Characterization of lactogen receptor-binding site 1 of human prolactin.

Prolactin (PRL) binds to two molecules of PRL receptor (PRLR) through two regions referred to as binding sites 1 and 2. Although binding site 1 has been generally assigned to the pocket delimited by helix 1, helix 4, and the second half of loop 1, the residues involved in receptor binding have not yet all been precisely identified. In an earlier alanine-scanning mutational study, we identified three major binding determinants in loop 1 of human PRL (hPRL) (Goffin, V., Norman, M. & Martial, J. A.(1992) Mol. Endocrinol. 6, 1381-1392). Here we focus on the two other regions that form binding site 1, namely helices 1 and 4. Putative binding residues, selected on the basis of a three-dimensional model of hPRL constructed in this laboratory, were mutated to alanine, and recombinant hPRL mutants produced in Escherichia coli were tested for their ability to bind to the PRLR and to stimulate Nb2 cell proliferation. We thus identified nine single mutations (three in helix 1 and six in helix 4) whose effect was to reduce both binding and mitogenic activity by more than half as compared with wild-type hPRL, indicating the functional involvement of the corresponding residues. Adding these to the three binding determinants identified in loop 1, we now propose a complete picture of PRLR-binding site 1 of hPRL. As we earlier hypothesized, the binding site 1 determinants of hPRL differ from those of human growth hormone, a hPRL homolog.

Prolactin (PRL) 1 and growth hormone (GH) are homologous hormones primarily secreted by the pituitary gland in all vertebrates (for reviews, see Refs. 1 and 2). PRL is involved in a wide variety of biological functions, mainly related to reproduction, lactation, osmoregulation, and immunomodulation (reviewed in Ref. 3), while GH is involved primarily in growth and morphogenesis (4). The multiple bioactivities of these hormones are mediated by homologous membrane receptors, the prolactin or lactogen receptor (PRLR) and the growth hormone or somatogen receptor (GHR) (for reviews, see Refs. 5 and 6).
Both receptors belong to class I of the newly described cytokine receptor superfamily (7)(8)(9). These receptors are all activated by clustering of two or more membrane subunits (for reviews, see Refs. 10 -13). On the one hand, receptor activation can result from the hetero-oligomerization of different subunits, such as a ligand-specific binding subunit (the ␣-chain) and a common signal transducer subunit (the ␤-chain). On the other hand, activation of some receptors can also result from the homodimerization of two identical binding components. This has been reported for the receptors of erythropoietin (14,15), granulocyte colony-stimulating factor (16), GH (17,18), and PRL (19 -22).
The mechanism of activation of the human (h) GHR by hGH has been extensively studied, and a sequential dimerization model was proposed by Wells and co-workers in 1992 (17). According to their model of activation, the hormone first binds to its receptor through a set of amino acids forming the socalled binding site 1. The complex composed of one molecule of receptor and one molecule of hormone (H1⅐R1) remains inactive until it associates with a second (and identical) receptor molecule to yield an active H1⅐R2 complex. Receptor clustering leads to interactions between both receptors as well as between the hormone and the second receptor molecule (18). Binding of the two receptor molecules is thus sequential, and the set of amino acids of hGH interacting with the second hGHR molecule is called binding site 2. Interestingly, a sequential two-site model has also been hypothesized for interleukin-4, another fourhelix bundle cytokine (25).
Mutational (23,24) and crystallographic (18) studies of hGH have led to the identification of the amino acids belonging to both binding sites. Binding site 1 of hGH involves residues of helices 1 and 4 and of the second half of the long loop (loop 1) joining helices 1 and 2. On the other hand, binding site 2 is formed by residues belonging to the facing sides of helices 1 and 3 and a few residues in the small N-terminal loop. Due to the numerous structural and functional similarities between the PRL-PRLR and GH-GHR systems, we hypothesized earlier that the sequential receptor dimerization model described for the hGHR might also apply to activation of the PRLR (21,26,27). In agreement with this assumption, we showed that steric hindrance introduced in the helix 1/helix 4/loop 1 pocket (binding site 1) (28) or in the helix 1-helix 3 interface (binding site 2) of hPRL (21) is detrimental to activity. While these studies clearly indicate the general location of both binding sites on hPRL, not all residues involved in receptor binding have been identified. With respect to binding site 1, segment 58 -74 (loop 1) has been characterized through systematic alanine-scanning mutagenesis (26). To the best of our knowledge, however, no systematic mutational study has yet focused on helices 1 and 4, so these helical segments, strongly suspected of containing several residues critical for tight receptor binding, remain essentially uncharacterized. To date, Arg-177 is the only amino acid within these helical segments to have been unambiguously identified as very important for the mitogenic activity of bovine PRL (bPRL) toward Nb2 cells (29).
Growth hormones, and presumably PRL, are composed of four ␣-helices and adopt the "four-helix bundle" fold (18,27,30,31). In these proteins, the hydrophobic faces of amphiphilic helices form the hydrophobic core (18,27,30). Therefore, mutational analysis within ␣-helices must be conducted with caution since any mutation affecting the hydrophobic core can alter the global folding pattern, as reported for some bPRL mutants (32). To date, no crystallographic structure has been reported for PRL; this has prevented any structure-based mutational study. Therefore, we have recently developed a threedimensional model of hPRL (27), constructed on the basis of the crystallographic coordinates of porcine GH, the first elucidated structure for a protein of the PRL/GH family (30). On the basis of these data, we selected 7 residues in helix 1 and 10 residues in helix 4 whose side chain orientations were compatible with an involvement in the pocket of binding site 1 (see Fig. 1). Although meeting this criterion, Arg-177 was not considered again in the present study since its importance has been demonstrated by others (29). The 16 remaining amino acids were individually mutated to alanine, and the effect of each mutation was examined by measuring the binding and mitogenic activity of the hormone mutants. In agreement with our hypothesis, both helical regions contain several residues that are required for the hormone's biological potency. Linking the present study with our previous analysis of loop 1 (26) and with structural data now available for hPRL (27), we can provide a complete picture of receptor-binding site 1 of hPRL.

