(Received for publication, November 18, 1996, and in revised form, May 30, 1997)
From the Department of Biochemistry, Stanford
University School of Medicine, Stanford, California 94305-5307 and the
¶ Laboratory of Molecular Genetics, NICHD, National
Institutes of Health, Bethesda, Maryland 20892-2785
High levels of guanosine tetraphosphate (ppGpp) and guanosine pentaphosphate (pppGpp), generated in response to amino acid starvation in Escherichia coli, lead to massive accumulations of inorganic polyphosphate (polyP). Inasmuch as the activities of the principal enzymes that synthesize and degrade polyP fluctuate only slightly, the polyP accumulation can be attributed to a singular and profound inhibition by pppGpp and/or ppGpp of the hydrolytic breakdown of polyP by exopolyphosphatase, thereby blocking the dynamic turnover of polyP. The Ki values of 10 µM for pppGpp and 200 µM for ppGpp are far below the concentrations of these nucleotides in nutritionally stressed cells. In the complex metabolic network of pppGpp and ppGpp, the greater inhibitory effect of pppGpp (compared with ppGpp) leading to the accumulation of polyP, may have some significance in the relative roles played by these regulatory compounds.
Inorganic polyphosphate (polyP),1 a linear polymer of hundreds of phosphate residues linked by high-energy phosphoanhydride bonds, is ubiquitous having been found in all microbes, fungi, plants, and animals examined (1, 2). In Escherichia coli, polyP, which accumulates up to 20 mM (based on Pi residues) in stationary-phase cells,2 is produced from ATP by a membrane-associated enzyme, polyphosphate kinase (PPK) (3).
Mutants lacking PPK are deficient in polyP, fail to adapt to stress, and do not survive in stationary phase (4, 5). A regulatory function is one of the many possible effects of polyP that might account for this essential role. In this regard, the relationship of polyP levels to those of guanosine penta- and tetraphosphate ((p)ppGpp) deserves special attention. Levels of (p) ppGpp rise drastically in response to starvation for amino acids, carbon, or Pi (6, 7). In response to nutritional stress, accompanied by increased levels of (p)ppGpp, large accumulations of polyP have been observed in E. coli,2, 3 Myxococcus xanthus (8), and Pseudomonas aeruginosa3. E. coli mutants that fail to produce (p)ppGpp are also deficient in the accumulation of polyP.3
The present study explores the mechanism of polyP accumulation in nutritionally stressed E. coli and the relationships to (p)ppGpp.
Sources were as follows:
[-32P]ATP was from Amersham Corp.; nonradiolabeled ATP
and ADP and bovine serum albumin were from Sigma;
polyethyleneimine-cellulose (PEI) TLC plate was from Merck; and
creatine phosphate and creatine kinase were from Boehringer Mannheim.
The gppA::kan
deletion-insertion allele in strain CF3376 was constructed as follows.
A plasmid bearing the wild type gppA region was subcloned
from phage
1039 (9). Using synthetic primers, DNA-encoding, amino
acid residues 1-452 of the 496 total in GppA were deleted. A kanamycin
resistance (Km-r) cassette derived from plasmid pUC4K (Pharmacia
Biotech Inc.) was substituted for the deletion. The insertion-deletion
allele was recombined into phage
1039 by phage growth on a
plasmid-bearing strain and the lysate used to lysogenize a wild type
K-12 strain (MG1655) selecting for Km-r. Phage curing and recombinative
transfer of the
gppA allele from phage
to the
chromosome was by heat-pulse curing (10), yielding strain CF3376.
Verification of the deletion in genomic DNA was done by: 1) polymerase
chain reaction amplification from primers flanking the deletion site
and visualization of the shortening of the product due to the presence
of the kan cassette, 2) measurement of increased abundance
of pppGpp relative to ppGpp during the stringent response induced by
serine hydroxamate by a nonuniform labeling procedure (11),
i.e. pppGpp/(pppGpp + ppGpp) = 0.072 (± 0.017) for wild
type versus 0.60 (± 0.084) for mutant, and 3) observing
expected genetic linkage of the Km-r phenotype after P1 phage
cotransduction of Km-r with ilv500::Tn10. Strain
CF3382 is an example of a
gppA::ilv500::Tn10 recombinant.
