Divalent cations and polyamines bind to loop 8 of 14-3-3 proteins, modulating their interaction with phosphorylated nitrate reductase


  • Gurdeep S. Athwal,

    1. US Department of Agriculture, Agricultural Research Service, and Departments of 1Horticultural Science and 2Crop Science, North Carolina State University, Raleigh, NC 27695-7631, USA
    Search for more papers by this author
  • and 1 Steven C. Huber 2,

    Corresponding author
    1. US Department of Agriculture, Agricultural Research Service, and Departments of 1Horticultural Science and 2Crop Science, North Carolina State University, Raleigh, NC 27695-7631, USA
    Search for more papers by this author

* For correspondence (fax +1 919 856 4598; e-mail steve_huber@ncsu.edu).


Binding of 14-3-3 proteins to nitrate reductase phosphorylated on Ser543 (phospho-NR) inhibits activity and is responsible for the inactivation of nitrate reduction that occurs in darkened leaves. The 14-3-3-dependent inactivation of phospho-NR is known to require millimolar concentrations of a divalent cation such as Mg2+ at pH 7.5. We now report that micromolar concentrations of the polyamines, spermidine4+ and spermine3+, can substitute for divalent cations in modulating 14-3-3 action. Effectiveness of the polyamines decreased with a decrease of polycation charge: spermine4+ > spermidine3+ >>> cadavarine2+ ≈ putrescine2+ ≈ agmatine2+ ≈ N1-acetylspermidine2+, indicating that two primary and at least one secondary amine group were required. C-terminal truncations of GF14ω, which encodes the Arabidopsis 14-3-3 isoform ω, indicated that loop 8 (residues 208–219) is the likely cation-binding site. Directed mutagenesis of loop 8, which contains the EF hand-like region identified in earlier studies, was performed to test the role of specific amino acid residues in cation binding. The E208A mutant resulted in a largely divalent cation-independent inhibition of phospho-NR activity, whereas the D219A mutant was fully Mg2+-dependent but had decreased affinity for the cation. Mutations and C-terminal truncations that affected the Mg2+ dependence of phospho-NR inactivation had similar effects on polyamine dependence. The results implicate loop 8 as the site of divalent cation and polyamine binding, and suggest that activation of 14-3-3s occurs, at least in part, by neutralization of negative charges associated with acidic residues in the loop. We propose that binding of polyamines to 14-3-3s could be involved in their regulation of plant growth and development.


The 14-3-3 proteins are ubiquitous among eukaryotes and function as sequence-specific and usually phosphorylation-dependent binding proteins. In plants, the 14-3-3s bind to a variety of enzymes including nitrate reductase (NR; Bachmann et al., 1996b,c; Moorhead et al., 1996); sucrose-phosphate synthase (SPS; Moorhead et al., 1999; Toroser et al., 1998); the plasma membrane H+-ATPase (; Olsson et al., 1998); and several others including glutamine synthase 1/2 and trehalose-phosphate synthase (Moorhead et al., 1999). It appears that, in most cases, the 14-3-3s bind to proteins containing the sequence RXXpSXP, as originally described by Muslin et al. (1996). With NR, the binding of 14-3-3s to form an inactive complex requires phosphorylation of Ser543 in the hinge 1 region of NR (Bachmann et al; Douglas et al., 1995), which completes the canonical binding motif RTApS543 TP in the phosphorylated form of NR (phospho-NR).

However, phosphorylation of NR may not be the only factor controlling the binding of the 14-3-3. Specifically, binding of 14-3-3s to phospho-NR requires millimolar concentrations of a divalent cation such as Mg2+, Mn2+ or Ca2+ (Kaiser and Brendle-Behnisch, 1991). Binding of the divalent cation to the 14-3-3 causes a conformational change that can be monitored as an increase in surface hydrophobicity with the fluorescent probe bis-ANS [4,4′-bis(1-anilinonapthalene 8-sulfonate)] (Athwal et al., 1998a). We postulated that the conformational change reflects an ‘opening’ of the ligand-binding groove, because divalent cations strongly stimulated the binding of 14-3-3s to synthetic phosphopeptides free in solution or surface-immobilized (Athwal et al., 2000). The binding of divalent cations to 14-3-3s has been directly shown. Lu et al. (1994) reported that recombinant Arabidopsis GF14ω bound an equimolar amount of Ca2+ with an apparent binding constant of 5.5 × 104 m−1. Subsequently, induced Tb3+ fluorescence was also demonstrated with GF14ω and yeast BMH1, providing additional evidence for a divalent cation-binding site on 14-3-3s (Athwal et al., 2000). In this system, Tb3+ binds to the divalent cation-binding site and can be induced to fluoresce at 545 nm by energy transfer from a nearby aromatic residue that has been excited (McNemar et al., 1991).