Materials
Restriction enzymes and DNA ligase were purchased from Boehringer Mannheim (Mannheim, Germany), Amersham International (Buckinghamshire, United Kingdom), Life Technologies, Inc., and Eurogentec (Seraing, Belgium). IODO-GEN was purchased from Sigma, and carrier-free Na 125 I was obtained from Amersham International. Ampholytes (pH range of 5-7) and pI protein markers were from Pharmacia (Uppsala). Oligonucleotides were from Eurogentec. Culture media and sera were purchased from Life Technologies, Inc.

Oligonucleotide-directed Mutagenesis
All mutated hPRL cDNAs (33) were constructed as previously reported (21,26) by the oligonucleotide-directed mutagenesis method of Sayers et al. (34). The vector used was single-stranded M13. We used the oligonucleotide-directed mutagenesis system of Boehringer Mannheim and strictly followed the manufacturer's instructions. Clones containing the expected mutation were identified by DNA sequencing; the mutated cDNAs were digested with HindIII and NdeI (helix 1 mutants) or with HindIII and BglII (helix 4 mutants); and isolated fragments (661 and 222 base pairs, respectively) were reinserted into the pT7L expression vector (35). The sequences of the mutated oligonucleotides are as follows (5Ј 3 3Ј-noncoding strands; mutated codons are underlined).

Production and Purification of Proteins
Recombinant native hPRL and hPRL mutants were overproduced in 500-ml cultures of Escherichia coli BL21(DE3) cells and purified as described previously (35). Purity was assessed by SDS-polyacrylamide gel electrophoresis according to Laemmli (37).

Quantification of Proteins
Proteins were quantified physically by weighing the lyophilized powder on a precision balance (Electrobalance, Cahn 26) and chemically by the Bradford method (36). The disparity between weight and chemical measurements never exceeded 20%.

Isoelectrofocusing
The isoelectric point of the hPRL mutants was estimated by isoelectrofocusing (pH range of 5-7) as described previously (21).

Structural Analyses
Circular Dichroism-Lyophilized proteins were resuspended in 50 mM NH 4 HCO 3 , pH 8, at a concentration ranging from 300 to 500 g/ml. Spectra were recorded with a CD6 dichrograph (Instruments SA-JO-BIN YVON, Longjumeau, France) linked to a personal computer for data recording and analysis (dichrograph software, Instruments SA-JOBIN YVON). For each protein, four spectra recorded between 195 and 260 nm were averaged. Measurements were performed in a 0.1-cm path length quartz cell. The helicity was calculated at 222 nm according to Chen et al. (38).
Apparent Molecular Mass-The apparent molecular masses of all the hPRL mutants were measured by high pressure liquid-gel filtration chromatography. 100-l samples (500 g/ml) were loaded on a Superose 12 molecular sieve (Pharmacia) equilibrated in 20 mM Tris-HCl, pH 8, 100 mM NaCl. Elution was carried out in the same buffer at a constant flow rate of 0.5 ml/min, and protein elution was monitored at 280 nm. The column was calibrated with several molecular mass markers: dextran blue (void volume), bovine serum albumin dimers (136 kDa), bovine serum albumin (68 kDa), ovalbumin (45 kDa), carbonic anhydrase (30 kDa), and myoglobin (17.5 kDa).

