GEORGE WALD
The Biological Laboratories, Harvard University, Cambridge, Massachusetts
It is one thing to ask questions and another to answer them; but of the two, the former is by far the rarer and more difficult - that is, if the questions are of the right sort, capable of being answered, and leading to the next questions. Szent-Györgyi has been having such a dialogue with Nature all his life; from inside, of course, not from outside, since he is part of Nature, like all of us. It is hard to know which is Socrates and which the Athenians, which is gadfly and which stung; but if one stays within earshot, one is almost sure to be stung too.
Some time ago, talking with Szent-Györgyi, I was a little startled to hear him ask, why phosphorus plays the peculiar role it does in organisms. It was a question I had asked myself, as part of a larger question: why any of the atoms that compose organisms occupy the position they do.
This is clearly a proper question, for there is nothing random about the choices. There are the atoms that form the bulk of the organic molecules: C, H, N, and O, and in special instances S and P; there are the major monatomic ions Na+, K+, Mg++, Ca++, and Cl-; there are the trace elements, mostly transition elements, and hence adapted to fill the roles in which we mainly find them, as nuclei and ligands in metallo-organic complexes and oxidoreduction enzymes: Fe, Mn, Co, Cu, Zn. All these elements are relatively light: the 16 elements mentioned fall within the first 30 in the Periodic System. Of the remaining 62 natural elements, only two - I and Mo - have restricted roles as trace elements in certain organisms. By and large, the lightest elements tend to be the most available; but except for the dominant position in the sea of the monatomic ions mentioned above, I think it is clear that something other than simple availability has governed the choice of the elements of which organisms are principally composed.
That "something other" is fitness, the peculiar combination of properties that makes these elements most suitable to play the parts that organisms demand of them. As one pursues this argument, it is to realize, I think, that organisms have had little choice among the elements. For the most part, they had to end up as they did. For that reason I think that not only the Periodic System, but the constitution of organisms, is probably about the same everywhere in the universe.
The question we ask is therefore meaningful; and it is within this larger context that I came to ask the question Szent-Györgyi asked: why phosphorus? - except that, as we shall see, I would prefer to ask at once, why phosphorus and sulfur? - for sulfur is part of the same question. And why not silicon? - that's part of the question too. It's all a matter finally of being in the Second and not in the Third Period of the Periodic System.[2]
About 99% of the living parts of living organisms are made of the four elements, H, O, N, and C. Some time ago I asked myself, why these? - and after a time thought I knew the answer. These are the smallest elements in the Periodic System that achieve stable electronic configurations by adding respectively 1, 2, 3, and 4 electrons. C, N, and O follow one another directly in the Second Period. Next comes fluorine, F; but fluorine, though it too gains a stable electronic configuration by adding one electron, has no part in making organisms, because in any of its properties that might matter for this reason it is outdone by the smaller element, hydrogen. Adding electrons is part of the story, but smallness is the rest. The only thing better than being in the Second Period is to be in the First.[3]
What it means to add electrons is clear enough. Adding electrons by sharing them with other atoms is the mechanism for forming chemical bonds, and hence molecules. But what has smallness to do with it? Two things: (1) the smallest atoms ordinarily form the tightest, most stable bonds; and (2) they alone regularly form stable multiple bonds. Many years ago G. N. Lewis (1923, p. 94) commented that "the ability to form multiple bonds is almost entirely, if not entirely, confined to elements of the first period of eight, and especially to carbon, nitrogen, and oxygen." Indeed he suggested that were it not for these elements, the concept of the multiple bond might never have been invented. Incidentally the rate occasions when one does encounter multiple bonds outside this trio involve most frequently sulfur and phosphorus.
All of this comes to a head in the comparison of carbon with silicon. In those parts of the earth even remotely available to living organisms, silicon is about 135 times as plentiful as carbon. In the surface layers of the earth, including the atmosphere and hydrosphere, silicon constitutes 16.08% of all atoms, carbon only 0l.119%. Coming directly under carbon in the Periodic System, it shares with carbon the property of tending to gain four electrons, and so to form four covalent bonds by sharing electrons with other atoms. Why then is life on earth based upon the relatively rare element carbon rather than on the prevalent silicon?
First, the strength and hence stability of binding. These are shown in the following table, from which it is clear that the interatomic distance (bond length) is much smaller in a C-C than in an Si-Si bond, and the bond energy of the former is almost twice that of the latter:
It is clear that to the degree that the business of making organisms includes the capacity to form tight, stable bonds, and eventually long, stable chains of atoms, carbon has a large intrinsic advantage over silicon.
