Two Independent Histidines, One in Human Prolactin and One in Its Receptor, Are Critical for pH-dependent Receptor Recognition and Activation*

Human prolactin (hPRL), a member of the family of hematopoietic cytokines, functions as both an endocrine hormone and autocrine/paracrine growth factor. We have previously demonstrated that recognition of the hPRL·receptor depends strongly on solution acidity over the physiologic range from pH 6 to pH 8. The hPRL·receptor binding interface contains four histidines whose protonation is hypothesized to regulate pH-dependent receptor recognition. Here, we systematically dissect its molecular origin by characterizing the consequences of His to Ala mutations on pH-dependent receptor binding kinetics, site-specific histidine protonation, and high resolution structures of the intermolecular interface. Thermodynamic modeling of the pH dependence to receptor binding affinity reveals large changes in site-specific protonation constants for a majority of interface histidines upon complexation. Removal of individual His imidazoles reduces these perturbations in protonation constants, which is most likely explained by the introduction of solvent-filled, buried cavities in the crystallographic structures without inducing significant conformational rearrangements.

lular studies of hPRL signaling provide a molecular basis for the hPRL role in breast cancer proliferation (6 -10). Recent epidemiological studies have strengthened the link between elevated hPRL and breast cancer risk (10). Consequently, detailed molecular understanding of hPRL⅐receptor recognition and activation is critical to aid efforts to develop potential cancer intervention strategies.
hPRL recognizes the hPRLr extracellular domain (ECD) and recruits two molecules of hPRLr via two distinct binding sites on the hPRL surface, i.e. site 1 (high affinity) followed by site 2 (low affinity) (11). Crystallographic structures of 1:1 (12) (hPRL antagonist⅐hPRLrECD) and 1:2 (13) (so called "affinity matured" hPRL⅐hPRLrECD) complexes reveal molecular details of the ϳ20% of the combined surface area buried within the high affinity binding interface. Cumulative work with mutagenic variants of hPRL further reduces the number of critical residue contacts for receptor binding (14 -16). Associated with formation of the heterotrimeric 1:2 (ligand⅐receptor) complex, induced conformational changes in the receptor intracellular domain brings two receptor-associated cytosolic tyrosine kinase molecules of Janus kinase-2 (JAK2) in close proximity. Sequential phosphorylation of specific Tyr residues of JAK2 and hPRLr results in recruitment and phosphorylation of Signal Transducer and Activator of Transcription-5 (STAT5) (17)(18)(19). Various other cellular signal transduction pathways including MEK-ERK, Phosphoinositide 3-AKT, and AP1 are also activated in response to hPRLr activation (20 -22). Although cross-talk between pathways is thought to determine the eventual phenotypic response of target cells, phosphorylation of STAT5 is a hallmark of hPRLr activation.
We have previously reported a dramatic difference in the pH dependence of hPRLr recognition by hPRL and hGH and believe this has important functional effects upon endocytosis. The affinity of hPRL for hPRLrECD decreases 500-fold as the pH decreases from pH 8 to 6, whereas the binding of hGH is unchanged over this same pH range (23). A triad of interacting histidines, His-27, His-30, and His-180, occupy the high affinity receptor binding site of hPRL (24), with His-188 contributed by hPRLr. As the overall charge of the histidine imidazole ring varies over the physiological pH range from 5 to 8, we hypothesized that pH-dependent hPRLr recognition is a function of coordinated protonation of these specific His residues. Using various biophysical and cellular methods, we show that His-180 in hPRL and His-188 of hPRLr are critical for pH-dependent receptor binding and activation, with His-30 playing an additional role. This is the first report to systematically dissect the structural and functional roles of specific histidines within the intermolecular interface of a hematopoietic cytokine-receptor pair.

Plasmids, Recombinant Protein Purification, and Cell
Culture-A modified vector, pT7L, was used to express and purify WT and site-directed mutants of hPRL from the BL21(DE3) strain of Escherichia coli as described previously (24). The hPRL antagonist (hPRL ⌬1-14 G129R) and site-directed mutants of the antagonist were also purified using the same protocol. Purification of the WT hPRLrECD expressed from pET11b (Invitrogen) in BL21(DE3) E. coli has also been previously described (25). The site-directed mutant of hPRLr-ECD, H188A was purified using the same protocol. The parent vector encoding hPRLr (pCDNA3-hPRLr) was a generous gift from Dr. Linda Schuler (University of Wisconsin at Madison). The mammalian expression vector, pEF1 V5-His A-hPRLr, was generated by subcloning (GenScript, Piscataway, NJ), the cDNA encoding full-length hPRLr. All site-directed mutants (including those of the hPRL antagonist and hPRLr), except H27A, H30A, and H180A of hPRL (obtained from Dr. Charles Clevenger), were generated by PCR mutagenesis using the QuikChange site-directed mutagenesis kit (Agilent technologies, La Jolla, CA) according to the manufacturer's protocol.
