604th MEETING, CAMBRIDGE
155
Regulation of sugar transport by the bacterial phosphoenolpyruvate:glucose
phosphotransferase system
D. FOX, M. KUKURUZINSKA, K. D.-F. LIU,
N. D. MEADOW, D. SAFFEN and S. ROSEMAN
Department of Biology and the McCollum-Pratt Institute,
The Johns Hopkins University, Baltimore, MD 21218,
U.S.A.
It is now generally accepted that the bacterial PTS serves
several important and diverse roles in bacterial cell
physiology (for a review see Postma & Roseman, 1976). The
best-defined function of the PTS is to translocate its sugar
substrates across the cytoplasmic membrane with concomitant phosphorylation of these sugars. In addition, the
system is involved in chemotaxis toward its substrates.
Finally, the PTS regulates adenylate cyclase and certain
non-PTS permeases, and through this type of regulation it
controls the induction of synthesis of the corresponding
permeases and catabolic enzymes such as those required for
lactose uptake and metabolism in Escherichia coli.
The PTS is a complex system of interacting cytoplasmic
and membrane proteins, and it is likely that this degree of
complexity is required in order to make it responsive to
stringent regulation per se and for the PTS, in turn, to
regulate other systems. For instance, it has been shown by
Mason et al. (1981) that there is a very rapid uptake of the
PTS sugar, glucose, when the sugar is added to starved
Streptococcus lactis cells, but this uptake proceeds for only
about 1-1.5min and then stops for several minutes. The
cells then enter a third phase, when they slowly take up the
glucose once again.
While there are variants, four PTS proteins are generally
required for the transfer of the phosphoryl group from PEP
to sugar, the first two proteins being Enzyme I and HPr,
while the second two proteins are a sugar-specific pair
called Enzyme 11. The sugar-specific complex involves not
only a pair of proteins but also lipid and divalent cations.
The sugar receptor in each Enzyme I1 complex is always an
integral membrane protein, called 11-B;there is no evidence
that the 11-B proteins are themselves phosphorylated. The
other protein component of the sugar-specific Enzyme I1
complex is either an integral membrane protein, 11-A, or a
soluble protein, 111, and each of these proteins acts as a
phospho-carrier.
With homogeneous proteins, we have shown that the
sequence in the phosphotransfer reactions is :
PEP-+EnzymeI+HPr-+III (or 11-A)+sugar
The last step is catalysed by the sugar receptor, 11-B.
The growth of bacterial cells is determined by the rate of
sugar uptake, i.e. sugars are generally metabolized more
rapidly than they are transported. From the experiments of
Mason, et al. (1981) mentioned above, as well as uptake
kinetic studies with non-metabolizable sugar analogues
(Simoni & Roseman, 1973), it is very clear that the PTS is
stringently regulated, but the mechanisms underlying
regulation are not known. Some possibilities are considered
below :
(a) The PTS is regulated as the system approaches
thermodynamic equilibrium. This is one mechanism that
we can unequivocally rule out (Weigel et al., 1982). In the
reaction :
PEP,, + sugarout= sugar-P,, + pyruvate,,
Abbreviations used: PEP, phosphoenolpyruvate; PTS, phosphoenolpyruvate :glucose phosphotransferase system.
Vol. 12
the calculated equilibrium constant is about lo8. Thus, at
1 p~ external sugar concentration, the internal sugar-P
concentration would be about 1 0 0 if~ the pyruvate and PEP
concentrations were approximately equal. Net sugar uptake
in fact stops when the internal sugar-P concentration is less
than 7 orders of magnitude below thermodynamic equilibrium, and this is also true in the case of membrane vesicles
where the internal PEP concentration exceeds pyruvate and
where PEP itself is not limiting in the system. As an aside, it
is worth noting that PEP and the phosphoproteins of the
PTS have the highest phosphate-transfer potentials of all
known naturally occurring phosphoryl compounds. The
apparent standard free energies of hydrolysis of these
phosphoproteins are about twice that of ATP. Our
equilibrium studies (Weigel et al., 1982) also showed that
the very large change in the apparent standard free energy
of hydrolysis in the sugar phosphorylation reaction occurs at
the last step, i.e. the transfer of the phosphoryl group from
phospho-111 to sugar as the sugar is translocated across the
membrane.
