Download The phosphorylation of proteins: a major mechanism for biological

The phosphorylation of proteins: a major mechanism for
biological regulation
EDWIN G. KREBS
Howard Hughes Medical Institute and the Department of Pharmacology, University of Washington School of Medicine,
Seattle, WA 981 95,U. S. A .
I am very. appreciative
of the honour that has been
._
bestowed upon me by the Biochemical Society in inviting
me t o give the fourteenth Hopkins Memorial Lecture. When
1 took mv first course in biochemistrv at the Universitv of
Illinois i;l 1940, it was fashionabie for professors to
acquaint students with the names of the great pioneers in
the field. so at that time 1 became aware of some of
Hopkins' contributions t o nutrition, particularly with
respect t o the vitamins. Later, however, I came to realize
the breadth of his work and thinking and the influence that
he exerted on the development of many other aspects of
biochemistry.
My lecture this evening will be divided into several
sections. First. I will present a brief historical account of
the developnient of our present knowledge of protein
phosphorylation. After that I would like to discuss the
protein kinases, which constitute key elements of the
protein phosphorylation-dephosphorylation machinery.
Then, briefly, I will consider what is known about the
effect of phosphorylation on the structure and function
of proteins. This will lead into a section on the biological
significance of protein phosphorylation, which is the part
most closely related to the title of the lecture. Finally, I
will close by attcnipting to forecast future developments
in tlie field.
Historv of protein phosphorvlation
A textbook of biochemistry published in 1947 (Hawk
ct a/.. 1947) had this to say about phosphoproteins, 'The
phosphorus in these proteins is bound in ester linkage to
the hydroxyamino acids, serine and threonine . . . .
Examples of pliosplioproteins include casein from milk,
ovovitellin from egg yolk. and other proteins associated
with the feeding of the young.' The nutritional function of
phosphoproteins is still considered to be important, of
course, but these proteins are now known to have
additional roles. One of these relates to the formation of
phosphoproteins as transient intermediates involved i n
the mechanism of action of certain enzymes and transport
proteins: but the most noteworthy function of phosphoproteins is their role in dynamic regulatory processes within
the cell. It is with the last aspect that this lecture will be
conce rned.
A realization that the phosphorylation and dephosphorylation of cellular proteins constitutes a major
regulatory process was relatively slow in developing. Unlike
the sudden and universal acceptance of allosteric control,
which followed tlic exciting revelations by Monod and his
co-workers (Monod etal., 1963) in the mid-l960s, a general
appreciation of tlie significance of phosphorylationdephosphorylation occurred only after several decades of
work carried out in a number of different laboratories. Let
us go back to early studies on glycogen metabolism, which
served to set the stage for the development of this field.
I n tlie early 1940s the Coris' and Arda Green showed
that the enzyme phosphorylase exists in two forms. which
they designated as phosphorylase b and phosphorylase a
(Cori & Green, 1943; Cori & Cori, 1945). Phosphorylase b
Vol. 13
Fourteenth sir Frederick Gowland Hopkins
Lecture
Delivered on 20 December 1984 at St. Georgets
Hospital Medical School, London
DOCTOR EDWIN G. KREBS
required high concentrations of 5'-AMP for activity and
was thought to be a physiologically inactive form, whereas
phosphorylase a was highly active in the absence of
cofactors, and was considered t o be the active form of the
enzyme. The Coris' showed that the two forms of phosphorylase are interconvertible in the intact cell, and they
postulated that the interconversion reactions constituted a
physiologically significant regulatory mechanism. It was
difficult for them t o be very precise as t o how the
regulatory system would work, because at that time the
existence of a separate biosynthetic pathway for glycogen
formation was unknown, and it was believed that phosphorylase catalysed glycogen biosynthesis as well as
glycogen degradation. Certi Cori died during tlie 1950s,
but Carl Cori passed away only this past October. As my
former mentor he had a major influence on my decision to
follow a career in biochemistry rather than internal
medicine, and at this time I would like to pay tribute to
him. Cori, like Hopkins, was one of the giants of 20th
century biochemistry. His active research career overlapped
that of Hopkins and those of all of us here today.
An understanding of the mechanism involved in the
81 3
8 I4
BIOCHEMICAL SOCIETY TRANSACTIONS
interconversion reactions of phosphorylase came approximately 10 years after their discovery, when, in the 1950s,
independent work on liver phosphorylase by Earl
Sutherland (also a Cori student) and muscle phosphorylase
by E. H . Fischer and myself (Fischer & Krebs, 1955;
Sutherland & Wosilait, 1955) showed that enzymecatalyscd protein phosphorylation-dephosphorylation
reactions were involved. The enzyme catalysing the
phosphorylation reaction was named phosphorylase kinase
and that catalysing the dephosphorylation step was called
phosphorylase phosphatase (Fig. 1). In 1959, evidence
was obtained that phosphorylase kinase, like phosphorylase
itself, is also regulated by phosphorylation-dephosphorylation and that cyclic AMP is somehow involved in the
conversion of the non-activated form of the kinase t o a
more active form (Krebs et al., 1959). In the early 1960s,
Joseph Larner and his collaborators showed that the then
recently discovered enzyme, glucogen synthase, constituted
a third enzyme that undergoes reversible phosphorylation-dephosphorylation (Friedman & Larner, 1963). To this
point the three known examples of enzyme regulation by
phosphorylation-dephosphorylation, phosphorylase, phosphorylase kinase and the synthase, all involved glucogen
metabolism, and it was considered by some that perhaps
this was an esoteric type of control system restricted t o
one limited area of carbohydrate metabolism.
