The Enzymes of Nucleotide Biosynthesis
Key enzymes in the biosynthesis of purines and pyrimidines: their regulation by
allosteric effectors and by phosphorylation
E. A. Carrey
Department of Biochemistry, University of Dundee, Dundee DD I 4HN, Scotland, U.K.
CAD, the key enzyme in pyrimidine
biosynthesis
The first step dedicated to pyrimidine biosynthesis in mammals is catalysed by carbamoylphosphate synthase I1 (CPS 11; EC6.3.5.5). This
activity is regulated through feedback inhibition
by UTP, the end-product of the pathway, and
activated by phosphoribosylpyrophosphate (PRib-PP), a substrate of a later step in the pathway [l]. The CPS I1 activity is part of the
240 kDa multienzyme polypeptide CAD, named
after the activities it contains in discretely folded
domains: CPS 11, aspartate transcarbamoylase
(ATCase; EC 2.1.3.2) and dihydro-orotase
(EC 3.5.2.3).
Phosphorylation of CAD
Protein kinase A (PKA; cyclic AMP-dependent
protein kinase) phosphorylates CAD in vitro at
two sites [2]. We have mapped site 1 to the Cterminal region of the CPS I1 domains (Figure
l), and site 2 to the linker between the ATCase
Abbreviations used: AMP-PK, AMP-activated protein
kinase; ATCase, aspartate transcarbamoylase; CPS 11,
carbamoyl-phosphate synthase 11; PKA, protein kinase
A; P-Rib-PP, phosphoribosylpyrophosphate.
a99
and DHOase domains [3]. We believe that site 1
is the regulatory site, because the stoichiometry
of labelling of site 1 is more directly related to
the properties of the phosphorylated CAD, which
are as follows: an increased affinity of CPS I1 for
ATP and magnesium ions; an antagonism of the
effect of UTP on CPS 11; increased susceptibility
of CAD to proteolytic enzymes [3].
Allosteric effectors
The antagonism between the effects of UTP and
of phosphorylation may result from direct steric
interference between the bound U T P and phosphorylation site 1. The presence of UTP prevents phosphorylation of CAD, but not peptide
substrates, by PKA [4] ; thus UTP is acting upon
the substrate protein to diminish the accessibility
of the sites to PKA. It has been proposed that
the 20 kDa C-terminal region of the CPS I1
domains in CAD (Figure 2) is an allosteric
domain, by analogy to other CPS enzymes [S]
and because the UTP effect is lost when the
region is deleted [6]. However, previous work (E.
A. Carrey, unpublished work) indicates that the
phosphorylated CAD can still bind UTP, so it is
possible that phosphorylation prevents the effect
of UTP, not its binding. We are currently isolating a peptide from the allosteric domain of CAD
Figure I
Phosphorylation sites in CAD proteins
The sequences corresponding to the sites that are phosphorylated in mammalian CAD are
compared with the consensus RRXS which is phosphorylated in numerous target enzymes by
PKA The proteins from hamster and yeast have been investigated experimentally for phos
phorylation by the kinase [2.3,10] Serine residues that are known to be phosphorylated are
underlined
ELLQ.2
Yeast
Slime W U l d
Fruit fly
Hameter
[lo]
Ill]
I111
191
protein kiname
A
LYINLP-SANRF-RRPASWSKQYKTRRLAVDY
LFINLP-SK~Y-RRPSSPWSRQYSLRRVAIDF
LVINLPWSSQQVPRRVSSPWTHOYRTRR-AVBY
LVINSLW-RGAOORRLkS~KQYRTRRLAADP
RRxa
conmenmutt
aiu
R=S
Yeast
81WUld
F a i t fly
~am8t.r
1101
[I21
[ll]
191
NPSPASITASIULQSTSAIRP~ITBEAIA
TKNQTTNITSPSLISDSPNKAINKIKSTST
TTWPWTPLPASSPPKDPOOWPYTBYRPKVP
RFHLPPRIHWDPQLP-BPKEKPSRKW
R R U
I995
Biochemical Society Transactions
Figure 2
Domains and phosphorylation sites in mammalian CAD
The functional domains are shown as stylized circles approximately in proportion to the sizes
of fragments liberated by limited proteolysis. GAT, glutamine amidotransferase; CPS, duplicated ATP-binding synthase domains; 2OK, domain of unknown function; allo, putative allosteric domain; DHO, dihydro-orotase domain; ATC, ATCase domain. The inter-domain linker
region is shown as an exposed polypeptide chain. The putative site for binding UTP is
indicated in the allo domain. The phosphorylation sites are indicated as follows: A l , A2, sites
phosphorylated by PK4; D?. sequence homologous to a site for DNA-dependent protein
kinase.