The metal-binding site has been generally localized to the last 60 amino acid residues of the C-terminus of the protein (Lu et al., 1994). The native 14-3-3 molecule is a dimer, with each monomer consisting of nine antiparallel helices (Liu et al., 1995; Xiao et al., 1995). The region encompassing loop 8 (between helices 8 and 9) was identified in early studies as having an EF hand-like region that might be responsible for the Ca2+ binding that was observed (Lu et al., 1994). The loop contains five acidic residues, as well as seven oxygen-containing residues that could participate in co-ordination of the divalent cation. The objective of the present study was to generate and use C-terminal truncations and directed mutants of residues within loop 8 in order to determine whether this region is actually involved in cation binding. The residues selected for mutagenesis were Glu208 and Asp219, because these residues occur at the x and –z co-ordinating positions (using the convention applied to EF hands; da Silva and Reinach, 1991), and are present close to the beginning and end of the loop. Using directed mutagenesis, we converted these two residues, both singly and in combination, to alanine to produce three mutant proteins (E208A, D219A, E208A/D219A) to test the role of loop 8 in cation-dependent responses. The results obtained are consistent with the notion that the loop functions in cation binding, and suggest that at least part of the significance of cation binding may be neutralization of negative charges associated with certain acidic residues of the loop. We also report that the naturally occurring polyamines spermidine3+ and spermine4+ can substitute for divalent cations in promoting the inactivation of phospho-NR by 14-3-3s. The polyamines also appear to interact with loop 8 of GF14ω. We suggest that polyamines may express as least part of their biological action by binding to 14-3-3s, thereby promoting their interactions with various target polypeptides.


Polyamines promote inhibition of phospho-NR by 14-3-3s

At pH 7.5 the inhibition of phospho-NR by 14-3-3 proteins requires millimolar concentrations of a divalent cation such as Mg2+ (Kaiser and Brendle-Behnisch, 1991; Kaiser et al., 1991). The naturally occurring polyamines spermidine4+ and spermine3+ can substitute for divalent cations in facilitating the inhibition of pNR by 14-3-3s. As shown in Figure 1(a), spermidine3+ had no effect on activity of dephosphorylated (dephospho-) NR in the presence or absence of GF14ω, but inhibited activity of phospho-NR in a GF14ω-dependent manner (Figure 1b). The maximum inhibition of phospho-NR activity observed (≈70%; Figure 1b) was similar to that achieved with excess Mg2+ (not shown), and presumably reflects the relative amount of the phosphorylated form of NR in the preparation. Thus spermidine3+ could mimic divalent cations in promoting the 14-3-3-dependent inhibition of phospho-NR activity. Identical results to those shown in Figure 1(a,b) were obtained using a mixture of endogenous 14-3-3 isoforms partially purified from spinach leaves (data not shown). All the experiments conducted with polyamines included 1 mm EDTA to ensure that the inhibition was not caused by contaminating divalent cations. Our results are consistent with those of a recent study (Provan et al., 2000) that reported spermidine-dependent inhibition of tobacco NR in the presence of yeast 14-3-3.

Figure 1.

Polyamines promote the 14-3-3-dependent inhibition of phospho-NR activity.

(a) No effect of spermine4+ on activity of dephospho-NR in the presence or absence of GF14ω.

(b) Spermine4+-dependent inhibition of phospho-NR activity in the presence of GF14ω.

(c) Polyamine specificity for inhibition of phospho-NR in the presence of GF14ω.

(d) Structures of polyamines used. Put, putrescine2+ Spd, spermidine3+ Spn, spermine4+ Ace, N1-acetyl spermidine2+ Agm, agmatine2+.

The specificity of the polyamine activation was examined. As shown in Figure 1(c), putrescine2+, N1-acetylspermidine2+ and agmatine2+ were significantly less effective than spermidine3+ and spermine4+. Thus there was considerable specificity apparent, suggesting that the action of the polyamines was not simply the result of binding of a positively charged molecule to the 14-3-3 protein in a non-specific manner.

C-terminal truncations of GF14ω identify loop 8 as the cation-binding site

While polyamines mimic the action of divalent cations (Figure 1), a priori it is not clear whether the cations bind to the 14-3-3 protein and/or to phospho-NR. As discussed in the Introduction, there is considerable evidence that the divalent cation-binding site on GF14ω is contained within the C-terminal 60 amino acid residues (Lu et al., 1994). In order to further localize the cation-binding site, we produced several C-terminal truncations and tested the ability of the mutant proteins to bind the lanthanide cation Tb3+. Binding of Tb3+ was monitored as the induction of fluorescence from the bound lanthanide cation upon excitation of a nearby aromatic residue (Athwal et al; McNemar et al., 1991). For effective resonance energy transfer, the aromatic donor residue must be within about 15 nm; of the bound Tb3+ (Snyder et al., 1990). Mutant proteins were produced from which progressively larger portions of the C-terminus were removed (Figure 2a). Mutant proteins terminating at Trp234 (T-1 truncation); Asp219 (T-2 truncation); or Glu208 (T-3 truncation) were produced; the truncations progressively removed the C-terminal tail, helix 9, and loop 8, respectively. The location of the two tryptophan residues in GF14ω and the three C-terminal tyrosine residues are also identified in Figure 2(a). When the 14-3-3 proteins were incubated with Tb3+ and irradiated with 273 nm light to excite tyrosine residues, fluorescence emission at 545 nm from bound Tb3+ was observed for the wild type and the T-1 and T-2 truncation mutants, but not from the T-3 truncation (Figure 2b). These results suggest that Tyr217 (missing only from the T-3 truncation) was the primary energy donor to bound Tb3+ when excitation was at 275 nm. Furthermore, the results suggest that loop 8 also contains the cation-binding site. If the cation-binding site was actually in residues N-terminal to loop 8 (for example, helix 8), it would be expected that fluorescence energy transfer from Tyr184Tyr185 would be observed, and it was not. When tryptophan residues were excited with 295 nm light, energy transfer to bound Tb3+ was observed for the wild type and T-1 truncation mutant, but not for the T-2 or T-3 truncations (Figure 2b). These results support the notion that Trp234 (missing from the T-2 and T-3 truncations) was the primary energy donor to bound Tb3+ when excitation was at 295 nm.