Preliminary reports involving possible examples of
phosphorylatable enzymes in pathways other than those
of glycogen metabolism appeared in the mid-1960s (Rizak,
1964; Medicino et al., 1966), but it was not until 1969,
with the finding from Lester Reed’s laboratory that
pyruvate dehydrogenase is regulated by phosphorylationdephosphorylation (Linn et al., 1 9 6 9 ~b. ) , that the field
broke out of the more restricted area. This event more or
less coincided with the finding that the enzyme that
catalyses the phosphorylation and activation of phosphorylase kinase is a cyclic AMP-dependent protein kinase
having a relatively broad specificity, thus making it a likely
mediator for the many functions of cyclic AMP (Walsh et
al., 1968; Kuo & Greengard, 1969). A period of rapid
growth in the number of reports of enzymes undergoing
phosphorylation ensued, and by 1984 at least 50 enzymes
in this category were known. It should be emphasized, however, that some reports of enzymes ‘regulated’ by phospliorylation are based solely on work in vitro and niany of
them do not meet the criteria that need to be satisfied
before it is established that a given enzyme phosphorylation
is physiologically significant (Krebs et al., 1985). In
addition, the protein kinases themselves, essentially all of
which undergo autophosphorylation reactions, are included
in this number. and in most instances there is n o known
Phosphorylase kinase
8
.
Mq2’ A T P
Phorphorylare phosphatase
Phosphorylase a
lacttvel
physiological significance of these phosphorylations. Nonetheless, perhaps at least 20-25 enzymes can be listed tor
which good evidence exists that their activities are regulated
physiologically by this process.
If one were t o tabulate the number of non-enzymic proteins known t o undergo dynamic pliosphorylation-dephosphorylation, the totals would be far greater than for
the number of enzymes. Many of these proteins are known
only by their M, values and may, in fact, turn out to be
enzymes. Others, however, are well-defined non-enzymic
proteins having a variety of biological functions. They arc
found in virtually all parts of the cell.
The protein kinascs
If one considers the general process of protein phosphorylation-dephospliorylation as exemplified by the
activation-inactivation cycle of phosphorylase (Fig. 1)
it is clear that in order for protein phosphorylation to
serve in a regulatory capacity, factors must be involved in
regulating the forward and reverse reactions so that the
proportion of the protein in its phosphorylated form can
be varied. Regulation is exerted at the kinase level and at
the phosphatase level, but in general we know more about
the protein kinases and their control than we d o about the
phosphatases. This picture may change, however, because
a number of laboratories, most notably that of Philip
Cohen in Dundee, are bringing the latter set of enzymes t o
the fore. Today, however, I would like t o restrict my
discussion to the protein kinascs and consider sonic of
their properties.
Classification of protein kinases. How many different
protein kinases are there and how are they classified‘?
With respect t o the latter question. the Nomenclature
Committee of the International Union of Biochemistry
has been struggling with the problem. and. indeed, has
appointed a study panel t o make recommendations in this
area. Because a given protein kinase usually catalyses the
phosphorylation of a number of different proteins. as
well as the fact that more often than not one protein
serves as a substrate for more than one kinase. classical
conventions for naming these enzymes are not applicable.
As we will see in a moment, the protein kinases are usually
classified and named o n the basis of what regulates their
activities. As to how many protein kinases exist, I think
it is much too early t o tell, but some idea as t o their number
and diversity can be seen in Table 1 and Table 2.
Perhaps the best-characterized protein kinases are the
family of cyclic AMP-dependent kinases. These exist in
several isoenzymic forms designated as type I. type II and
type 11’. These enzymes share a common catalytic subunit
and all of them exhibit the same specificity. There is also a
cyclic GMP-dependent protein kinase. No isoenzymic forms
for this kinase have been described. A very important group
is made up of the Ca2+ (calmodu1in)-dependent protein
kinases. These are distinct enzymes having different
specificities. Of the enzymes in this category listed here, I
have already mentioned phosphorylase kinase, an enzyme
that is regulated not on1 by pliosphorylation~depliosphorylation but also by Cax ions. Then there are two types
of myosin light chain kinase (MLCK), an enzyme first
discovered by Pircs et al. (1974). Finally, there is a sonicwhat less specific Ca2+ (calmodu1in)-dependent protein
kinase, which is designated as the multifunctional kinasc
by many investigators. This latter enzyme was probably
first detected by Schulnian & Grcengard (1978) in brain
and other tissues and by Payne & Soderling (1980) as a
glycogen synthase kinase. Another Ca2+ ion-dependent
protein serine kinase, but one that does not utilize
calmodulin, is protein kinase C, discovered by Niskizuka
and co-workers (Inoue et al., 1977; Takai et al., 1977) and
1985
FOURTEENTH HOPKINS MEMORIAL LECTURE
815
Table 1 , Protein seririe (thrtwnine) kiriases
and insulin-like growth factor (IGF-I) receptors. In addition t o the identified viral and cellular oncogen-encoded
protein tyrosine kinases and the receptor kinases, investigators have noted the presence of protein tyrosine kinases
in various animal tissues such as spleen and liver (Swarup
etal., 1983; Wong & Goldber. 1983).
Structural homologies betweeti proteiri kinases. AS a
collaborative project involving several laboratories at the
University of Washington, including those of Kenneth A.
Walsh, Koiti Titani, Edniond H. Fischer and my own, the
primary structures of a group of protein serine kinases are
being examined. Today I would like t o summarize the
results that have been obtained. The protein kinases that we
have been interested in, each o f which has been sequenced
at least in part, are shown in Table 3.
