Quantitative estimation of channeling from early glycolytic
intermediates to CO
2
in intact Escherichia coli
Georgia Shearer, Jennifer C. Lee, Jia-an Koo and Daniel H. Kohl
Department of Biology, Washington University, St. Louis, MO, USA
The idea that intermediates in many metabolic path-
ways are ‘channeled’ from one pathway enzyme to the
next is widely [1,2], but not universally, accepted. One
reason for the controversy is that ‘many of the enzyme
complexes are dissociated during isolation owing to
dilution effects’ [3]. Srere, in his authoritative 1987
review [4], critically examined the evidence to that
date. For more recent reviews, see [1,5]. Contrarians,
such as Gutfreund and Chock [6], interpret their
kinetic data, compiled from experiments with pure
enzymes of the glycolytic pathway in dilute solution,
to be compatible with a diffusion model without need
to invoke channeling.
Atkinson [7] was influential in preparing the ground
for the idea of channeling. He pointed out that there is
not enough water in the cell to support uniform con-
centrations of all pathway intermediates at K
M
, the
approximate concentration traditionally assumed to be
necessary to permit pathways to function optimally.
Along with other considerations, this led Srere to
postulate the existence of ‘metabolons’, transient asso-
ciations of pathway enzymes in addition to stable
complexes (e.g. cytochrome complexes of the electron

neled all the way to CO
2
, whereas fructose-6-phosphate was not. Because
the signature of channeling is lost if any downstream intermediate prior to
CO
2
equilibrates with molecules in the aqueous cytosol, it was not possible
to evaluate whether glucose-6-phosphate was channeled in its transforma-
tion to fructose-6-phosphate. The data also suggest that, in addition to
pathway enzymes being associated with one another, some are free in the
aqueous cytosol. How sensitive the degree of channeling is to growth or
experimental conditions remains to be determined.
Abbreviations
Fru1,6P
2
, fructose-1,6-bis phosphate; Fru6P, fructose-6-phosphate; F
ch
, fraction of total flux that is channelled; Glu6P, glucose-6-
phosphate; OPPP, oxidative limb of the pentose phosphate pathway; PFK, phosphofructokinase; PGI, phosphoglucoisomerase;
TCA, tricarboxyclic acid.
3260 FEBS Journal 272 (2005) 3260–3269 ª 2005 FEBS
the product of the first enzyme to have an advantage
in competition for the active site of the second enzyme
compared with the same molecular species within the
aqueous cytoplasm of the cell. That is, intermediates
just made in a pathway are not part of the same pool
as are identical molecules within the cell. Intermediates
produced within the pathway are ‘channeled’ to the
next enzyme.
Evidence for channeling

graphs showing colocalization of enzymes of the urea
cycle across the mitochondrial membrane [13] are con-
sistent with the proximity of sequential enzymes that is
evoked as a necessary (although not sufficient) condi-
tion for channeling. (c) NMR. Incubation of yeast in
[4-
13
C]glutamate did not result in the randomization of
the label in aspartate formed from it as would be
expected if the symmetric intermediates, succinate
and ⁄ or fumarate, dissociated from their enzymes and
were free to rotate [14]. In addition
19
F NMR studies
of citrate synthase 1 tagged with 5-fluorotryptophan
showed motional restriction in vivo [15]. (d) The use of
stable isotopes. Clegg and Jackson [16] compared the
specific activity of
14
C-labelled glycolytic intermediates
with that of pyruvate. These studies resulted in much
less dilution of the radioactivity in pyruvate than
would be expected if intermediates dissociated from
their enzymes and entered the cytosol. (In order to
facilitate uptake of intermediates these investigators
permeabilized the cells with dextran sulfate.)
The body of work cited above is strong evidence of
channeling in a number of pathways; glycolysis, the
TCA cycle, the oxidative limb of the pentose phos-
phate pathway (OPPP), the urea cycle. However, these

with the introduced aminoacyl-tRNA; i.e. there was
perfect channeling from free amino acids to protein. In
this experiment, it was not necessary to calculate the
percentage of the flux that was channeled, because the
unchanneled flux was essentially zero. However, had
channeling been less than 100%, the data collected
would have enabled this calculation.
In this study, we describe experiments aimed at cal-
culating the fraction of the flux from early glycolytic
intermediates to CO
2
in intact Escherichia coli. Cells
incubated with [
14
C]glucose made
14
C-labeled glyco-
lytic intermediates. When the incubation mix also
included a [
12
C]intermediate, assuming that this inter-
mediate entered the cell, there was a competition
between the intermediate just produced in the pathway
and the same molecular species in the aqueous cytosol.
To the degree the latter was successful, the amount of
14
CO
2
evolved was decreased. Cells were also incuba-
ted in [

