INTRODUCTION

In the yeast Saccharomyces cerevisiae, the purine-cytosine permease (PCP)¹ mediates cotransport through the plasma membrane of proton and purine bases (adenine, hypoxanthine, and guanine) or a pyrimidine base (cytosine) (1-6). The FCY2 gene, encoding for PCP, has been cloned (7) and sequenced (8), thus enabling studies of the relationships between the deduced amino acid sequence of the permease (a polytopic protein consisting in 533 amino acid residues with a molecular mass of 58.2 kDa) and the transport mechanism.

An important question for the mechanism of base transport through the plasma membrane is the characterization of the amino acid residues involved in the binding of ligands. It has been shown that the 4 bases bind to the same site with very similar affinities and that their uptake occurs with equivalent efficencies (3, 9). To identify the binding site, an interesting approach involves the isolation of mutant strains with altered affinity constants for the various ligands, as already described in the case of lactose permease of Escherichia coli (10-12). Screening of PCP mutants (3), based on competition on either adenine and cytosine uptake in an Ade strain (fcy2-21 and fcy2-20 alleles), or cytosine and 5-methylcytosine in an Ura strain (fcy2-10) allowed the identification of a region of the protein whose mutations led to strains with altered K_t(app) (apparent Michaelis constant of transport) of purine bases and cytosine. This region, according to a hypothetical topologic model (8), is located in a loop (367-399) connecting two putative transmembrane  helices and facing the periplasmic space. The cloning and sequencing of these mutants (13) revealed that one of them (fcy2-21 allele) had extensive nucleotide replacement, resulting in three amino acid changes (I371V, I375V, and N377G), whereas two others (fcy2-10 and fcy2-20 alleles) displayed only one amino acid change (N377I and N374I, respectively). In addition, to analyze the contribution of each amino acid residue change to the phenotype of the triple mutant fcy2-21, single (N377G) and double (I371V,I375V) mutants have been constructed by site-directed mutagenesis (13). All these mutants displayed modifications of the uptake parameters for the different bases. We have already extensively described the triple mutant strain and shown that modifications of equilibrium binding parameters were responsible for the uptake phenotype (9).

The purpose of the present study was to investigate in greater detail the contribution of individual amino acid residue changes localized in the 371-377 hydrophilic segment to the phenotype of the various mutant strains, especially on the base binding step. This was done by measuring, for adenine, cytosine, and hypoxanthine, the uptake activity on intact cells and equilibrium binding parameters on plasma membrane enriched fractions. Moreover, dissociation constants for hypoxanthine and proton binding were calculated by measuring the amount of specifically bound hypoxanthine as a function of the external pH. This approach made it possible to investigate the role of the 371-377 segment in the active conformation of the functional PCP.

EXPERIMENTAL PROCEDURES

Adenine, hypoxanthine, and cytosine were purchased from Aldrich. Radioactive [8-¹⁴C]adenine (1.96 GBq·mmol¹) and [8-¹⁴C]hypoxanthine (1.88 GBq.mmol¹) were from DuPont NEN and [2-¹⁴C]cytosine (1.96 GBq.mmol¹) from Moraveck. Restriction endonucleases and enzymes used for constructions and analyses of plasmids were purchased from New England Biolabs, Amersham Corp., and Life Technologies, Inc. and used as recommended by the suppliers. Prestained molecular mass markers were obtained from Sigma. Horseradish peroxidase-conjugated IgG were from Pierce. ECL Western blotting kit and hyperfilm^TM ECL were obtained from Amersham. Protein A-Sepharose beads were purchased from Pharmacia Biotech Inc. LB and yeast nitrogen base without amino acid were from Difco. Scintillation mixtures were Ready Safe from Beckman. All other materials were of analytical grade.

E. coli HB101 (supE44, hsdS20(r_Bm_B), recA13, ara-14, proA2, lacY1, galK2, rpaL20, xyl-5, mtl-1) (14) was used as carrier for the plasmids described. Characterization and isolation of the yeast strains used in this work have already been described (8, 15). All are derivatives of the NC233-10B (MAT a, fcy1-1, fcy2(BglII-KpnI), leu2) strain. NC233-10B[pJDB] is a permease-null strain. Strain NC233-10B[pAB4] is a permease-proficient strain, carrying plasmid-encoded multiple copies of the FCY2 gene. NC233-10B[pAB25] is a permease-proficient strain, carrying plasmid-encoded multiple copies of the mutated fcy2-21 allele.