They include type I and type I1 cyclic AMP-dependent
protein kinases, cyclic GMP-dependent protein kinase,
phosphorylase kinase and skeletal muscle MLCK. The
cyclic AMP-dependent protein kinases are made up o f
distinct cyclic AMP-binding regulatory and catalytic subunits, and have the quaternary structure R 2 C 2 . The Mr
values of the several isoenzymic forms of the regulatory
subunit (R) are all about 4 5 0 0 0 ; the catalytic subunit,
which is identical for all types of cyclic AMP-dependent
protein kinases, has an M, of 40000. When activated by
cyclic AMP the inactive holoenzyme form undergoes dissociation t o yield the active catalytic subunit. For the
closely related cyclic GMP-dependent protein kinase the
regulatory and catalytic domains are on the same polypeptide chain, and this enzyme is activated simply by
binding cyclic GMP. These relationships for the cyclic
nucleotide-dependent kinases are shown in Fig. 2 .
In this model the catalytic subunits of the cyclic AMPdependent protein kinase are shaded, as are the catalytic
domains of the cyclic GMP-dependent protein kinase
chains. Regulatory subunits or regulatory domains are
unshaded. On binding of cyclic nucleotides dissociation and
activation of the inactive holoen~yme occurs with the
cyclic AMP-dependent kinase, but with the cyclic GMPdependent enzyme actication occurs without dissociation.
In each case 2 mol of cyclic nucleotide are bound per mol
of regulatory subunit or regulatory domain. Corbin ct a / .
(1978) were the first t o show the correct stoichiometry for
cyclic nucleotide binding to these enzymes.
Phosphorylase kinase (Table 2 ) is a large molecule made
up of four different types of subunits. This enzyme has an
a4/34y464structure. The y-subunit is the catalytic subunit
of this kinase; its M, value is 40 000. the same as that of the
catalytic subunit of the cyclic AMP-dependent protein
kinase. The 6-subunit is calmodulin. as shown by Cohen e l
al. (1978). The a- and /3-subunits are involved in the activation of the enzyme by phosphorylation-dephosphorylation. In addition t o activation as a result of the binding of
Regulatory agent(s)
Cyclic AMP
Cyclic GMP
Ca2+(calmodulin)
Diacylglycerol (Ca2+,
phospholipid)
Double-stranded RNA
Haemin
AcctylCoA. N A D H , pyruvate,
ADP
Unknown or nonexistent
Specific enzymes
Type I , type 11, and Type 11'
cyclic AMP-dependent
protein kinases
Cyclic GMP-dependent
protein kinase
Phosphorylase kinasc
Skeletal muscle MLCK
Smooth muscle MLCK
Multifunctional Ca2+
(ca1modulin)-dependen t
protein kinase
Protein kinase C
eIF2 kinase 1
elF2 kinase 2
Pyruvic dehydrogenase
kinase
Casein kinase I
Casein kinase 11
Glycogen synthase kinases
3 and4
Growth-associated
protein kinase
Others
very much in the news nowadays (for a review, see
Nishizuka, 1983). The principle effector controlling the
activity of protein kinase C is diacylglycerol. Phorbol esters
have been shown t o mimic diacylglycerol in this respect
(Castagna ct al., 1982). There are other protein serine
kinases regulated by various effectors including doublestranded RNA, haeniin, and metabolites involved in the
pyruvate dehydrogenase reaction. Finally, there is a fairly
large group of protein serine kinases for which no
regulatory agents are known. These so-called 'independent'
protein kinases include those listed in Table 1 as well as a
number of others. It is not unlikely that regulators for
iiiany of the enzymes in this group will eventually be discovered, o r it may turn out that these enzymes are
regulated by other kinases.
I n 1979 it was recognized that there are protein kinases
(Table 2 ) that catalyse the phosphorylation of tyrosine
residues i n proteins (for a review, see Sefton & Hunter,
1984). The first of thcm to be described was pp6OSm, the
protein product of the Rous sarcoma oncogene, src. Later i t
was found that at least six other viral oncogenes also
encode protein tyrosine kinases, as d o their cellular
counterparts. All of the enzymes in this group are struc~ . major group o f
turally related t o the ~ ~ 6 0A" second
protein tyrosine kinases consist of a set of hormones o r
growth factor receptors, four of which are currently
known; the enzynies in this group are the insulin, epidermal
growth factor (EGF), platelet-derived growth factor (PDGF)
Table 3. Propcrties of protein kinases
Quaternary
structure
Mechanism of
activation by
effectors
K,C,
Dissociation
Table 2. Protein tyrosine kiriusrs
Regulatory agent(s)
Specific enzymes
Enzyme
Mr
~~
Insulin
FG I:
PDGI:
IGF-I
Unknown or nonexistent
Vol. 13
Insulin receptor
EGF receptor
PDGF receptor
IGF-I-recep tor
pp6V" and related
oncogeneencoded kinases
Tissue kinases
Cyclic AMP-dependent
protein kinase
Cyclic GMP-dependent
protein kinase
Phosphorylase
kinase
MLCK
1.7 x l o s
1.6 x l o 5 Jlomodimer
Binding
1.3 x 10'
Binding
o(qp4y.164
7 X lo4 Monomer
Binding
816
BIOCHEMICAL SOCIETY TRANSACTIONS
Cylic A M P
h
(Inactive)
.4
Cyclic GMP-
c-
Fig. 2 . Activation of the qrclic AMP- and cyclic CMPdrpcndent protcin kinases
Shaded portions represent catalytic subunits or domains.