The importance for organisms of the capacity to form multiple bonds can best be illustrated by comparing CO2 with SiO2. In CO2, carbon is bonded to each of the oxygen atoms by double bonds, each involving the sharing of two pairs of electrons. By this means, each of the atoms of CO2 achieves the complete octet of outer shell electrons found in the neighboring inert gas, neon. All the combining tendencies are satisfied, and the molecule, free and independent, goes off in the atmosphere as a gas, and readily dissolves in and combines with water, the forms in which living organisms obtain and use it.
In SiO2, on the contrary, Si is joined to oxygen by single bonds, leaving two unpaired electrons on the silicon and one on each of the oxygen atoms. Unable to pair by forming multiple bonds, these pair instead with the unpaired electrons on neighboring molecules of silicon dioxide. This process, repeated endlessly, ends in a huge polymer, in a sense a huge supermolecule of silicon dioxide. This is the essential structure of quartz - an extraordinarily dense, hard, inert material, which can be broken only by breaking covalent bonds.
Silicon has a third fundamental disability relative to carbon. Like carbon, silicon has a strong tendency to combine with itself to form chains. Potentially it should be possible for silicon compounds to exist in a variety and complexity rivaling the carbon compounds. From the point of view of life, however, the former have a fatal disability: Si-Si bonds are unstable in the presence of water, ammonia, or oxygen.
The reason for this is fundamental to our discussion. In such a Second-Period element as carbon, the outer shell contains one 2s and three 2p orbitals, each potentially capable of holding a pair of electrons of opposed spin, so completing an octet. (In forming four tetrahedral covalent bonds, the four second-shell orbitals hydridize to form four identical hybrid sp orbitals.) The formation of four covalent bonds - which whatever mixture of single and multiple bonds - in a sense completes the octet, filling all available second-shell orbitals, and that is the end of it.
In silicon, however, the outermost electron shell is the third. When the one 3s and three 3p orbitals have been filled, as the result of chemical combination, the atom has achieved a measure of stability, its electron configuration - though distorted in that the atomic orbitals have been replaced by molecular orbitals - approaching that of the inert gas argon. The completion of an octet of electrons, however, does not saturate the third shell. It possesses in addition five 3d orbitals, holding potentially 5 further pairs of electrons. That is, though when the third shell is outermost its stable number of electrons is 8, as in argon, the formation of a fourth shell, as in the Fourth Period, ends in bringing the third shell number from 8 to 18, as in the elements from zinc to krypton. After silicon has in effect completed an octet by forming four covalent bonds, its outer-most shell can still potentially accept further electrons in the empty 3d orbitals.
The molecules to which silicon chains are most susceptible have the particular characteristics of small size and the possession of lone pairs of electrons. The relatively large interatomic spacing of silicon chains (see above) seems to allow such molecules to come in close enough for their lone electron pairs to occupy empty 3d orbitals, and to disrupt the chain. Indeed much the same attack is made on linkages between silicon and other atoms than itself. For example, whereas methane (CH4) is stable to water and sodium hydroxide, silane (SiH4) is attacked by these substances to form sodium silicate and gaseous hydrogen (SiH4 + 2 NaOH + H2O -> Na2SiO2 + 4 H2).[4]
We can state therefore three powerful reasons, one relative and two absolute, why silicon is unsuitable as a basis for forming organisms: (1) It forms much weaker bonds than carbon, both with itself and with other atoms. (2) Its reluctance to form multiple bonds results in the formation of huge, inert, covalently bonded polymers, so removing all but traces of silicon from circulation. (3) The instability of silicon chains and compounds in the presence of oxygen, ammonia, and water should suffice to disqualify them. I cannot imagine life existing at all apart from water, or going very far without oxygen; and both conditions a priori rule out basing the constitution of living organisms upon silicon.
It is precisely these disabilities of silicon that help to create a special opportunity for phosphorus and sulfur. I am thinking here of the opportunity for these elements to play their most distinctive role in organisms, as agents of group and energy transfer. They have of course other functions, but no others that make such special demands, or that single these atoms out so particularly.
Two properties should most facilitate this type of function: on the one hand the capacity to form a wide variety of linkages of small and large energy potential, i.e., an adequate coinage; and added to this, an intrinsic instability of linkage, so facilitating exchange. The first property lies in the province of thermodynamics; as for the second, thermodynamics sets the stage, but what actually happens lies in another province, that of reaction mechanisms and kinetics. We will be mainly concerned with this latter type of consideration in discussing the special virtues of phosphorus and sulfur.