E. coli strains were grown at 37 °C, in LB liquid or solid medium, with or without ampicillin. Yeast strains were grown at 28 °C under agitation in a yeast nitrogen base without amino acid medium (6.75 g·l¹) containing 2% glucose (w/v) and 25 mM sodium phtalate, pH 4.5.

Accurate determination of uptake and binding parameters can be performed only on overexpressed PCP-containing strains. Thus, all the FCY2 alleles (wild type and mutated) were subcloned onto the multicopy pJDB207 plasmid (16). Plasmids pTF1 and pTF2 were constructed by replacing the restriction fragment BglII-SphI (1.92 kilobase pairs) in the FCY2 open reading frame of pAB4 (8) by the corresponding fragment from the pJC5 or pJC6 plasmids (13), respectively. Plasmids pTF2-10 and pTF2-20 were constructed by replacing the restriction fragment MluI-ApaLI (1.76 kilobase pairs) of pAB4 by the corresponding fragment from the plasmids pPZ1-10 or pPZ1-20 (13), respectively. Double-stranded plasmid DNA prepared by Wizard^TM Midipreps (Promega) were sequenced using the dideoxynucleotide termination method (17) and synthetic sequencing primers after alkaline denaturation (18).

E. coli cells were transformed according to Cohen et al. (19) and yeast cells according to Gietz et al. (20).

To simplify the text, the NC233-10B transformed yeast clones used in this study will be referred to only by the plasmid they harbor.

Plasma membrane-enriched fractions (PMF) were isolated as described previously (21). Cells (2.8-3.2×10⁷ cells·ml¹) were disrupted with glass beads, and the homogenate was submitted to differential centrifugations. Mitochondrial material was eliminated by acid precipitation (22).

Initial rates of uptake (adenine, hypoxanthine, and cytosine) were measured at various solute concentrations by a filtration technique, essentially as described previously (9). The maximal rate of uptake (V_max) and the apparent Michaelis constant of transport (K_t(app)) were calculated by nonlinear regression analysis of the initial rates of uptake versus the solute concentration, using the method of Cleland (23).

Ligand binding measurements were performed by a centrifugation technique (9) in 50 mM sodium citrate, pH 4.5, containing 100 mM NaCl. The maximal amount of specifically bound ligand (B_max) and the apparent half-saturation constant of solute binding (K_d(app)) were calculated by nonlinear regression analysis of the saturation curve.

Guanine was not used in this study because of its poor solubility under our experimental conditions.

Hypoxanthine binding measurements as a function of the external pH were done as described previously (24), with slight modifications. pH variations from 3.5 to 6.2 were obtained by mixing two buffer solutions (50 mM citric acid containing 100 mM NaCl, pH 1.9, and 50 mM sodium citrate containing 100 mM NaCl, pH 8.1) at a given hypoxanthine concentration (11-79 µM depending on the strain). Under these conditions, the ionic strength of the incubation medium varies between 150 and 350 mM. In control experiments, it was found that, at a given pH, these ionic strength variations in the incubation medium had no effect on the amount of specifically bound hypoxanthine.

In a previous work (24), it was shown that a random mechanism with a stoichiometry 1/1/1 (proton/base/site) was the best model to fit the experimental data obtained for the binding of proton and hypoxanthine to plasma membrane fractions. In this random process (Fig. 1), four constants are defined. K_d and K_d, are respectively the dissociation constants of hypoxanthine from the protonated and deprotonated PCP. K₁ and K₂ are, respectively, the acidic dissociation constants of the proton from a group of the free and the liganded PCP. Calculations of these constants were done using the following equations.

(Eq. 1)

where B and B_max are, respectively, the amount and the maximal amount of specifically bound hypoxanthine (pmol·mg protein¹), [L], the free hypoxanthine concentration and K_d(app), the apparent dissociation constant of hypoxanthine binding at a given pH.