Non-shaded portions represent regulatory subunits or
dorn ains.
Ca2+ ions t o the endogenous calmodulin, this enzyme can
bc further activated through the binding of additional
Ca*+--calmodulin complex. No dissociation of phosphorylase kinase occurs when it is activated. Skeletal muscle
MLCK is a simple monomeric enzyme having an M, of
about 7 0 0 0 0 , which is activated as a result of binding the
Ca'+-calmodulin complex.
Let us look first at the cyclic nucleotide-binding
domains of the cyclic AMP-dependent and cyclic GMPdependent protein kinases, which I will hereafter refer t o
simply as the cA and CC kinases respectively (Fig. 3). In
Fig. 3 bovine cG kinase has arbitrarily been chosen as the
basis for sequence comparison with the other proteins
(Takio et aL, 1984). This enzyme is depicted diagrammatically as a linear bar. The two cyclic nucleotide-binding
domains are indicated by the shaded sections labelled A and
B. The bars under the CC kinase represent the regulatory
85
cGK I
04
cAKRI
I
V
A
8
10 0
13 8
/
/
/
/
/
subunits of bovine types I and type I1 cA kinase and the
bars under them the cyclic AMP-binding protein from
Escheridria coli known as CAP, i.e. the catabolite gene
activator protein. The numbers in Fig. 3 refer t o alignment
scores, which serve as a useful index for indicating amino
acid sequence homology (Dayhoff, 1983; Reimann et al.,
1984). Without going into the derivation of these scores,
suffice it t o say that an alignment score of 2 corresponds
t o a probability o f 0.023 that random chance would lead
t o a higher score. A score of 3 corresponds t o a probability o f 0.0014. Higher scores are clearly indicative of
very significant homology. What has been done in Fig. 3
is t o use the second (B) cyclic GMP-binding donlain of the
CG kinase as a basis for comparison, first with the other
cyclic GMP-binding domain in the same polypeptide
chain (A) and then with the cyclic AMP-binding sites in
the other proteins. A score of 8.5 for domain A as compared with B indicates strong homology between the two
domains in the cG kinase. Aligment scores of 13.8 and
10.2 show an even higher degree of homology of this
segment with the corresponding segments of type I and
type I1 R. Lesser, but still strong, homology of this domain
with the first cyclic AMP-binding sites in R, and R!, is also
apparent. Interestingly, the cyclic nucleotide-binding sites
in the mammalian protein kinases are homologous t o the
cyclic AMP-binding domain in C A P (Weber et al., 1982).
This homology is greater when the B domain of CG kinase
is used as a basis for calculating the scores rather than the
A domain.
Turning now t o the relationship that exists between the
structures of the catalytic subunits or catalytic domains of
protein kinases, we find that all of those that have been
sequenced t o date are homologous. In Fig. 4 the catalytic
subunit of the cA kinase is used as the basis for calculating
alignment sources. The subunit has been divided more or
less arbitrarily into three segments, each of which is compared with segments of other proteins lined up beneath
it. Alignment scores of 17.3, 22.9 and 13.2 indicate very
strong homology with the corresponding segments of the
catalytic domain of the c C kinase (Takio et al., 1984).
The catalytic or y-subunit of rabbit muscle phosphorylase
kinase, sequenced by Erwin Reimann while on sabbatical
leave in Kenneth Walsh's laboratory, is homologous t o the
catalytic subunit throughout part of its structure but not
in its C-terminal portion (Reimann et al., 1984). The
catalytic domian of rabbit skeletal muscle MLCK, which
resides in the C-terminal half of that molecule, is
homologous to the catalytic subunit (K. Takio, D. K .
Z
c A K catalytic subunit
cAK RI' 1
05
10 2
81
V
/
,
Z
/
4
1
17 3
cGK
62
I
W
13 2
229
/
i
Z
l
1 1
CAP
20 3
48
06
Phosphorylase kinase catalytic subunll ~
1
1
1
30'
1
-09'
CAP
MLCK
v
I
nucleotide-binding domains of' various proteins
The second cyclic GMP-binding domain (B) of the cyclic
GMP-dependent protein kinase (cGK) is compared with the
first binding domain (A) of the same enzyme and with cyclic
AMP-binding domains in other proteins. RI and RII refer t o
type 1 and type 2 regulatory subunit of the cyclic AMPdependent protein kinase, respectively. CAP refers t o the
catabolite gene activator protein. The numbers are Dayhoff
alignment scores. *Alignment scores relevant t o segment A
of cG kinase.
P P E ~ " 'r
I
p
64
V
0 3
13 0
66
Fig. 3 . A mi n o acid sequence homologies between cyclic
94
/
/
21
2
Fig. 4.Amino acid sequence homologics hctwivn catal.vtic
domains of'various protein kinases
cAK, catalytic subunit of cyclic AMP-dependent protein
kinase; cGK, cyclic GMP-dependent protein kinase. MLCK,
skeletal muscle myosin light chain kinase. The numbers are
Dayhoff aligment scores. Segments of cAK are compared
with segments of other proteins immediately under them.
1985
FOURTEENTlI HOPKINS MEMORIAL LECTURE
817
Blumenthal, A. M., Edelman, K. A. Walsh, E. G. Krebs,
& K . Titani, unpublished work). Again, however, this
homology is lost in the C-terminal portion. Finally,
amino acid sequence homology exists between the catalytic
subunit of tlie cA kinase and pp6OSm (Barker & Dayoff,
1982). Not shown are comparisons with other oncogeneencoded protein tyrosine kinases or the EGF receptor.
