J. theor. Biol. (1996) 182, 437–447

0022–5193/96/190437+11 $25.00/0 7 1996 Academic Press Limited

Triosephosphate Isomerase Deficiency: Predictions and Facts*

F O,† B G. V,† S H,‡ M H‡ J O†§

†Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences,Budapest, H-1518, P.O. Box 7, Hungary and the ‡National Institute of Haematology,

Blood Transfusion and Immunology, Budapest. H-1113, Daroczi ut 24, Hungary

Deficiencies in around 20 enzymes, associated with widely different degrees of severity and complexity,have been identified for human erythrocytes. The fact that glycolysis is crucial for erythrocyte functionis reflected by the large number of inherited glycolytic enzymopathies. Triosephosphate isomerase (TPI)deficiency, a rare autosomal disease, is usually associated with nonspherocytic hemolytic anemia,progressive neurologic dysfunction, and death in childhood. The two affected Hungarian brothersstudied by us have less than 3% TPI activity and enormously (30–50-fold) increased dihydroxyacetonephosphate (DHAP) concentration in their erythrocytes.

The well-established concept of the metabolic control theory was used to test the contribution of TPIand some related enzymes to the control of a relevant segment of the glycolytic pathway in normal anddeficient cells. Deviation indices, DJ

E =(DJ/DE) Er/Jr, which give a good estimation of flux controlcoefficients using a single large change in enzyme activity, were determined from the fluxes in the absenceand presence of exogeneous enzymes. We found that PFK and aldolase are the enzymes thatpredominantly control the flux, however, the quantitative values depend extensively on the pH: DJ

E

values are 0.85 and 0.14 at pH 8.0 and 0.33 and 0.67 at pH 7.2 for aldolase and PFK, respectively.Neither the flux rates nor the capacities of the enzymes seem to be significantly different in normal andTPI deficient cells.

There is a discrepancy between DHAP levels and TPI activities in the deficient cells. In contrast tothe experimental data the theoretical calculations predict elevation in DHAP level at lower than 0.1%of the normal value of TPI activity. Several possibilities suggested fail to explain this discrepancy.Specific associations of glycolytic enzymes to band-3 membrane proteins with their concomitantinactivation have been demonstrated. We propose that the microcompartmentation of TPI that couldfurther decrease the reduced isomerase activity of the deficient cells, is responsible for the high DHAPlevel.

7 1996 Academic Press Limited

Glycolysis in the Red Blood Cell

The mature red cell has to depend almost solely onanaerobic glycolysis to produce the energy requiredfor its functions (cf. Fig. 1). The process of extractingenergy from glucose and the utilization of this energyis carried out by a large number of enzymes. Thepathways of energy and redox metabolism of

erythrocytes have been identified and it has also longbeen recognized that the rates of all the metabolicprocesses of the cell depend on the properties of theenzymes that catalyse each of the required reactions,the number and quality of enzyme molecules present,the temperature and the concentration of substrates,cofactors, activators, inhibitors and the pH within thecell. Attempts have been made to construct computermodels that simulate this network of reactions in thered cell (Heinrich et al., 1977; Heinrich & Rapoport,1974; Rapoport et al., 1974). The glycolytic flux was

* This paper is dedicated to the memory of Henrik Kacser.§ Author to whom correspondence should be addressed.E-mail: ovadi.enzim.hu

437

. E T A L .438

found to be controlled by hexokinase (HK) andphosphofructokinase (PFK). The control strengthsfor the HK and PFK were calculated to be 0.69–0.73and 0.31–0.27 at pH 7.2 and 0.87–0.90 and 0.13–0.1at pH 8.2, respectively (Rapoport et al., 1974). Themodel was appropriate to describe the glycolytic fluxusing kinetic parameters determined in dilutedsystems with isolated enzymes. In addition, itdescribes the time courses of changes in concen-trations of glycolytic intermediates following changesin substrate concentrations (Schauer et al., 1981), andon other effects [for references see Keleti & Ovadi(1988)]. Additional experiments from other labora-tories which demonstrate the regulation of glycolysisthrough the change of energy charge by affecting theactivity of PFK and pyruvate kinase (Yoshino &Murakami, 1985) or PFK by ATP and/or HK byglucose 6-phosphate (Ataullakhanov et al., 1981;Melendez-Hevia et al., 1984) and of PFK and fruc-tose 1,6-bisphosphatase by fructose 2,6-bisphosphate(Hers & Van Schaftingen, 1982) are also consistentwith the model. These analyses suggest that ATPconcentration is kept constant in the cell by amechanism in which 2,3-bisphosphoglycerate by-passacts as an ‘‘energy buffer’’ (cf. Fig. 1), it acts as energysource and ATP change is buffered by variations inthe accumulation rate of fructose 1,6-bisphosphateand triose phosphate.

The role of PFK in the control of red cell glycolysishas been extensively discussed (Bosca & Corredor,1984; Fell, 1984). The published data suggest theimportance of PFK in the control of erythrocyteglycolysis; however, its control depends on the type ofcells which have various PFK isoforms. For example,in tumor cells or in yeast, a relatively high ratio offructose-1,6-bisphosphate to fructose-6-bisphosphatewas found indicating that PFK was not rate-limitingunder glucose utilizing steady-state conditions. There-fore, the contribution of the enzymes to the controlof glycolytic flux is varied by a number of effectsincluding isoforms and cellular conditions.