Figure 2.

Localization of the divalent cation-binding site to loop 8 of GF14ω.

(a) Schematic representation of the recombinant truncation mutants used in the present study. Black blocks under the wild-type sequence identify the nine α-helices; stippled blocks indicate the location of the EF hand-like domain (loop 8). The locations are identified of the two tryptophan residues in the molecule and tyrosine residues in the C-terminal 75 residues of GF14ω. As described in Experimental procedures, we constructed three truncation mutants: T-1, wild-type sequence lacking the C-terminal tail; T-2, lacking helix 9 and the C-terminus; and T-3, lacking loop 8 and residues beyond.

(b) Fluorescence emission at 545 nm (relative units) of wild-type GF14ω and truncation mutants complexed with Tb3+ upon excitation of tyrosine (Tyr Ex) or tryptophan (Trp Ex) residues with 273 and 295 nm light, respectively. The 14-3-3 protein concentration was 0.5 µm; Tb3+ concentration was 20 µm. For tyrosine excitation, slits were set at 1.5 nm for excitation and 3.0 nm for emission. For tryptophan excitation, emission was much weaker and slits were increased to 10 nm for both excitation and emission. Thus there was a 20-fold difference in sensitivity for the two measurements. In each case, values were corrected for the low intrinsic fluorescence of the 14-3-3 protein in the absence of Tb3+.

The fluorescence intensity from bound Tb3+ was considerably greater with the wild-type GF14ω when the protein was excited with 273 nm light compared to 295 nm light. Consequently, in the experiment presented in Figure 2(b), slits on the spectrofluorophotometer were set to give a 20-fold difference in sensitivity for the two wavelengths studied. This is consistent with the notion that loop 8 contains both the cation-binding site and Tyr217, which is the primary residue responsible for energy transfer when excitation is with 273 nm light. Moreover, as will be shown elsewhere (G.S.A. and S.C.H., unpublished results), substitution of Tyr217 by directed mutagenesis to produce the Y217S or Y217A mutant proteins dramatically reduces Tb3+ fluorescence, confirming Tyr217 as the most efficient energy donor. It is also interesting to note that fluorescence from Tb3+ bound to the T-1 truncation was significantly higher compared to the wild type when energy transfer was from Trp234, but was identical to wild type when energy transfer was from Tyr217 (Figure 2b). This may indicate that the absence of the C-terminal tail in the T-1 truncation mutant allows for a structural or conformational change that enhances energy transfer from Trp234 to Tb3+. Collectively, the results are consistent with the postulate that loop 8 is the cation-binding site.

Loop 8 is essential for divalent cation- and polyamine-induced conformation change of 14-3-3s

We have reported that addition of Mg2+ to recombinant GF14ω resulted in a conformational change that increased surface hydrophobicity, as monitored by increased fluorescence of the environmentally sensitive probe, bis-ANS (Athwal et al., 1998a). This provides an additional functional assay for the binding of divalent cations that could be used in the analysis of the truncation mutant proteins to localize the cation-binding site. As shown in Table 1, the wild type and T-1 and T-2 truncation mutants responded to addition of 5 mm Mg2+ with an increase in bis-ANS fluorescence of 42–49%. However, the bis-ANS fluorescence of the T-3 truncation was essentially unaffected by addition of Mg2+. The basal bis-ANS fluorescence of all three truncations (in the absence of Mg2+) was similar to, or slightly higher than, the wild type (Table 1). Therefore it appears that the bis-ANS binding site(s) themselves were not removed in the T-3 truncation mutant. Rather, these results indicate that residues between Glu208 and Asp219 (loop 8) are required for binding of Mg2+ and/or for the Mg2+-induced conformational change. The former explanation is likely, as loop 8 was also identified as the probable cation-binding site by analysis of Tb3+ fluorescence (Figure 2).

Table 1.  Relative enhancement by Mg2+ and spermine4+ of bis-ANS fluorescence in wild type and C-terminal truncation mutants of GF14ω
14-3-3bis-ANS fluorescence
(–Mg2+)a(+Mg2 ± Mg2+)b(+Spn/–Spn)b
  1. a Fluorescence emission in the presence of bis-ANS but in the absence of Mg2+ and spermine 4+ (Spn), expressed in relative units.

  2. b Relative enhancement of bis-ANS fluorescence emission after addition of 5 mmMg2+ or 1 mm spermine4+ (Spn) as indicated (dimensionless units). Reaction mixtures contained 0.1 µm 14-3-3 protein, and 1.0 µm bis-ANS, at pH 7.5. Excitation was at 385 nm and emission was measured at 480 nm with 5 nm slits.