900
M
U P
Figure 3
Putative sites for the regulation of key enzymes in
purine biosynthesis
The similar sequences implicated in P-Rib-PP binding are shown
for a number of phosphoribosykransferases (PRTases) [ 14, 17, I81
and a human P-Rib-PP synthase [21], and compared with the
consensus seauence for DhOSDhor/latlon bv AMP-PKr161. The R
residue may also be K,' and T may also' be S; the @ symbol
represents non-polar residues such as M, L, I and V The putative
site for phosphorylation in the PAP 39 protein is also shown [22]
HGPRTase, hypoxanthine PRTase; Ade PRTase, adenine PRTase.
L
V n t h x e 'I1(human)
Uracil PRTase (E. coli)
Orotate PRTase (human)
HGPRTase (human)
Ade PRTase (human)
Gln arnidoPRTase (rat)
AMP-PK consensus
PAP 39 insert
El1
>
D R V A I L ~ ~ C V T I C -
[I81 ERHALIVpPYIATPQSV
[17] QETcLIIMvvTsQssvL
[17] ~ K N V L I ~ I I D T Q K T M Q
171
QQRVVVIQDLLATQQTYH
141
PKRIVLIpPSIVRQHTISPIIK
PZ]
PDADQRHSPPMVKHATVHPQL
c61
@XRXXTXXX@
to which UTP is covalently bound and attempting to obtain the sequence.
The activator P-Rib-PP activates CPS I1 by
increasing the affinity for ATP and magnesium,
similarly to phosphorylation. The P-Rib-PP site
will be more difficult to isolate chemically
because of the lability of the reagent. T h e consensus P-Rib-PP binding site that is seen in
phosphoribosyltransferases (Figure 3) is not
found in the CAD sequence.
Allostery in CPS II and 'reciprocal
allostery' in CAD
We have proposed that the allosteric effectors
and phosphorylation have a common mechanism
in altering the distribution of CPS I1 between
Volume 23
two states or conformational forms. T h e P state
is stabilized by P-Rib-PP or by phosphorylation,
and has a high Vma.and a low K,,, for ATP and
for magnesium
ions. T h e U state is stabilized by
binding of UTP, and has a low
and high Km
for ATP and magnesium [7]. These two states
resemble the R and T States respectively Of C h s sical allosteric theory. The - c P S structure
includes two duplicated domains with substantial
sequence identity, each catalysing a different
ATP-dependent part-reaction. T h e allosteric
transition may thus entail altered interactions
between the two ATP binding domains within
the CPS I1 region rather than between two or
more separate molecules of CAD. This interpretation is implicit in the model for the CPS
structure, which places the allosteric domain in
contact with each of the ATP-binding domains
v,,,,
~51.
The presence of the three different enzyme
activities in the multienzyme polypeptide poses
some risks to the organism, notably the possibility of losing all three enzyme activities through
a single polar mutation. On the other hand, the
potential for enhanced activity or novel sites for
regulation has often been speculated upon. We
have recently proposed a novel regulatory mechanism that justifies the multienzyme structure
for the CPS I1 and ATCase reactions. We have
called the phenomenon 'reciprocal allostery',
since the substrates for ATCase and CPS I1 each
activate the other activity (H. S. Irvine, S. M.
Shaw and E. A. Carrey, unpublished work).
We measured the incorporation by purified
CAD of radioactive bicarbonate into carbamoyl
aspartate or carbamoyl phosphate, and thus demonstrated the inhibition of carbamoyl phosphate
The Enzymes of Nucleotide Biosynthesis
synthesis in the presence of excess unlabelled
carbamoyl phosphate (product-inhibition). T h e
overall synthesis of carbamoyl aspartate was
measured by a colorimetric assay. Two phenomena indicate reciprocal interactions between
the folded CPS I1 and ATCase domains of the
multienzyme polypeptide. First, even in the
presence of approx. 1 mM unlabelled carbamoyl
phosphate, endogenous carbamoyl phosphate is
favoured for the synthesis of radiolabelled carbamoyl aspartate when CPS I1 is in the liganded or
P state. That is, the binding of aspartate to
ATCase protects the active site of the liganded
form of CPS I1 from inhibition by its product,
and ensures that the carbamoyl phosphate is
channelled to the ATCase active site. Secondly,
when substrates (glutamine, ATP-Mg and
bicarbonate) bind to the U state of CPS 11, they
act allosterically to increase the affinity of the
ATCase for carbamoyl phosphate and for aspartate. When P-Rib-PP is present with low concentrations of ATP and magnesium ions, conditions
that would be met during active biosynthesis in
the cell, both effects operate, to protect the CPS
I1 from product-inhibition and to channel the
carbamoyl phosphate to a high-affinity site in
ATCase. Similar ‘reciprocal allosteric regulation’
of the active sites is seen in the increased affinity
of the synthase domains of CPS I1 for ammonia
when glutamine binds to the N-terminal glutamine amidotransferase domain of CAD; the
glutamine is the source of the ammonia, and the
allosteric response to its binding ensures that
ammonia is used efficiently by the synthase
domains. T h e mechanism ensures that carbamoyl phosphate is efficiently synthesized and is
dedicated to the second step of pyrimidine
biosynthesis.