Nb2 Cell Culture and in Vitro Bioassay
The bioactivity of the hPRL mutants was estimated by their ability to stimulate growth of lactogen-dependent Nb2 lymphoma cells (39). The procedure used (40) has been previously detailed (21,26). Briefly, cells were cultured in Fisher's medium containing 10% horse serum and 10% fetal calf serum. 24 h before the bioassay, the cells were synchronized in culture medium containing only 1% fetal calf serum. Bioassays were performed in fetal calf serum-free Fisher's medium (starvation medium). Various amounts of hPRL samples diluted in starvation medium (from 25 to 100 l) were added to 2.5 ml of cells (1-2 ϫ 10 5 cells/ml) plated in 6-well Falcon plates. Nb2 cells were counted with a Coulter counter (Coulter Electronics Ltd., Harpenden, Hertfordshire, United Kingdom) after 3 days. Two to four experiments were performed in duplicate for each mutant. The ED 50 value (amount of hormone needed to achieve half-maximal cell growth) was calculated, and the relative bioactivity of each mutant with respect to native hPRL was estimated as the ratio of the native versus mutant ED 50 values.

Binding Experiments
Binding of hPRL mutants to the lactogen receptor was performed as reported earlier (21,26,28). Briefly, homogenates from 3 ϫ 10 6 Nb2 cells were incubated for 16 h at 25°C with 30,000 -50,000 cpm 125 I-hPRL in the presence of increasing amounts of unlabeled native hPRL or hPRL analog (final reaction volume of 0.5 ml). The assay was terminated by addition of 0.5 ml of ice-cold buffer (0.025 M Tris-HCl, 0.01 M MgCl 2 , pH 7.5) followed by centrifugation (5 min, 11,000 ϫ g). The supernatants were removed carefully, and the radioactivity of the pellets was analyzed in a ␥-counter (Hybritech 002011B).
Each mutant was tested at least three times in duplicate. Specific binding was calculated as the difference between radioactivity bound in the absence (B 0 , maximal binding) and in the presence (nonspecific binding) of 2 g of unlabeled native hPRL. In the different experiments, nonspecific binding never exceeded 20% of maximal binding. Data are presented as percentages of specific binding. Competition curves were analyzed with the LIGAND PC program (41). The relative binding affinity of each mutant was estimated as the ratio of the native versus mutant IC 50 values.

Structure-based Design of the Mutational Study
Residues to be mutated were selected on the basis of a three-dimensional model of hPRL constructed in our laboratory (27), which is to date the only atomic structure available for any PRL. Selection was based on two criteria. First, residues had to be located on the exposed (hydrophilic) faces of helices 1 and 4. Second, only residues whose side chain orientations were compatible with an involvement in the predicted binding site 1 were selected. We thus did not consider residues such as the helix 1 residues pointing toward the helix 1-helix 3 interface, which forms binding site 2 (21). From this structure-based prediction of binding determinants, a set of 17 residues was selected: for helix 1, Arg-16, Val-23, Ser-26, His-30, Ser-34, Phe-37, and Ser-38; and for helix 4, Tyr-169, His-173, Arg-176, Arg-177, His-180, Lys-181, Asn-184, Tyr-185, Leu-188, and Arg-192.
To assess the functional involvement of each residue, we used the alanine-scanning approach since this strategy appears to be the most appropriate for identifying binding residues in both hGH and hPRL (24,26,42). Arg-177, previously reported to be extremely important for bPRL mitogenic activity (29), was not retested. To evaluate the cumulative effect of the most effective mutations in helix 1, a triple mutant was constructed (V23A/H30A/F37A). Finally, Lys-181 was also mutated to Glu in order to evaluate the effect of an opposite charge at this position. The hPRL mutants are designated by a letter referring to the mutated residue, followed by a number referring to its position, followed by a letter referring to the substitute residue (i.e. R16A designates the analog in which Arg-16 is replaced with Ala). The 16 preselected residues are shown in Fig. 1.