Compartmentation of Glycolytic Enzymes:

New Concept

The mathematical models including the recent oneshave ignored the experimental observations thatcertain glycolytic enzymes associate with the erythro-cyte membrane and these specific associations altertheir catalytic properties. For example, convincingstudies with PFK indicated that the activity curveshifts from a sigmoidal shape to a rectangular hyper-bola on binding of the enzyme to the erythrocytemembrane (Karadsheh & Uyeda, 1977). Up to now

several experiments using a variety of techniquesindicated isotonic binding of glycolytic enzymes.Activity of glyceraldehyde-3-phosphate dehydro-genase (GAPD) in vivo was measured by using 1HNMR to monitor non-invasively a couple 1H2Hexchange reactions in which the enzyme was involved(Brindle et al., 1982). The study showed that theenzyme was totally inhibited when bound to the band3 membrane protein of erythrocytes. In a series ofstudies Steck and his co-workers have demonstratedthat glycolytic enzymes bind specifically to the acidicN-terminal region of human erythrocyte band 3(Jenkins et al., 1985; Tsai et al., 1982). From theseenzymes aldolase and GAPD effectively compete withPFK for binding to band 3 protein and release thebound PFK (Higashi et al., 1979). More recently, arigorous test was developed providing direct evidencefor control of glycolysis by binding to the cytoplasmicextension of the anion transporter, band 3 proteinin vivo. The glycolytic flux was found to be modulatedover 30-fold by controlling the availability ofglycolytic enzyme binding sites at extreme N terminusof the anion transporter, band 3 (Low et al., 1993).By regulating the occupancy of the enzyme bindingsite at the N terminus of the anion transporter, the cellhas the potential to adjust its glycolytic flux over awide range.

The role of the isoenzyme-specific interactions inthe regulation of glycolysis in various tissues has beenextensively emphasized (for review see Ovadi, 1995).The presence of isoenzymes in several metabolicsteps can keep multiple pools of intermediates byisoenzyme-isoenzyme associations as demonstratedfor conversion of glucose-6-phosphate to pyruvate orglycogen via glycolytic or gluconeogenic pathways(Ureta, 1978, 1991).

Recently we provided evidence that the differentisoforms of brain PFK exhibited different affinitytowards MAP-containing microtubules (Vertessyet al. 1996). C-type PFK that predominantly occursin brain and tumor cells has much lower affinity toMTs than M-type has. The muscle type PFK underidentical conditions binds to MTs while the bindingof C isoform is not significant. The binding of muscleenzyme reduces the overall activity of the kinase sincethe inactive dissociated form of the enzyme associateswith MTs. This finding may have physiological relev-ance and it may partly explain the high uncontrolledglycolytic rate in tumor cells.

The data available about binding of glycolyticenzymes to microtubule suggest that the glycolyticenzymes with high isoelectric point bind to the acidicC-terminal ‘‘tail’’ of a subunit of tubulin (Carr &Knull, 1993; Itin et al., 1993; Volker & Knull, 1993).

GLUCOSE

G6P

F6P

FDP

ATP

ADP

*Hexokinase

*Glucosephosphateisomerase

ATP

ADP

*Phosphofructokinase

GAP

*Aldolase

DHAP*Triosephosphateisomerase

Glyceraldehyde-3-phosphatedehydrogenase

Pi NAD

NADH

1,3-DPG

*Diphosphoglyceratemutase

ATP

ADP*Phospho-glyceratekinase

2,3-DPG

*Diphosphoglyceratephosphatase

*Phosphoglyceratemutase

3PG

Pi

2PG

PEP

*Enolase

*Pyruvate kinase

*Lactate dehydrogenase

PYRUVATE

LACTATE

ADP

ATP

NADH

NAD+

439

F. 1. The glycolytic pathway in human erythrocytes. Enzymes whose deficiency have been demonstrated are indicated by an asterisk. Forsimplicity all reactions are denoted with single arrows. The portions of glycolysis studied in this paper are indicated by solid or dotted lines.

. E T A L .440

This C-terminal binding domain of a tubulin sharesmany properties with the N-terminal binding domainof human erythrocyte band 3 including sequencehomology (Knull & Walsh, 1992). Since it has beendemonstrated that glycolytic enzymes specificallybind to the acidic N-terminal region of band-3membrane protein in RBC, it appears, therefore, thatthe binding of glycolytic enzymes to domains ofeither MTs or red cell membrane is highly specific andit produces similar functional consequences. Thesemacromolecular associations have yet to be takeninto account in order to understand the regulation ofenergy production in the RBC and the molecularalterations of diseases caused by inherited or acquiredenzyme deficiency.

Enzyme Deficiency

Extensive evidence indicates that the metabolism ofcells can be impaired if the activity of only one of theparticipating enzymes is altered by spontaneousmutations (inherited or acquired enzymopathies) orby the administration of toxic drugs or for any otherreason (Schuster & Holzhutter, 1995). The fact thatglycolysis is crucial for RBC function is reflected bythe large number of inherited glycolytic enzymo-pathies found to result in hemolysis or other aber-rations (Tanaka & Zerez, 1990). Based on the energydependence of mature erythrocytes on glycolysis, thedepletion of ATP has been proposed to be the causeof the shortened life span in deficiencies of theglycolytic enzymes (Valentine et al., 1984). However,low red cell ATP levels are not invariably associatedwith loss of viability, and circulating ATP levels arenot necessarily diminished in patients with glycolyticenzymopathies (Beutler, 1980).