WT73.9 ± 1.61.49 ± 0.041.78 ± 0.03
T-185.2 ± 1.81.42 ± 0.021.64 ± 0.03
T-2161.2 ± 2.11.48 ± 0.021.68 ± 0.01
T-3118.6 ± 1.41.05 ± 0.011.07 ± 0.01

Addition of spermine4+ to wild-type 14-3-3 also resulted in an increase in bis-ANS fluorescence. This indicates that spermine4+ does in fact bind to the 14-3-3s, and caused a conformational change similar to that induced by divalent cations. A similar level of enhancement of bis-ANS fluorescence was observed for the wild type and T-1 and T-2 truncation mutants (about 70%), but that of the T-3 truncation was much less affected (Table 1). Thus loop 8 (missing only from the T-3 truncation) was also essential for the spermine4+-induced conformational change. Collectively, these results are consistent with the suggestion that the cation-binding site is contained within loop 8.

The C-terminal tail is required for inhibition of phospho-NR

Surprisingly, all the C-terminal truncation proteins were dramatically less inhibitory to phospho-NR compared to the wild-type protein (Figure 3). The T-1 truncation, which lacks only the C-terminal tail, produced less than 12% of the inhibition obtained with the wild-type 14-3-3 protein. Further truncation of 14-3-3 reduced the inhibition slightly further. Thus the C-terminal tail (residues 235–259) was essential for proper interaction of the 14-3-3 protein with the target protein, phospho-NR. In other experiments we added a synthetic peptide (corresponding to residues 234–253 of the C-terminal tail of GF14ω) to the T-1 truncation protein, in an attempt to determine whether the unattached sequence could restore inhibition of phospho-NR. The synthetic peptide, which had no effect on phospho-NR activity by itself, did increase the inhibition of the T-1 truncation (to approximately 30% of the inhibition caused by the wild-type protein). However, addition of the peptide to wild-type GF14ω also resulted in an equivalent increase in inhibition of phospho-NR activity (data not shown). Thus the presence of the synthetic peptide caused a general increase in the inhibitory action of the 14-3-3 proteins that was not specifically related to the lack of this region in the T-1 truncation. We tentatively conclude that the C-terminal tail is essential for inhibition of phospho-NR, but that the sequence must be attached to helix 9 in order to function properly.

Figure 3.

Inhibition of phospho-NR activity by the wild-type and C-terminal truncation mutants of GF14ω.

Relative inhibition, expressed as a percentage of the inhibition produced by the wild type 14-3-3, is shown parenthetically. The 14-3-3s were present at a concentration of 1 µm, and activity of phospho-NR was assayed at pH 7.5 in the presence of 5 mm Mg2+.

Deletion of the C-terminal tail did not result in gross structural changes in GF14ω

In the case of mutant proteins that had a reduced ability to inhibit phospho-NR, such as the T-1 truncation mutant, it is essential to demonstrate that gross structural changes in the 14-3-3 protein were not responsible for the lack of inhibition. With the 14-3-3s, two aspects are of concern: the ability of the subunits to dimerize (Liu et al., 1996); and the α-helical content of the protein (Wang et al., 1998). We used CD spectroscopy to assess possible differences in protein secondary structure caused by truncation. The three-dimensional crystal structure (Liu et al., 1995; Xiao et al., 1995) and subsequent CD spectroscopic analysis of mammalian 14-3-3ζ protein (Wang et al., 1998; Zhang et al., 1997) has confirmed that α-helices are the major structural component of the protein. Consistent with these results, the far-UV CD spectra of the wild type and T-1 truncation mutant were superimposable, with ellipticity minima observed at ≈208 and ≈220 nm (Figure 4). The α-helical contents of the proteins were calculated to be about 39%. In contrast, the α-helical content of the T-2 and T-3 truncation mutants was reduced slightly (to about 30%), which may reflect some alteration in structure (Figure 4).

Figure 4.

Far-UV CD spectra of wild-type GF14ω and C-terminal truncation mutants.

The wild-type and T-1 mutant spectra were identical; and the T-2 and T-3 spectra were similar and distinct from the wild type, suggesting some change in secondary structure.

Previous studies (Liu et al., 1998) have demonstrated that dimerization of subunits is essential for 14-3-3 binding to ligands. Consequently, it was important to determine whether any of the mutations that affected inhibition of phospho-NR also disrupted dimerization. We examined dimer formation using native PAGE. In most cases, the dominant Coomassie-stained protein band corresponded exclusively to the dimeric form of the protein, and the proteins were >95% homogeneous (data not shown). The only exceptions were the T-2 and T-3 truncation mutants, which also occurred, paradoxically, as higher molecular-weight forms that remained completely soluble but clearly did not reflect the native conformation of the protein (data not shown). However, the T-1 truncation mutant was similar to the wild type with respect to dimerization. Thus, for the T-1 truncation mutant, the disrupted interaction with phospho-NR cannot be attributed to gross structural changes. However, some structural alterations may have contributed to the inability of the T-2 and T-3 truncation mutants to inhibit phospho-NR. But it is important to note that the structural changes associated with the T-2 and T-3 truncation mutants cannot explain the loss of Tb3+ fluorescence (Figure 2) or bis-ANS fluorescence (Table 1) associated specifically with removal of loop 8.