T47D human breast cancer cells were obtained from the Yale Cancer Center Tissue Culture Core Facility, and the CHO-K1 (ATCC# CCL-61) cell line was obtained from American Type Culture Collection (ATCC, Manassas, VA). T47D and CHO-K1 cells were maintained in RPMI 1640 and Kaighn's modified F-12K media (Invitrogen) respectively, both supplemented with 10% heat-inactivated fetal bovine serum (Invitrogen), 500 units/ml penicillin, 500 g/ml streptomycin (Invitrogen) and incubated at 37°C and 5% CO 2 . Transfection of CHO-K1 cells was carried out using Superfect transfection reagent according to the manufacturer's protocol (Qiagen, Valencia, CA).
Measurement of Receptor Binding Kinetics for hPRL by Surface Plasmon Resonance (SPR)-SPR experiments were done on a Biacore T100 (Biacore AB, Uppsala, Sweden). The WT and mutant forms of hPRL were immobilized on a GLM sensor chip using standard amine coupling. Surfaces were activated for 7 min with sulfo-NHS/EDC, and the hPRL proteins were coupled using a concentration of 3 M in 10 mM sodium acetate, pH 5.0, for 1 min. Surfaces were then blocked with a 3-min injection of ethanolamine. Receptor-ECD was tested at a high concentration of 1 M followed by a 3-fold dilution series. Injection of ECD at each concentration was repeated twice. Surfaces were regenerated with a 20-s pulse of 1/1000 dilution of phosphoric acid. The running buffer contained 25 mM KPO 4 , 50 mM NaCl, 0.1 mg/ml BSA, and 0.005% P20. The samples were tested from low pH (5.75) to high pH (8.0) in quarter-unit increments. All of the response data were globally fit to a 1:1 interaction model to extract the rate constants at 25°C using the software program Scrubber-2 (BioLogic Software Pty Ltd.).
X-ray Crystallography-Purified recombinant proteins were mixed in a 1:1 molar ratio at a final concentration of ϳ9.5 mg/ml and incubated at room temperature for 15 min for complex formation followed by buffer exchanging into 2 mM ammonium bicarbonate, pH 7.9. Complexes were crystallized by the hanging drop method using a Mosquito liquid handling robot (TTP Labtech, Cambridge, MA) to dispense various ratios of protein:precipitant (3.7 M NaCl and 0.1 M Hepes buffer, pH 7.5, 3-6% polyethylene glycol monomethyl ether 550). Hexagonal crystals were flash-frozen in liquid nitrogen and transferred to the Ϫ180°C cold stream for data collection. Diffraction data were collected at National Synchrotron Light Source at the Brookhaven National Laboratories (Upton, NY) on beamline X29A using an ADSC Quantum 315 detector. Data were processed and scaled in HKL2000 (26). Molecular replacement was performed in Phaser (27) using the published structure PDB code 3D48. Phenix.refine (28) was used for further refinement followed by multiple rounds of adjustment and model building using Coot (29). Manual refinement was done using composite omit maps from CNS (30). Final refinement of all structures was done using restrained refinement in refmac (31). Table 2 presents statistical data obtained from all crystal structures.
NMR Spectroscopy-Preparation of isotopically labeled proteins, collection and processing of NMR spectroscopic data, and pK a calculations for individual histidine residues have been described previously (24,25).
Modeling-Thermodynamic modeling of the pH dependence to hPRL⅐receptor binding affinity derived from the SPR pH-dependent hPRLr Binding Requires hPRL His-180 and hPRLr His-188 DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 results was performed using MATLAB, Version R2010a (The Mathworks Inc., Natick, MA), with custom scripts and functions written in-house (provided as supplemental data). The "apparent" receptor binding affinity, K D,app ϭ 1/K a,app , is expressed as a function of a theoretical "basic state" association constant (with all histidines unprotonated), K a,basic , and the partition functions (or binding polynomials) for protonation of all histidines in both the free (Z H,PRL and Z H,ECD ) and bound (Z H,bound ) states, K a,app ϭ K a,basic ͑Z H,bound /(Z H,PRL ⅐Z H,ECD )͒. For n-independently titrated histidines, the partition function for protonation takes the form, where K 1 … K n represents the site-specific proton-binding association constants (K ϭ 10 pKa ), and H is the molar proton concentration (H ϭ 10 ϪpH ).