(Eq. 2)

and

(Eq. 3)

Equation 2 can be written in a logarithmic form (as follows).

In this work, we have used the plot pK_d(app) versus pH, which is very convenient to visualize directly the effects of the mutations. This curve admits two asymptotic values (the plateaus), one at very acidic pH values (pH  pK₁ and pK₂) whose value is pK_d and the other at very basic pH values (pH  pK₁ and pK₂) whose value is pK_d + pK₁-pK₂. The midpoint is observed when pH = (pK₁+ pK₂)/2. The slope at this midpoint is correlated with the difference between pK₁ and pK₂ and shows a lower limit value of 1, i.e. when pK₂  pK₁ (which means at least a 3 pH unit difference between the two pK_a values or K₁/K₂ > 1000) and an upper limit value of 0 (when pK₂  pK₁). The drawn tangent at the midpoint gives intercepts with the upper and lower asymptotes that are pK₁ and pK₂, respectively.

PMF samples (1 µg of protein) were separated on Tris/Tricine SDS-polyacrylamide gel electrophoresis on 12.5% polyacrylamide resolving gels (25) and then transferred to polyvinylidene difluoride membranes (Amersham) for Western blot analyses (26). The blots were incubated with either purified anti-N-terminal PCP (27) or anti-H⁺-ATPase (a generous gift of J. Nader, University of Louvain-la-Neuve) as primary antibodies. Peroxidase-conjugated anti-rabbit IgG were used as secondary antibodies. Reacting materials were visualized using the ECL^TM Western blotting protocol according to the manufacturer's guidelines (Amersham).

Cell concentrations were determined by measuring the turbidity of the culture medium at 550 nm (an absorbance unit corresponding to 2 × 10⁷ cells·ml¹). Protein concentrations were determined by the method of Lowry (28) using bovine serum albumin as standard.

RESULTS

The NC233-10B strain, carrying a chromosomal deletion at the FCY2 locus, was transformed with the plasmids presented in Table I.

Table I. Plasmids used in this study All the FCY2 alleles were introduced within the pJDB207 multicopy plasmid. This shuttle vector harbors the yeast 2-µm fragment and the LEU2 gene with a defective promoter, leading to a high copy number and a resulting overexpression of the FCY2 gene product. Plasmid name FCY2 gene Amino acid replacement Ref. pJDB207 Beggs (16) pAB4 Wild type Weber et al. (8) pAB25 fcy2-21 I371V,I375V,N377G Weber et al. (8) pTF1 Site-directed mutagenesis I371V,I375V This work pTF2 Site-directed mutagenesis N377G This work pTF2-10 fcy2-10 N377I This work pTF2-20 fcy2-20 N374I This work

Expression of the relative amounts of PCP among the various strains was checked by using Western blot detection of both PCP and plasma membrane H⁺-ATPase on PMF samples (Fig. 2). With anti-N-terminal PCP antibodies, a unique band was detected in the range of 47-52 kDa for the overexpressed wild type strain (Fig. 2, lane 7) and all of the overexpressed mutant strains (Fig. 2, lanes 1-5). This band, which was not visible for the permease-null strain (Fig. 2, lane 6), corresponds to PCP. Thus, the different mutations did not seem to affect PCP targeting to the plasma membrane.

Scanning of the bands corresponding to PCP and to H⁺-ATPase, used as an internal standard, showed that there was no significant difference in the overexpression level of PCP (40-fold overexpression as compared with the wild type) among the various multicopy strains studied (data not shown).

The uptake constants are presented in Table II.