Directed mutants of GF14ω loop 8 alter interactions with cations

Directed mutants of loop 8 were produced to test the role of specific residues (Glu208 and Asp219) in binding polycations (Figure 5a). These residues were selected because Glu208 and Asp219 correspond to the x and –z co-ordinating positions using the convention typically applied to EF hands. The mutant proteins were tested for their ability to inhibit phospho-NR in the presence and absence of Mg2+. In the absence of Mg2+, the wild-type 14-3-3 caused no inhibition, as expected. Of the three mutants, the most surprising result was that the E208A mutant caused significant Me2+-independent inhibition. The D219A mutant caused only a slight inhibition in the absence of Mg2+ (Figure 5b). The presence of Mg2+ increased the inhibition caused by all the 14-3-3s. Relative to the wild type, inhibition by the E208A mutant was slightly greater and that of the D219A mutant was slightly less (Figure 5c). The conservative substitution of a glutamate for an aspartate residue at the –z position in the D219E mutant did not alter the inhibition of phospho-NR activity, relative to wild-type protein, in the absence or presence of Mg2+ (data not shown). However, the E208A/D219E double mutant inhibited phospho-NR activity approximately 25% in the absence of Mg2+ (data not shown). Thus the characteristic effect of the E208A mutation was also apparent when incorporated as a double mutant with the D219E mutation. These results confirm the observation that the E208A mutation results in significant Mg2+-independent inhibition of phospho-NR. The reduction in Mg2+-dependence by substitution of Ala for an acidic residue in loop 8 suggests that the loop is, indeed, involved in divalent cation-binding, and that the significance of cation-binding may be to neutralize the negative charge associated with the acidic residue(s).

Figure 5.

Directed mutants of loop 8 of GF14ω alter the divalent-cation dependent inhibition of phospho-NR activity.

(a) Schematic representation of loop 8 showing the x,y,z co-ordinates assigned to functional EF hands, and the substitution mutations made in the present study.

(b) Effect of wild-type and mutant GF14ω proteins on phospho-NR activity assayed in the absence of Mg2+.

(c) Effect of 14-3-3 proteins on phospho-NR activity assayed in the presence of 5 mm Mg2+.

The loop 8 directed mutants were tested for their dependence on polycation concentration for the inhibition of phospho-NR. As shown in Table 2, the wild-type 14-3-3 inhibited phospho-NR, and the concentration of Mg2+ required for half-maximal inhibition (I50) was approximately 4.4 mm. Lower concentrations of the polyamines were required for half-maximal inhibition. With spermidine3+ the I50 was ≈0.6 mm, and with spermine4+, ≈0.05 mm. Thus I50 decreased in the order: spermine4+ < spermidine3+ < Mg2+, with a roughly 10-fold change in apparent affinity for each additional positive charge. Relative to the wild-type 14-3-3, the E208A mutant protein had a several-fold reduction in the concentration of Mg2+ or polyamine required for half-maximal inhibition of phospho-NR. In contrast, the opposite was observed for the D219A mutant, which required roughly two- to threefold higher concentrations of the polycations for half-maximal inhibition of phospho-NR activity. The E208A/D219E double-mutant protein generally behaved similarly to the E208A single mutant (Table 2). The most important point is that the apparent affinity for all three cationic activators was affected in similar manner by the non-conservative substitutions at Glu208 and Asp219. This suggests that the polyamines, like Mg2+, bind directly to the 14-3-3 protein, and furthermore that the binding sites are either identical or overlapping and involve acidic residues in loop 8.

Table 2.  Directed mutagenesis of loop 8 residues of GF14ω affects the concentration of polycation required to produce half-maximal inhibition of phospho-NR activity (I50)
14-3-3I50 for cofactor (mm)
  1. Reactions contained 0.5 µg of the indicated 14-3-3 protein at pH 7.5. Spd, spermidine; Spn, spermine.

WT4.4 ± 0.30.62 ± 0.040.053 ± 0.003
E208A1.0 ± 0.10.12 ± 0.010.023 ± 0.003
D219A9.5 ± 0.41.9 ± 0.10.170 ± 0.006
E208A/D219A1.7 ± 0.20.12 ± 0.020.016 ± 0.001

The loop 8 mutations that affected the apparent affinity for Mg2+ or polyamines for inhibition of phospho-NR activity also affected induced Tb3+ fluorescence. At a Tb3+ concentration of 50 µm, fluorescence emission at 545 nm was reduced 65% by the E208A mutation and 87% by the D219A mutation relative to the wild-type 14-3-3 protein (data not shown). These results support the conclusion that loop 8 is the polycation-binding site.


The results obtained in the present study using directed mutants and C-terminal truncations have identified residues and regions on GF14ω that are important for the inactivation of phospho-NR. Three main conclusions emerge: (i) polyamines bind to 14-3-3s, which may be one mechanism by which they control gene expression (Yamakawa et al., 1998) and thereby act as plant growth regulators (Kumar et al., 1997; Walden et al., 1997); (ii) loop 8 of GF14ω is the likely polycation-binding site, involved in binding both divalent cations and polyamines; and (iii) the C-terminal tail of GF14ω is required for inhibition of phospho-NR activity.