For human erythrocytes, deficiencies of about 20enzymes, associated with widely different degrees ofseverity and complexity have been identified so far(Fuji & Miwa, 1990; Valentine & Paglia, 1984).Nevertheless, quantitative relationships between thedegree of enzyme deficiency and the extent ofmetabolic dysfunction are very difficult to establishexperimentally. For most enzymopathies, the exper-imental and clinical observations can be satisfactorilyrationalized by the computational results [Schuster &Holzhutter (1995) and references therein]. The modelsfor the main metabolic pathways of the humanerythrocyte were successfully employed to describestationary and time-dependent metabolic states of thecells under normal physiological conditions as well asin the presence of enzyme deficiencies. Recently amathematical model was evaluated for predicting themetabolic effect of large-scale enzyme activity alter-

ations. This model was applied for study of enzymedeficiencies of RBCs (Schuster & Holzhutter, 1995).

Triosephosphate isomerase (TPI) deficiency is arate autosomal disease. There are only some 30 casesreported so far [Hollan et al., (1993) and referencestherein]. Although several isoenzymes have beenidentified in normal tissues, none of them have beenobserved to be specifically associated with a func-tionally deficient state. Clinically significant TPIdeficiency is usually associated with non-spherocytichemolytic anemia, progressive neurologic dysfunc-tion, and death in childhood (Hollan et al., 1993).

TPI, notable for its high catalytic efficiency,enhances the movement of a single proton to inter-convert DHAP and GAP in glycolysis and gluco-neogenesis by a factor of about 1010 (Nickbarg &Knowles, 1988). The rate of catalysis is diffusionlimited (Rose et al., 1990), and the equilibriumfavours the formation of DHAP by 20:1 (Lolis &Petsko, 1990). In fact, in the erythrocyte the moststriking metabolic abnormality is the 20- to 60-foldincrease in the concentration of DHAP, the substratefor the enzyme, suggestive of an almost completemetabolic block at this step (Valentine & Paglia,1984). Therefore, the disease could be a consequenceof an increased concentration of DHAP. Little orno modifications occur in the levels of ATP and2,3-diphosphoglycerate (Hollan et al., 1993).

In Hungary a 13-year-old boy (B.J. Jr.) withcongenital hemolytic anemia and hyperkinetic torsiondyskinesia was found to have severe TPI deficiency(cf. Table 1). One of his two brothers, (A.J.), a23-year-old amateur wrestler has hemolytic anemia aswell, but no neurological signs or symptoms. Both arecompound heterozygotes and have equally less than3% TPI activity in their red cells. Both parents anda third brother are healthy heterozygote carriersof the defect. The main characteristics of theTPI-deficient Hungarian family are summarized inTable 1. A.J. represents a unique phenotype from thepoint of view that all published homozygotes andcompound heterozygotes had severe neurologicalalterations from infancy or early childhood. Incontrast to the two affected Hungarian brothers,apart from one patient (Harris et al., 1970), allcompound heterozygotes died under the age of 6years. The dramatic decrease of TPI activity occursduring the time course of biological evolution dueto spontaneous mutations affecting the amino acidsequence, and, thus, the spatial arrangement ofenzyme molecules. The Hungarian family is charac-terized by two mutations: one is mis-sense mutationwithin codon 240 (Chang et al., 1993). The othermutation has been recently localized (Hollan et al.,

441

T

1C

hara

cter

istic

feat

ures

inth

em

embe

rsof

the

TP

Ide

ficie

ntfa

mily

Hem

olyt

icN

euro

logi

cD

HA

Ple

vel

Cha

ract

eristics

ofT

PI

Enz

yme

activi

ty(V

max)

U/g

hem

oglo

bin

anem

iaa

diso

rder

anm

ol/m

lR

BC

isol

ated

ghos

tsth

erm

olab

ility

aT

PI

PF

KA

ldol

ase

Nor

mal

00

20–2

7pe

arly

whi

teN

one

1051

–184

211

.12

2.1

3.22

0.9

Pro

positu

s+

++

+85

42

1pi

nkT

otal

31.5

21.

012

.32

3.5

3.32

0.2

(B.J

.Jr

.)B

roth

er+

054

62

35ye

llow

gel-lik

eT

otal

40.0

21.

7N

TN

T(A

.J.)

Bro

ther

00

NT

NT

Par

tial

745a

NT

NT

(T.J

.)M

othe

r0

071

pear

lyw

hite

Par

tial

436a

NT

NT

(Mrs

.B

.J.)

Fat

her

00

44pe

arly

whi

teN

one

644a

NT

NT

(B.J

.)

DH

AP

leve

lin

who

lebl

ood

was

dete

rmin

edas

in(H

olla

net

al.,

1993

).E

nzym

eac

tivi

ties

wer

em

easu

red

acco

rdin

gto

(Beu

tler

etal

.,19

77).

TPI

activi

tyw

asm

easu

red

with

GA

Pas

asu

bstr

ate.

Gho

sts

wer

epr

epar

edac

cord

ing

to(V

erte

ssy

&St

eck,

1989

).N

T,

not

test

ed;

a ,da

tata

ken

from

(Hol

lan

etal

.,19

93).

10Time (min)

[NA

DH

] (µ

M)

30

20

10

2 4 80 6

TPI

PFKControl

Aldolase

10Time (min)

[NA

DH

] (µ

M)

30

20

10

2 4 80 6

TPIPFKControl

Aldolase

. E T A L .442

unpublished). The modified TPI termed ‘‘deficient’’has a slightly higher Michaelis constant for DHAPbut normal Michaelis constant for GAP. The mutantenzyme is heat unstable and has slower electrophor-etic mobility as compared with the normal enzyme(Hollan et al., 1993). In spite of the considerableamount of knowledge accumulated about this raregenetic defect, the pathomechanism of both thehemolytic anemia and the neurological symptoms isstill obscure.