(b) Regulation is effected via Enzyme I. This enzyme has
an M , of approx. 58000, and self-associates in a temperature
and concentration-dependent manner (Kukuruzinska et al.,
1982). Although not proven, it appears likely that the
monomer is catalytically inactive (Weigel ef a/., 1982), and
that the association of monomers is much slower than the
catalytic reaction. In the presence of excess PEP, 2
phosphoryl groups are introduced per mol of Enzyme I
dimer. From these facts, it is possible to postulate different
mechanisms for the sequence of phosphoryl group transfer
from PEP to HPr through Enzyme I involving associationdissociation of certain of the Enzyme I species (phosphoprotein and non-phosphoprotein). The major point of these
speculations is that the normal transfer sequence involves
both catalytically active and inactive Enzyme I species. If
the ratios of species were in turn affected by other proteins
or catabolites, then it is possible to control the PTS at this its
first step.
( c ) Regulation is effected by the tricarboxylic acid cycle
and the ratioof ADP/ATP. We have recently found that the
PTS interacts with the enzyme acetate kinase (Fox &
Roseman, 19831, and this interaction may provide a direct
link between the tricarboxylic acid cycle and sugar
transport via the PTS. Actate kinase catalyses the reaction:
ATP +acetate
= ADP
+ acetyl-P
The equilibrium for this reaction lies well to the left. The
enzyme is important historically, since it led to the discovery
of acetyl-SCoA (Lipmann,l944). Since that time it has been
shown that the enzyme is itself phosphorylated (Anthony &
Spector, 1972), but there is controversy (Knowles, 1980)
concerning the role of the phosphoenzyme, i.e. whether or
not it is an obligatory intermediate in the catalytic pathway.
Acetate kinase had not been purified to homogeneity before
the present studies. For phospho-transfer experiments,
homogeneity is an obvious requirement, and the enzyme
was therefore isolated in pure form and found to donate its
phosphoryl group to Enzyme I in a reversible manner. In
addition, this phosphoryl group (derived from ATP or
GTP) must be transferred to the active site of Enzyme I
since it can be subsequently transferred to glucose when the
reaction mixture is supplemented with HPr, I 1 P c , and 11BC". The rate of glucose phosphorylation with acetate
kinase and GTP is about 20% of that obtained with PEP.
These results suggest that the levels of ADP, ATP, GDP,
GTP, acetate, acetyl-SCoA, etc. can potentially regulate the
156
BIOCHEMICAL SOCIETY TRANSACTIONS
activity of the PTS in sugar transport. Studies with
appropriate mutants are in progress to test this idea.
(d) Regulation is effected by feed-back inhibition of the
PTS. Studies with membrane vesicles which transport PTS
sugars suggest that regulation may involve feed-back
inhibition by the transport product, the sugar-P, although
this is not seen with purified proteins or with toluenetreated membrane vesicles which are permeable to the
sugar-P (Liu & Roseman, 1983). It also appears possible
that regulation is much more subtle than we now realize,
perhaps involving association4issociation of the various
soluble proteins with the integral membrane protein 11-B.
As noted above, and by Kornberg et al. (1984), the PTS
not only regulates the rate of uptake of its sugar substrates
but also regulates certain non-PTS permeases, such as
those responsible for the uptake of glycerol, melibiose and
maltose in Salmonella typhimurium and E. coli (also lactose
in the latter organism). PTS-mediated control over these
permeases, as well as over adenylate cyclase, explains the
phenomenon of diauxic growth (cells prefer to grow on PTS
sugars and utilize the non-PTS sugars only in the second
phase of growth).
PTS-mediated repression of synthesis of the non-PTS
catabolic systems is dependent upon two factors: one is the
ratio of the PTS to the non-PTS proteins, and the other is the
product of a gene called crr. Inhibition of the non-PTS
permeases by PTS sugars depends on the ratio of the PTS to
the non-PTS proteins (Mitchell et al., 1982). When the ratio
is low, then glucose or other PTS sugars do not significantly
inhibit the uptake of the non-PTS sugars. These results
imply some sort of stoichiometric interaction between the
two systems.
The second factor in PTS-mediated repression is the crr
gene. We had shown in earlier work that PTS-mediated
repression requires a functioning crr gene, and that the crr
mutants contained normal levels of PTS proteins with the
exception of one, IIIGIC, i.e. crude extracts contained low
levels of IIIG~cwhen assayed for the activity of this protein.
However, this evidence was insufficient to conclude that the
crr gene codes for IIIG1csince the crr product, whatever it
was, has such a broad pleiotropic effect on cell physiology.
In order to establish the relationship between the crr gene
and IIIG1c,the following experiments were performed :
IIIGICwas isolated in homogeneous form from an S.
ryphimurium crr- mutant (Meadow er al., 1982~)
and
was found to have different physical properties from
wild type IIIG1' and only slight activity as a
phosphate-carrier in the sugar phosphorylation
assay.