2
(presumably via mixed acid fermentation) be 100%
channeled. We found no significant channeling of
Glu6P or Fru6P to CO
2
. However, because Fru6P
was not channeled, the signature of any channeling of
Glu6P that might exist would be lost.
Results
The cells grew well on all carbon sources of interest to
us, including the 6-carbon sugar phosphates, Glu6P,
Fru6P, and Fru1,6P
2
. The doubling times for Glu6P
and Fru6P were comparable with the doubling time
of the parental strain growing on glucose (Table 1).
Doubling times when grown on glycerol and Fru1,6P
2
were considerably longer.
Rate of
14
CO
2
evolution and growth rate were
poorly correlated. In particular, the rate of
14
CO
2
evo-
lution from Fru1,6P

CO
2
evolved in all incubations
were highly correlated with the total amounts of
14
C
taken up into the cell (three experiments, four treat-
ments in each experiment, three or four replicates each
treatment ¼ 40 data points; R
2
¼ 0.97, data not
shown). The total amount of
14
C entering the cell was
taken to be the sum of the
14
CO
2
evolved plus the
amount retained by the cell.
Saturation of
14
CO
2
evolution rate by the mutant
strain as a function of concentration depended on sub-
strate. The rate of
14
CO
2

Experiments to investigate channeling of early
glycolytic intermediates to CO
2
Figure 1 illustrates our experimental paradigm. In this
example, the cells are incubated in [
14
C]glucose and
[
12
C]-Fru6P (the challenger). This creates a competi-
tion for E3 (phosphofructokinase; PFK, EC 2.7.1.10).
The measure of channeling is the degree to which the
Fru6P just made in the pathway is disproportionately
successful in being the substrate for PFK. Invoking the
usually proposed mechanism for channeling, the pref-
erence for the intermediate just made in the pathway is
a consequence of the interaction of PFK and the prior
enzyme in the pathway, phosphoglucoisomerase (PGI,
EC 5.3.1.9). If all Fru6P molecules just made in the
pathway dissociate from PGI and equilibrate with the
pool of Fru6P in the aqueous cytoplasm, then there
would be no channeling. When there is no channeling,
the result of the competition for binding to PFK will
be proportional to the number of [
14
C]Fru6P mole-
cules made in the pathway and the [
12
C]Fru6P in the
aqueous cytoplasm. In principle, the challenger can be

)1
)
Glucose RK 9118 parent 2.62
Glucose RK9117 mutant 2.23 ± 0.32 0.283 ± 0.014
Glycerol RK9117 mutant 3.61 ± 0.42
Glu6P RK9117 mutant 1.77 0.053 ± 0.004
Fru6P RK9117 mutant 2.17 0.042 ± 0.001
Fru1,6P
2
RK9117 mutant 4.05 ± 0.72 0.092 ± 0.001
Quantitation of channeling in intact E. coli G. Shearer et al.
3262 FEBS Journal 272 (2005) 3260–3269 ª 2005 FEBS
enzymes, there are free enzymes whose activity would
result in unchanneled flux even if every intermediate
just made in the pathway remained within the putative
complex. Finally, if channeling is being assessed by
measuring the isotopic composition of a downstream
compound (such as CO
2
), the signature of channeling
will be lost if any step between the reaction being con-
sidered and the downstream intermediate is not chan-
neled. Thus, an experiment with our design will result
in an evaluation of the minimum channeling in any
intervening step.
Figure 2 shows the results of experiments in which
cells were incubated with [U-
14
C]glucose alone (black
bars) and with [U-

(8.2 ± 1.0 vs. 5.3 ± 0.6). By contrast, exogenous
[
12
C]Fru6P had a clear impact on the counts in CO
2
(21.2 ± 2.2 vs. 7.9 ± 0.8) originating from [
14
C]glu-
cose. This suggests that Fru1,6P
2
is strongly channeled
all the way to CO
2
, that Glu6P is modestly channeled
and that Fru6P is channeled to an even lesser degree.
But two additional, essential pieces of information are
required before even such a qualitative conclusion con-
cerning the degree of channeling can be drawn. Also
these data by themselves do not allow us to calculate
the fraction of the flux that is channeled. The first
additional requirement is to show that the exogenous
intermediate from the incubation mix entered the cell.
Clearly, if the exogenous intermediate did not enter
the cell, it could not compete for the active site of the
enzyme for which it is substrate. Data relevant to this
are shown in Fig. 3. These data establish that all of
the exogenous intermediates entered the cells (black
bars) and evolved CO
2
, although with varying degrees