In the final round of modeling considering all the SPR, the known protonation constants (and cooperativity constants) for unbound (i.e. free) hPRL and hPRLrECD were fixed, whereas the bound state pK a values were varied. For WT hPRL, the partition function (binding polynomial) for protonation is expressed as, We have experimentally measured all the associated protonation and cooperativity constants for WT hPRL at both 25 and 35°C. We have also determined the site-specific protonation constants for the H30A and H180A single-site hPRL mutations at 35°C. In our simulations we have assumed an equivalent perturbation of His-27, His-30, and His-180 pK a values due to the single-site His-to-Ala mutations (in the unbound state) at 25°C along with the loss of the associated cooperativity constants, as was measured at 35°C. As our intention for performing the thermodynamic modeling is only to identify the overall changes in protonation constants upon complexation required to explain the SPR results and because the simulations have proven insensitive to the relatively minor changes in pK a values between 25 and 35°C, the extensive effort required for experimental measurement of the exact site-specific protonation constants for all His mutations at both temperatures is not warranted.

RESULTS
Effect of Single-site His-to-Ala Mutations in hPRL on pH-dependent Receptor Recognition-We originally hypothesized that protonation of one (or more) of the three His residues located in the intermolecular binding interface (His-27, His-30, or His-180) is responsible for the observed pH dependence. A primary contribution of residues in the hPRLrECD was initially excluded given the lack of pH dependence for hGH binding to the same receptor molecule. To assess whether loss of any His residue in hPRL affects binding to the hPRLrECD, we performed SPR analysis of WT hPRLrECD binding to WT and mutant hPRLs. Supplemental Table 1a lists the derived equilibrium constants for pertinent interface mutants across a range of pH values (detailed results are available in supplemental Table 2 and Fig. 1). Surprisingly, removal of the imidazole rings for His-27 and His-30 had essentially no effect on receptor binding, with the exception of a moderate loss in pH dependence for H30A compared with WT hPRL beginning at pH 6.25. In contrast, mutation of His-180 results in an ϳ100-fold lower receptor binding affinity from pH 7.25 to 8.00 compared with WT hPRL, followed by a similar drop in affinity with increasing acidity beginning around pH 7.0. However, unlike the WT hormone, the receptor binding affinity of H180A hPRL begins to normalize toward a loss of pH dependence around pH 6.25 analogous to H30A. Single-site mutations of the remaining six histidines in hPRL result in closely overlapping receptor binding affinities with the WT hormone, with the exception of a minor loss of neutral state affinity for H173A (supplemental Fig. 2). Last, mutation of His-180 to an Asp, as is seen for the analogous position in hGH (Asp-171), completely abolishes detectable receptor binding (supplemental Fig. 3). In total, these results demonstrate that no single histidine in hPRL is wholly responsible for the observed pH dependence, with only residues H30A and H180A appearing to partially mitigate the steep pH dependence of the WT-WT hormone⅐receptor interaction. Also, the neutral state of His-180 contributes significantly to the receptor binding affinity at high pH, whereas His-27 and His-30 appear to contribute little to receptor recognition despite their location within the high affinity receptor binding interface.
The pH dependence to any biochemical equilibrium can be modeled based on the thermodynamic linkage of the protonation reactions for each of the corresponding equilibrium states. In the case of protein-protein association reactions, the pH dependence of the apparent binding affinity can be completely described by changes in site-specific protonation constants (i.e. individual residue pK a values) between the uncomplexed and complexed protein states. Fig. 1 presents "best-fit" simulations of the experimental binding affinities for WT hPRL and three corresponding His mutants binding to the WT hPRLrECD as a function of pH. These simulations establish that a minimum of two protonation events are required to describe the pH dependence for WT, H27A, and H30A hPRL. Alternative models with more than two proton binding reactions fit the data equally well; however, a single protonation event is clearly incompatible with the experimental results. An additional restriction resulting from the initial decline in affinity starting around pH 7 requires a pK a Ͼ 7.0 for at least one of the involved titrated residues. In contrast, simulation of the SPR results for H180A hPRL requires only a single protonation event with a pK a Ͼ 7.0. In our previously reported analysis of site-specific protonation in uncomplexed hPRL (24), no histidine titrated with an apparent pK a Ͼ 7.0. The triad of His-27, His-30, and

pH-dependent hPRLr Binding Requires hPRL His-180 and hPRLr His-188
His-180 located in the high affinity receptor binding site have initial pK a values of 6.7, 6.3, and 6.1, respectively, with negative cooperativity constants in their titration reactions (c 27,30 ϭ 0.29, c 30,180 ϭ 0.18), acting to further inhibit the effective mass action for protonation. Attempts to simulate the pH dependence of receptor binding affinities using our experimentally measured pK a values unambiguously revealed a requirement for an unidentified residue with a more basic protonation constant, specifically with a pK a Ն 7.4. Initial speculation for the origin of this residue included either a basic amino acid in hPRL (e.g. an Arg or Lys) with a dramatically perturbed protonation constant or a titratable residue contributed by the hPRLrECD, which had been initially excluded due to the lack of the observed pH dependence for hGH binding to the same receptor molecule.