Polyamines bind to 14-3-3s

The polyamines – naturally occurring polycations – can mimic divalent cations and promote the 14-3-3-dependent inhibition of phospho-NR (Figure 1). Like the divalent cations, the polyamines bind to the 14-3-3s and induce a conformational change that is manifested as an increase in surface hydrophobicity (Table 1). Not all polyamines are effective. Specificity studies (Figure 1c) indicated that effective compounds must possess two charged primary amino groups as well as at least one charged secondary amine group. The absence of a secondary amine group, as in putrescine2+, rendered the compound ineffective. Similarly, substitutions on the primary amine group, as in N1-acetyl spermidine2+ and agmatine2+, dramatically reduced the effectiveness of the polyamine. The requirement for charged amine groups for the 14-3-3-dependent inhibition of phospho-NR activity suggests that electrostatic interactions between the polyamine and 14-3-3 protein may be important. However, it is also likely that other interactions, such as hydrogen bonding, may also be involved, as demonstrated for the binding of polyamines to tRNA (Frydman et al., 1992). Thus the notion that polyamines are acting only as ‘organic cations’ is probably too simplistic.

Spermine4+ was effective at micromolar concentrations (Table 2), suggesting that interactions with 14-3-3s could occur at concentrations that have physiological relevance (Kumar et al., 1997). Whether polyamines interact with 14-3-3s in vivo might also be a function of cytosolic [Mg2+], as both polycations appear to bind to the same site (discussed below). Although precise measurements of cytosolic [Mg2+] in plant cells are lacking, there is a general notion that intracellular free Mg2+ is maintained below 1 mm in eukaryotic cells (Grubbs and Maguire, 1987). These in vivo estimates are below the concentrations required for half-maximal inactivation of phospho-NR by wild-type GF14ω (≈4 mm; Table 2). Thus one of the targets of polyamines in vivo could be the 14-3-3 proteins, which interact with a variety of metabolic enzymes and serve as components of trans factor complexes (Chung et al., 1999). Conceivably, binding of 14-3-3s could be one component of polyamine action. In plants, polyamines have also been demonstrated to inhibit lipoxygenase-1 (Maccarrone et al., 1998); to activate glucan synthase (Kauss and Jeblick, 1985); to bind to several plasma membrane proteins (Tassoni et al., 1998); and to inhibit the KAT1 K+ channel of guard cell plasma membranes (Liu et al., 2000) and vacuolar ion channels (Dobrovinskaya et al., 1999). These may also be important sites of action of polyamines in vivo. In many cases, effectiveness of the polyamines reflected their charge, and concentrations required for half-maximal effects decreased in the order: spermine4+ < spermidine3+ < putrescine2+ ≈ cadavarine2+ (Dobrovinskaya et al., 1999; Maccarroneet al., 1998; this study).

Loop 8 of GF14ω is the likely cation-binding site

Early studies with GF14ω (Lu et al., 1994) determined that the protein binds Ca2+ at a single site with mm affinity, and that the site resides in the C-terminal 59 amino acids (residues 200–259). Our results suggest that the site can be localized to loop 8 (residues 208–219). This evidence includes: (i) the site is in the vicinity of Trp234 (C-terminus) and not Trp63 (N-terminus) (Figure 2); (ii) loop 8 was essential for Tb3+ fluorescence (Figure 2) and for the divalent cation- or polyamine-induced conformational change (Table 1); and (iii) directed mutagenesis of loop 8 residues in GF14ω altered the cation dependence of inhibition of phospho-NR (Figure 4; Table 2) and reduced Tb3+ fluorescence (data not shown). The simplest explanation that is fully consistent with all the results is that polycations (divalent cations and polyamines) bind to residues contained within loop 8. Importantly, the binding site for Mg2+ appears to be identical to, or overlapping with, the site for spermine4+ and spermidine3+, and both involve loop 8 of GF14ω. Because the most effective polyamines, spermine4+ and spermidine3+, are long but flexible molecules, it is likely that they interact in a folded or bent manner with acidic groups in loop 8 of GF14ω. Polyamine-binding sites have been identified in several other proteins. In the case of the K+ channel IRK1 (Taglialatela et al., 1995) and the regulatory subunit of casein kinase-2 (Leroy et al., 1995), the polyamines are thought to bind to two acidic residues: for IRK1, these two acidic residues are distant in the primary structure (Asp172 and Glu224), whereas for the regulatory subunit of casein kinase, two glutamate residues, separated by three intervening amino acids, constitute the binding site. At least in the case of IRK1, the acidic amino acid residues involved in polyamine binding also form a binding pocket for Mg2+. The 14-3-3s appear to be another example where a divalent cation-binding site functions to bind polyamines. Additional mutagenesis of loop 8 residues will be required to completely elucidate the residues involved in binding Mg2+ and polyamines. It is conceivable that the sites are overlapping but not identical, because Mg2+ has a point charge whereas the polyamines have spatially separated charges. It will also be of interest to determine whether the nature of the polycation (divalent cation versus polyamine) affects the specificity of different 14-3-3 isoforms for their target proteins. There appears to be no difference for the binding of GF14ω to phospho-NR, but conceivably other combinations may be influenced differently.