Table II. Purine and cytosine uptake parameters for the various strains NC233-10B cells were transformed with the corresponding plasmids. Cells harvested in the exponential phase (3-3.2 × 107 cells·ml1) were washed in 50 mM sodium citrate, pH 4.5, containing 2% glucose at 30 °C. Uptakes were started by addition in cellular suspensions (5-7 × 106 cells·ml1) of [8-14C]adenine (176-198 MBq·mmol1): 0.24-40 µM for pTF1 and pTF2, 5-120 µM for pTF2-10, and 0.62-50 µM for pTF2-20; [2-14C]cytosine (180-205 MBq·mmol1): 0.25-40 µM for pTF1, 0.5-80 µM for pTF2, 2-200 µM for pTF2-10, and 4.2-168 µM for pTF2-20; [8-14C]hypoxanthine (140-187 MBq·mmol1): 0.25-40 µM for pTF1, 1-150 µM for pTF2, 6-230 µM for pTF2-10 and 20-220 µM for pTF2-20. The initial rates of uptake were obtained as described under "Experimental Procedures." Kt(app) (µM) and Vmax (nmol·107 cells·ml1) were calculated by nonlinear regression of the saturation curves obtained with at least 10 different solute concentrations. All experiments were done in duplicate. Solutes Uptake parameters Plasmids and resulting amino acid changes [pAB4]a [pAB25] (I371V, I375V, and N377G)a [pTF1] (I371V and I375V) [pTF2] (N377G) [pTF2-10] (N377I) [pTF2-20] (N374I) Adenine Kt(app) 1.8  ± 0.2 1.9  ± 0.2 2.4  ± 0.2 1.8  ± 0.2 42  ± 9 7.6  ± 1.2 Vmax 9.2  ± 0.2 8.5  ± 0.2 10.5  ± 0.4 8.1  ± 0.2 4.2  ± 0.4 6.6  ± 0.6 Hypoxanthine Kt(app) 2.5  ± 0.5 35  ± 17 2.7  ± 0.3 64  ± 20 47  ± 8 153  ± 40 Vmax 11.7  ± 0.8 8.5  ± 1.0 9.5  ± 0.8 5.7  ± 1.3 11.5  ± 0.7 6.0  ± 0.5 Cytosine Kt(app) 1.8  ± 0.2 10.3  ± 0.7 2.8  ± 0.2 11.6  ± 0.8 15.6  ± 1.9 47  ± 7 Vmax 8.5  ± 0.2 15.4  ± 1.0 14  ± 1 20.5  ± 1.0 17.9  ± 0.8 9.2  ± 0.6 a The values for the pAB4 and the pAB25 strains are from Brèthes et al. (9).

Despite a quite similar expression level of PCP among the various strains of this study (Fig. 1), significant variations were observed for V_max. For a given strain, these variations could be in relation to the base transported; and for a given base, variations could be in relation to the strain. In addition, these V_max variations, in contrast to those observed for the K_t(app) values (see below), could be either increases or decreases and, for a given strain, could be more or less pronounced depending on the transported solute. For example, if one compares the pTF2-10 cells and the wild type, there was a V_max decrease for adenine (2-fold) and a V_max increase for cytosine (2-fold) for the mutant strain, whereas for hypoxanthine, V_max remained the same.

Since V_max is a complex parameter related to the efficiency of different steps of the solute translocation process, it is not yet possible to explain these variations.

The main effect of the triple substitution (I371V, I375V, and N377G) in pAB25 cells, as compared with the pAB4 strain, was a large increase in K_t(app) for two solutes (6- and 14-fold for cytosine and hypoxanthine, respectively) and no change for adenine (9, 24).

For pTF1, which has two (I371V, I375V) of the mutations of the pAB25 strain, the K_t(app) for the 3 bases were quite similar to those observed for the wild type. The main effect of the two amino acid residue changes was a slight increase (1.5-fold) in the K_t(app) for cytosine. For the pTF2 simple mutant strain (N377G), K_t(app) values for cytosine and hypoxanthine were dramatically increased (6.4- and 25.6-fold, respectively), whereas the K_t(app) for adenine remained the same, as compared with the wild type. Thus, the simple mutant pTF2 behaves like the triple mutant strain pAB25, whereas the double mutant pTF1 resembles the wild type.

For pTF2-10 cells, the substitution N377I resulted in a large increase in the Kt(app) for the 3 bases, as compared with the pAB4 wild type strain (9-, 19-, and 23-fold for cytosine, hypoxanthine, and adenine, respectively). K_t(app) values for cytosine and hypoxanthine of this mutant were of the same order of magnitude as those observed for the pTF2 strain, the mutant for which the same Asn residue was replaced by Gly instead of Ile.