Flux Studies in Normal and TPI Deficient

Hemolysates

In the cases where clinically manifested TPI defectwas identified the TPI activities in erythrocytesvaried between 1.6–28% of the normal value. In mostcases (from 28 diseased persons) the TPI activity wasless than 20.0% (Eber et al., 1991). The two affectedHungarian brothers have less than 3% activity intheir RBCs (cf. Table 1). Since the activity of TPI isthe highest of any glycolytic enzyme in RBCs, it hasvirtually no control property in normal cells and thusone can ask whether the reduced TPI occurring indeficient cells is able to sustain the normal glycolyticflux to produce the appropriate energy for red cellfunctions.

The influence of particular enzyme activities on theflux of metabolites in a pathway can be estimated by‘‘modulating’’ enzymes and measuring the response inselected parts of the system. In this particular case weanalysed a relevant segment of glycolysis includingTPI and some related enzymes but not HK which isknown to be by far the enzyme of lowest capacity.The well-established concept of the metabolic controltheory (Heinrich & Rapoport, 1974; Kacser & Burns,1973) was used in our laboratory to test thecontribution of TPI to the control of the segments ofglycolytic pathway in TPI deficient cells in compar-ison to normal cells. In the first type of experimentthe basic flux of normal and TPI deficient cells wereanalysed with excess Fru-6-P, Mg-ATP and NADHas substrates (cf. Fig. 2). In this case the formation ofDHAP produced by the PFK/aldolase/TPI catalysedconsecutive reactions (Fig. 1) was monitored bycoupling them with GDH as auxiliary enzyme. In asecond type of experiment DHAP was an inter-mediate of the pathway and NADH produced inequimolar amount with diphosphoglycerate, aproduct of GAPD reaction (Fig. 1) was monitored inthe presence of excess Fru-6-P, Mg-ATP, NAD andarsenate as substrates (Fig. 3). For these studies thehemolysates of the isolated RBCs were used, preparedas described in Hollan et al. (1993). The amount of

F. 2. Fructose-6-phosphate conversion in the consecutivereactions catalysed by PFK/aldolase/TPI in hemolyzed red bloodcells from a normal individual (upper panel) and from thePropositus (B. J., Jr.) (lower panel). Packed red blood cells(prepared from the washed isotonic red blood cell preparation bya final centrifugation at 5000 g, 4°C, 20 min) were lysed by dilutingthem fourfold into 10 mM Tris/HCl buffer, pH 8.0, containing1 mM EDTA and 5 mM mercaptoethanol, followed by three cyclesof freezing in liquid N2 and thawing [c.f. Hollan et al. (1993)]. Theselysed cells were used as hemolyzate in 150-fold dilution. ExogenousGDH was added to 0.06 mg ml-1 final concentration and fluxwas measured by monitoring NADH consumption at 340 nm inthe presence of 1 mM MgATP and 25 mM NADH in 100 mMTris/HCl buffer, pH 8.0 at 25°C. The reaction was started by theaddition of 1 mM Fru-6-P. Exogenous enzymes PFK, aldolase orTPI, where indicated, were added separately to the assays at0.02 mg ml−1.

hemolysates (extracts) for the assays was limited bytheir turbidities. Accordingly, the kinetics weremeasured with 150-fold diluted hemolysates of thenormal and deficient cells, hemoglobin content ofwhich varied between 0.5 and 0.8 mg ml−1. As shownin Fig. 3, in the second type of experiment in whichTPI catalyses the physiological DHAP4 GAPconversion the basic fluxes catalysed by endogenousenzymes were very low. When DHAP was coupledwith exogeneous GDH as indicated in the first type ofexperiment then the flux (NADH consumption) waswell-detectable (cf. Fig. 2) and it corresponded to1.7 mM NADH min−1 (3.1 U g-1 hemoglobin). Thedata were similar whether the extract of normal orTPI deficient cell was tested. These observations

10Time (min)

[NA

DH

] (µ

M)

30

20

10

2 4 80 6

TPIGAPD, PFKControl

Aldolase

10Time (min)

[NA

DH

] (µ

M)

30

20

10

2 4 80 6

GAPDPFKControl,TPI

Aldolase

443

F. 3. Fructose-6-phosphate conversion in the consecutivereactions catalysed by PFK/aldolase/TPI/GAPD in hemolyzedred blood cells from a normal individual (upper panel) and fromthe Propositus (B. J., Jr.) (lower panel). Hemolyzates prepared asdescribed in Fig. 2 were used at 150-fold dilution in the cuvettes.Flux was measured by monitoring NADH production at 340 nmin the presence of 1 mM MgATP, 4 mM NAD, 10 mM sodiumarsenate in 100 mM Tris/HCl buffer, pH 8.0 at 25°C. The reactionwas started by the addition of 1 mM Fru-6-P. Exogenous enzymesPFK, aldolase, TPI or GAPD, where indicated were addedseparately to the assays at 0.02 mg ml−1.

substantial increase in TPI activity caused a marginalincrease in the flux catalysed by deficient cells (Fig. 3).In the normal cells the addition of exogeneous TPIdid not alter the basic flux. The observation that inthe normal cell TPI virtually does not controlglycolysis is in agreement with the expectation. Whencomparing the data of normal and deficient cells it canbe argued that TPI activities around the in vivo leveldo not appear to limit significantly the hexose andtriosephosphate conversion, even in the deficient cells.

