carbethoxyimidazole moiety on His-4 or His-8 may disrupt receptor interactions through steric or conformational effects more drastically than the substitution or deletion mutants. However, data supporting an intact secondary and quaternary structure of DEP-modified H75A/H84Q include the relatively small change in CD spectra upon treatment with DEP (Fig. 6B) and the lack of increased dissociation by the nonreducing SDS-PAGE method (Fig. 7). Nevertheless, we cannot rule out the possibility of a local conformational change, not detectable by our methods, that disrupts the ability of DEP-modified protein to bind to either receptor.
Conclusion-Mutagenesis of His-75 and His-84 has shown that each residue contributes about a 10-fold effect to Trk binding and activation that is not additive between the two residues. These His residues also appear to play a significant structural role in stabilizing the NGF dimer. Dual roles of residues in both structure and function could make a significant contribution to the specificity of a neurotrophin. His-4, but probably not His-8, similarly contributes to binding and activity as shown by mutagenesis (35) and by chemical modification of the H75A/H84Q double mutant (Table I), but the contribution of the N terminus to structural stability is minimal (33). NH 2 -terminal sequence and amino acid composition. We also appreciate the helpful comments of Dr. R. A. Bradshaw during preparation of this paper.