chapter 9

ON THE ACCURACY
OF
PROTEIN SYNTHESIS

Except for the relatively minor influence of environ-
mental factors, the phenotypic potentialities of an individual are de-
termined by the genetic information in his chromosomes. In this
book we have taken as a working hypothesis an even more specific
proposition, namely, that the phenotypic picture presented by an or-
ganism is the summation of the effects, physical and catalytic, pro-
duced by the complement of protein molecules characterizing the
species in question. That is to say, we assume that the genetic in-
formation available to an organism is first translated into protein
structure, each gene determining specific details of a corresponding
protein molecule, either alone or in collaboration with other genes.
For example, we may think of the structure and physiological prop-
erties of hemoglobin as being determined by a single hemoglobin
gene or “cistron” (see Chapter 4) or perhaps by several cooperating
genes having to do with each of the peptide chains of hemoglobin
and with the folding and cross attachments between them. The
biosynthesis and arrangement of nonprotein substances such as car-

185
bohydrates and fats, and the distribution and integration of these sec-
ondary products into characteristic cellular systems, might then be
considered to be the business of the proteins as agents of the geno-
type, acting as enzymes, hormones, and structural subunits of mor-
phology.

Since we know that even a single gene mutation, causing a very
limited change in the structure of a single protein species (e.g. the
sickle-cell hemoglobin case discussed in the previous chapter), can
induce a marked variation in phenotypic behavior, it becomes of
prime importance in these considerations to examine the accuracy of
the mechanisms by which proteins are synthesized and to try to ap-
praise the degree to which errors in these mechanisms may occur.
If a change in the structure of a protein involving a single amino
acid residue can cause a marked change in function, we may justifi-
ably be rather concerned by the influence of random heterogeneities
in structure caused by biosynthetic errors. When we speak of the
“control” of protein synthesis by genes, we would like to have some
idea of the limits within which this control is exercised.

The highly developed techniques now available for the physico-
chemical study of proteins have made it possible to detect inhomo-
geneities in protein preparations caused by differences as small as the
presence or absence of a single amide nitrogen group in a fraction of
a population of molecules in solution. The apparent inhomogeneity
of “pure” proteins may become even more marked as time goes on.
In spite of such elegant tools as ion exchange chromatography, coun-
tercurrent distribution, and refined electrophoresis machines, we must
recognize that the criteria of homogeneity are only relative and that
heterogeneities not observable by presently available methods may,
tomorrow, be detected on the basis of new, more discriminating
procedures.

Since proteins are very complicated organic chemicals, they may
naturally exist in a number of isomeric forms. We can conveniently
classify such variations under two major headings, using terms sug-
gested by D. Steinberg and E. Mihalyi in their recent review on pro-
tein chemistry." Variations in the structure of a protein may be at-
tributed to sequential or to configurational isomerism. The former
classification refers to differences in amino acid sequence between in-
dividual molecules and, for convenience, is defined to include other
aspects of structure which involve covalent bonds such as disulfide
bridges, phosphate ester linkages, and amide nitrogen groups. These
are the stable, black-or-white parameters of structure, not modified
by the ordinary methods of handling proteins during purification and

186 THE MOLECULAR BASIS OF EVOLUTION

storage. Configurational isomerism, on the other hand, refers to dif-
ferences in the mode of coiling of peptide chains or to the location,
frequency, and stability of noncovalent bonds. Such isomerism is to
be expected in aqueous solutions since the bonds responsible for the
secondary and tertiary structure of proteins are strongly affected by
the acidity, polarity, and temperature of the environment.