In pTF2-20 cells (N374I), K_t(app) values for the 3 bases were also increased, more dramatically than already observed for the pTF2-10 cells. In fact, this mutation was the amino acid residue change that induced the largest increases in K_t(app) values for cytosine and hypoxanthine (26- and 61-fold, respectively).

It is noteworthy that, for fcy2-10 and fcy2-20 alleles, the K_t(app) of adenine was affected, although this was not the case for the fcy2-21 allele and its derivatives.

For all strains and solutes tested, saturation curves showed only one class of binding site (data not shown). The K_d(app) and B_max, obtained by nonlinear regression of the saturation curves, are displayed in Table III. With a given strain, the B_max measured is quite similar for the 3 bases tested. On the contrary, B_max values obtained for the different strains showed significant variations. The dispersion of this parameter reflects the poor reproducibility we observed on numerous large-scale PMF preparations. Thus, owing to this heterogeneity of PMF preparations, no comparison will be made for this parameter among the various strains.

Table III. Equilibrium binding parameters on PMF from the various strains PMF were prepared as described under "Experimental Procedures." Total binding measurements were performed as described in 50 mM sodium citrate, pH 4.5, containing 100 mM NaCl with the following concentrations of radioactive [8-14C]adenine (1167 MBq·mmol1): 0.15-26 µM for pTF1 and pTF2, 6-400 µM for pTF2-10 and 3-400 µM for pTF2-20; [2-14C]cytosine (1233 MBq·mmol1): 0.7-35 µM for pTF1, 20-360 µM for pTF2, 4-200 µM for pTF2-10 and 20-300 µM for pTF2-20; [8-14C]hypoxanthine (1100 MBq·mmol1): 2-150 µM for pTF1, 20-380 µM for pTF2, 5-310 µM for pTF2-10, and 7-300 µM for pTF2-20. In each case, nonspecific binding measured in the presence of 4 mM adenine, was 22 ± 2 pmol·mg protein1·µM1 of free ligand. Specific binding was calculated by subtracting the nonspecific binding to the total binding. Kd(app) (µM) and Bmax (pmol·mg protein1) were calculated by nonlinear regression of the saturation curves, obtained with at least 14 different concentrations of ligand. All experiments were done in duplicate. N. D. non-measurable. Ligands Binding parameters Plasmids and resulting amino acid changes [pAB4]a [pAB25] (I371V, I375V, and N377G)a [pTF1] (I371V and I375V) [pTF2] (N377G) [pTF2-10] (N377I) [pTF2-20] (N374I) Adenine Kd(app) 1.3  ± 0.2 3.9  ± 0.5 1.3  ± 0.2 3.6  ± 0.5 132  ± 38 138  ± 35 Bmax 1186  ± 36 495  ± 50 2750  ± 73 1485  ± 131 2250  ± 350 2150  ± 350 Hypoxanthine Kd(app) 7.1  ± 0.8 206  ± 60 11.1  ± 0.9 170  ± 50 >700 ND Bmax 893  ± 34 460  ± 110 1888  ± 86 1007  ± 203 4350  ± 1950 Cytosine Kd(app) 4.4  ± 0.7 42  ± 10 10.9  ± 1.6 106  ± 31 144  ± 32 ND Bmax 1000  ± 40 520  ± 52 2280  ± 138 1298  ± 324 2390  ± 330 a The values for the pAB4 and the pAB25 strains are from Brèthes et al. (9).

In terms of K_d(app) and as already observed with base uptake, the pTF2 mutant (N377G) behaves in a way very similar to that of the triple mutant, while the pTF1 double mutant (I371V,I375V) behaves like the wild type, even though K_d(app) values for cytosine and hypoxanthine are slightly increased.

With pTF2 PMF, N377G replacement resulted in a small but significant increase in K_d(app) for adenine (2.8-fold) and a large increase for cytosine and hypoxanthine (24-fold for these 2 bases) as compared with the wild type.

For pTF2-20 and pTF2-10 (N374I and N377I, respectively), K_d(app) were dramatically increased for the 3 bases tested. K_d(app) for adenine was the same in these two strains and was 2 orders of magnitude higher than that of the wild type. This decrease in affinity was so great for pTF2-20 PMF that any specific binding could hardly be measured for cytosine and hypoxanthine.