Regulation by phosphorylation in vivo
T h e phosphorylated CAD that is obtained from
cell extracts exposed to radioactive ATP is
labelled at sites 1 and 2, suggesting that PKA is
the physiological kinase. However, substantial
labelling still takes place in the presence of the
PKA inhibitor peptide, and CAD has also been
shown to be a substrate for S6 (ribosomal protein) kinase, whose specificity is similar to that of
PKA (E. A. Carrey, unpublished work). Figure 1
compares the sites in CAD [9] with the typical
target sequence that is phosphorylated by PKA.
Similar sequences are found in the homologous
enzymes in yeast, slime mould and fruit fly
[lo-121, as shown in Figure 1. T h e CPS I1
activity in the Saccharomyces cerevisiae URA 2
protein is not regulated by phosphorylation [lo];
the labelled site corresponds not to site 1 but to
site 2 in mammalian CAD.
We have identified a short CAD sequence
resembling the site in p53 that is labelled by the
DNA-dependent protein kinase [ 131. CAD is
phosphorylated in vitro by this kinase (S. Jackson, personal communication). T h e putative site
(shown in Figure 2) is in the 20 kDa domain in
the N-terminal half of the CPS I1 region of CAD,
a domain to which no function has so far been
ascribed. T h e usual substrates for this kinase are
transcription factors, and phosphorylation is
usually favoured by the presence of doublestrand breaks in DNA, suggesting a role in DNA
repair. If CAD is a genuine target for the DNAdependent protein kinase, its role might be
rationalized as a supplier of nucleotides for DNA
repair, and the kinase might regulate the enzyme
by tethering it close to the area that requires the
nucleotides.
Regulation of purine biosynthesis by
phosphorylation
T h e symmetry of the pathways for the biosynthesis of purines and pyrimidines, and the
requirement for approximately equal amounts of
the nucleotides in the cell, might lead us to look
for more enzymes that are regulated by phosphorylation in response to external signals.
T h e first step dedicated to purine biosynthesis is catalysed by glutamine phosphoribosylpyrophosphate amidotransferase (amidophosphoribosyltransferase; EC 2.4.2.14), which is
known to be inhibited by AMP, the end-product
of the pathway. T h e suggestion that this enzyme
is phosphorylated by PKA [14] is not borne out
by the sequence [14,15], which contains no sites
corresponding to the well known consensus
sequence for this kinase. Intriguingly, the
sequence that corresponds to a binding site for
P-Rib-PP is overlapped by a consensus sequence
for the AMP-activated kinase (AMP-PK), as
shown in Figure 3 [16]. We have not demonstrated that glutamine P-Rib-PP amidotransferase is a substrate for AMP-PK, but it is an
attractive possibility for two reasons. First, the
closeness of the phosphorylation site to the position at which P-Rib-PP binds immediately suggests that the phosphorylated enzyme will be
inactive because it will be unable to bind its
substrate. In the crystallographic structure of the
similar enzyme from B. subtilis [ 191, the P-Rib-
I995
90 I
Biochemical Society Transactions
902
PP site can be seen overlapping a site at which
AMP binds, inhibiting the binding of P-Rib-PP.
Perhaps the phosphorylation by AMP-PK in
mammals replaces the direct binding of AMP.
Secondly, the AMP-PK is activated, as its name
suggests, by AMP, and phosphorylation inactivates the key enzymes in the biosynthesis of fatty
acids and cholesterol. A similar inactivation of
purine biosynthesis would be a logical response
to a rise in AMP concentrations, or to any of the
stressful conditions that are implicated in the
control of AMP-PK [20].