The thermodynamic aspects of energy and group transfer - the business of "low-energy bonds" and "high-energy bonds", as biochemists commonly employ these terms - have been carefully explored in the twenty years since Lipmann (1941) first formulated this type of concept (cf. Kalckar, 1946, 1947; Oesper, 1950; Hill and Morales, 1951; George and Rutman, 1960; Huennekens and Whiteley, 1960; Rutman and George, 1961). It is by now well recognized that these are poor terms; that one is dealing here, not with the energy localized in a bond - a bond energy in the strict sense - but with the change in free energy that accompanies a transfer reaction, i.e., with the differences in free energy between the reactants and their products. This free energy change in any given instance is compounded from changes in resonance stabilization, ionization, electrostatic forces - indeed all the consequences of reaction under the particular circumstances in which it occurs. It is in this sense that high- and low-energy compounds are at present defined on the basis of the standard free energy change that accompanies the hydrolysis of a particular bond; a [[Delta]]Fdeg. of -1 to -3 kcal per mole characterizes a low- and -5 to -10 kcal per mole a high-energy bond.
By far the widest variety of energy and group transfer reactions in biological systems is carried out by organic phosphates. Sulfur forms three types of "high-energy" complex, one consisting of acyl esters of thiols (e.g., acetyl CoA); another of mixed anhydrides of phosphoric and sulfuric acids [e.g., adenosine-phosphoryl-sulfate for solfonations (cf. Lipmann, 1958)]; and a third of sulfonium compounds [e.g., S-adenosyl methionine for methylations (cf. Cantoni, 1953, 1960)]. This short list almost exhausts the known categories of "high-energy" compounds. One must add an - as yet - very limited class of acyl imidazoles, in which the acyl group is attached to a nitrogen atom (Stadtman and White, 1953; Stadtman, 1954); and the highly important classes of activated amino acids, in which the amino acid carboxyl group is joined in ester linkage with the 2' - or 3'-OH of ribose in the terminal adenylic acid of transfer-RNA (Zachau, et al., 1958; Hecht et al., 1959; Preiss et al., 1959). These latter instances are as yet the only ones I know in which group transfers are negotiated by something other than sulfur or phosphorus compounds.
What lends sulfur and phosphorus this special position? Or to put this question a little more in the context of our general argument, what properties do sulfur and phosphorus have that oxygen and nitrogen - their congeners in the Periodic System - lack, that fit them better to perform this type of function?
I think the answer to this question is to be sought in three directions, all multiply interconnected: (1) S and P form more open and in general weaker bonds that O and N. (2) S and P can expand their covalent linkages beyond 4 on the basis of their 3d orbitals. (3) To a unique degree among Third-Period elements, S and P retain the capacity to form multiple bonds.
(1) The interatomic distances (bond lengths) are larger in covalent linkages of S and P than in those of O and N. In making such comparisons one must consider comparable molecules, since the length of any particular bond varies somewhat with the remaining structure of the molecule. A first approach to this situation is gained in such tabulations of covalent bond radii (Table I) as given by Pauling (1960, pp. 224-228).
The interatomic distances of bonded atoms, obtained by adding together the appropriate bond radii, are reasonably reliable for molecules in which the atoms in question possess the numbers of covalent bonds ordinarily associated with their position in the Periodic System (i.e., C, 4; O, 2; P, 3; etc.), and in which the bonds do not possess too much ionic character or in which the bond order does not depart too greatly from a whole number.
In general the larger bond radii encountered in the Third Period go also with the formation of weaker bonds - bonds of lower energy - than those formed by their congeners in the Second Period. This point was made earlier in comparing Si-Si with C-C. So, for example, the bonds formed by these atoms with hydrogen have the energies shown in Table II (Pauling, 1960, p. 85):
The first point in our argument therefore is that in going from the Second to the Third Period, from N and O to P and S, one goes in general to more widely open and weaker bonds, hence bonds intrinsically more susceptible to attack, and more ready to undergo cleavage and exchange reactions.