C]intermediates with [
12
C]glucose.
Whereas the [
12
C]glucose tended to decrease
14
CO
2
when [
12
C]glucose was coincubated with a [
14
C]inter-
mediate, the radioactivity in the CO
2
evolved was
significant compared to that in the absence of
[
12
C]glucose; 42, 72, and 67% for [
14
C]Glu6P,
[
14
C]Fru6P and [
14
C]Fru1,6P
2
, respectively. This is

to
14
CO
2
in the presence of
[
12
C]glucose.
The data in Figs 2 and 3, considered separately,
are insufficient to determine the predicted result if
Table 2. Fraction of the total flux to
14
CO
2
via the OPPP. Note that only one of the six carbon atoms in [1-
14
C]glucose and [6-
14
C]glucose
are labeled, whereas the label in [U-
14
C]glucose is divided among all six atoms. Glc, glucose; % of total flux to
14
CO
2
via the OPPP ¼
100 · 3.42 ⁄ 79.5 ¼ 4.3%.
[1-
14
C]Glucose [6-

produced 5.80 2.38 3.42 79.5
G. Shearer et al. Quantitation of channeling in intact E. coli
FEBS Journal 272 (2005) 3260–3269 ª 2005 FEBS 3263
channeling were zero. When the data from both figures
are combined, we are able to calculate the expectation
if the fraction channeled were zero (F
ch
¼ 0). When
F
ch
¼ 0, the relative success of binding to the enzyme
of intermediates just made in the pathway and those in
the aqueous cytoplasm will be proportional to their
relative numbers. A method for calculating the fraction
of the flux that is channeled from each intermediate to
CO
2
, using the data of Figs 2 and 3, is developed
below.
Discussion
The first task is to calculate the relative amounts of
intermediate made in the pathway vs. the same inter-
mediate made from exogenous sources. In this regard,
Fig. 2. Effect of unlabeled challenging compound (Glu6P, Fru6P or
Fru1,6P
2
) on the quantity of
14
CO
2

alone; white bars represent the quantity evolved when cells were
incubated with
14
C-labeled intermediate and unlabeled challenging
glucose. For experiments with Fru6P, cells were grown with a
trace of Fru6P (25 l
M) in addition to 0.2% (v ⁄ v) glycerol as the car-
bon source.
Fig. 1. Glycolytic pathway in an E. coli cell coincubated with
[
14
C]glucose and a [
12
C]intermediate (Fru6P in this example).
[
14
C]Glucose is converted to [
14
C]Glu6P as it enters the cell via the
phosphotransferase system (PTS) and thence to [
14
C]Fru6P by E2
(phosphoglucoisomerase; PGI). In the absence of channeling,
[
14
C]Fru6P will equilibrate with [
12
C]Fru6P from the exogenous
source, competing with [
14

14
CO
2
from the earliest
time point (see Results). Also recall from the Results
that the CO
2
evolved from [
14
C]glucose was strictly
proportional to the total uptake of glucose. Thus, we
can use CO
2
evolved as a reliable proxy for the total
uptake.
To facilitate the following discussion, let A ¼ the
14
CO
2
evolved when cells were incubated with [
14
C]glu-
cose alone; B ¼ the
14
CO
2
evolved when cells were
incubated with [
14
C]glucose plus [

C]glucose as the result of coincubation of
[
14
C]glucose and [
12
C]intermediate is proportional to
‘B’ as defined above. Likewise the amount of
[
14
C]Fru6P converted to
14
CO
2
from exogenous
[
14
C]Fru6P when coincubated with [
14
C]Fru6P is pro-
portional to ‘D’.
Consequently in the absence of channeling (i.e. when
the fraction of the flux that is channeled is zero,
F
ch
¼ 0), the expected dilution of
14
CO
2
originating
from