Correlation of the pH Dependence of hPRL Function in Cellular Assays-In our original description of the pH-dependent properties of hPRL (24), we reported an apparent pH dependence of hPRL function in two cellular assays; they are [ 3 H]thymidine incorporation into rat lymphoma Nb2 cells and a luciferase reporter assay of STAT5 transcription factor activation in the human T47D breast cancer cell line. Although the results of both experiments support a functional consequence of the pH dependence to hPRL receptor recognition, the long timescale of both studies (4 -24 h) complicates their interpretation due to the uncertain effects of prolonged cellular exposure to an acidic environment. Here, the pH dependence of hPRL function is assayed based on a shorter (10-min) exposure to decreased pH and with the more direct (i.e. receptor proximal) indicator of STAT5 phosphorylation. Fig. 2a presents the result of exposing human T47D cells, known to express high levels of hPRLr, to increasing concentrations of recombinant hPRL across a range of pH. At lower concentrations of hPRL (1 nM), STAT5 phosphorylation is strongly pH-dependent; however, as the concentration of hPRL is increased to 100 nM, the pH dependence of STAT5 phosphorylation is lost. This is expected due to saturation of the receptor with ligand at higher hPRL concentrations, even at the lower receptor binding affinities associated with acidic pH. Hence, these results further support the conclusion that the pH-dependent loss of hPRL function results from decreased receptor occupancy and not from a more general cellular effect.
Using a similar T47D-based assay of STAT5 phosphorylation, we have tested a variety of site-directed His-to-Ala mutants of hPRL for impaired receptor activation (Fig. 2b). Analogous to the changes in receptor binding affinity from the SPR studies (at pH Ͼ 7), removal of the imidazole ring for a majority of histidines in hPRL has a negligible effect on STAT5 phosphorylation, with the exception of a significant loss of function for H180A hPRL. The overall affinity of H180A hPRL for WT hPRLrECD at neutral and basic pH is ϳ100-fold lower than that of WT (or H27A and H30A) hPRL (Fig. 1). Consequently, the level of JAK2 and Stat5 phosphorylation is lower in The affinity of H180A hPRL for hPRLrECD is ϳ100-fold lower than that of hPRL at high pH, whereas a pH-dependent response is maintained. Best-fit simulation using observed binding affinity (by SPR) of WT or single site histidine mutants, H27A, H30A, and H180A, over a range of pH 5.5-8.0 are indicated by solid lines. The dashed line is an attempt to fit the WT data to a single pK a model with the starting (i.e. high pH) fixed to the best-fit value from the two pK a fit. FIGURE 2. Functional analysis of WT and single site histidine mutants of hPRL to evaluate their efficacy in pH-dependent hPRLr activation. T47D breast cancer cells, grown to ϳ80% confluency in conditioned media, were treated with 1, 10, or 100 nM recombinant WT or mutant hPRL at the indicated pH for 10 min. Total protein cell lysates were used for immunoblotting to determine relative levels of total and phosphorylated STAT5, hPRLr, and ␣-tubulin. a, a saturable pH-dependent effect on STAT5 phosphorylation was observed over the pH range of 5. H180A hPRLr-treated cells at neutral and basic pH; however, acidification of the extracellular environment leads to a pH-dependent decrease in its affinity for WT hPRLr. This is in good agreement with the SPR results. A minor discrepancy is seen for H30A hPRL, for which a small but consistent loss of STAT5 phosphorylation is seen at higher pH despite similar binding affinities to WT hPRL. This finding may implicate His-30 in the structural mechanism for receptor activation or communication between the high and low affinity receptor binding sites, which has been previously proposed (32). Interestingly, when the multiply mutated variants are considered, the combination of H27A with the H180A mutation causes a consistently larger loss of function than the corresponding addition of H30A even though the isolated H27A mutation had little effect on function. Such compensatory changes are not unexpected given the strong interacting free energies identified between the triad of His imidazoles within the receptor binding interface (23,24). Last, the concentration and pH-dependent changes in receptor activation for all possible combinations of His-to-Ala mutations for the His-27, His-30, and His-180 triad are shown in Fig.  2c, revealing preservation of some degree of pH dependence for all hPRL variants even when all three imidazoles are removed (H27A/H30A/H180A). Hence, the functional studies further support the requirement for an additional titratable residue in regulating the pH dependence of hPRL⅐receptor recognition. His-188 in the hPRLrECD Has a Site-specific pK a Ն 7.5-The hPRLrECD contains six histidines, but only one, His-188, is found within the high affinity receptor binding interface. Using uniformly 13 C, 15 N-labeled hPRLrECD, we have monitored sitespecific protonation of His imidazoles from 13 C, 1 H HSQC NMR spectral data using a protocol previously reported for hPRL (12,24). Five independent imidazole 3 C ⑀1 , 1 H ⑀1 HSQC correlation peaks are observed for the hPRLrECD, displaying typical pH-dependent 1 H ⑀1 and 13 C ⑀1 NMR chemical shifts (supplemental Fig. 4). Immediately evident in the derived titration profiles is a single His with a perturbed pK a of 7.7. As site-specific NMR chemical shift assignments have not yet been reported for the hPRLrECD, we have inferred the identity of His-188 in our NMR spectra using a pH titration of 13 C, 15 Nlabeled hPRLrECD and a second titration of labeled hPRLrECD in the presence of unlabeled hPRL antagonist (selective ligand for the high affinity binding site). The results are unambiguous; only one 13 C ⑀1 , 1 H ⑀1 HSQC correlation peak is perturbed by the addition of ligand, that with the unusual pK a of 7.7. In the presence of ligand, this residue also demonstrates a large shift in its apparent pK a toward a more typical value of 7.0. However, it must be emphasized that this apparent pK a in the presence of ligand cannot be taken to represent the true protonation constant for the complexed state, as saturated complexation of the hPRLrECD cannot be guaranteed across the entire pH range. Hence, the apparent pK a seen during the titration with ligand most likely represents a population-weighted average of the free and bound states of hPRLrECD. Last, poor dispersion and broad line widths of NMR signals for mixtures of hPRL and the hPRLrECD at lower temperatures necessitated the use of 35°C during these NMR studies, in contrast to the 25°C used for the remainder of experiments.