(2) We have already pointed out with reference to silicon that elements of the Third Period possess d in addition to s and p orbitals, and so have place to hold electrons beyond the normal outse-shell octet. It is this property that most distinguishes S and P from their congeners in the Second Period. The five 3d orbitals of S and P could hold potentially 5 pairs of electrons. It seems however that the possibilities of forming stable linkages are more restricted. Ordinarily phosphorus does not go beyond 6, as in SF6. Some formalisms prefer to represent such molecules otherwise than possessing 5 or 6 equivalent covalent bonds; but it is significant that, however represented, they call upon properties not shared by N and O, which do not form molecules of comparable valence.
We can expect therefore that as with compounds of silicon, the presence of unoccupied 3d orbitals in S and P invites attack by molecules possessing lone pairs of electrons, the one pairs occupying in an intermediate stage the 3d orbitals, with an exchange reaction as the eventual result.
(3) Sulfur and phosphorus, to a unique degree among Third Period elements, retain the tendency to form multiple bonds, otherwise so characteristic of carbon, nitrogen, and oxygen. Having stressed earlier in this paper that silicon does not markedly display this property, it is interesting to ask why sulfur and phosphorus possess it.
The tendency of elements to form multiple bonds seems to me to be associated somehow with small size as such. I have not been able to find a clear formulation of this view in the literature, nor can I formulate it myself. Coulson (1953, p. 178) ascribes "the experimental fact that multiple bonds are practically confined to the first two rows of the Periodic Table" to increasingly strong repulsive forces between nonbonding electrons as atoms grow larger. If one thinks of the bond as a pair of electrons shared between two atoms, this demands that the bonded atoms approach each other closely. Under such circumstances the electrons not engaged in bonding repel one another strongly, and the larger the number of such electrons the stronger the mutual repulsion. This is thought to be the reason for the relative weakness of bonds between the heavier elements.[5] A multiple bond, since it involves still closer approach between atoms than a single bond, evokes still stronger repulsions. For this reason only atoms of small kernel, containing relatively few electrons, might permit the formation of stable multiple bonds.
This cannot be the whole story, however, for it would seem to imply that in the series Si, P, S, both bond strengths and the tendency to form multiple bonds should decline continuously, since the number of nonbonding electrons and hence the mutual repulsions are rising; whereas just the opposite is the case. The energy values of single bonds are: Si-Si, 42.2 P-P, 51.3; S-S 50.9; and Cl-Cl, 58.0 kcal per mole; and S and P have more, not less tendency than Si to form multiple bonds.
It may be therefore that a small atomic radius as such promotes the tendency to form multiple bonds. If so, that rises an interesting consideration. The atomic radii of the elements do not rise continuously with atomic number; on the contrary, they decline within each period of the Periodic System, jumping to a higher level at the opening of each new period. The reason for this is that the atomic radius is set principally by the number of electron shells, which of course does not change within each period. As one ascends each period, however, the increasing positive charge on the nucleus draws the electrons in closer toward it. For this reason within each period the atomic radii grow smaller as the atomic number rises.
It is not as strange as it first seems therefore that sulfur and phosphorus may make double bonds more readily than silicon, for they are smaller. The order of covalent bond radii in Second and Third Period elements is as given in Table III (Pauling, 1960, pp. 224-228):
Selenium has the same single and double bond radii as silicon, bromine has somewhat smaller radii. All the other elements have larger bond radii. One might suppose from such a series that phosphorus is the largest (not heaviest!) element that readily forms multiple bonds; and that from silicon on only vestiges of this tendency remain.
Occasionally one encounters the notion that it is more rigorous to write the P-O linkages in H3PO4 as three single covalent bonds and one dative bond, rather than three single bonds and one double bond. [6] This is however an almost meaningless distinction, and has little basis in fact or theory. I think it arises in part through a supposed analogy with nitrogen, which after forming three covalent bonds, as in trimethylamine, (CH3)3N, forms a fourth dative bond as in trimethylamine oxide, (CH3)3N -> O. The analogy is not apt, however, since unlike nitrogen, phosphorus can form a fifth covalent bond by accepting a pair of electrons in its 3d orbitals. A double bond in addition to three single bonds is clearly possible in this instance. Indeed the measurements of bond length make this the best single formula that can be written with two single and two double S-O bonds, making six covalences in all (cf. Pauling, 1959, p. 240).
Cruickshank (1961).