D=C ¼ 1; and F
ch
¼ 1
Values of F
ch
between 0 and 1 can be calculated by
interpolation as illustrated in Fig. 4. An important
caveat, if one is to make this interpolation, is that a
substantial quantity of the exogenous intermediate
must enter the cell during coincubation with glucose.
Clearly, the exogenous intermediate cannot compete
with the intermediate just made in the pathway for
occupancy at the active site of the appropriate enzyme
if it does not get into the cell. In our notation, if not
much of the intermediate gets into the cell when glu-
cose is present, then B » D and B ⁄ (B + D)  1. If the
expected value when there is no channeling is close to
1 and the value for 100% channeling is also 1, then
there is no ‘room’ to interpolate as required by the
procedure outlined in Fig. 4.
The data in Figs 2 and 3 permit calculation of
B ⁄ (B + D), the ratio by which
14
CO
2
evolved from
[
14
C]glucose would be affected by coincubation with
an unlabeled intermediate, if no channeling occurred,

absence of channeling [B ⁄ (B + D)] compared to the observed
effect (B ⁄ A). Black bars; the expected effect ¼ B ⁄ (B + D). White
bars; the observed effect ¼ B ⁄ A.
G. Shearer et al. Quantitation of channeling in intact E. coli
FEBS Journal 272 (2005) 3260–3269 ª 2005 FEBS 3265
[B ⁄ (B + D)] (Fig. 5). This result is consistent with the
Fru1,6P
2
just made in the pathway being channeled.
By contrast, for Glu6 P and Fru6P, Fig. 5 shows no
significant difference between the observed effect
(Fig. 5, Glu6P or Fru6P, white bar) and the value
expected if there were no channeling (Fig. 5, Glu6P or
Fru6P, black bar). This is a necessary but not suffi-
cient indication of the absence of channeling. The data
in Fig. 3 rule out the exogenous intermediate not
entering the cell as an explanation. There is a second
alternative explanation for the apparent absence of
channeling of Glu6P. Glu6P just made in the pathway
could be strongly channeled to Fru6P, but the signa-
ture of that channeling would be lost because Fru6P
was apparently not channeled.
The data in Table 3 show the method for and the
results of calculating the fraction of the total flux that
is channeled.
The most striking result is that essentially all of the
flux from Fru1,6P
2
to CO
2

pathway enzyme responsible for its entry into glycoly-
sis. However, the data [D ⁄ (B + D) vs. D⁄ C] provide
no evidence for channeling from any exogenous Glu6P
or Fru1,6P
2
to CO
2
. Although there was a significant
difference between D ⁄ (B + D) vs. D ⁄ C for Fru6P, the
fraction channeled was small (0.16 ± 0.01).
The evidence presented here for channeling of
Fru1,6P
2
to CO
2
is consistent with the existence of a
glycolytic complex that holds together long enough for
the Fru1,6P
2
that binds to it to be converted through
Table 3. Calculation of the fraction of the total flux that is channeled (F
ch
). Values are means ± SE.
Challenging intermediate Glu6P Fru6P Fru1,6P
2
No. experiments 6 2 5
Carbon source in growth medium 22 m
M glycerol 22 mM glycerol
+25 l
M Fru6P

14
CO
2
evolved from incubation
with
[14
C]intermediate plus unlabeled glucose)
2.7 ± 1.0 15.8 ± 2.6 21.6 ± 5.9
Evaluation of channeling of intermediate just made in the pathway
B ⁄ (B + D) (expected effect of unlabeled
challenger on
14
CO
2
evolution from
[
14
C]glucose if no channelling
0.695 ± 0.090 0.333 ± 0.004 0.311 ± 0.050
B ⁄ A (observed effect) 0.707 ± 0.119 0.358 ± 0.001 0.961 ± 0.104
F
ch
0.103 ± 0.256 (NS) 0.036 ± 0.012 (NS) 0.994 ± 0.158
(P ¼ 0.005)
Evaluation of channeling of exogenous intermediate
D ⁄ (B + D) (expected effect of unlabeled
challenger on
14
CO
2

. Whe-
ther channeling over such a large number of intermedi-
ates is sensitive to growth conditions or other details of
the experimental protocol remains to be determined.
Experimental procedures
Materials
[U-
14
C]-, [1-
14
C]- and [6-
14
C]glucose, [U-
14
C]Glu6P, [U-
14
C]
Fru6P, and [U-
14
C]Fru1,6P
2
, as well as all other biochemi-
cals, were obtained from Sigma (St. Louis, MO). Our colla-
borator, Robert Kadner (University of Virginia), made an
E. coli mutant (RK9117) that was engineered to take up
4-, 5- and 6-carbon sugar phosphates constitutively. Table 4
gives the genotype of RK9117 and the parent from which it
was made (RK9118).
Bacterial growth
Cells were grown at 37° in a rotary shaker in a defined