The ppkx::kan deletion-insertion allele in
strain CF5802 consists of a deletion of the C-terminal portion of
ppk fused with a N-terminal deletion of ppx,
again with a substitution of a kanamycin resistance cassette. In the
ppkx operon this deletion is deduced to remove 98.5% of PPK
and 78% of PPX. The source of the residual ppk sequences
was plasmid pBC29 (after SacI cleavage), while the source of
the residual ppx sequences was the plasmid pBC6 (after PvuII cleavage). After insertion of the pUC4K kan
cassette, the plasmid was linearized and used to transform a
recBC sbc strain of E. coli selecting for Km-r,
yielding strain CF5772. The
ppkx deletion-insertion was
transferred into strain MG1655 by transduction with phage P1, selecting
for Km-r recombinants, which yielded strain CF5802. Verification of the
replacement of wild type ppkx locus by the
deletion-insertion allele in CF5802 was by: 1) polymerase chain
reaction with flanking primers, as above, 2) observation of
dramatically decreased levels of PPK and PPX enzymatic activities in
extracts, and 3) observing the expected genetic linkage of Km-r after
phage P1 transduction to recipients bearing
quaBA::Tn10 (86%) or
zff-208::Tn10 (10%).
Although the gppA and
ppkx alleles are both Km-r, they
could be combined in a single strain by transduction of CF5802
with phage P1 grown on a
gppA ilv500::Tn10
donor (CF3382), selecting for Rc-r and screening recombinants for the
presence of high pppGpp:ppGpp ratios (11), yielding strain CF5986. The
ilv500::Tn10 present in CF5986 was removed by
transduction with phage P1 grown on a wild type MG1655 parental strain,
selecting for prototrophs and screening recombinants for retention of
the gpp-phenotype, yielding strain CF6032
ppkx
gppA.
Cells were
grown at 37 °C on LB medium or MOPS medium (12) and harvested by
centrifugation. The cell pellet was resuspended in 100 µl (per 1-ml
cell culture) of 50 mM Tris-Cl (pH 7.5) and 10% sucrose,
frozen in liquid nitrogen, and stored at 80 °C. Lysozyme (250 µg/ml) was added to the thawed cell suspension and incubated at
0 °C for 30 min. The cells were lysed by exposure to 37 °C for 4 min, followed by immediate chilling in ice water. The lysate was
further homogenized by sonication (Kontes, ultrasonicator, 50 W, twenty
1-s pulses on ice).
The production of [32P]polyP
from [-32P]ATP was measured by using glass filters
(3). The other method to detect [32P]polyP, using PEI-TLC
plates, is reliable only with purified PPK. The reaction mixture
contained PPK, PPK buffer (3), 1 mM
[
-32P]ATP (0.01 µCi/nmol), and 2 mM
creatine phosphate and creatine kinase (20 µg/ml) as an
ATP-regenerating system. The mixture was incubated at 37 °C for 10 min or the time indicated. 1-µl samples were spotted on a PEI-TLC
plate, which was developed with 1.5 M
KH2PO4 (pH 3.5). The dried plate was exposed to
a screen and visualized in a PhosphorImager scanner (Molecular
Dynamics). The ratio between the image intensity of the origin (polyP)
and the total was calculated. One unit of activity incorporated 1 pmol of phosphate into [32P]polyP/min (3). The specific
activity of the purified PPK was 7 × 107 units/mg of
protein.
The assay for PPX measured Pi released from [32P]polyP, prepared as described previously (3). The reaction mixture contained PPX, PPX buffer (13), and 0.1 mM [32P]polyP (based on Pi residues). The mixture was incubated at 37 °C for 10 min or the time indicated and then 1-µl samples were spotted on PEI-TLC plates. The development and calculation were described in the assays for PPK. One unit of activity liberated 1 pmol of [32P]Pi/min (13).
Purification of E. coli Exopolyphosphatase (PPX)Cells were lysed by freezing and thawing, followed by lysozyme treatment, and sonication as described above. The lysate was centrifuged, and the supernatant was precipitated by 60% ammonium sulfate. The precipitate was dissolved in the HEPES buffer (13) and then purified by DE52, S-Sepharose, and Mono Q columns as described (13). The homogeneity of purified PPX was checked by SDS-polyacrylamide gel electrophoresis stained with Coomassie Blue.