Production and Purification Yields
The yield of overproduced protein was about the same for all 16 hPRL analogs as for native hPRL (Ϯ150 mg/liter) (35). Routinely, ϳ30 mg of purified monomeric hPRL can be recovered per liter of culture. Similar amounts of monomer were recovered for all the hPRL mutants, which indicates a behavior similar to that of native hPRL during renaturation.

Structural Characterization of hPRL Mutants
Since all residue changes reported in this study affect regular secondary structures (␣-helices), each mutant was first structurally characterized by circular dichroism and chromatography on a molecular sieve to assess its proper folding. The isoelectric point was also determined for each mutant.
Isoelectric Point-The major isoform of purified recombinant hPRL exhibits a pI of 6.2 (21,35). Introduction or removal of charged residues alters the net protein charge. Accordingly, the pI values of hPRL mutants R16A, R176A, K181A, K181E, and R192A were 0.2-0.3 units lower than normal (data not shown), thus confirming at protein level the presence of the mutations.
Circular Dichroism-hPRL has an ␣-helix content of 45 Ϯ 5% (21,26,28) as determined by circular dichroism. In agreement with the spectrum of native hPRL, all mutants produced in this study displayed the typical curve of all ␣-proteins, with two minima at 222 and 208 nm and a maximum at 195 nm. The calculated helicities are reported in Table I. They are all in the range of 45 Ϯ 5%, suggesting no significant alteration of the overall secondary structure content.
Apparent Molecular Mass-The apparent molecular mass of each mutant was estimated from its retention time on a high pressure molecular sieve. The calculated apparent molecular masses are reported in Table I. In agreement with the theoretical molecular mass of hPRL (23 kDa), the estimated molecular masses of all the mutants are in the range of 22 Ϯ 2 kDa; they correlate with the data obtained by CD analysis.  (50)). Putative binding determinants were selected on the basis of two criteria. First, residues located on the exposed faces of the helices 1 and 4 were listed. Then, among this set of amino acids, we considered only those whose side chain orientations were compatible with an involvement in the binding site 1 pocket. The side chains of the 16 residues that were studied by mutagenesis are colored in green (helix 1) and blue (helix 4).

TABLE I Structural analysis of hPRL mutants
The helical content of each alanine substitution mutant was measured by circular dichroism and calculated at 222 nm according to Chen et al. (38). The apparent molecular mass of each mutant was estimated by gel filtration as described under "Experimental Procedures." The calculated apparent molecular masses are presented. Since no significant alteration of the global structure was detected by either procedure, we conclude that any alteration of the biological properties (see below) reflects the functional involvement of the mutated residue rather than an unexpected effect of the mutation on protein folding.

Biological Analysis of hPRL Mutants
We have previously reported that recombinant native hPRL stimulates Nb2 cells as efficiently as pituitary-purified hPRL, with half-maximal growth at ϳ200 pg of hPRL/ml (ED 50 ) (26). Therefore, recombinant wild-type hPRL was used as a reference for estimating the bioactivity of all the hPRL mutants. Nb2 cells contain ϳ12,000 PRL receptors/cell (43). As described earlier (21,26,28), Nb2 cells were also used in binding assays of hPRL mutants. Typical Nb2 cell proliferation and binding curves are represented in Fig. 2. Binding and cell proliferation data are summarized in Fig. 3.