Loss of Both His-180 in hPRL and His-188 in hPRLr Causes
Loss of pH-dependent Receptor Binding and Activation-The exciting identification of an unusually basic protonation constant for His-188 in the hPRLrECD prompted the investigation of its role in hPRL⅐receptor recognition. To test the functional consequence of the H188A mutation, we monitored hPRL-induced STAT5 phosphorylation in CHO cells transfected with an expression vector for either WT or H188A-full-length hPRLr. The addition of recombinant hPRL demonstrated pHand concentration-dependent STAT5 phosphorylation analogous to the results seen for T47D cells, shown above (Fig. 2). We subsequently tested the pH dependence of STAT5 phosphorylation along with the more receptor-proximal phosphorylation of JAK2 for all possible His mutations of the His-27/His-30/ His-180 triad in hPRL with both WT and H188A hPRLr (Fig. 3). The H188A mutation in the hPRLr has no effect on pH dependence of receptor activation when exposed to WT hPRL. However, when cells expressing H188A hPRLr are treated with any hPRL variant harboring the H180A mutation, the pH dependence is lost, due primarily to a decrease in receptor activation at high pH. These findings strongly suggest that pH-dependent receptor recognition involves a highly cooperative interaction between His-188 in the receptor and the localized triplet of clustered histidines in hPRL, with His-180 playing a dominant role.
X-ray Crystallographic Structural Analysis of hPRL Receptor Binding-To address whether specific structural rearrangements either globally or within the intermolecular binding interface are induced by removal of specific His imidazoles, we have crystallized and determined high resolution structures of numerous hPRL⅐receptor complexes. This work follows the lead of a previous crystal structure of a 1:1 complex between the hPRLrECD and an hPRL variant (12) (PDB code 3D48) that combines the hPRL G129R mutation with deletion of the hPRL N terminus to abolish lower affinity, site 2 receptor binding. Using similar crystallization conditions, we determined the structures of the hPRLrECD complexed with an analogous hPRL variant (⌬1-14, G129R-hPRL), hereby referred to as the "WT antagonist." Well diffracting crystals have been found for the following combinations of hPRL (antagonist)/hPRLrECD: WT⅐WT, H27A⅐WT, H30A⅐WT, H180A⅐WT, WT⅐H188A, H27A⅐H188A, and H30A⅐H188A. X-ray diffraction data have been collected for these complexes, and their corresponding structures have been determined by molecular replacement using the PDB code 3D48 structure as an initial search model for the WT⅐WT structure. The final WT⅐WT structure was used as a search model for the remainder of mutant combinations. Details are given in Table 1 (and representative electron density is shown in supplemental Fig. 5). As expected, our structural model of the WT⅐WT complex agrees well with 3D48, which superpose with an root mean square deviation of ϳ0.5 Å for the backbone atoms. Additionally, our 2.0 Å electron density maps for WT⅐WT complex show good density for regions missing in the 3D48 structure, which was modeled to 2.5 Å resolution, including the two long loops in hPRL and some of the extended loops between ␤-strands in the hPRLr-ECD. In hPRL, these long loops have been associated with post-
Using UCSF Chimera (33), we calculated the solvent-accessible surface area for all five complexes and compared them to their isolated counterparts (supplemental Table 3), revealing ϳ1194 Å 2 of buried surface area within the WT hPRL⅐hPRLr complex, in good agreement with the reported burial of 1180 Å 2 in PDB code 3D48 (12). A total of 52 residues have greater than 10% of their solvent-accessible surface area buried within the intermolecular interface and are displayed in the upper left panel of Fig. 4. Both the total amount of buried surface area and the percent burial for individual residues are largely unchanged with all of the single-site His to Ala substitutions studied here. In general, imidazole removal resulted in cavity formation within the intermolecular