The E. coli crr gene was cloned on a 1.35 kilobase
DNA fragment in the plasmid pBR322 and found to
be responsible for the synthesis of only one protein,
I I P C(Meadow et al., 19826).
The technique of y. 6 insertions (Guyer, 1978) was
used to mutagenize the cloned crr gene. Some of these
PIS/
Hpal EcoRl
Hpal
1
0.72
10.134
--Sequenced +
ptsH
Pstl
I
Kph
Kpnl
0.88
ptsl
4
/ig
(iex, cysK)
0.67
44
\
Hindlll
0.6
1
-
Fig. 1. Restriction map of the PTS region of the E . coli genome
The circle at the top of the figure represents the E. coligenetic map and indicates the
positions of some of thepts genes. The relative positions are given for theprs operon
(prsH, prsI), crr (encodes for IIIGIC),ptsG (structural gene for II-BGlc),and prsM
(structural genes for IIMa"complex, II-A/II-BMan).
The expanded section of the map
in the centre of the figure gives the order of the pts and neighbouring genes as
determined by genetic mapping (Epstein er al., 1970; N. M. Kredich, personal
communication), and analysis of cloned DNA segments (Britton et al., 1983). The
map at the bottom of the figure gives the positions ofptsH (structural gene for HPr)
and ptsl (structural gene for Enzyme I) with respect to the cutting sites of various
restriction enzymes. Genetic studies (Cordaro et al., 1974) show that the structural
genes for HPr and Enzyme I constitute an operon with transcription proceeding
from ptsH through prsl, while the gene crr is transcribed independently. In the
nomenclature of Kornberg and associates, crr = gsr. The DNA regions indicated
have been sequenced.
I984
157
604th MEETING, CAMBRIDGE
crr mutants produced no detectable I 1 P c at all, while
others yielded peptides of lower relative molecular
mass than IIIciic that were still antigenically active
with anti-III[ilcpolyvalent antibodies.
(4 The E. coli crr gene has been completely sequenced,
and the DNA sequence gives an amino acid sequence
which corresponds (except for N-terminal methionine) to the N-terminal amino acid sequence of S .
typhimurium IIIG1cinsofar as it has been established
(first 53 residues).
From these data, we conclude that crr codes for IIICic,and
therefore that PTS-mediated repression of the non-PTS
systems is effected through the protein IIIG1c.The protein
itself is interesting in that it exists in two electrophoretically
distinguishable forms called IIIgEwand 11192, (Meadow &
Roseman, 1982). The difference between the two is that the
electrophoreticallyfast form lacks the N-terminal heptapeptide of the slow form. We have recently found that this
cleavage of the slow form to the stable heptapeptide and fast
IIIG'c is mediated by what appears to be a specific
membrane-bound protease.
III@kwis the active form of the protein in sugar
phosphorylation. While IIIg.&, can accept the phosphoryl
group from phospho-HPr as easily as I I I ! & ,(Meadow
,
&
Roseman, 1982). phospho-IIIgk, is almost without activity
as a phosphoryl donor to methyl a-glucoside, the reaction
mediated by II-BGIC.We infer that the N-terminal heptapeptide 'tail' is essential for binding of the phosphoprotein to
the membrane receptor, II-BGlC.
At this point we are investigating which of the various
forms of IIIG1c(fast, slow, phosphorylated, unphosphorylated) are active in binding to and inhibiting homogeneous,
reconstituted lactose permease (in collaboration with M.
Newman, H. R. Kaback and T. H. Wilson) and whether
any of these or the other soluble PTS proteins affect
the activity of adenylate cyclase (in collaboration with A.
Peterkofsky). These experiments would permit a critical
test of the hypothesis (Roseman, 1977) that the permeases
are inhibited by IIIG1c, but not by phospho-IIIG1c.
The N-terminal heptapeptide discussed above has been
synthesized (A. Komoriya & C. Anfinsen, unpublished
work). Since it is stable when incubated for prolonged
periods with S . typhimurium membranes, it seems possible
that the heptapeptide has a physiological function, and
experiments in progress are aimed at testing this idea as
well.
As a final point, many of the structural genes for the PTS
proteins have been isolated and cloned into plasmids, and
either completely or partially sequenced. These results are
summarized in Fig. I , and when completed these experiments will have a number of ramifications. They should
allow us to study regulation of the synthesis of the PTS
proteins, and the primary amino acid sequences, derived
from the DNA sequences, should permit us to begin a
prolonged study on the structure-function relationships
between these proteins and also between the PTS and the
non-PTS proteins.