Structure-based Prediction of Putative Binding
Residues-A few years ago, Luck et al. (29,32,44) reported the effects of point mutations on the mitogenic activity of bPRL. Since no structure was available for any PRL at that time, the residues to be mutated were selected mainly on the basis of sequence comparisons between members of the PRL/GH family. In many cases, point mutations either proved ineffective (44) or were assumed to affect biological properties solely as a result of altered global protein folding (32). Consequently, only a very few residues, such as Arg-177, could be clearly identified as functionally required for bPRL bioactivity (29).
To circumvent the lack of a three-dimensional structure of PRL, we have recently constructed a three-dimensional model of hPRL (27) based on the crystallographic structure of porcine GH, the first structure of a member of the PRL/GH family that has been determined experimentally (30). Thanks to this model, we were able to formulate hypotheses concerning the interaction between PRL and its receptor, notably with regard to the location of both binding sites and to the residues that form them (27).
Analysis of sequence-structure-function relationships in PRL led us to propose that binding site 1 involves the pocket delimited by helix 1, helix 4, and loop 1 (26 -28). In agreement, mutational analysis of loop 1 clearly demonstrated the involvement of this region in bioactivity (26). To confirm our hypothesis, we decided to further characterize binding site 1 by scanning the two remaining regions, namely helices 1 and 4. As reported earlier, helix 1 is involved in both binding sites since residues facing helix 3 are involved in binding site 2, whereas residues facing helix 4 are predicted to be part of binding site 1 FIG. 2. A, competition curves for the displacement of 125 I-labeled native hPRL by unlabeled native hormone and the K181A mutant. The figure represents a typical experiment, in which nonspecific binding was 22% of B 0 . The curves presented in this figure are taken from the same experiment and are presented as percentages of specific binding. Each point is the average of duplicate measurements; maximal disparity between duplicate values is 7% of specific binding. All the hPRL mutants were tested in at least three independent experiments. B, mitogenic activity toward Nb2 cells of native hPRL and the K181A mutant. The relative mitogenic potency of each mutant was estimated as the amount of native versus mutant hPRL required to produce half-maximal proliferation of Nb2 cells (ED 50 ). Each mutant was tested at least three times in duplicate. A typical experiment is presented. (21,27). Therefore, we chose in this study to mutate only those residues that point toward the binding site 1 pocket. Helix 4, on the other hand, is at the center of binding site 1; there were more residues in this region (10 residues) to be investigated in order to assess their involvement in the protein's biological properties.
Structural and Biological Analysis of hPRL Mutants-Because all the selected residues are predicted to be located on exposed faces of helices, they should not affect protein folding. Accordingly, our structural analyses of the various mutants failed to detect any significant alteration of the global protein conformation. Although CD analysis and estimation of the apparent molecular masses are probably not sufficiently sensitive methods for detecting subtle and local structural changes, it should be stressed that by combining these methods, we have previously been able to identify misfolded hPRL mutants and to eliminate them from our study (21,26). Moreover, we know of no reported structurally disruptive alanine substitution of any residue located on an exposed face of PRL or GH, with the sole exception of the cysteines involved in disulfide bonds (21,24,26,42).