Equilibrium binding measurements of hypoxanthine at different pH values were done on PMF from pTF1, pTF2, and pTF2-10, at a constant hypoxanthine concentration (11, 71, and 79 µM, respectively).

As already observed for pAB4 and pAB25 strains (24), the amount of hypoxanthine specifically bound to PMF from pTF1, pTF2, and pTF2-10 increased continuously when the proton concentration was varied from 0.6 to 350 µM (Fig. 3). Since B_max remained the same at any proton concentration tested (data not shown), these variations in the amount of specifically bound hypoxanthine reflected K_d(app) variations. Moreover, these K_d(app) variations cannot be ascribed to a significant change in the hypoxanthine ionization level in the range of pH tested.

1

1

1

Plots of pK_d(app) versus pH are presented in Fig. 4. When possible, dissociation constants of the hypoxanthine binding model were estimated by nonlinear regression using Equation 2 and are displayed in the inset of Fig. 4.

1

1

1

For pAB4, we have shown that K_d was 1.7 µM and that an amino acid residue, displaying a pK₁ = 3.8, might be involved in PCP protonation (24); this pK_a was shifted to a more alkaline value (pK₂ = 4.8) when hypoxanthine was bound. For pAB25, the main effects of the triple mutation were: (i) a large decrease in the affinity of PCP for hypoxanthine (K_d of 14.4 µM); (ii) a shift in the pK₁ of the amino acid residue toward a more acidic pH (pK₁ < 3.1).

Fig. 4 clearly shows that the sole substitution N377G (pTF2) can account for the phenotype of the triple mutant pAB25. This modification induced such a shift of the curve toward a more acidic pH that the top plateau could not be reached under our experimental conditions. Because of the flocculation of PMF, binding measurements could not be made below pH = 3.5. Consequently, accurate values of K₁ and K_d for pTF2 were quite difficult to estimate. Anyway, the data show that pK₁ was shifted toward more acidic pH (pK₁ < 2.7) than in the case of the wild type, whereas pK₂ remained unchanged (pK₂ = 4.3). K_d did not appear to be significantly modified by the N377G substitution, and K_d was dramatically increased.

Owing to the poor binding affinity of PCP for hypoxanthine with PMF from pTF2-10 strain (displaying the N377I substitution) at pH 4.5 (Table II) and since K_d(app) increased with increasing proton concentration, specifically bound ligand could not be accurately measured with this strain at pH > 4.5. This, together with the impossibility of doing binding measurements at pH below 3.5, made the calculations of the dissociation constants for hypoxanthine binding highly speculative. Clearly, horizontal plateaus were not reached. Anyway, as compared with the wild type, the curve was clearly shifted toward more acidic pH. In addition, experimental data were scattered in a narrow window rendering impossible any fit with the theoretical curve. Nevertheless, within this window, pK_d(app) varied almost linearly as a function of pH and with a slope of about 0.8 (i.e. closed to 1). This suggests that the difference between pK₂ and pK₁ is at least equal to 2.5 pH units and that pK₁ is lower than 3. Thus, the hypoxanthine binding profile observed with pTF2-10 PMF is partially due to a modification of the acidic dissociation constants of the proton (K₁ and/or K₂) of PCP. This behavior is very similar to that induced by the N377G substitution in the pTF2 strain.

Since the affinity of the PCP for hypoxanthine was too low (Table III), these experiments were not feasible with PMF from the pTF2-20 strain.

DISCUSSION

The study of uptake and binding parameters on several PCP mutants displaying amino acid changes in the 371-377 segment has allowed us to show clearly that the observed phenotype of the double mutant strain pTF1 (I371V,I375V) is the same as the wild type one and that the simple mutant pTF2 (N377G) has the phenotype of the triple mutant pAB25 (I371V,I375V,N377G). Therefore, the phenotype of the triple mutant is only due to the substitution N377G. This amino acid residue change leads to PCP being able to discriminate the 3 bases, and this effect seems, at least partially, to be due to modification of the binding step for cytosine and hypoxanthine. Moreover, as the N377G substitution did not dramatically modify the K_d(app) for adenine, we may assume that the amido group of this Asn residue does not play a crucial role in the binding of adenine.