Colvin, Smith and Cook, in their review? written in 1954, have
presented a list of examples from the literature in which inhomogene-
ity has been demonstrated in protein preparations, presumably of a
high degree of purity. Thus, to quote only a few examples, lyso-
zyme, ribonuclease, and ovalbumin showed reversible boundary
spreading upon electrophoretic analysis, human gamma globulin ap-
peared heterogeneous both by electrophoretic and by ultracentrifu-
gal criteria, and insulin could be resolved into two components by
countercurrent distribution techniques. These authors have chosen
to interpret the experimental results in terms of a “microheterogene-
ity” in structure and suggest that “it seems more correct to describe a
native protein, not in terms of a finite number of definite chemical
entities, but as a population of closely related individuals which may
differ either discretely or continuously in a number of properties.”
They have suggested that, should “microheterogeneity” be observed
for all native proteins, the cellular mechanisms for the synthesis of
proteins need not be specific and rigid, and that there might exist a
broad spectrum of individual protein “subspecies” within any single
“species,” differing in enzymic, hormonal, or physicochemical prop-
erties. This conclusion was certainly an understandable one on the
basis of the information available in 1954. It is now apparent, how-
ever, that many of the examples which supported the concept of
“microheterogeneity” could be included in the list quoted only be-
cause of the inadequacy of the available knowledge about the chem-
istry of these proteins. In the case of beef pancreatic ribonuclease,
for example, we can separate on proper chromatographic columns
two major and two very minor components. A similar family of
ribonucleases may be shown to be present in sheep pancreas. These
four components appear to be present normally in pancreas tissue
since they may be separated both from purified, crystalline starting
materials and from crude extracts of pancreas glands. Electrometric
titrations on the two major components isolated from beef pancreas
have suggested that they differ by a single carboxyl group (or con-
versely by a single amide nitrogen group). Rather than assume a
broad microheterogeneity in the sense of a spectrum of related ma-
terials, we might equally well conclude that “ribonuclease” is not a

ON THE ACCURACY OF PROTEIN SYNTHESIS 187
statistical population of related proteins but rather a limited group of
well-defined chemical entities.

The complete absence of sequential isomerism in any sample of
protein can only be established by the quantitative recovery of all the
fragments on which the sequence reconstruction is based. In prac-
tice the optimal situation has never been reached, and, for the pro-
teins and polypeptides that have been studied in detail hitherto, se-
quential purity can only be inferred from the fact that aberrant se-
quences of amino acid residues, not fitting into the final reconstruc-
tion, have not been observed. In the studies of Sanger and his col-
leagues on the structure of insulin, for example, an enormous array
of peptide fragments was examined, and none of these was found to
be incompatible with the sequences of the two chains as they finally
emerged. Such studies give strong presumptive evidence for the se-
quential purity of the starting material. In other work, such as that
of Shepherd and his collaborators on the structure of ACTH,* and in
the careful chromatographic separation of the enzymatically pro-
duced fragments of ribonuclease by C. H. W. Hirs, S. Moore, W. H.
Stein and L. Bailey ( Chapter 5), careful balance sheets of recoveries
were kept, and for many portions of the over-all sequences recov-
eries were quantitative within the experimental error of determination.
In the case of ACTH none of the fragments was isolated in less than
70 per cent yield, and total recovery, on the average, was 93 per cent
of the total starting material. In those fragments recovered in yields
less than quantitative it was shown, by paper chromatography, by
terminal amino acid analysis, and by the presence of integral molar
ratios of constituent amino acids, that sequential purity was extremely
likely.

These examples illustrate two of the ways in which we may ap-
praise the homogeneity of a protein; first, by an examination of the
internal consistency between the sequences of a large number of
small peptide fragments in terms of the final reconstructed sequence
of the polypeptide chain representing the common denominator and,
second, by a consideration of the completeness of recovery of these
fragments. Both methods are as good as the accuracy of methods of
detection or analysis available for peptides and give a lower limit for
the degree of purity. Except for special sorts of inhomogeneity
which we shall discuss more fully later, most proteins or polypeptides
that have been examined will probably appear to be at least 90 per
cent or more pure by these criteria. The last 5 to 10 per cent (or 3
to 5 per cent when analysis is done by careful ion exchange chroma-
tography) remains an unknown quantity and might conceivably ob-

188 THE MOLECULAR BASIS OF EVOLUTION

scure the presence of small amounts of physically similar but struc-
turally different protein material.

Another more worrisome factor in this regard has to do with the
extent to which closely related proteins might be removed from the
major fraction during isolation procedures. The purification of a
protein such as hemoglobin presents no problem of this sort, since
the starting material, washed red cells, contains only insignificant
amounts of other proteins and the yield of hemoglobin can be made
nearly quantitative by careful experimental manipulation. Most en-
zymes and hormones, however, must be concentrated many hundreds
or even thousands of times from their initial in vivo state of purity.
Modern purification procedures are tremendously discriminating.
Under circumstances in which two forms of the same protein, differ-
ing only by a carboxylic acid group, may be separated, it is not un-
likely that very minor variations in charge or polarity within a fam-
ily may result in the complete separation of related molecules.