It is becoming increasingly clear that the binding of 14-3-3s to their target proteins is dependent on polycations at pH 7.5 (this study and others), and is thus regulated by factors in addition to phosphorylation of the target protein. The physiological significance of regulation by divalent cations and polyamines remains an important topic for future studies. It is generally recognized that the polyamine content of plant cells changes developmentally and in response to stress (Evans and Malmberg, 1989; Kumar, 1997; Walden et al., 1997). For example, K+ deficiency (Smith, 1985); salinity (Edei et al., 1990); and osmotic stress (Edei et al., 1990; Flores and Galston, 1984) elevate polyamine levels. It has been proposed that the production of polyamines is associated with specific responses of plants to stress stimuli, and is a key factor in plant cellular protection against these stresses (Flores and Galston, 1984). Nothing is known about the compartmentation of polyamines in plant cells, and it is also not known how much of the total polyamine pool is free and not complexed to cellular constituents. However, the apparent high affinity of 14-3-3s for spermine4+ (I50 = 50 µm) is of the same order of magnitude as the spermine4+ concentration in barley leaves (Flores and Galston, 1984; Smith, 1985). It is tempting to speculate that polyamine accumulation under conditions of excess reduced nitrogen (Slocum and Weinstein, 1990) might function to bolster the 14-3-3-dependent inhibition of phospho-NR and thereby prevent additional nitrate reduction.

The C-terminal tail of GF14ω is required for inhibition of phospho-NR

It is recognized that the fundamental structure of the 14-3-3s is a conserved central core with highly divergent N- and C-termini (Chung et al., 1999). This suggests that the termini may play an important role in specialization or diversification of function among isoforms. Consistent with this notion, the T-1 truncation mutant, which lacked only the C-terminal tail, was largely ineffective in inhibition of phospho-NR activity (Figure 3). The T-1 truncation mutant was expressed as a dimer, with secondary structure identical to wild-type GF14ω (Figure 4). The T-2 and T-3 truncation mutants were also unable to inhibit phospho-NR, but with these mutants there were some structural alterations to the 14-3-3 proteins that could also contribute to the inability to properly interact with phospho-NR. Other studies have also demonstrated a role for large portions of the C-terminus of 14-3-3 proteins in binding to target proteins (Ichimura et al., 1995; Luo et al., 1995). Of particular relevance to the present study was the earlier report (Liu et al., 1996) that the 15 C-terminal residues of 14-3-3 τ were required for binding to three different target proteins – Cbl, Raf-1 kinase, and phosphatidylinositol 3-kinase. In contrast, removal of the 33 C-terminal residues from 14 to 3-3 η did not prevent binding to Raf-1 and Bcr protein kinases (Ichimura et al., 1996). Thus the requirement for C-terminal residues may vary with the target protein and the 14-3-3 isoform. Additional C-terminal truncation mutants of GF14ω are being prepared to further clarify the role of the C-terminus in binding to targets such as phospho-NR, and in ‘activation’ by cations such as Mg2+ and polyamines.

Experimental procedures

Plant material

Spinach (Spinacia oleracea L. cv. Bloomsdale and Tyee) plants were grown in soil in a greenhouse during the winter months. Mature leaf tissue was harvested in the morning directly into liquid N2 and stored at −80°C. Experiments were usually carried out two or three times. Typical results from a representative experiment, or mean values (±SE) from several experiments, are presented.

Preparation of phospho-NR and spinach leaf 14-3-3 proteins

Frozen leaf tissue was powdered in a mortar and then extracted [2 ml (g FW)−1] in cold buffer, exactly as described previously (Athwal et al., 2000). Briefly, protein that precipitated from the clarified crude extract between 3 and 13% (v/v) polyethylene glycol-8000 was subjected to fast protein liquid chromatography (FPLC) anion-exchange chromatography using a 50 ml Source 15Q media column (Pharmacia, Piscataway, NJ). Fractions were assayed for maximum NR activity (Bachmann et al., 1995), and the co-eluting NR-kinase activity was assayed using the NR6 synthetic peptide as substrate (Bachmann et al., 1996a). As isolated from illuminated leaves, the NR was largely in the dephosphorylated form. To prepare the phosphorylated form, the NR peak fractions were desalted by centrifugal filtration and incubated in a mixture containing 1 mm ATP, 4 mm MgCl2, 0.1 mm CaCl2, 2 mm DTT and 1 µm microcystin-LR (Calbiochem, La Jolla, CA) in 50 mm 3-[N-morpholino]propanesulfonic acid (MOPS)–NaOH pH 7.5, at 25°C for 1 h. The phosphorylation of Ser543 was monitored as inhibition of NR activity in the presence of excess 14-3-3 proteins and Mg2+ (Bachmann et al., 1996b). The phospho-NR was dialysed into 20 mm MOPS–NaOH pH 7.5, and stored at 4°C prior to use in testing the ability of recombinant 14-3-3 wild-type and mutant proteins to bind and inactivate the enzyme.

NR inhibition assays

Phospho-NR was pre-incubated with the indicated amount of 14-3-3 protein (0–2 µg protein) prior to assay at pH 7.5, in the presence of 0–5 mm MgCl2 or polyamines as specified in the text.