In additional experiments the concentrations ofPFK, aldolase and GAPD of hemolysates weremodulated by separate addition of the exogeneousenzymes and the fluxes were analysed in extractsfrom normal and deficient cells. The data from thesestudies allowed us to compare the contributions of theenzymes with the control of a segments of glycolysis,where the significant control effect of HK can be dis-regarded. We compared the measure of the sensitivityof fluxes to the change in the enzyme activity at PFK,aldolase and GAPD catalysed steps. The results ofthe titrations with excess activities of the enzymes areshown in Figs. 2 and 3.

In one of his last papers, Henrik Kacser (Small &Kacser, 1993) developed a method for unbranchedchains to estimate the response of metabolic systemsusing a single large change in enzyme activities. Adeviation index, DJ

E , (Small & Kacser, 1993) is intro-duced which gives a measure of the relative change ina flux:

DJE =(DJ/DE)Er/Jr

where DJ is calculated from the fluxes measured in theabsence and presence of exogeneous enzymes; DE isthe difference of the activities (aldolase or PFK) of thehemolysates before and after addition of exogeneousenzymes. The ratio of enzyme activity/flux (Er/Jr) wascalculated at the ‘‘new point’’, after addition ofexogeneous enzymes: 4 U and 0.08 U for PFK andaldolase, respectively. According to the experimentaldata presented in Fig. 2 and Table 1. in the case ofthe normal cells, at pH 8.0, DJ

E values are 0.852 0.1and 0.142 0.05 for aldolase and PFK catalysed reac-tions, respectively. Since there is a direct relationshipbetween the deviation indices and control coefficients(Small & Kacser, 1993), it can be concluded from ourquantitative data that under our experimental condi-tions aldolase has more significant control on the fluxthan PFK. A qualitatively similar result was observedin the case of the TPI deficient cells. It has to be addedthat the ratio of DJ

aldolase/DJPFK depends extensively on

the pH, it decreases from 5.7 to 0.5 by decreasing pHfrom 8.0 to 7.2. These data suggest that aldolase andPFK beside HK are important control enzymes of the

indicate that although TPI activity of the deficientcells is only 3% or even lower than the normal RBC,the rate of DHAP formation is not limited by the TPIactivity occurring in the deficient cells.

To get additional data for the control role of TPIin deficient cells exogeneous TPI was added to theassays and the fluxes of both NADH consumption(GDH coupled reaction) and NADH production(in the presence of NAD and arsenate) were analysed.As shown in Fig. 2, if DHAP was coupled by excessexogeneous GDH (first type of experiment) then TPIcaused additional increase in the fluxes in bothnormal and deficient cell hemolysates. This findingindicates that TPI activity is not in enough excess ascompared to the activity of auxiliary enzyme, GDH.When TPI catalysed the DHAP4 GAP conversionin the reaction sequence (second type of experiment)which is approximately 20-fold lower than theconversion rate of the reverse direction, a quite

100

5000

0

0.01TPI Vmax (% normal activity)

Met

abol

ite

con

cen

trat

ion

(%

nor

mal

val

ue)

4000

3000

2000

1000

0.1 1 100

DHAP calculated

DHAP measured

ATP

. E T A L .444

RBC glycolysis and that the control properties of theenzymes extensively depend on the conditions.

In order to compare directly the endogenousaldolase and PFK activities of the normal anddeficient cells the initial rates of the reactions weremeasured in the hemolysates at substrate saturationswith excess auxiliary enzymes. As shown in Table 1.the Vmax of aldolase is significantly lower than thatof PFK in both normal and deficient cells at pH 8.0,that predicts lower capacity for aldolase than for PFKduring the glycolysis. Vmax values of TPI reactionmeasured in normal and deficient hemolysates (cf.Table 1) refer to GAP4 DHAP conversion fromwhich Vmax values for the reverse direction can becalculated assuming constant equilibrium, Kequ=[DHAP]/[GAP]=20, for both systems. This value(50–90) for normal cell is still far above the Vmax valuesof both aldolase and PFK, however, for propositus(1.6) it becomes comparable with that of aldolasewhich is the slowest enzyme of this segment. Althoughfrom the Vmax values of the sequential enzymes mustnot be directly concluded for the capacities of theenzymes in the sequence, nevertheless, in the light ofthe Vmax data it is not surprising that the exogeneousTPI could enhance the flux to some extent in deficientcell if the reaction was coupled with GAPD.

Why is the DHAP Level so High in the

Deficient Cells?

A distinctly elevated DHAP level was detected inthe erythrocytes of all patients with defective TPIalthough the activity of the isomerase was variedwidely. The concentration of DHAP in erythrocytesof the Hungarian family is also extensively increasedand it is extremely high in the two affected brothers(cf. Table 1). The concentration of the GAP isnormal and, in addition, that of ATP and2,3-diphosphate does not differ significantly fromthat of normal cells (Hollan et al., 1993). Althoughmathematically oriented, theoretical research haspredicted in many cases the metabolic changes causedby changing the activity of a given enzyme in themetabolism of RBC (Schuster & Holzhutter, 1995),considerable discrepancies can be found in the DHAPmetabolism. According to the theoretical calculationsa few per cent of TPI activity from the normal valueshould not result in any elevation in DHAP level(cf. Fig. 4).