interface, identified using the SURFNET (34) module of UCSF Chimera (33,35). In some cases, solvent (water) molecules are evident in the electron density for these cavities, and we generally assume they are occupied by buried solvent in aqueous solution. Fig. 4 allows for clear visualization of all histidines present within the WT⅐WT intermolecular interface and the structural consequences of systematic substitution. In the upper right panel, hydrogen bonds for His imidazoles in the high affinity binding site are identified; accuracy has been enhanced by the incorporation of their experimentally measured tautomeric states (24). The lower set of panels compare the binding interfaces for the His mutant complexes superposed onto the WT⅐WT complex. Overall, structural changes are minimal. Mutation of His-27 displays the least structural rearrangements. For H30A⅐WT, removal of its imidazole results in a "downward" motion of the C␣-C␤ bond along with a uniform displacement of the helical backbone for H30A and nearby residues; however, positions of the associated side chains are mostly unaffected. A similar effect is seen for WT⅐H188A and H180A⅐WT complexes in the lower right panel, with the additional perturbation of multiple side-chain positions in H180A hPRL. Conformational rearrangement of the protein backbone upon removal of the His-30 and His-180 imidazoles suggests a degree of backbone structural strain in the WT protein consistent with the previously reported stabilization of prolactin global fold induced by the H30A and H180A mutations (36). Sidechain rearrangement in H180A may relate lower receptor binding affinity. Last, an interesting heterogeneity in side-chain conformation is found for Asn-31 of the antagonist in the WT⅐WT complex, for which two alternate conformations were observed. One state, shown for Asn-31 in the upper right panel of Fig. 4, allows hydrogen bonding with the His-27 imidazole; in the other state, the Asn-31 side chain rotates toward solvent. However, in each of the mutated complexes, Asn-31 resides in only a single rotameric state; in H30A⅐WT, H180A⅐WT, and WT⅐H188A, Asn-31 is solvated, and in H27A⅐WT, Asn-31 adopts the state previously associated with hydrogen bonding to His-27. The rationale behind these structural variations in Asn-31 is currently unknown. However, we note that Asn-31 is the site of a well established N-glycosylation event in prolactin (37), whose physiologic significance has not been clearly elucidated.
Re-evaluation of pH-dependent Binding Affinities by SPR with Inclusion of H188A hPRLrECD-To confirm an apparent role for His-188 in the pH dependence to hPRL⅐receptor recognition, we repeated SPR studies with both WT and H180A hPRL binding to the H188A receptor mutant. Listed in supplemental Table 1b (with details in supplemental Table 2) and shown in Fig. 5, there is a 100-fold decrease in affinity at high pH for H188A hPRLrECD binding to WT hPRL. Note that Fig.  5 also includes SPR results from our original analysis of His mutations in hPRL, shown previously in Fig. 1. As these two sets of studies involved entirely independent preparations of FIGURE 3. Immunoblot analysis to determine the role of His-188 of hPRLr in pH-dependent receptor activation. CHO cells grown to ϳ60% confluency were transfected with mammalian expression vectors for hPRLr (WT or H188A) as indicated. At ϳ80% confluency (24-h post-transfection) cells were treated with WT or mutant hPRL (1 nM unless stated otherwise) at the indicated pH for 10 min. Total protein cell lysates were used for immunoblotting to determine relative levels of total and phosphorylated JAK2, total, and phosphorylated STAT5, hPRLr and ␣-tubulin. a, consistent pH-dependent JAK2 and STAT5, phosphorylation is observed only when the ligand⅐receptor combination retains His-180 of hPRL and His-188 of hPRLr. b, the pH-independent receptor activation (phosphorylated JAK2 and phosphorylated STAT5 levels) of the ligand⅐receptor pair lacking both His-180 of hPRL and His-188 of hPRLr is maintained at higher ligand concentration. Conc, concentration; Lig, ligand.