Site-directed mutagenesis and recombinant protein purification

All constructs utilized the pET15b plasmid (Novagen, Madison, WI), containing either the wild-type or mutated Arabidopsis GF14ω gene (Lu et al., 1997). Site-specific and C-terminal truncated mutants were generated by direct mutation using the QuikChange TM Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The GF14ω sequence was analysed for codon usage prior to designing and obtaining both synthetic sense and antisense oligonucleotide primers (IDT, Inc. Coralville, IA). The sense sequences were: 5-CT CTC TGG TGA TCT GAT ATG-3 for the T-1 truncation mutation; 5-TCA TAC AAA GAC TGA ACC TTG ATC ATG-3 for the T-2 truncation mutation; 5-GCG ATT GCA GAG TGA GAC ACT CTT GG-3 for the T-3 truncation mutation; 5-GAG TCA TAC AAA GAG AGT ACC TTG ATC-3 for the D219E mutation; 5-GAG TCA TAC AAA GCT AGT ACC TTG ATC-3 for the D219A mutation; 5-GAG GCG ATT GCA GAT TTG GAC ACT CTT G-3 for the E208D mutation; and 5-GAG GCG ATT GCA GCT TTG GAC ACT CTT G-3 for the E208A mutation. The underlined codons show the nucleotide triplet changes resulting in the amino acid substitutions, or for the truncations the Umber stop codon. The corresponding antisense sequences are available upon request. Briefly, mutations were generated by PCR amplification using the appropriate primers. The PCR reactions were carried out in a final volume of 50 µl containing 10 mm KCl, 10 mm (NH4)2SO4, 20 mm Tris–Cl pH 8.8, 2 mm MgSO4, 0.1% Triton X-100, 0.1 mg ml−1 nuclease-free BSA, 20 ng dsDNA template, 50 µm dNTP mix, 125 ng oligonucleotide primers, and 2.5 units of PfuTurbo DNA polymerase. Amplification was performed over 16 repeated cycles using a Sprint PCR machine (Hybrid, Ashford, Middlesex, UK). The cycling conditions for the PCR reaction following a pre-incubation of 30 sec at 95°C were: denaturation at 95°C for 30 sec, annealing at 55°C for 1 min, and extension at 68°C for 11 min. Unwanted PCR products (methylated and hemimethylated parental supercoiled DNA) were digested by incubation with Dpn1 endonuclease, 10 units, for 60 min at 37°C. The nicked vector DNA incorporating the mutation was then transformed into Epicurian coli XL1-Blue supercompetent cells. These were grown on Amp selective media, and viable colonies were miniprepped (WizardPrep, Stratagene) and the entire GF14ω coding region sequenced (DNA Sequencing Facility, Duke University) and analysed to verify production of only the desired mutation, prior to transformation into BL12 (DE3) (Novagen).

The recombinant proteins were expressed and purified as described previously (Athwal et al., 1998b). Briefly, standard manufacturer's procedures were followed (Novagen) for the expression and purification on an Ni2+-charged iminodiacetic acid-Sepharose column (Hi-Trap Chelating, Amersham Pharmacia Biotech). In initial experiments, removal of the His-tag by thrombin cleavage was found to have no effect on inactivation of phospho-NR, and consequently was not employed routinely. The final purification step involved FPLC-Resource Q chromatography, prior to dialysis into 20 mm MOPS–NaOH pH 7.5 and storage at −80°C until used.

Circular dichroism spectroscopy

Far-UV CD spectra of wild-type GF14ω and mutant proteins were acquired on a PiStar-180 Spectrometer (Applied Photophysics, Leatherhead, Surrey, UK), equipped with a thermally regulated cell holder maintained at 22°C. Spectra were obtained by scanning the region from 190 to 260 nm, with 0.2 nm increments and 1 sec response times using a rectangular fused quartz cuvette with a 0.1 cm optical path. Mean residue molar ellipticities were calculated as described (Johnson, 1990). The protein concentration of all samples was 65 µg ml−1 dissolved in 1 mm MOPS–NaOH buffer pH 7.4, containing 50 mm KCl. All spectra were obtained as the average of three independent scans. Each spectrum was corrected for solvent contribution (scanned in an identical manner and using the same buffer).

Fluorescence spectroscopy

All fluorescence measurements were performed using a Shimadzu RF-5301 PC spectrofluorophotometer, as described previously (Athwal et al., 1998a). Briefly, metal-free wild-type and mutant GF14ω proteins were prepared by extensive dialysis against 10 mm MOPS–NaOH pH 7.5, containing 2.5 mm DTT, 5 mm EDTA and 5 mm EGTA. The protein was then dialysed against the same buffer without EDTA or EGTA. A stock solution of 100 µm bis-ANS (Molecular Probes, Eugene, OR) was prepared and diluted to a final concentration of 1 µm during measurements (unless otherwise stated). A stock solution of 0.5 mm Tb3+ in 10 mm MOPS–NaOH pH 7.5, was made from the highest grade TbCl3.6H2O (Aldrich, Milwaukee, WI) and used without further purification. The final concentrations of GF14ω used for fluorescence measurements are stated in the text, as are the excitation and emission wavelengths used for each chromogenic probe.


The authors thank Dr Ashtosh Tripathy, Macromolecular Interactions Facility, University of North Carolina, Chapel Hill, NC, for advice and assistance in obtaining the CD data presented. This research represents co-operative investigations of the US Department of Agriculture (USDA), Agricultural Research Service, and the North Carolina Agricultural Research Service. The research was supported in part by funds from USDA-NRI (Grant 2001-35318-10185). Mention of a trademark or proprietary product does not constitute a guarantee or warranty of the product by the USDA or the North Carolina Agricultural Research Service and does not imply its approval to the exclusion of other products that might also be suitable.