Another selected mutant showing the N377I substitution (pTF2-10) was characterized by larger increases in both K_t(app) and K_d(app) for hypoxanthine than that observed for the N377G substitution and, in addition, a large increase in these parameters for adenine. Thus, even though the amido group of Asn-377 does not play a crucial role in the binding of adenine, its substitution by a large hydrophobic residue like Ile had a significant effect on the K_d(app) of the purine bases. This suggests that steric hindrance of the Ile residue would affect the access of adenine to its binding site. Finally, whereas the substitutions N377G and N377I had distinct effects on purine bases, they still had the same effect on cytosine binding. A possible explanation for these results could be that owing to its structure, cytosine might not show the same sensitivity to steric hindrance of the residue 377.

Interestingly enough, another spontaneous mutant (pTF2-20, displaying the N374I substitution) also showed an Asn change in the same region. If we compare the results obtained with the pTF2-10 mutant, it appears that N374I and N377I displayed quite similar phenotypes, N374I being the change for which the K_t(app) and K_d(app) for cytosine and hypoxanthine were the most dramatically increased. These results raise the question as to whether these two residues belonging to the sequence segment 371-377 (I-A-N-N-I-P-N) play a similar role for the proper conformation of the active PCP.

Moreover, all the results obtained with these mutants were modifications of uptake and binding constants, which implied changes of the tertiary structure of the active PCP. We have investigated this hypothesis by studying the effects of these two Asn residue replacements on the binding of hypoxanthine as a function of external pH.

Again, our results on pTF1 and pTF2 confirm that the sole N377G substitution can account for the phenotype of the triple mutant pAB25.

Surprisingly, for pTF2, K_d does not seem to be significantly modified. The main effect of the N377G substitution is a dramatic increase in K₁ and K_d. Thus, it seems that this amino acid residue change has a dramatic effect on the free, deprotonated mutated PCP, which binds both hypoxanthine and proton with lower affinities than the wild type forms.

The nature of the PCP amino acid residue whose pK_a is modified by the binding of the base remains to be elucidated. However, from the calculated pK₁ for the wild type strain (pK₁ = 3.8), it should be a carboxylic group. Since the pK_a value of an amino acid residue reflects its environment, the shift of pK₁ due to N377G substitution may be explained by a modification of the environment of an amino acid residue of free and deprotonated PCP. This shift could be due to a conformational change (nearby or at a distance) induced by the replacement of Asn-377 by Gly. Many kinds of electrostatic interactions may be involved in the pK_a variations of an amino acid residue in a protein, as compared with its pK_a in pure water (29, 30). The shift of pK₁ toward more acidic pH may reflect the protection of the assumed carboxylic group from complete protonation, owing to stabilization of its basic form. In this hypothesis, a possible explanation for the observed phenomenon is that the N377G substitution may induce exposure of this ionizable group toward a more hydrophilic environment. However, we still do not know whether this group is located near or far from the Asn-377 residue. In contrast to what is observed for pK₁, it should be pointed out that pK₂ is not modified by the N377G substitution. This means that hypoxanthine binding restores in the mutated PCP an environment of this ionizable group very similar to that of the wild type.

The effect of N377I change on the proton dissociation constants of the PCP is not well determined. Nevertheless, since pK₁ is largely shifted toward acidic pH (pK₁ < 3), it seems that this mutation also modifies the environment of the ionizable group. However, in this case, it is not possible to know if this replacement affects only the carboxylic group environment in the free permease and/or also in the liganded permease, since pK₂ could not be estimated.

Taken altogether, these results argue for the role of the 371-377 segment in the maintenance of a functional conformation of the PCP. This is strengthened by the presence of a Pro residue at position 376. Site-directed mutagenesis constructs in this region are under study and should allow a more accurate investigation of the role of this segment in the tertiary structure of the active PCP.

Our present aim is to determine whether or not this segment is part of the base binding site of PCP. Specific covalent labeling of PCP by 8-azido-[2-³H]adenine (7, 21) and subsequent isolation of radiolabeled peptides of the permease together with mutant strain analyses would allow better characterization of some of the constitutive elements of the binding site.