For these reasons we cannot be categorical about the absence of
microheterogeneity, even in the sense in which this term was used by
Colvin, Smith and Cook. There is, however, no evidence which re-
quires that we take the “broad spectrum” point of view, and it would
seem unnecessary at the present time to think of protein biosynthesis
as an inaccurate or arbitrary process.

A more optimistic alternative may be given as an explanation for
the bald fact that a well-defined sequential isomerism does occur in
certain proteins. Isomerism, as observed for the 8-lactoglobulins,
hemoglobins, serum haptoglobin, etc., may be thought of as the re-
flection of heterozygosity in the corresponding genetic material. In
those instances in which adequate genetic analyses have been carried
out, the occurrence of more than one form of a single protein has
been attributable to the presence of sets of allelic genes. In most
cases only two forms are observed. We do not find, for example,
more than one abnormal hemoglobin in any one individual (except
for varying amounts of the fetal form which many believe to be ex-
tremely similar or identical with a portion of adult homoglobin). On
the other hand, multiple forms of certain other proteins have been
observed in some instances. Proteins in the 8-globulin fraction of
the serum of cattle, for example, may be divided, by the sensitive
starch gel electrophoresis method of O. Smithies,® into four or five
well-separated subcomponents. It is important to emphasize, how-
ever, that these are separable and that what is obtained is not a
smear of overlapping substances but rather a well-defined and repro-
ducible pattern. By some techniques such as gradual salting-out

ON THE ACCURACY OF PROTEIN SYNTHESIS 189
precipitation, even highly purified human hemoglobin from normal
individuals appears to be subdivisible into several components as ob-
served by Roche and his colleagues.* Although most workers in the
field consider this phenomenon to be due to artifacts introduced by
interactions with salts and buffer molecules, it is still quite possible
that the effect is a real one, one that is simply not demonstrable by
the usual electrophoretic analysis employed for the study of the
hemoglobin series.

We have discussed, in a previous chapter, the concept of genetic
fine structure which suggests that each “gene” may be composed of a
large number of subgenic chemical units, each contributing a small
piece of information to the protein biosynthetic process. The occur-
rence of many abnormal forms of a protein in addition to the normal
one cannot be excluded if this concept is generally applicable. Thus,
the portion of the genetic material that has to do with a particular
protein molecule might involve more than one “cistron.” The syn-
thesis of separate chains, or even of parts of the same chain of a pro-
tein, might be controlled by different genetic functional units. If
this were the case, and if the mutations were not lethal ones and still
permitted the synthesis of a functionally adequate protein, we can
easily see that a multiplicity of closely related proteins could result
through the cooperation of the several cistrons involved. The
hemoglobins that have so far been separated and studied have dif-
fered in electric charge. Since electrophoresis would not distin-
guish between two hemoglobins differing, let us say, by the substitu-
tion of valine for isoleucine or of serine for threonine, other tech-
niques for fractionation will be required to test this possibility.

The Significance of Amino Acid Analogue Incorporation

A large number of structural analogues of amino acids have been
synthesized (Figure 89) and tested for their utilizability in protein
biosynthesis. These include the methionine analogues, selenomethi-
onine and ethionine, the phenylalanine analogues, o- and p-fluoro
phenylalanine and §8-2-thienylalanine, and the tryptophane analogue,
7-azatryptophane. They may be synthesized in radioactive form
and thus furnish a powerful and sensitive tool for testing whether
the mechanisms of protein biosynthesis are absolutely precise and
specific or whether alternative structures may be formed by replace-
ment of natural amino acid residues with man-made substitutes.