This study confirms the assumed involvement of both helices 1 and 4 in binding site 1 of hPRL. As previously observed for hGH (Refs. 24, 45, and 46; for review, see Ref. 47), binding site 1 is centered on helix 4 since this segment contains not only the greatest number of binding determinants, but also those whose replacement with alanine is the most detrimental to both binding and mitogenic activity. The most effective mutations in helix 1 (V23A and F37A) cause only a 5-fold reduction of the biological activity, compared with the 100-fold decrease in FIG. 3. Bioactivity and affinity for the PRLR of the different hPRL mutants. The mitogenic activity and the affinity for the Nb2 receptor were determined as described under "Experimental Procedures" and shown in Fig. 2 (A and B). The values corresponding to wild-type hPRL are arbitrary set at 100%. A and B show mutations performed in helices 1 and 4, respectively. In each panel, the hPRL mutants are shown in three groups: first, those that have not significantly affected biological properties; second, those that have slightly decreased potency, although mutated residues cannot be considered as strong binding determinants; and third, mutants for which both binding and mitogenic activity are significantly altered, indicating that mutated residues are involved in receptor binding.
binding when Lys-181 is replaced with alanine. Even when the three most effective mutations in helix 1 are combined (V23A, H30A, and F37A), the mitogenic activity is diminished only 6-fold; this suggests a limited involvement of this helical segment in the interaction with the receptor. Considering that a Ͼ2-fold reduction of both binding and mitogenic activity reflects a significant functional involvement of an amino acid, we have identified 12 residues (referred to as "binding determinants") in receptor-binding site 1 of hPRL (Figs. 4 and 5): Val-23, His-30, and Phe-37 in helix 1 (this work); His-59, Pro-66, and Lys-69 in loop 1 (26); and Tyr-169, His-173, Arg-176, His-180, Lys-181, and Tyr-185 in helix 4 (this work). Furthermore, Luck et al. (29), using the same bioassay, found that mutating Arg-177 in bPRL drastically alters the bioactivity of the hormone (reducing it to 1.1% of the reference value). Since an Arg residue is found at this position in all PRLs, it is most likely that this residue is also a major binding determinant of hPRL.
Some hPRL mutants, such as S26A, S34A, S38A, and L188A, display lesser binding, but normal to slightly altered mitogenic activity. This might reflect the "spare receptor" phenomenon, in which maximal biological activity occurs at submaximal receptor occupancy. In the Nb2 system, maximal cell growth has been reported to occur at 35% of maximal binding (42). Alanine substitution of Lys-187 is reported to halve the mitogenic activity of bPRL, but since this position remained almost insensitive to other mutations (replacement with Leu, Asn, or Arg) (29), it seems unlikely that Lys-187 is a major determinant of receptor binding. Finally, Luck et al. (29,32) reported that mutation of Arg-21 or Tyr-28 to various amino acids reduces the effect on Nb2 cells by a factor of 2-5. Our threedimensional structural model of hPRL (27) suggests that the side chains of both these residues point outside binding site 1; we anticipate that they belong to binding site 2, which involves the facing sides of helices 1 and 3 (21,27). For these various reasons, we do not consider any of the residues just mentioned to belong to binding site 1. Fig. 4 shows the spatial distribution of the 13 determinants of binding site 1 of hPRL. Although all potential binding determinants were selected on the basis of our three-dimensional model, we cannot rule out the involvement of other residues not tested in this study, although it does appear unlikely.
Binding site 1 of hPRL contains both hydrophobic (Phe-37, Tyr-169, and Tyr-185) and hydrophilic (Lys-69, Arg-176, Arg-177, and Lys-181) residues. The importance of charged residues in hormone-receptor binding has been emphasized in a comprehensive energetic study of cytokine-receptor interactions, showing complementary electrostatic potentials on the binding surfaces of the two interacting proteins (31). In the case of hPRL, we have demonstrated the critical role of charged residues by replacing Lys-181, the strongest binding determinant of hPRL, with a negatively charged glutamic acid residue. The K181E mutant was biologically 10 times less active than the K181A mutant (the former displayed only 0.3% of the activity of wild-type hPRL), suggesting that introducing a negative charge where a positive one is required is detrimental to efficient receptor docking. Luck (27) is shown with binding site 1 facing the viewer (see Fig. 1 for details). The side chains of the residues identified as strong determinants of hPRL binding to the PRLR are colored in green (helix 1), blue (helix 4), and red (loop 1). The side chain of Arg-177 (29) is also represented.