Guanosine Pentaphosphate Hydrolase Assay(p)ppGpp and
3-
[32P]pppGpp were prepared by the procedure of
Krohn and Wagner (14). The final concentration was determined at 252 nm
(
252 = 13,100 M
1
cm
1 (15)). The reaction mixture containing PPX buffer
(13) or 40 mM Tris acetate (pH 8.0), 1 mM
dithiothreitol, 10 mM magnesium acetate, 30 mM
ammonium acetate, 0.2 mM EDTA, and the indicated concentrations of [32P]pppGpp were incubated at 37 °C,
and 1-µl samples were applied to PEI-TLC plates. The development and
calculations were described in the assays for PPK. The ratio between
the image intensity of ppGpp and ppGpp plus pppGpp was calculated by
the PhosphorImager scanner.
Accumulation of polyP by E. coli lacking or
overproducing PPK has established that this activity is responsible for
the synthesis of polyP (5). Similarly, the removal of polyP can be
attributed to the presence or absence of PPX, the principal
polyphosphatase of E. coli (13). Combination of these two
opposing activities can account for a turnover of polyP in growing
cells of 12 min or less.4 It
was paradoxical then that 100-fold increases in the level of polyP in
response to the nutritional stress of amino acid starvation were not
accompanied by significant changes in the activities of PPK or PPX as
measured in crude cell extracts (Fig. 1).
Adding increased amounts of pppGpp and ppGpp had no influence on the activity of PPK: at 100 µM pppGpp and at 400 µM ppGpp, neither the synthesis of polyP from ATP nor the
conversion of polyP to ATP were detectably (±5%) affected. However,
the effects of these compounds on PPX activity were strikingly
inhibitory (Fig. 2). At 100 µM, pppGpp inhibited PPX by 90%; ppGpp also inhibited
PPX, but less strongly.
Kinetic Features of Inhibition of PPX by pppGpp and ppGpp
Inhibition of PPX, examined at several levels of pppGpp,
was consistent with its behavior as a competitive inhibitor of the polyP substrate with a Ki value of 10 µM (Fig. 3). A similar
Lineweaver-Burk plot for inhibition of PPX by ppGpp yielded a
Ki value of 200 µM. Binding of
pppGpp to PPX was judged to be reversible in that a prior incubation of
the nucleotide at 400 µM with the enzyme (200 ng) for 10 or 20 min at 37 °C yielded the expected level of inhibition upon a
subsequent 10-fold dilution for assay of enzyme activity.
Stimulation of polyP accumulation by pppGpp and ppGpp could be
demonstrated in the course of polyP synthesis by PPK and hydrolysis by
PPX (Fig. 4). The presence of either of
these nucleotides from the outset or an addition later in the reaction
led to a large increase in the amount of polyP produced. Further
addition of 500 µM ppGpp did not lead to a significant
increase in the amount of polyP produced in the presence of 100 µM pppGpp (data not shown).
PPX Possesses pppGpp Hydrolase Activity
An awareness of the
profound inhibitory effect of pppGpp on PPX activity and the far lesser
effect of ppGpp makes it important to examine the enzymatic routes
whereby pppGpp is hydrolyzed to ppGpp. The principal known route is the
action of GppA (16), which also proved to have exopolyPase activity
(17). In view of the considerable sequence homology between GppA and
PPX (18), the latter was examined and also proved to have pppGpp
hydrolase activity (Fig. 5). The
Km value of 7 µM for pppGpp as substrate (Table I) as expected was
virtually identical to its Ki value of 10 µM as an inhibitor of polyPase activity and even lower
than the Km of 110-130 µM pppGpp
determined for GppA (16, 17). Yet, as indicated by the
kcat value, PPX still qualifies more as a
polyPase than as a pppGppase (Table I).
|
To
evaluate the relative activities of GppA and PPX in the hydrolysis of
pppGpp, extracts of several strains with various levels of these
activities were compared. The strain lacking both ppkx and
GppA was strikingly deficient in the hydrolytic activity compared with
the strain that lacked only ppkx (Fig.
6). However, gppA and
ppx mutants complemented with multicopy plasmids bearing ppx did recover pppGppase activity. Thus, GppA appears to be
the major source of pppGppase activity, but PPX may function in an auxiliary role.