The TPI activity of the members of the Hungarianfamily are lower than the normal values but differentfrom each other, thus, by determining the in vivoDHAP concentrations in their blood (Hollan et al.,1993), we were able to construct an experimental

F. 4. The dependence of stationary DHAP concentrations onthe activity (Vmax) of TPI. TPI activity (in washed red blood cells)and ATP and DHAP levels (in whole blood) were determinedaccording to Hollan et al. (1993). Solid circles represent ATP levels,solid and open squares represent DHAP levels measured in thepresent study or taken from Hollan et al. (1993), respectively. Solidcurves are theoretical curves computed by Schuster et al. (1995).

curve of DHAP concentration vs. TPI activity. Thiscurve is compared with the theoretical one computedfor a wide range of TPI activity (Schuster &Holzhutter, 1995). As shown in Fig. 4, according tothe computation model the extensive reduction ofVmax of TPI (to about 0.1% of the normal value)results in only a two-fold elevation in the DHAP level.In contrast to the theoretical predictions, a 45-foldincrease of DHAP concentration was measured atabout 3% of the normal TPI activity in the patient.There are other examples that illustrate the dis-crepancy between DHAP level vs. TPI activity. Forexample, in the case of a Turkish girl, DHAPconcentration is 18-fold normal while the TPI activitywas reduced to only 28% of the normal (Eber et al.,1991). The ATP level determined in the RBCs(Hollan et al., 1993) appears to be consistent with thecomputed data (Schuster & Holzhutter, 1995) andindependent of the actual values of TPI activities(cf. Fig. 4).

The possible explanations for the discrepancy ofDHAP level vs. TPI activity are summarized asfollows:

(1) The problem in obtaining experimental datafor mature erythrocytes in the case of severe enzymedeficiencies originates from the fact that thepopulation of red cells may contain large amounts ofreticulocytes which generally have higher enzymeactivities as well as higher intermediate concentra-tions than mature erythrocytes (Piomelli & Seaman,1993). The ratios of the erythrocytes which are near orbeyond the critical threshold of loosing cell integrity

445

and of reticulocytes are different in the normal andthe TPI deficient cells. However, the reticulocytecontents of the patient and his compound hetero-zygote brother were only slightly elevated (3% vs. therange of 0.8–1.0 per cent) (Hollan et al., 1993).

(2) The discrepancy for TPI deficiency may alsooriginate from the remarkable instability of almost allabnormal enzyme variants (Tanaka & Zerez, 1990).Unstable proteins may even be more degraded duringthe experimental procedures and thus the activitiesmeasured may not reflect the critical level of activitiescausing cell damage.

(3) Comparing the results of computerized calcu-lations (Schuster & Holzhutter, 1995) with measuredmetabolite concentrations (cf. Fig. 4), discrepanciesmay originate from differences between kinetic para-meters determined in normal and deficient cells. Infact, the KM of DHAP to the mutant TPI occurringin deficient cells is higher than that of the normal,while the KM of GAP is the same (Hollan et al., 1993).

The velocities of the DHAP4 GAP conversioncatalysed by TPI in normal and deficient cells canbe calculated according to the Michaelis–Mentenequation:

vnorm

vdef =Vnorm

max [DHAP]norm([DHAP]def +KdefM )

Vdefmax [DHAP]def([DHAP]norm +Knorm

M )

The KM values are 1.5 mM for normal and 2.8 mMand 3.0 mM for deficient TPIs from the two affectedbrothers (Hollan et al., 1993). [DHAP] for normaland deficient cells are given in Table 1. Vmax valuesof the DHAP4 GAP conversion for normal anddeficient TPIs were calculated from data presentedin Table 1 using the Halden relationship assumingidentical equilibrium constant for DHAP/GAP con-version catalysed by normal or mutant TPIs. Theratios of the velocities of the normal and mutant TPIcatalysed reactions are 1.0 and 1.1 for B.J. Jr. andA.J., respectively. The fact that these ratios corre-spond to the unity suggests that in deficient cells thereduction of Vmax and increase of KM are compensatedby the elevation of DHAP concentration; thus TPImay not limit the endogenous flux.

(4) In general, it is postulated that the steady-statemetabolite level is adjusted by its production/conversion. DHAP is not an inert metabolite in cellswhich are active in lipid synthesis, since it is anessential precursor of ether lipids. However, there isno lipid synthesis in mature red blood cells. In fact,in the hemolysates we were not able to detect GDHactivity, which would be responsible for the DHAPconversion to lipid synthesis (data not shown).

(5) Additional reason for discrepancies betweenthe measured and the theoretically postulated DHAP

levels and TPI activities may be due to the fact that anumber of yet unknown structural kinetic parametersare not taken into consideration. One of these para-meters is the specific association of glycolytic enzymesto the N-terminus of band-3 membrane protein underphysiological conditions (Harrison et al., 1991;Jenkins et al., 1985; Low et al., 1993; Rogalski et al.,1989; Tsai et al., 1982). There are data that aglycolytic enzyme complex from TPI to pyruvatekinase is bound in vivo to the cytoplasmic domainof band 3 (Fossel & Solomon, 1978). Due to thesespecific associations some glycolytic enzymes areinactivated, only the unbound enzymes exhibitcatalytic activities (Low et al., 1993). The binding ofthe enzymes is reversible. Therefore, it seems to be aplausible explanation for the discrepancy of TPIactivity and DHAP level that under in vivo conditionsthe TPI activity in the deficient cells is further reduceddue to the binding of the isomerase to band 3 in thered cell membrane or to other cytoplasmic enzymes.Thus a decrease of TPI activity of deficient cells couldeasily reach the limit value which is not able to ensurethe rapid equilibrium of the triosephosphates andresults in the enormous accumulation of DHAP.