Just before each experiment, cells were resuspended in a vol-
ume of growth medium (lacking any carbon source or Mg
2+
or Ca
2+
) such that the absorbance was  10 at 600 nm.
About 20 l L of the concentrated cells were added to 230 lL
of incubation medium in order to incubate cells at an absorb-
ance at 600 nm of 0.8. Each incubation mixture contained
appropriate carbon sources, one labeled with
14
C and, when
required by the experimental design, a second unlabeled car-
bon source. Incubations were carried out in 25 mL vials that
were sealed with a rubber septum fitted with a straight pin. A
small strip of filter paper was placed on the pin and wetted
with 10 lL 10% (w ⁄ v) NaOH. Incubations were carried out
for 30 min at 37 °C in a rotary shaker water bath. Incuba-
Table 4. Genotypes of E. coli strains used in this report.
Strain Genotype
RK9117 D(argF-lac)U169 araD139 thi gyrA219 relA rpsL150 non polA1 D(ilvBN-uhpABCT ¢)2095 zig621::Tn10 uhpA
+
B
+
C91::4 (Con)T
+
Rk9118 D(argF-lac)U169 araD139 thi gyrA219 relA rpsL150 non polA1 D(ilvBN-uhpABCT ¢)2095 zig621::Tn10 uhpA
+
B
+

10 No No Yes 8648 9232 8719 ± 88 9302 ± 94
11 No No Yes 8616 9186
12 No No Yes 8894 9488
G. Shearer et al. Quantitation of channeling in intact E. coli
FEBS Journal 272 (2005) 3260–3269 ª 2005 FEBS 3267
tions were terminated by adding 100 lL 70% (v ⁄ v) HClO
4
.
The outgassed CO
2
was captured on the base impregnated fil-
ter paper. After incubation, the filter paper was removed and
eluted with 100 lL water. Three milliliters of scintillation
cocktail (Ecolite, ICN Biomedicals, Irvine, CA) were added.
Aliquots of the incubation mixture were also counted in
order to calculate percent conversion of the added
14
C-
source. The rate of CO
2
production was calculated as
nmolsÆmin
)1
¼ nmol equivalents of
14
C source · percentage
conversion ⁄ (100 · min).
The term ‘nmol equivalents of
14
C-source’ is included in

, any CO
2
produced via the oxidative limb of the OPPP must be cor-
rected for or shown to be small enough to neglect. Catabol-
ism of glucose via the OPPP results in CO
2
being evolved
from the C-1 position of glucose. When glucose is catabo-
lized completely to CO
2
in either the TCA cycle or by a
branch of mixed acid fermentation, then both the C-1 and
the C-6 positions of glucose give rise to CO
2
. Thus the
amount of CO
2
produced in the OPPP is CO
2
from
[1-
14
C]glucose minus CO
2
from [6-
14
C]glucose. The total
amount of CO
2
produced by the OPPP plus that produced

plus [
12
C]glucose (1 mm).
Acknowledgements
This work was supported by an SGER NSF grant:
NSD Grant # MCB-02004900.
References
1 Agius L & Sherratt H, eds. (1997) Channelling in Inter-
mediary Metabolism. Portland Press, London.
2 Ovadi J & Srere P (2000) Macromolecular compart-
mentation and channelling. Int Rev Cytol 192,
255–280.
3 Velot C & Srere P (2000) Reversible transdominant
inhibition of a metabolic pathway. J Biol Chem 275,
12926–12933.
4 Srere P (1987) Complexes of sequential metabolic
enzymes. Annu Rev Biochem 56, 89–124.
5 Milani M, Pesce A, Bolognesi M, Bocedi A & Ascenzi
P (2003) Substrate channeling: molecular bases. Biochem
Mol Biol Educ 31, 228–233.
6 Gutfreund H & Chock P (1991) Substrate channeling
among glycolytic enzymes – fact or fiction. J Theor Biol
152, 117–121.
7 Atkinson D (1977) Conservation of solvent capacity. In
Cellular Energy Metabolism and its Regulation, pp. 13–31.
Academic Press, New York.
8 Masters CJ (1977) Metabolic control and the microenvi-
ronment. Curr Topics Cell Regul 12, 75–105.
9 Beeckmans S (1999) Chromatographic methods to study
protein–protein interactions. Methods 19, 278–305.

17 Negrutskii BS & Deutscher MP (1991) Channeling of
aminoacy-tRNA for protein synthesis in vivo. Proc Natl
Acad Sci USA 88, 4991–4995.
18 Mowbray J & Moses V (1976) The tentative identifica-
tion in Escherichia coli of a multienzyme complex with
glycolytic activity. Eur J Biochem 66, 25–36.
G. Shearer et al. Quantitation of channeling in intact E. coli
FEBS Journal 272 (2005) 3260–3269 ª 2005 FEBS 3269