An inactivation of glutamine P-Rib-PP
amidotransferase alone might lead to an increase
in the concentration of P-Rib-PP in the cell, and
hence to increased production of pyrimidines
through activation of CPS 11. T h e enzyme that
synthesizes P-Rib-PP (P-Rib-PP synthase) has at
least three mammalian isoenzymes, containing
the P-Rib-PP binding consensus. An overlapping
AMP-PK site is not found in these sequences.
An inhibitory P-Rib-PP synthase-associated protein (PAP 39) has recently been described [22]
which has substantial sequence identity with PRib-PP synthase, b u t no enzyme activity. An
inserted non-homologous sequence in PAP 39
contains a consensus site for AMP-PK, and the
protein is indeed phosphorylated by the kinase
(S. Ishijima and M. Tatibana, personal communication). O u r proposed mechanism here
would involve the enhancement by phosphorylation of the ability of PAP 39 to inhibit the activity
of P-Rib-PP synthase, or perhaps the targeting of
PAP to the enzyme after phosphorylation. Inactivation of the synthesis of P-Rib-PP by AMP-PK
would have a fundamental role in shutting down
the de novo biosynthesis of purines and pyrimidines, and t h e recycling of nucleotides by phosphoribosyltransferases.
This work was supported by the Medical Research
Council and Accion Integrada 28B/94 (British Council-Spanish Ministry of Education and Science) with
Dr. V. Rubio.
1 Shoaf, W. T. and Jones, M. E. (1973) Biochemistry
12,4039-405 1
2 Carrey, E. A., Campbell, D. G. and Hardie, D. G.
(1985) EMBO J. 4, 3735-3742
Volume 23
3 Carrey, E. A. and Hardie, D. G. (1988) Eur. J.
Biochem. 171, 583-588
4 Carrey, E. A. (1989) Biochem. SOC. Trans. 17,
76 1-762
5 Rubio, V. (1994) in Carbon Dioxide Fixation and
Reduction in Biological and Model Systems,
Proceedings of the Royal Swedish Academy of
Sciences Nobel Symposium 1991 (Branden, C.4.
and Schneider, G., eds.), pp. 249-264, Oxford
University Press, Oxford
6 Liu, X., Guy, H. I. and Evans, D. R. (1994) J. Biol.
Chem. 269,27747-27755
7 Shaw, S. M. and Carrey, E. A. (1992) Eur. J.
Biochem. 207,957-965
8 Reference deleted
9 Simmer, J. P., Kelly, R. E., Rinker, A. G., Jr.,
Scully, J. L. and Evans, D. R. (1990) J. Biol. Chem.
265, 10395-10402
10 Denis-Duphil, M., Lecaer, J.-P., Hardie, D. G. and
Carrey, E. A. (1990) Eur. J. Biochem. 193, 581-587
1 1 Freund, J. N. and Jarry, B. P. (1987) J. Mol. Biol.
193, 1-13
12 Faure, M., Camonis, J. H. and Jacquet, M. (1989)
Eur. J. Biochem. 179, 345-358
13 Anderson, C. W. (1993) Trends Biochem. Sci. 18,
433-437
14 Iwahana, H., Oka, J., Mizusawa, N., Kudo, E., Ii, S.,
Yoshimoto, K., Holmes, E. W. and Itakura, M.
(1993) Biochem. Biophys. Res. Commun. 190,
192-200
15 Zhou, G., Dixon, J. E. and Zalkin, H. (1990) J. Biol.
Chem. 265,21152-21159
16 Dale, S., Wilson, W. A., Edelman, A. M. and
Hardie, D. G. (1995) FEBS Lett. 361, 191-195
17 Suttle, D. P., Bugg, B. Y., Winkler, J. K. and
Kanalas, J. J. (1988) Proc. Natl. Acad. Sci. U.S.A.
85, 1754- 1758
18 Andersen, P. S., Smith, J. M. and Mygind, B.
(1992) Eur. J. Biochem. 204,51-56
19 Smith, J. L., Zaluzec, E. J., Wery, J.-P., Niu, L.,
Switzer, R. L., Zalkin, H. and Satow, Y. (1994)
Science 264, 1427-1433
20 Corton, J. M., Gillespie, J. G. and Hardie, D. G.
(1994) Curr. Biol. 4, 315-324
21 Taira, M., Iizasa, T., Shimada, H., Kudoh, J.,
Shimizu, N. and Tatibana, M. (1990) J. Biol.
Chem. 265, 16491-16497
22 Kita, K., Ishizuka, T., Ishijima, S., Sonoda, T. and
Tatibana, M. (1994) J. Biol. Chem. 269,
8334-8340
Received 16 June 1995