TPI Deficiency and Enzyme Compartmentation

It is a widely discussed issue that the low TPIactivity in deficient cells leads to a metabolic block inthe glycolytic pathway that results in an increasedconcentration of DHAP in the erythrocytes. Never-theless, the relationship between high DHAP levelsand increased hemolysis is unclear, since accumula-tion of this compound occurs in cases of diphospho-glycerate mutase deficiency without hemolysis (Rosaet al., 1978). Also, there are no indications thatDHAP may inhibit regulatory enzymes by itsincreased concentration (Eber et al., 1991).

The main objective of this study was to assess theseverity of cellular dysfunction associated with TPIdefect. Since there is no reliable model which de-scribes the relation between TPI activity and DHAPlevel characteristic for deficient cells we propose thatthe microcompartmentation of TPI, which may ormay not be different in normal and TPI deficient cells,could be responsible for the altered metabolism in thedeficient RBCs. Different binding affinity of normaland mutant isomerase molecules to the red cell mem-branes may originate from differences in the tertiarystructure of the mutant enzyme as well as fromchanges in membrane fluidity (Hollan et al., 1995).Direct binding studies on this issue are in progress inour laboratory. Concerning the physiological relev-ance of the computerized models for the RBC

. E T A L .446

metabolism we agree with the argument of Beutler(1980) that ‘‘the usefulness of the models has beenlimited by the fact that all the in vivo interactions arestill not entirely understood’’. In fact he referred tointeractions of enzymes with metabolites, allostericligands etc. Now, in the light of recent data theimportance of the macromolecular interactions hasto be underlined. These regulatory mechanisms areprobably different in normal and deficient cells.

This work was supported by grants from the HungarianNational Science Foundation, OTKA, T-5412, T-6349 andT-17830 to J.O and F 017392 to B. G. V. We thank EmmaHlavanda for her expert assistance.

REFERENCES

A, F. I., V, V. M., Z, A. M.,P, A. V., P, O. V., K, B. N. &E, L. I. (1981). The regulation of glycolysis in humanerythrocytes. The dependence of the glycolytic flux on the ATPconcentration. Eru. J. Biochem. 115, 359–365.

B, E. (1980). A commentary. Blood Cells. 6, 827–829.B, E., B, K. G., K, J. C., L, G. W., R, B.

& V, W. N. (1977). International Committee forStandardization in Haematology: recommended methods forred-cell enzyme analysis. Br. J. Haematol. 35, 331–340.

B, L. & C, C. (1984). Is phosphofructokinase therate-limiting step of glycolysis? Trends Biochem. Sci. 9, 372–373.

B, K. M., C, I. D. & S, R. J. (1982). A 1Hn.m.r. study of the kinetic properties expressed by glyceraldehydephosphate dehydrogenase in the intact human erythrocyte.Biochem. J. 208, 583–592.

C, D. & K, H. (1993). Aldolase-tubulin interactions:removal of tubulin C-terminals impairs interactions. Biochem.Biophys. Res. Commun. 195, 289–293.

C, M.-L., A, P. J., W, X., H, S., L A.& M, L. E. (1993). Human triosephosphate isomerasedeficiency resulting from mutation of phe-240. Am. J. Hum.Genet. 52, 1260–1269.

E, S. W., P, A., B, A., G, M., K,W. K. G., K, J., M, R. & S, W. (1991).Triosephosphate isomerase deficiency: haemolytic anaemia,myopathy with altered mitochondria and mental retardation dueto a new variant with accelerated enzyme catabolism anddiminished specific activity. Eur. J. Pediatr. 150, 761–766.

F, D. A. (1984). Phosphofructokinase and glycolytic flux.Trends Biochem. Sci. 9, 515–516.

F, E. T. & S, K. (1978). Ouabain-sensitive interactionbetween human red cell membrane and glycolytic enzymecomplex in cytosol. Biochim. Biophys. Acta. 510, 99–111.

F, H., M, S. (1990) Recent progress in the molecular geneticanalysis of erythro-enzymopathy. Am. J. Hematol. 34, 301–310.

H, S. R., P, D. E., J, E. R., V, W. & K,R. L. (1970). Triosephosphate isomerase deficiency in an adult.Clin. Res. 18, 529–535.

H, M. L., R, R. A, P., G, R. L. &L, P. S. (1991). Role of band 3 tyrosine phosphorylationin the regulation of erythrocyte glycolysis. J. Biol. Chem. 266,

4106–4111.H, R., R, S. M. & R, T. A. (1977).

Metabolic regulation and mathematical models. Prog. Biophys.Mol. Biol. 32, 1–82.

H, R. & R, T. A. (1974). A linear steady-statetreatment of enzymatic chains. General properties, control andeffector strength. Eur. J. Biochem. 42, 89–95.

H, H.-G. & V S, E. (1982). Review article:fructose 2,6-bisphosphate 2 years after its discovery. Biochem. J.206, 1–12.

H, T., R, C. S. & U, K. (1979). The interactionof phosphofructokinase with erythrocyte membrane. J. Biol.Chem. 254, 9542–9550.