However, inasmuch as all of the protein tyrosine kinases
whose sequences are known show homology with pp6Oflcc,
they are all related to the catalytic subunit (reviewed in
Sefton & Hunter, 1984). It can be anticipated that tlie
insulin receptor will also fall into this category.
Phosphorylase kinase and MLCK are both regulated by
CaZ+ and calmodulin, whereas the cA and cG kinases are
regulated by cyclic nucleotides. It was hoped at first that
a unique sequence, common to the first pair of en7ynies
but lacking in the second, might stand out to help in the
identification o f the calmodulin-binding site on these
enLynies. At first glance this did not occur, although a
very low degree of homology (alignment of 2.4) was
found between the C-terminal segments of the y-subunit
of phosphorylase kinase and MLCK. Other approaches
were successful, however, in defining and localizing a
calmodulin-binding site on the latter enzyme.
If skeletal muscle MLCK is partially degraded by using
proteases one can obtain catalytically active fragments, as
has also been demonstrated by Mayr & Heilmayer (1983).
The size and location in the molecule of some of the fragments generated by chyniotrypsin and trypsin are shown
in Fig. 5. With chymotrypsin a fragment is obtained that is
calniodulin-independent and has a molecular mass of about
35 000 daltons. This fragment contains the catalytic domain (shaded), i.e. the protein homologous t o the catalytic
subunit of the cA kinase, as well as having small portions
which show n o honiology with the catalytic subunit of the
cyclic AMP-dependent protein kinase. With trypsin one
generates catalytically active calmodulin-dependent fragments having molecular masses of 40000 and 60000
daltons. These fragments extend almost all the way t o the
C-terminus of MLCK and also have regions beyond the
catalytic domain that extend toward the N-terminus.
Based on these findings. it was tempting t o postulate that
the calmodulin-binding site in MLCK resides somewhere
in the non-catalytic portions of the tryptic fragments that
are missing in the chymotryptic fragment, i.e. either at
their N-terminal or C-terminal ends. The latter turned out
to be the case.
Localization and definition of the calniodulin-binding
site was finally determined by testing the ability of CNBr
C 3 51K
T40K
T-60 K [
I
I
-&
"
I
A/ -V
M-13
0
Fig. 5. Frugmcnts derived from skeletal muscle M L C K
Upper bar represents intact MLCK in which the shaded
portion is that part of the molecule homologous with the
catalytic subunit of the cyclic AMP-dependent protein
kinase. C-35K, an M,-35 000 fragment generated by using
ehymotrypsin; T-40K, a n M,40 000 fragment generated by
using trypsin; C-60K, an M,-60 000-fragment generated by
using trypsin; M-13, the C-terminal peptide from MLCK
generated by using cyanogen bromide.
Vol. 13
fragments of MLCK to inhibit MLCK activity under conditions in which added calmoddin is limiting. When a total
CNBr digest of non-proteolysed MLCK was examined it was
found to be inhibitory at low molar concentrations, halfmaximal inhibition inhibition occurring at a molarity of
4-5 nM, based o n the original MLCK concentration. A
digest of a mixture of the 40 and 60 kilodalton tryptic
fragments was similarly inhibitory. By contrast, the CNBr
digest of the calmodulin-independent 35 kilodalton fragment showed a marked reduction in its ability t o inhibit
activity, and the slight inhibition that was seen could have
been due to a low level of contamination of the calmodulinindependent fragment with dependent fragment(s). When
CNBr fragments from MLCK were exaniined individually,
only one of them, fragment 13, consisting of the C-terminal
27 amino acids of the enLyme, was found to be inhibitory.
Inhibition by this fragment, M-13, was only slightly less
than that of the digest of intact MLCK. The adjacent
fragment, M-12, or a mixture of M-l 1 and M-12, was not
inhibitory. Donald Blumenthal from any laboratory has
synthesized a peptide corresponding to M-13 and has determined that it is even more potent in inhibiting MLCK
activity than M-13 generated by CNBr digestion. The lower
inhibitory potency of M-13 relative to the synthetic peptide
is probably due to modification of sensitive amino acid side
chains in the peptide by the conditions used for CNBr
digestion. Drs Blumenthal, Edelman, Takio and others
involved in this work will be reporting on thc structure and
properties of the calmodulin-binding site at tlie Biophysical
Society Meeting in Baltimore in February (Blumenthal et
al., 1985).
Effect of phosphorylation on protein structure
For protein phosphorylation to function in a regulatory
capacity it is clear that the introduction or removal of covalently bound phosphate must cause structural changes in
protein substrates that in turn lead t o functional changes.
How much d o we really know about the chemical nature of
this process? The demonstration of functional changes
accompanying phosphorylation has been the sine qua non
among investigators for acceptance that a given protein
phosphorylation event is of regulatory significance, and
such changes are readily seen, but in only one instance is
much known about structural changes that underlie
functional change. This exception is for the original
phosphorylation-dephosphorylation enzyme, namely glycogen phosphorylase.
As a result of X-ray crystallography studies o n the
structure of phosphorylase b , carried out by Louise
Johnson at Oxford University, and similar studies on the
structure of phosphorylase a by Robert Fletterick and Neil
Madsen in Canada, and their respective collaborators, we
now have one example in which information is available
with respect t o the actual structural change brought about
by phosphorylation. In the conversion of phosphorylase b
to phosphorylase a a single phosphate group is introduced
on Ser-14, very close t o the N-terminus of each
phosphorylase subunit, which contains 841 residues.