Anthony, R. S. & Spector, L. B. (1972).J . Bwl. Chem. 247,2120-
2125
Britton, P., Boronat, A., Hartley, D. A., Jones-Mortimer, M. C.,
Kornberg,H. L. & Parra, F. (1983)J. Gen. Microbwl. 129,349-
358
Cordaro, J. C., Anderson, R. P., Grogan, E. W., Wenzel, D . J.,
Engler, M. & Roseman, S. (1974)J . Bacteriol. 120, 245-252
Epstein, W., JewettLS. & Fox, C. F. (1970)J. Bacteriol. 104,793-
797
Fox, D. K. & Roseman, S . (1983)Fed. Proc. 42, 1942
Guyer, M. S.(1978)J . Mol. Biol. 126, 347-365
Knowles, J. R. (1980)Ann. Rev. Biochem. 49,877-919
Kornberg, H. L., Britton, P., Jones-Mortimer, M. C. & Lee, L. G .
(1984)Biochem. Soc.Tram. 12, 157-159
Kukuruzinska, M. A., Harrington, W. F. & Roseman, S. (1982)J.
Biol. Chem. 257, 1447&14476
Lipmann, F. (1944)J. Bwl. Chem. 155, 55-70
Liu, K.D.-F. &Roseman,S. (1983)Proc. Natl. Acad. Sci. U.S.A. in
the press
Mason, P. W., Carbone, D. P., Cushman, R. A. & Waggoner, A.
(1981)J . Bwl. Chem. 256, 1861-1866
Meadow, N . & Roseman,S. (1982)J.Bwl. Chem. 257,14526-14537
Meadow, N., Rosenberg, J., Pinkert, H. & Roseman, S. (1982a)J .
Bwl. Chem. 257, 14538-14542
Meadow, N.,Saffen, D. W., Dottin, R. P. & Roseman, S. (198%)
Proc. Natl. Acad. Sci. U.S.A. 79, 2528-2532
Mitchell, W. J., Misko, T. P. & Roseman, S. (1982)J . Bwl. Chem.
257, 14553-14564
Postma, P. W. & Roseman, S. (1976)Biochim. Biophys. Acra. 457,
21 3-257
Roseman, S . (1977)Fed. Eur. Bwchem. Soc. Symp. (Bed.)42,582-
597
Simoni, R. D.& Roseman, S. (1973)J . Bwl. Chem. 248,966-976
Weigel, N., Kukuruzinska, M. A., Nakazawa, A., Waygood, E. B.
& Roseman, S. (1982)1. Bwl. Chem. 257, 14477-14491
The role of the phosphoenolpyruvatephosphotransferase system in inducer exclusion
H. L. KORNBERG, P. BRITTON, M. C. JONESMORTIMER and LYNNE G. LEE
Department of Biochemistry, Tennis Court Road, Cambridge
CB2 IQW, U . K .
Monod (1947) first demonstrated that glucose inhibits the
induction of fl-galactosidase by Escherichia cofi so effectively that organisms placed in a medium containing glucose at
a concentration of 0.4mg/ml and lactose at a concentration
of 2mg/ml completely utilize the glucose and cease to grow
until the lactose now induces the components of the lactose
operon (Epstein et al., 1966). This phenomenon is complex
and involves at least three effects of glucose. (i) An
inhibition of adenylate cyclase activity, leading to a
lowering of the intracellular concentrations of cyclic AMP.
This in turn results in (ii) repression of the synthesis of flAbbreviations used: a-MG, methyl a-glumside; PTS, phosphoenolpyruvate-phosphotransferasesystem.
VOl. 12
galactoside permease and /3-galactosidase. However, as
Cohn & Horibata ( 1 9 5 9 ~demonstrated,
)
glucose also acts
(iii) to exclude lactose from the cells: if the &galactoside
permease is pre-induced, or if the lactose operon is
expressed constitutively (Cohn & Horibata, I9596), flgalactosidase is formed despite the presence of glucose,
though [in consequence of the effects (i) and (ii)] at a rate
only about half of that noted in its absence.
It is now generally accepted that inducer exclusion by
glucose involves the PTS and, specifically, the factor IIIclc
that interacts with the Enzyme 11 for glucose specified by
ptsC. The main evidence for this view can be summarized as
follows:
(a) Mutants devoid of Enzyme I activity (ptsl) or lacking
the histidine-containing protein HPr ( p t s H ) grow on
many non-PTS sugars and C4 components only if
inducers of the appropriate uptake system, or cyclic
AMP, are added (Berman et a f . , 1970).