Actually of course no single formula properly represents these molecules. They are resonance hybrids in which all the P-O and S-O bonds are considerably contracted compared with single covalent bonds. Cruickshank (1961) has estimated the amount of such contraction in the tetrahedral oxyanions of Third-Period elements (Table IV).[7] The single bond lengths were calculated here on the basis of the Schomaker-Stevenson formulation (1941), in which a correction term is introduced for the partial ionic character of the bond, owing to differences in electronegativity of the participant atoms. The further contraction of the bonds observed here is taken to be evidence of distributed double-bond character, and involves in each instance the 3d orbitals of the central atom.
Cruickshank (1961) has also estimated the rearrangement of bond lengths that accompanies the attachment of another atom or group to one or more of the oxygen atoms in such ions. He states it to be "a simple empirical rule" that the average X-O distances in XO4 tetrahedral ions remain equal to the "observed" distances in Table IV. The attachment of another atom or group to O may lengthen that X-O bond by amounts up to 0.15 Å, but the other X-O bonds simultaneously contract so as to preverse the average.
So for example in the tetrahedral ion PO42-, each P-O bond is 1.54 Å long. In H3PO4 however, the P-OH bonds are about 1.57Å long, the P-O bond 1.52 Å. If instead of H a large organic radical is attached to O, this produces a much larger asymmetry. In serine phosphate, for example, the P-O bond lengths are as in Scheme I.
Sulfate exhibits similar relationships. In the ion, SO42-, the S-O bonds are 1.47 Å long. An S-OH bond is longer, 1.56 Å for example in KHSO4. In the ethyl sulfate ion, however, (C2H5)-O-SO3-, the S-O(C2H5) bond is 1.603 Å long, the three S-O bonds 1.464 Å long.
Molecules in which phosphate P is bonded directly to N, as in phosphocreatine and phosphaorginine, exhibit similar relationships. So, for example in the phosphoammonium ion fond lengths are distributed as shown in Scheme II.
In the substituted phosphates and sulfates, therefore, the lengthened bond where an organic or other radical is attached to O makes a preferred opening; the unoccupied 3d orbitals on P or S provide a berth. Any molecule that can donate a lone pair of electrons should find easy access, and a ready means of attachment to the S or P nucleus. This is a condition that invites attack by water, ROH, RNH2, and similar molecules; and to some degree this vulnerability must be shared by all organic compounds of sulfur and phosphorus.
This discussion implies that in general the exchange reactions in which organic S and P compounds participate begin - whatever rearrangements may follow - with the addition of a lone electron pair to the S or P nucleus, occupying an empty 3d orbital. The hydrolysis of an S or P bond for example would begin with the attachment of a lone electron pair of oxygen to S or P. Barring further rearrangements one would expect that in the products of such a hydrolysis, a hydroxyl group derived from water would remain attached to S or P.
Apparently all known enzymatic hydrolyses of organic phosphates do end in this way, as has been shown by carrying out such reactions with H2O18. All the known kinase reactions, in which molecules with alcohol-OH groups replace water as reagents, also yield this result. In all such reactions the existing P-O or P-N bond is broken, and the -OH or -OR of the attacking molecule remains attached to P (cf. Cohn, 1959).
The reactions of ATP seem to involve a similar mechanism. Here enzymatic hydrolysis (ATPase reaction) and exchange (kinase) reactions with ROH or RNH2 cleave the terminal phosphate bond, the HO-, RO-, or RN- remaining with the amputated phosphoric acid. In another important class of reactions involving the activation of fatty and amino acids, ATP is cleaved so as to separate inorganic pyrophosphate, yielding the fatty acid or amino acid adenylate as the other product of the exchange (Scheme III).
In all these cases the reaction is probably initiated by a lone pair of O or N attaching to P. Why this attack involves the first phosphoric acid in some cases and the terminal phosphoric acid in others is not understood. (No case of pyrophosphoryl transfer has yet been observed). It should not be forgotten that these are enzymatic reactions, and the enzyme presumably has much to do with where the attach occurs.
To leave the subject at this point, as I am about to do, is hardly to begin it. However such exchange reactions may in general be initiated, they do not always end as simply as just described. So, for a prominent example, acyl-S-CoA transfers the acyl group so as to leave, not a donator of lone pairs, but a hydrogen atom on S: the product of course is HS-CoA. None of this is especially mysterious, though it does present further problems. The mechanisms of numbers of enzymatic reactions involving the cleavage of organic phosphates have been carefully studied by Cohn, Koshland, and others (cf. Cohn, 1959); and noneyzmatic transfer reactions involving acyl phosphates have recently been explored by DiSabato and Jencks (1961). Clearly also one should have to invoke quite different mechanisms from those I have discussed to deal with the transfer reactions of acyl imidazoles and the amino acid-ribosyl linkages on transfer-RNA.