Phosphorylation of the protein results in the formation
of three new salt bridges, one of which is between the
phosphate group and Arg-69 of the same subunit. Two of
the salt bridges are intersubunit bridges. Fletterick &
Madsen have calculated that the binding energy of these
salt bridges is translated into a highly significant decrease in
the allosteric constant for phosphorylase a as compared
with phosphorylase b . They have spoken of the phosphate
group as a 'covalently bound allosteric ligand', which, like
5'-AMP and other positive effectors, promotes a T to R
conformational change (Sprang & Fletterick, 1980; N. B.
Madsen, personal communication).
818
BIOCHEMICAL SOCIETY TRANSACTIONS
The majority of proteins that are regulated by
phosphorylation-dephosphorylation are oligomeric proteins, which, like phosphorylase. are known t o be regulated
by the non-covalent binding o f various ligands. It is not
unlikely that the findings with respect to the effect of
phosphorylation on phosphorylase can be extended to
other systems regulated in this manner.
The biological significance of protein phosphorylation
Given the tremendous potential for the regulation of
protein function provided by allosteric control involving
non-covalently bound ligands, why is another regulatory
mechanism such as phosphorylation-dephosphorylation
needed? Investigators puzzling over this question have
invoked a number of possible reasons. For example, some
think of the various forms of regulation on a time scale and
consider allosteric control as being rapid and transient,
whereas regulation due t o protein phosphorylation is
viewed as being slower and niore enduring. Disregarding
these temporal considerations, I like t o think of the two
control systems as having evolved to serve two different
situations, allosteric control being primarily designed to
handle signals arising from within the cell, whereas control
by phosphorylation is geared to the handling of signals
arising from outside the cell. According t o this general
concept, messages carried by hormones and other extracellular stiniuli would be transmitted as a result of the
phosphorylation o f key functional proteins ultimately
responsible for the cellular functions known t o be affected
by the various outside agents (Fig. 6). The idea that protein
phosphorylation-dephosphorylation is largely concerned
with the handling of external signals is in keeping with the
fact that protein phosphorylation is much more abundant
in eukaryotic cells than it is in prokaryotic cells, suggesting
that the process evolved to meet the needs of more complex organisms in which extracellular signals are especially
important. By contrast, allosteric control mechanisms
are fully developed in prokaryotes.
Let us look at some of the spccit’ic systems in which
protein phosphorylation serves such a function:
Systems in which cyclic AMP is involved. The prototypical model for the regulation of cellular activity by
hormones is the cyclic AMP system (Fig. 7). Here i t is
known that one group o f hormones, including, for example.
adrenocorticotropin, glucagon, P-adrenergic catecholamines
and the gonadotropins, acting through their specific
receptors, stimulate adenylate cyclase, which leads t o
elevated cyclic AMP levels. A second group of horinones,
exemplified by adenoisine and a2-adrenergic agents cause
inhibition of adenylate cyclase leading t o decreased cyclic
AMP levels. The changes in cyclic AMP concentration in
turn alter the activity of the cyclic AMP-dependent
protein kinase, thus far the only receptor protein defined
for cyclic AMP in eukaryotic cells. Changes in the activity
of the cyclic AMP-dependent kinase lead t o an increase or
a decrease in the phosphorylation state of key cellular
proteins. The last change then promotes some physiological
Hormone
I
IIIIITrIIu:;ln~I1I:I::::::::.:::::::.::i
0
Cyclic A M P
Cycl~
AMP-deprndsnt
prolein kinas?
I
0
Phosphorylailon
o f key p r o i e i r ~ s
I
~
+ Physiological
evenis
Fig. I . General scheme for the rcgulation o j ’ cellular
functions by hormones that cause an increase or a dccrease
in cvclic A M P levels within the cell
ATP
Cyclic A M P
4j
i
Cyc18~AMP-dependent profejn kinabe
n
/-----\
Altered cellular functions
Glycogen
1
Glucose 1 phosphate
I
Fig. 6 . llormones and extracellular stimuli transmit their
signals through changes in the state of’ phosphorylation of’ Fig. 8 . Regulation o j glycogenol.vsis
muscle
cellular proteins
by adrenalitw in
1985
FOURTEENTH HOPKINS MEMORlAL LECTURE
event. A specific example of a cyclic AMP-niediated process involving protein phosphorylation is the classical
glycogcnolytic cascade as it occurs in skeletal muscle
stiniulated by adrenaline (Fig. 8). The particular target for
tlie cyclic AMP- dependent protein kinase in the regulation
of glycogenolysis is phosphorylase kinasc, which, as
mentioned earlier, is activated by phosphorylation.
Activated phosphorylase kinasc catalyses the conversion of
pliospliorylasc h t o phosphorylase a and this event triggers
glycogenolysis providing glucose 1-phosphate for glycolysis
and ATP production.
Sj~stcnisin w/ric/r Cu2+ions arid diacjl/g!jwrol scwc as
scwrid rncsserrgers of honiional action. A second major
system in which hormonal signals generate cellular
1-csponscs as a result of changes in the state of phospliorylation of key proteins involves second messengers that are
fornied as a result of stimulation of the phosphatidylinositol (Ptdlns) cycle (Fig. 9). According to current concepts,
many different extracellular agents. including, for example,
niuscarinic cholincrgic hormones, a-adrenergic Iiorniones,
5-liydroxytryptaminc, histamine. and angiotensin trigger
tlic breakdown of pliosphatidylinositol bisphosphatc and
acceleration o f tlic Ptdlns cycle. This leads to elevated
levels of two intracellular second messengers, inositol
triphosphate and diacylglycerol. The first of. these niesscngcrs is responsible for the mobilization of free intracellular C;i2+ ions and the subsequent activation o f a set of
at least four Ca2+ (caliiiodu1in)-dependent protein kinases
and one phosphoprotein phosphatase. Diacylglyccrol
accumulation causes activation of protein kinasc C. These
cnzynies then niodulate protein phosphorylation i n a
nicaningful nianner so as to produce tlie physiological
effect of tlic hormone in question (for reviews. see
Berridge. 1984: Nishizuka, 1984).