Whereas the polyP levels in E. coli may increase 1000-fold in response to nutritional stress, the levels of the enzyme activities responsible for polyP synthesis (PPK) and polyP degradation (PPX) hardly change at all.2 This was shown to be true when the stringent response was provoked using an amino acid analog serine hydroxamate (Fig. 1). In this study, we show that this phenomenon can be explained by the profound and singular inhibition of PPX by the stress-responsive nucleotides, ppGpp and pppGpp, without any effect on PPK (Figs. 2 and 3). Thus, the continued synthesis of polyP without degradation results in its extensive accumulation.
In keeping with these in vitro findings are several in vivo observations. (i) Accumulation of polyP follows the buildup of ppGpp and pppGpp in response to amino acid starvation.2, 3 (ii) In amino acid-starved cells, the levels of ppGpp increases from 20 µM to 1 mM and that of pppGpp from 0.5 µM to 200 µM; these elevated levels far exceed those required to inhibit PPX in vitro (i.e. a Ki value of 200 µM for ppGpp and 10 µM for pppGpp (Figs. 2, 3)). The fact that the activity of PPX was unchanged in extracts of stressed cells can be attributed to the large dilution of ppGpp and pppGpp suffered in the preparation of the cell extracts. (iii) Turnover of polyP (resulting from its cyclic synthesis by PPK and hydrolysis by PPX) was found to be 12 min or less,4 a result simulated by the accumulation of polyP in a mixture containing purified PPK and PPX responding to the presence of ppGpp and pppGpp (Fig. 4). (iv) Finally, mutants that fail to produce ppGpp and pppGpp (e.g. relA and spoT) also fail to accumulate any polyP2, 3 when treated with serine hydroxamate.
While ppGpp is more abundant than pppGpp, the relative effects of these two regulatory nucleotides have not been thoroughly examined. In the case of polyP accumulation, pppGpp appears to be more effective than ppGpp (Fig. 2). The levels of these nucleotides depend on the synthesis of pppGpp by RelA and SpoT, the hydrolysis of pppGpp to ppGpp, and the conversion of ppGpp to GDP by SpoT. With regard to the generation of ppGpp from pppGpp, the action of the GppA enzyme is likely the principal route (Fig. 6). However, the facts that GppA is an exopolyphosphatase and also bears strong amino acid homologies to PPX promoted our discovery that PPX, like GppA, can also hydrolyze pppGpp to ppGpp (Fig. 5).
Based on studies with extracts of various mutants, the PPX generation of ppGpp from pppGpp may be auxiliary to the action of GppA (Fig. 6). Other possibilities for minor contributions toward the formation of ppGpp from pppGpp include the translation elongation factors, EF-G and EF-Tu (6).
The kinetic parameters of PPX actions (Table I) show the same Km value for pppGpp as a substrate as the Ki value as an inhibitor of exopolyphosphatase, indicating that the same active center is employed for both activities.
A scheme of polyP metabolism (Fig. 7),
based on information mentioned in this report, is surely incomplete.
For example, the role of RpoS, the sigma factor induced by (p)ppGpp and
responsible for the expression of some 50 genes important in the
response to starvation, needs to be included. Complementation of
ppk mutants by a multicopy rpoS plasmid restored
hydrogen peroxidase II activity, indicating an interaction relevant to
polyP metabolism (4). Another example is the behavior of
phoB, the regulatory gene of the phosphate regulon (19).
Mutants of phoB, which produce (p)ppGpp in a stringent
response, nevertheless fail to accumulate polyP.2
Attempts to reconstitute transcriptional systems with an RNA polymerase
holoenzyme containing RpoS acting on an RpoS-activated gene (e.g.
katE, the hydroperoxidase II gene) have resulted in a profound
inhibition by polyP rather than the anticipated
activation.5 One must
conclude that factors, such as a polyP-binding protein, may be
operating in vivo and are lacking in the reconstituted systems. Clues might also be supplied from studies of other microbial systems, such as Myxococcus xanthus (8) and
Pseudomonas aeruginosa,3 in which starvation
responses, heralded by increased levels of (p)ppGpp are followed, as in
E. coli, by a large accumulation of polyP.
We thank Dr. H.-Y. Kim for the use of unpublished data, Dr. N. N. Rao for technical help and discussion, and L. Bertsch for critical reading of this manuscript.