H, S., F, H., H, A., H, K., K, H., M,S., H, V., G, E. & I-K, M. (1993).Hereditary triosephosphate isomerase (TPI) deficiency: twoseverely affected brothers one with and one without neurologicalsymptoms. Hum. Genet. 92, 486–490.

H, S., D, I., S, L., H, M., M, M.,H, V. & F, T. (1995). Erythrocyte lipids intriose-phosphate isomerase deficiency. PNAS 92, 268–271.

I, C., B, Y., C U. & D, H. (1993). Selectiveinhibition of Plasmodium falciparum aldolase by a tubulinderived peptide and identification of the binding site. Mol.Biochem. Parasitol. 58, 135–143.

J, J. D., K, F. J. & S, T. L. (1985). Mode ofinteractions of phosphofructokinasae with the erythrocytemembrane. J. Biol. Chem. 260, 10426–10433.

K, H. & B, J. A. (1973). The control of flux. In: RateControl of Biological Processes (Davies, D. D. ed), pp. 65–104.Cambridge: Cambridge University Press.

K, N. S. & U, K. (1977). Changes in allostericproperties of phosphofructokinase bound to erythrocytemembranes. J. Biol. Chem. 252, 7418–7420.

K, T. & O, J. (1988). Control mechanism by dynamicmacromolecular interactions. Curr. Top. Cell. Regul. 29, 1–33.

K, H. R. & W, J. L. (1992). Association of GlycolyticEnzymes with the Cytoskeleton. Curr. Top. Cell. Regul. 33,

15–30.L, E. & P, G. A. (1990). Crystallographic analysis of the

complex between triophosphate isomerase and 2- phosphoglyco-late at 2.5 A� resolution: implications for catalysis. Biochemistry.29, 6619–6625.

L, P. S., R, P. & H, M. L. (1993). Regulationof glycolysis via reversible enzyme binding to the membraneprotein, Band 3. J. Biol. Chem. 268, 14627–14631.

M-H, E., S, J. M. & P, J. A. (1984). Studieson glycolysis in vitro: role of glucose phosphorylation andphosphofructokinase activity on total velocity. Int. J. Biochem.16, 476–496.

N, E. B. & K, J. R. (1988). Triosephosphateisomerase: energetics of the reaction catalysed by the yeastenzyme expressed in Escherichia coli. Biochemistry. 27,

5939–5947.O, J. (1995). Cell architecture and Metabolic Channeling.

Heidelberg: Springer-Verlag.P, S. & S, C. (1993). Mechanism of red blood cell

aging: Relationship of cell density and cell age. Am. J. Hematol.42, 46–52.

R, T. A., H, R., J, G. & R, S.(1974). A linear steady-state treatment of enzymatic chains.A mathematical model of glycolysis of human erythrocytes.Eur. J. Biochem. 42, 107–120.

R, A. A., S, T. L. & W, A. (1989). Associationof glyceraldehyde-3-phosphate dehydrogenase with the plasmamembrane of the intact human red blood cell. J. Biol. Chem. 264,

6438–6446.R, R., P, M-O. & B, Y. (1978). The first case of a

complete deficiency of diphosphoglycerate mutase in humanerythrocytes. J. Clin. Invest. 62, 907–915.

R, I. A., F, W. J. & W, J. V. B. (1990). Proton diffusionin the active site of triosephosphate isomerase. Biochemistry. 29,

4312–4317.S, M., H, R. & R, S. M. (1981). Mathe-

matische Modellierung der Glykolyse und des Adeninnukleo-tidstoffwechsels menschlicher Erythrozyten II. Acta Biol. Med.Germ. 40, 1683–1697.

S, R. & H, H.-G. (1995). Use of mathematicalmodels for predicting the metabolic effect of large-scale enzyme

447

activity alterations. Application to enzyme deficiences of redblood cells. Eur. J. Biochem. 229, 403–418.

S, J. R. & K, H. (1993). Responses of metabolicsystems to large changes in enzyme activities and effectors. 1. Thelinear treatment of unbranched chains. Eur. J. Biochem. 213,

613–624.T, K. R. & Z, C. R. (1990). Red cell enzymopathies of

glycolytic pathway. Sem. Hematol. 27, 165–185.T, I., M, S. N. P. & S, T. L. (1982). Effect of red cell

membrane binding on the catalytic activity of glyceraldehyde-3-phosphate dehydrogenase. J. Biol. Chem. 257, 1438–1442.

U, T. (1978). The role of isozymes in metabolism: A modelof metabolic pathways as the basis for the biological role ofisozymes. Curr. Top. Cell. Reg. 13, 233–251.

U, T. (1991). The role of isoenzymes in metabolite channeling.J. theor. Biol. 152, 81–84.

V, W. N. & P, D. E. (1984). Erythrocyte enzym-opathies, hemolytic anemia, and multisystem disease: anannotated review. Blood. 64, 583–591.

V, B. G., K, J. & O, J. (1996). Specificcharacteristics of phosphofructokinase-microtubule interaction.FEBS Lett. 379, 191–195.

V, B. G. & S, T. L. (1989). Elasticity of the human redcell membrane skeleton. Effects of temperature and denaturants.Biophys. J. 55, 255–262.

V, K. W. & K, H. R. (1993). Glycolytic enzyme-tubulininteractions: role of tubulin carboxy terminals. J. Mol. Recognit.6, 167–177.

Y, M. & M, K. (1985). AMP deaminase reactionas a control system of glycolysis in yeast. J. Biol. Chem. 260,

4729–4732.