Ssstcwrs iri which Ca" ion niobilization as a result oj
electrical cxcitatioii resiiits in protein phosphorylation.
Horniones arc not the only extracellular stimuli that give
rise to enhanced intracellular protein phosphorylation.
This plicnonienon is also seen in connection with the
electrical stimulation of cells. Consider, for example, the
protein phosphorylation events that occur in skeletal
819
muscle stimulated to contract as a result of the nerve
impulse (Fig. 10). When muscle is stimulated electrically or
by nerve inipulse Ca2+ ions are released from the sarcoplasinic reticulum and this leads t o contraction asa result o f t h e
interaction of Ca2+ with the troponin- tropoinysin complex
on thin filaments. The rise in muscle Ca2+ also activates two
Ca2+ (ca1iiiodulin)-dcpendciit protein kinases. One of these
enLynies. phosphorylasc kinasc, promotes glycogenolysis
through tlie familiar pathway involving phosphorylase
activation. Glycogenolysis, which ensues, serves to supply
ATP that is needed for muscular work. The other protein
kinase activated by Ca2+ is MLCK. Activation of this
enzyme leads to phosphorylation of myosin light chains,
which exerts a regulatory influence on actin -myosin
interaction (Stull et ul., 1982).
Mechanisnis oj' action of horniones whose receptors arc
protein tyrosine kirmcs. The final system that I would like
t o consider is the mechanism of intracellular transmission
of signals generated by those hormones whose receptors are
protein tyrosine kinases. When it became known that the
EGF, PDGF, insulin and IGF-I receptors are all protein
tyrosine kinases (reviewed in Hunter & Cooper, 1985), it
was generally assumed that the mechanism of action of this
set of horniones would be elucidated quickly. The obvious
inference was that appropriate cellular proteins, which
could account for the plciotropic actions of these
hormones, would soon be identified as substrates for the
receptor kinases and that at last the long sought (particularly in reference to insulin) niechanisni of action would
be solved. Several years have gone by, however, and
investigators have not as yet identified any universally
accepted substrates for the receptor protein tyrosine
kinases. For this reason some investigators are now having
doubts as t o whether or not the protein kinase activity of
the receptors is really important. My own bias remains that
it is very important. A catalytic function o f a protein is not
a trivial property.
In work being carried out in my own laboratory, we
have been guided by the hypothesis that protein tyrosine
Extracellular
stimulus
1
Phosvhorvlase
Muscle i h t n
MLCK
Phosphory lase
I
I
Fig. 9. Mechuriisni o j actiori of horniories o r othc-r c x t r u ccllitlor s t i m u l i that u c t through sccond rncsscwgcJrs of
pro tciri pli o s p h o rjila t io ti getwra t cd by t h e p h o s p h t i d y lit1 o sitol (Ptdlri s ) i:v cle
Vol. 13
9-
act in-myosin
M~~,~,,,
interaction
G lycoqenulysis
Fig. 10. Role of protcin pliosphorvlatiori iri mediating t h c
rcsporisc to ric'rvc s t i m u l i iri skelcJtul rnusclc
820
BIOCHEMICAL SOCIETY TRANSACTIONS
Insulin
I
0
t
I /
:,Ynraqsdne
I
1
I
I
D ~ v r r s r ,m ~ t ~ ~ b eof fle~
c tcs
Fig. 1 1. Ilortnoncs whose r e c e p t o r s are p r o t e i n t j frosine
kinases a c t through the regulation of p r o t e i n sarine a n d
tlireonine p h o s p l i o r y l a t i o n
phosphorylation feeds into or regulates the much more
abundant serine (threonine) phosphorylation that occurs
in cells. This is illustrated using the insulin receptor as an
example in Fig. 11. One mechanism by which this might
occur would be a ‘mixed cascade’ systeni in which a
specific protein serine kinase or a phosphatase serves as a
substrate for a protein tyrosine kinase. Activation of the
kinase (or phosphatase) as a result of its being phosphorylated on a tyrosine residue could then lead t o alteration
of serine phosphorylation within the cell. Support for a
communication between protein tyrosine kinases and
changes in protein serine phosphorylation has been
obtained by a number of laboratories in intact cell systems.
For example it is known that insulin stimulates the serine
phosphorylation of proteins such as ribosomal protein S6,
acetyl-CoA carboxylase, and ATP-citrate lyase. Insulin
stimulates the dephosphorylation of other proteins such as
glycogen synthase and pyruvate dehydrogenase. Despite a
considerable amount of work, however, we have not been
successful in reconstructing a mixed cascade system using
isolated components. In this connection it is of interest,
however, that at this meeting there is a report (Tavare
cf al., 1985) that insulin stimulates the phosphorylation
of acetyl-CoA carboxylase on serine residues in Triton
extracts of placental membranes. Elucidation of the steps
involved in this and other insulin-responsive cell-free
systems should eventually lead t o an understanding of how
tyrosine phosphorylation affects serine phosphorylation.
Barker, W. C. & Dayhoff, M. 0. (1982) Proc. Natl. Acad. Sci.
lJ. S. A. 79,2836-2839
Berridge, M. J. (1984) Biochem, J . 220,345-360
Blunienthal, D. K., Takio, K., Edelnian, A . M., Charbonnc;iu.