FIG. 5. Distribution of the determinants on hPRL for binding to the PRLR and on hGH for binding to the PRLR and GHR.
Amino acid sequences forming binding sites 1 of hPRL and hGH are aligned. Numbers above and below the sequences correspond to hPRL and hGH, respectively. The first line represents the binding determinants on hPRL for binding to the PRLR (Ref. 26 and this study). Identification of Arg-177 as an important residue is from Luck et al. (29). The second line refers to binding determinants on hGH for binding to the PRLR, identified by Cunningham and Wells (42) using the hPRLbp; and the third line represents the determinants on hGH for binding to the hGHR, identified by means of the hGHbp (24). that replacing Arg-177 with an alanine or a glutamate also decreases bPRL bioactivity to 1.1 and 0.3% of the reference value, respectively. Finally, Clarckson and Wells (46) proposed, on the basis of an energy analysis of the hGH-hGHbp interface, that electrostatic interactions might contribute to determining the binding specificity; this is also possible in the case of PRL (see below).
Comparison of Binding Sites 1 of hPRL and hGH-It is usually assumed that homologous proteins exert a common activity through identical or very similar mechanisms. In the present context, hPRL and hGH might thus be expected to bind by the same mechanism to the lactogen receptor. The available data indicate otherwise. First, whereas tight binding of hGH to the hPRLbp requires mediation by a zinc ion, hPRL binding to the hPRLbp is zinc-independent (48). Second, as shown in Fig.  5, hGH and hPRL clearly appear to bind to the PRLR via different sets of amino acids (Refs. 26 and 42 and this work). For example, Lys-69, Tyr-169, and His-180 play a major role in hPRL binding, whereas their hGH counterparts (Arg-64, Tyr-160, and Asp171) can be mutated without significantly affecting binding to the hPRLbp (Table II). In contrast, Ile-58, Ser-62, Glu-65, and Arg-183 are required in hGH, while their hPRL counterparts (Leu-63, Glu-67, Glu-70, and Arg-192) can be mutated without compromising receptor binding. Even when topologically equivalent residues are binding determinants for both hormones, they do not appear to be equally important (Table II). One of the rare similarities between binding of hGH and hPRL to the PRLR is the involvement of two basic residues: Arg-176 and Arg-177 in hPRL and their topological equivalents (Arg-167 and Lys-168) in hGH (see Fig. 5). In hGH, both residues are considered specificity determinants, meaning that they are crucial to hGH binding to the PRLR, but not to its binding to the GHR (42,47,49). As Arg-176 and Arg-177 in hPRL are also among the strongest binding determinants (Table II), one would expect these residues to be a characteristic requirement for PRLR binding, in agreement with the proposed role of charged residues in determining the binding specificity (46). The disruptive effect of mutating Arg-167 (in hGH), however, has been partly linked to an indirect alteration of the zinc-binding pocket conformation (Table II) (42), so the role of this arginine is likely to differ from that of Arg-176 (in hPRL) since receptor binding to PRL is zinc-independent (48). The same applies to His-30 in hPRL and its counterpart (His-21) in hGH since the former is a real binding determinant in hPRL, while the latter is involved only in zinc chelation in hGH (42,49).
Among the features common to the interaction of hGH with the hGHbp (18) and with the hPRLbp (49) are the major contacts involving two Trp residues found in both receptors (Trp-104 and Trp-169 in the hGHR and Trp-74 and Trp-139 in the hPRLR (Refs. 18 and 49; for a review, see Ref. 47). In the hGH-hGHR interaction, these Trp residues are buried in a hydrophobic environment formed by the alkyl portions of Lys-172 and Thr-175 surrounded by Asp-171 and Phe-176 (46), in keeping with earlier findings that these residues are among those accounting for the majority of the free energy of the hormone-receptor interaction (45). In hPRL, the topologically equivalent amino acids are Lys-181, Asn-184, His-180, and Tyr-185. With the exception of Asn-184, these residues are also strong binding determinants in hPRL (Table II), suggesting analogous interactions of helix 4 amino acids with Trp-74 and Trp-139 of the PRLR. As these two Trp residues are not found in the other cytokine receptors, it is likely that the network of hormone-receptor contacts involving these Trp residues is a characteristic feature of the interactions between PRL/GH hormones and their receptors.
Altogether, this study thus confirms our earlier hypothesis that hPRL and hGH bind to the PRLR via mechanisms with different requirements at the molecular/residue level (26,27). In contrast, structural analysis of the binding sites (27) led us to propose that there is a closer parallel between the mechanisms by which hPRL and hGH bind to their respective receptors. These interactions do indeed share some common features, such as the non-involvement of zinc ions (48) and a similar distribution of strong binding determinants (for example, His-59, Lys-69, His-180, and Tyr-185 in hPRL and their respective equivalents (Phe-54, Arg-64, Asp-171, and Phe-176) in hGH) (see Table II).
This work completes our picture of lactogen receptor-binding site 1 of hPRL. Combined with our previous work on binding site 2 (21), it provides a global view of the interaction between this important lactogenic hormone and its receptor.

TABLE II
Relative importance of binding determinants identified in hGH and hPRL The binding affinities of mutants in which the binding determinants were mutated to alanine are expressed as percentages of the affinity of the wild-type hormone. The first and last columns correspond to the aligned sequences of hGH and hPRL, respectively (see Fig. 5). All binding determinants for the three interactions hGH-hGHbp (second column; data from Cunningham and Wells (24)), hGH-hPRLbp (third column; data from Ref. 42), and hPRL-PRLR Nb2 (fourth column; data from Goffin et al. (26) and this study) are indicated. When no value is available for hPRL, results obtained for the bPRL mutants in the Nb2 proliferation assay are indicated and marked with an asterisk (29,32,44). The hGH-hPRLbp interaction is mediated by zinc chelation. In the third column, values in parentheses indicate the effects of mutations in the presence of EDTA, i.e. in the absence of zinc chelation (42). It appears that the effects of mutating His-18, His-21, Glu-174, and, to a lesser extent, Glu-167 are due mainly to alteration of zinc binding and/or pocket shape. Although binding of the L188A hPRL mutant is reduced by 33% (fourth column), this residue is not considered a strong determinant since its mitogenic activity was reduced to a much lesser extent (see text).