pH-dependent hPRLr Binding Requires hPRL His-180 and hPRLr His-188
DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 recombinant proteins, the comparison allows an assessment of reproducibility for the WT-WT and H180A-WT (hPRL⅐receptor) binding interaction. As can be seen in the figure, the two sets of results overlap well, and the pH-dependent trends are unchanged. A blunted transition to lower receptor binding affinity is observed with decreasing pH for WT hPRL recognition of H188A hPRLrECD. When the H180A hPRL and H188A receptor mutations are combined, essentially all pH dependence is lost, and the overall receptor binding affinity is greatly reduced. At pH ϳ 6 and below, the affinities for combination of H180A-WT-, WT-H188A-, and H180A-H188A-interacting pairs converge, indicating that the effect of protonation and loss of the His-180 and His-188 imidazoles are not thermodynamically independent. To gain further insight into the site-specific changes in protonation constants (i.e. pK a values) responsible for the pH dependence of hPRL⅐receptor recognition, thermodynamic modeling analogous to that described for Fig. 1 was performed for all the hormone⅐receptor interacting pairs. Summarized in Table 2, the results of the thermodynamic simulations are revealing. As previously determined for WT hPRL, the steep dependence of binding affinity on pH, apparently continuing on below pH 5.5, requires two or more histidines whose pK a values upon complexation shift toward very low values. Additionally, the original decline in affinity beginning around pH 7.5 necessitates the presence of at least one histidine with a free-state pK a Ͼ 7.5, served in this case by His-188. Removal of the His-27 imidazole has a minimal effect on the pH dependence of binding affinity, consistent with a minimal change in its pK a upon complexation. For the H30A mutation, whose pH-dependent loss of affinity appears to stabilize beginning around pH 6.25, the simulations reveal a less dramatic decrease in protonation constants for the remaining imidazoles upon complexation. Note that the best-fit pK a values for the bound state were all nearly exactly 6.0. Systematic variation of each value followed by re-minimization of the other two allowed a Ϯ0.5 variation around the best-fit values with a statistically insignificant effect on the fit. Hence, we conclude that the energetic strain induced by protonation of the imidazoles relaxes from an average pK a ϳ 4 for WT hPRL to an average pK a ϳ 6 for H30A hPRL, underscoring an important role of the His-30 imidazole and its strong thermodynamic interaction with nearby histidines. A similar conclusion can be reached for the pH-dependent binding of the H180A-WT and WT-H188A ligand⅐receptor pairs; their removal relaxes the energetic strain for protonation of the remaining imidazoles in the complexed state. Last, when both His-180 in hPRL and His-188 in the receptor are mutated, the pH dependence of binding is essentially lost completely, and correspondingly, the best-fit pK a values for the remaining His-27 and His-30 imidazoles are almost identical to the unbound state.

DISCUSSION
Previous work in our laboratory has demonstrated dependence of the functional and biophysical properties of hPRL on variation in the solution acidity across the physiologic pH range from 6 to 8, prompting us to investigate the site-specific properties of histidine residues in hPRL (24,36). In the current work we seek to ascribe pH dependence of hPRL⅐receptor recognition and activation to the behavior of individual His residues within the high affinity hPRL⅐hPRLr intermolecular interface. Initially, we did not expect a role for residues contributed by hPRLr in the molecular mechanism for pH dependence, as the binding of hGH to the same receptor is pH-independent. However, we show here that regulation of the hPRL⅐receptor interaction by solution pH depends critically on His-188 from the hPRLr, which demonstrates a dramatically perturbed titration profile. Additionally, the triad of histidines in hPRL (His-27, His-30, and His-180), also located within the high affinity binding interface, secondarily contributes to the regulation of pH dependence, especially at the more acidic end of the pH range. Within this group, His-180 appears to play a dominant role, as its removal essentially negates the binding-dependent perturbations in the site-specific titration profiles for His-27 and His-30. Although a bulk of the observed pH dependence is preserved in H30A, a small, but significant loss of the pH dependence to receptor binding affinity is observed below pH 6.5. In contrast, His-27 appears to be dispensable for receptor binding and makes minimal contributions to the mechanism for pH dependence, except upon mutation of either His-30 or His-180, when His-27 makes a larger contribution to regulation of pH dependence. Thus, similar to our previous investigations, the highly cooperative nature of this network of coupled histidines is evident. High resolution crystallographic structures of the hPRL⅐receptor interfaces, determined for the WT complex and all possible single-site His mutants, emphasize the importance of electrostatic forces in regulating the pH dependence of hPRL⅐receptor recognition while suggesting a lesser role for conformational effects. The introduction of polarizable water molecules into the intermolecular interface upon removal of a single His imidazole provides a likely explanation for the loss of binding-induced perturbations of site-specific pK a values for the remaining histidines (Table 2). Cellular studies presented here confirm a clear correlation between the pH dependence to receptor binding affinity and two receptor-proximal indicators of hPRLr function: JAK2 and STAT5 phosphorylation. To the best of our knowledge this is the first study that systematically evaluates the role of each histidine residue of hPRL. Previous studies have included only four single-site His mutations: His-59, His-30, His-173, and His-180 (15,16). The H59A and H30A mutations reduce Nb2 cell proliferation by ϳ50%, whereas the H180A and H173A mutations reduce Nb2 cell proliferation even further. Although our data are largely in agreement with these findings, the more subtle phenotypic differences are most likely a result of assay method. Nb2 proliferation in response to treatment with mutant hPRL is most likely a cumulative effect of the multiple pathways regulated by activated hPRLr, whereas our current results assay the most receptor-proximal step after efficient ligand binding and receptor activation. Last, although there are multiple studies involving mutagenic variants of hPRL, evaluation of the role of specific amino acids in the hPRL receptor is far less complete, and we have been unable to identify a previous report of H188A mutagenesis in hPRLr.