H., Walsh, K., Titani, K. & Krcbs, I<. G. (1985) Biophys. J . 47.
Abstract M-POS23
Castagna, M., Takai, Y., Kaibuchi, R., Sana, K., Kikkaw;), U . &
Nishizuka, Y. (1982)J. Biol. Chem. 257, 7847-7851
Cohen, P., Burchell, A., Foulkes, J . G. & Cohcn, P. T. W. (1978)
FEBS Lett. 92,287-293
Corbin, J . D., Sugden, P. H., West, L.. I:lockhart, I). A , , Lincoln,
T. M. & McCarthy, D. (1978)J. Biol. Chem. 253, 3997-4003
Cori, G. T . & Cori, C. 1:. (1945)J. Biol. Chem. 158, 321 -332
Cori,G. T. & Green, A. A. (1943)J. B i d . C/tein. 151, 31-38
Dayhoff, M. O., Barker, W. C. & Hunt. L. T. (1983) Methods
Enzymol. 9 1 . 5 24
Fischer, E. H . & Krebs, E. G. (1955)J. Biol. Cheni. 216. 121 --I32
Friedman, D. L. & Larner, J . (1963) Biochemistry 2.669-675
Hawk, P. B., Oser, B. L. & Summerson, W. tl. (1947) Practical
Physiological Chemistry. 12th edn., p. 169 and p. 178, The
Blakiston Co., Philadelphia
Hunter, T. & Cooper, J . A. (1985) Anrtu. Rev. Biochem. 54. 897930
Inoue, M.. Kishinioto, A., Takai, Y. & Nisliizuka, Y. (1977)J. Bid.
Chem. 252,7610-7616
Krebs, I:. G., Graves, D. J. & I:ishccr, E . 11. (1959) J . Biol. C I i ~ i n .
234,2867--2873
Krebs, 1;. G., Blunicmthal, D. K.. 1:dclman. A . M. 6; Hales, <’. N.
(1985) in Mechanisms of Receptor Re&atiori (Crookc. S. T. &
Poste, G., eds.), Plenum Publishing Co., New York, in the press
Kuo, J . 1.’. & Grccngard, P. (1969)J. Biol. Chem. 244, 3417-3419
Linn, T. C., Pettit, I:. H. Hucho, I:. & Reed. L. (1. ( 1 9 6 9 ~Proc.
)
Natl. Acad. Sci. U. S. A . 6 4 , 227-234
Linn, T. C . , Pettit, I:. H. & Reed, L. J . (1969b) Proc. Natl. Acad.
Sci. U. S. A. 6 2 , 234-241
Mayr, G. W . & Heiliiicycr, L. M . G., Jr. (1983) FEBS Lett. 157.
225 -23 1
Mcndicino, J . . Beaudreau, C. & Bhattachryya, R. N. (1966) Arch.
Biochem. Biophys. 116,436-445
Monod, J . , Changeus, J.-P. & Jacob. 1:. (1963)J. Mol. Biol. 6 . 3063 29
Nishizuka, Y . (1983) Trends Biocherri. Sci. 8. 13-16
Nishizuka, Y. (1984) Nature (London) 308.693-697
Payne, M . 1;. & Soderling, T. R. (1980) J. Biol. Chem. 258. 2376-2382
Pires, E., Perry, S. V. & Thomas, M. A. W . (1974) FEBS Lett. 4 1 ,
292-296
Reimann, F.. M . , Titani, K., Ericsson, L. H., Wade, R. 0 . . Fischcr,
I;. 11. & Walsh, K. A. (1984) Biochemistry 23.4185-4192
Rizak, M. A. (1961) J. Biol G e m . 239. 392-395
Schulman, H. & Greengard, P. (1978)Proc. Natl. Acad. Sci. U.S. A .
75,5432-5436
Sefton, B. M. & Hunter, T. (1984) Adv. Cyclic Nucleoride Prorein
Phosphorylation Res. 18. 195-226
Sprang, S . & Fletterick, R. J . (1980) Biophys. J. 32, 175--192
Stull, J . T., Blumenthal, D. K., Botterman, B. R., Klug, G. A.,
Manning, D. R. & Silver, P. J . (1982) in Colmodulin and Intracellular Ca‘+ Receptors (Kakiuchi, S . , Hidaka, H. & Means,
A,, eds.), pp. 219-238, Plenum Publishing Co., New York
Sutherland, E. W. & Wosilait, W. D. (1955) Nature (London) 175,
169-171
Swarup, G . , Dasgupta, J. D. & Garbers, D. L. (1983)J. Biol. Chem.
258,10341-10347
Takai, Y., Kishimoto, A,, Inoue, M. & Nishizuka. Y . (1977)J. Biol.
Chem. 252,7603-7609
Takio, K., Wade, R. D., Smith, S. B., Krcbs, E. G., Walsh. K . A. &
Titani, K. (1984) Biochemistry 23,4207-4218
Tavare, J . M., Smyth, J . E., Barthwick, A. C., Brownsey, R. W. &
Denton, R. M. (1985) Biochem. SOC. nuns. 13,734-735
Walsh, D. A., Perkins, J . P. & Krebs, E. G . (1968)J. Biol. Chem.
243,3763-3765
Weber, J . T., Takio, K. & Steitz, T. A. (1982) Proc. Natl. Acad. Sci.
U.S. A . 79,7679-7683
Wong, T . W. & Goldberg, A. R . ( 1 983) R o c . Natl. Acad Sci. U.S. A .
80,2529-2533
1985