To summarize the whole argument: I have tried to find a basis for the biological selection of S and P for group and energy transfer reactions in (1) the fact that they form more open and usually weaker bonds than their congeners in the Second Period, O and N; (2) their possession of 3d orbitals, permitting the expansion of their valences beyond four; and (3) their retention of the capacity to form multiple bonds, a property otherwise characteristic of C, N, and O.
The capacity to form multiple bonds contributes principally to the thermodynamics of energy transfer. Particularly when combined, as in P and S, with the possibility of forming 5 and 6 covalent bonds, this introduces a wide range of resonance possibilities among the precursors and products of exchange reactions that greatly increases the variety and extent of the energy changes that can occur. To use an earlier phrase, these properties ensure an adequate coinage.
The relatively wide spacing and weakness of S and P bonds, together with their tendency to add lone electron pairs in their unoccupied 3d orbitals, induces an intrinsic instability and vulnerability to attack by other molecules that promote exchange reactions.
I think that as with all the other elements of which organisms are principally composed, sulfur and phosphorus were selected on the basis of fitness: among all the elements of the Periodic System they apparently possess to a unique degree properties that lend themselves to group and energy transfer. It was to find those properties that organisms had to go from the Second into the Third Period.
References
Cantoni, G. L. (1953). J. Biol. Chem. 204, 403.
Cantoni, G. L. (1960). In "Comparative Biochemistry" (M. Flokin and H. S. Mason, eds.), Vol. I, p. 181. Academic Press, New York.
Cohn, M. (1959). In Symposium on Enzyme Reaction Mechanisms. J. Cellular Comp. Physiol. 43, Suppl. 1, 17.
Coulson, C. A. (1953). "Valence." Oxford Univ. Press, London and New York.
Cruickshank, D. W. J. (1961). J. Chem. Soc. p. 5486.
Di Sabato, G., and Jencks, W. P. (1961). J. Am. Chem. Soc. 83, 4393.
George, P., and Rutman, R. J. (1960). Progr. in Biophys. and Biophys. Chem. 10, 2.
Hecht, L. I., Stephenson, M. L., and Zamecnik, P. C. (1959). Proc. Nat'l. Acad. Sci. U.S. 45, 505.
Hill, T. L., and Morales, M. F. (1951). J. Am. Chem. Soc. 73, 1656.
Huennekens, F. M., and Whiteley, H. R. (1960). In "Comparative Biochemistry" (M. Florkin and H. S. Mason, eds.), Vol. I, p. 107. Academic Press, New York.
Kalckar, H. M. (1946). In "Currents in Biochemical Research" (D. E. Green, ed.), p. 229. Interscience, New York.
Kalckar, H. M. (1947). Nature 160, 143.
Lewis, G. N. (1923). "Valence." Chemical Catalog Co., New York.
Lipmann, F. (1941). Advances in Enzymol. 1, 99.
Lipmannn, F. (1958). Science 128, 575.
Oesper, P. (1950). Arc. Biochem. 27, 255 (1950).
Oesper, P. (1951). In "Phosphorus Metabolism" (W. D. McElroy and B. Glass, eds.), Vol. I, p. 523. Johns Hopkins Press, Baltimore, Maryland.
Pauling, L. (1959). "General Chemistry," 2nd ed. W. H. Freeman, San Francisco.
Pauling, L. (1960). "The Nature of the Chemical Bond," 3rd ed. Cornell Univ. Press, Ithaca, New York.
Preiss, J., Berg, P., Ofengand, E. J., Bergmann, F. H., and Dickmann, M. (1959). Proc. Nat'l. Acad. Sci. U.S. 45, 319.
Rutman, R. J. and George, P. (1961). Proc. Nat'l. Acad. Sci, U.S. 47, 1994.
Schomaker, V., and Stevenson, D. P. (1941). J. Am. Chem. Soc. 63, 37.
Stadtman, E. R. (1954). In "Mechanisms of Enzyme Action" (W. D. McElroy and B. Glass, eds.), p. 581. Johns Hopkins Press, Baltimore, Maryland.
Stadtman, E. R., and White, F. H. (1953). J. Am. Chem. Soc. 75, 2022.
Zachau, H. G., Acs, G., and Lipmann, F. (1958). Proc. Nat'l. Acad. Sci. U.S. 44, 885.