Kossiakoff and co-workers (38) proposed that recognition of the hGHr and hPRLr by hGH involves two fundamental steps; FIGURE 5. SPR and thermodynamic modeling of WT and H188A hPRLrECD to assess the role of His-188 in pH-dependent hPRL binding. SPR performed with WT or H180A hPRL bound to a GLM sensor chip by standard amine coupling and purified recombinant WT or H188A hPRLrECD. Binding kinetics of the hPRL⅐hPRLrECD interactions obtained in quarter unit increments in pH over a range of pH 5.75-8.00 are shown. Affinities of the H180A hPRL⅐WT hPRLrECD and WT hPRL⅐H188A hPRLrECD combinations are ϳ100fold lower than that of WT hPRL⅐WT hPRLrECD at high pH, whereas pH-dependent response is maintained. Loss of both His-180 and His-188 results in pHindependent receptor binding along with loss in affinity. Solid lines indicate best-fit simulation using observed binding affinity (by SPR) of WT hPRL or H180A hPRL for WT hPRLrECD or H188A hPRLrECD over a range of pH 5.5-8.0. The asterisk indicates data from Fig. 1, included to ensure reproducibility and consistent comparison.

pH-dependent hPRLr Binding Requires hPRL His-180 and hPRLr His-188
DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 the initial formation of a transitions state followed by structural fine-tuning of the intermolecular interface that creates a slowly dissociating complex. When residues within the structural epitope for high affinity receptor binding are systematically mutated and their binding kinetics analyzed, alterations in the equilibrium binding constant are generally entirely accounted for by variation in the dissociation rate constant with little or no change in the association rate constant. This is interpreted as a lack of contribution by these residues in forming the transition state complex. Analysis of our results agrees with this observation in that mutation of both His-180 and His-188 destabilizes the hormone⅐receptor complex at high pH by selectively increasing the dissociation rate constant. Similarly, for all of the complexes, loss of binding affinity with decreasing pH is caused predominantly by an increase in the dissociation rate constant, with little effect on association. Under more acidic conditions (pH Ͻ 6.5), changes in the association rate constant become more significant, suggesting that protonation of surface residues begins to affect the rate of transitions state formation.
The critical role ascribed to two histidine residues (one of the nine in the hPRL and one of six in hPRLr) of the site 1 binding interface has prompted us to also evaluate their evolutionary conservation. Alignment of hPRL and hPRLr sequences from diverse species ranging from Danio rerio (zebrafish) to Homo sapiens (human) reveals that the four histidines located in the site 1 receptor binding interface are much more highly conserved than the remaining histidines in both proteins (supplemental Table 4). The two residues appearing to dominate regulation of pH-dependent receptor recognition, His-180 and His-188, are conserved across all species investigated with the exception of Bos taurus (cow) and Xenopus laevis (frog), respectively. Interestingly, His-30 is the most conserved histidine among all of the hPRL histidines, which suggests a major functional role for this residue. Last, His-27 is also well conserved (except for Asp-27 of D. rerio), yet its mutation has a minimal effect on receptor binding and activation and does not appear to contribute significantly to regulation of pH dependence. Based on previous work, the imidazoles of all histidines in the triad appear to be net-destabilizing to the hPRL tertiary structure, as their removal significantly improved structural stability (36), further supporting their likely functional importance.
Upon endocytosis of the activated ligand⅐receptor complex, as the pH of endocytic vesicles decreases in a stepwise manner, acidity-induced dissociation of the internalized complex is thought to be a critical step in targeting the ligand for degrada-tion, whereas the receptor is selectively recycled back to the cell surface. The molecular mechanism for pH-dependent dissociation of the ligand⅐receptor complex has not been investigated for any other hematopoietic cytokine. This is the first report that systematically dissects the role of specific histidines involved in pH-dependent hPRL⅐receptor binding, potentially providing a basis for the design of pH-independent hPRL variants that could resist dissociation from the receptor in acidified endosomes. Such hPRL variants would serve as powerful experimental tools to investigate the functional role of endosome acidification in regulating trafficking of internalized receptors without introducing the broad cellular consequences of disrupting acidification itself. We hypothesize that a pH-independent hPRL variant may efficiently target hPRLr to the lysosome (or proteasome) for degradation and prevent recycling back to the cell surface. Alternatively, un-dissociated ligand⅐ receptor complexes may "stall" within the endocytic pathway, with unforeseen and likely widespread implications. Support for these hypotheses comes from previous studies performed with insulin (39) and the epidermal growth factor family of receptors (40 -47), where the apparent pH-dependent dissociation of endocytosed ligand⅐receptor complexes regulates their intracellular trafficking. In future studies we hope to compare the trafficking of pH-independent and pH-dependent hPRL⅐hPRLr complexes to elucidate the functional importance of acidity-induced dissociation.