The results of such tests are clear-cut. Abnormal amino acids

190 THE MOLECULAR BASIS OF EVOLUTION

can be used. Cowie and Cohen," for example, have grown a methi-
onine-requiring mutant of E. coli in a medium completely free of
methionine but containing selenomethionine instead. The cells were
able to synthesize certain enzymes in a relatively normal manner in
spite of the unusual nutritional circumstance, and the proteins of the
daughter cells were of the selenomethionine variety. In another
similar study, M. Gross and H. Tarver have shown that the proteins
of Tetrahymena pyriformis can incorporate C'4-labeled ethionine.
The incorporation represents true peptide bond formation since it
was found that ethionine-containing peptides could be isolated from
partial acid hydrolysates. An interesting experiment on analogue
incorporation has been carried out by D. Steinberg and M. Vaughan,
who studied the in vitro uptake of tritium-labeled o-fluorophenyl-
alanine (see Figure 89) into the proteins of the minced hen’s ovi-
duct.* Pure lysozyme was then isolated from the tissue and digested
with trypsin and chymotrypsin. The digest was subjected to finger-
printing as described in Chapter 7, and the peptides thus separated
were analyzed for the presence of aromatic amino acids. A small
proportion of the peptides that normally contained phenylalanine
were found to contain the radioactive analogue in place of the nat-
ural amino acid.

Although analogues may be incorporated into proteins, the com-
promise with normalcy does not seem to be a happy one, for most of
them also cause marked inhibition of growth. We can say nothing
at the moment about the mechanism of this inhibition. Studies are
now in progress in several laboratories to determine the efficiency’ of
incorporation of analogues as a function of their degree of dissimilar-
ity from the natural amino acid they mimic.

Nature appears to have been extremely clever in her choice of
standard amino acids and has managed to choose some twenty which
differ sufficiently to preclude mistakes in recognition. Valine and
isoleucine residues are extremely similar from the point of view of
three-dimensional structure, and it would not be too surprising to
find an occasional lapse in the precision of protein assembly at points
of protein structure involving one or the other of these amino acids.
To my knowledge, however (within the limits of analytical accuracy
mentioned earlier in this chapter), valine-isoleucine interchange has
not been observed, except in samples of the same protein or polypep-
tide isolated from two different species or from an individual who is
“heterozygous” for the material in question. On the other hand,
D. Cowie and his colleagues have recently shown that a considerable
amount of the methionine in E. coli proteins may be replaced by

ON THE ACCURACY OF PROTEIN SYNTHESIS 191
 

(d)

   

Figure 89. The molecular structure of some amino acids and amino acid ana- though norleucine is a chemical isomer of leucine and isoleucine, its molecular
logues. (a) Phenylalanine and p-fluorophenylalanine, (b) tyrosine and. o0-fluoro- shape is much more similar to that of the sulfur-containing amino acid methionine
phenylalanine, (c) norleucine and methionine, (d) isoleucine and leucine. Al- than to that of isoleucine or Jeucine.

192 THE MOLECULAR BASIS OF EVOLUTION ON THE ACCURACY OF PROTEIN SYNTHESIS 193
norleucine,® an amino acid which is not normally found in proteins
but which exhibits a remarkable similarity in molecular appearance
(Figure 89) to methionine.

Protein synthesis is not an absolutely precise process. Some amino
acid analogues can substitute for natural amino acids. Nevertheless,
the weight of evidence indicates that no mistakes are detectable
under normal circumstances and that the protein assemble mecha-
nism must have built into it an extraordinary capacity for structural
discrimination.

REFERENCES

. D. Steinberg and E. Mihalyi, Ann. Rev. Biochem., 26, 373 (1957).

. J. R. Colvin, D. B. Smith, and W. H. Cook, Chem. Revs., 54, 687 (1954).

. §. Aqvist and C. B. Anfinsen, J. Biol. Chem., 234, No. 5 (1959); C. B. Anfin-
sen, §. Aqvist, J. Cooke, and B. Jonsson, J. Biol. Chem., 234, No. 5 (1959).

4. R. G. Shepherd, S. D. Wilson, K. S. Howard, P. H. Bell, D. S. Davies, S. B.
Davis, E. A. Eigner, and N. E, Shakespeare, J. Am, Chem. Soc., 78, 5067
(1956).

. O. Smithies, Biochem. J., 61, 629 ( 1955),

» J. Roche, Y. Derrien, and M. Roques, Bull. soc. chim. biol., 35, 933 (1953).

D. B. Cowie and G. N. Cohen, Biochim. et Biophys. Acta, 26, 252 (1957).

» [am grateful to Dr. Daniel Steinberg and Dr. Martha Vaughan of the Na-
tional Heart Institute, Bethesda, Md., for information on these studies prior to
publication.

9. Personal communication from Dr. D. B. Cowie of the Carnegie Institution,

Department of Terrestrial Magnetism, Washington, D. C.

won =

BAD

194 THE MOLECULAR BASIS OF EVOLUTION