Browse Search Feedback Other Links Home Home
The Talk.Origins Archive: Exploring the Creation/Evolution Controversy

Apolipoprotein AI Mutations and Information

A reply to Answers in Genesis regarding the Apo AI Milano mutation

Copyright © 2003
[Posted: 20 April 2003, Revised 6 May]
[Links updated: 14 July 2005]
Version 1.5


allele: one of several possible forms of a gene.
amino acid: a small organic molecule used as a building block for proteins.
apolipoprotein: the protein component of a lipoprotein.
apolipoprotein AI (Apo-AI): the major protein for forming HDL cholesterol lipoproteins; that is, good HDL.
cholesterol: a lipid (fat) which is an important part of animal cells. It is transported through the bloodstream as lipoproteins; but in isolation it would congeal into a solid.
dimer: a molecule formed by two identical simpler units, called monomers.
gene: a segment of DNA which contributes to the function or form of an organism. Genes mostly work by coding for protein.
heterozygote: An organism that carries two different alleles of the same gene.
homozygote: An organism that carries two identical alleles of the same gene.
lipid: lipids (fats) are defined as substances which dissolve in alcohol but not in water. Fats are an important part of all cells, and a major source of fuel.
lipoprotein: a particle which combines lipids and protein. The proteins allow the particle to dissolve in water, which means that the lipid can be transported in the blood stream.
LDL, HDL: low density lipoprotein, high density lipoprotein. Lipoproteins are characterised by their density.
LDL cholesterol, HDL cholesterol: Cholesterol in LDL lipoproteins is called "bad" cholesterol, because the LDL can deposit cholesterol on artery walls, restricting blood flow.
Cholesterol in HDL lipoproteins is called "good" cholesterol, because HDL can capture and remove cholesterol from the arteries and get transported out of the blood stream through the liver. A good, simple explanation is provided by the University of Miami School of Medicine.
protein: a large organic molecule made up of amino acids linked in a chain.

In the 1980's a small Italian community was found to have a mutant version of a protein, called Apolipoprotein AI (Apo-AI), which is associated with a decreased risk of arteriosclerosis (clogged arteries), heart attack, and stroke (1). The reduction in risk in these people has been attributed to the mutant protein (known as the Apolipoprotein AI Milano allele, henceforth referred to as Apo-AIM), and Apo-AIM has often been used as an example of a beneficial mutation.

Apo-AI is a lipid-binding protein and is the major component of High Density Lipoprotein (HDL) particles, which play an important role in removing cholesterol from cells. Subsequent detailed research of the Apo-AIM mutation has demonstrated that it has improved biological function that directly contributes to lowering the incidence of cardiovascular disease in the individuals carrying it.

However, Answers in Genesis has a feedback response (last accessed 2/05/2003 [see Note]), which claims that the mutant provides no evidence for evolution as it has lost "specificity" (and by implication has lost "information"). AiG additionally claims that the mutant Apo-AIM is restricted in its ability to form useful HDL particles. The basis for these claims appears to be two press releases (2,3), rather than the primary literature.

These press releases are not entirely accurate, and they appear to have been substantially misinterpreted. Contrary to what is stated by AiG, Apo-AIM is a fully functional protein that lowers the risk of arteriosclerosis and cardiovascular disease by a number of mechanisms (4). In an experiment where reconstituted HDL particles made of Apo-AIM were infused into arteriosclerotic rabbits, the rabbits had fewer and less extensive plaques (4). There was decreased aortic cholesterol, and decreased cell proliferation (both of which improve cell wall flexibility (4)).

While not all of the anti-arteriosclerotic actions of Apo-AIM are fully understood, we now know of two major mechanisms involved in its action. Apo-AIM actively stimulates cholesterol removal from cells (5), and its antioxidant ability also prevents some of the inflammatory damage in arteriosclerosis (6). AiG has focused on the antioxidant activity (possibly because of its prominence in the U.S. Department of Energy (DOE) press release (2)), claiming that this represents a loss of specificity. In the following sections we will examine AiG's claims more closely.

Are Apolipoprotein AI-M dimers of restricted usefulness?

In the AiG page it is stated:

"One amino acid has been replaced with a cysteine residue in an enzyme that normally assembles high density lipoproteins (HDLs), which are involved in removing 'bad' cholesterol from arteries. The mutant form of the enzyme is less effective at what it is supposed to do, but it does act as an antioxidant, which seems to prevent atherosclerosis (hardening of arteries). In fact, because of the added -SH on the cysteine, 70% of the enzymes manufactured bind together in pairs (called dimers), restricting their usefulness."

This statement contains a number of inaccuracies and mistakes. Apo-AI is a lipid-binding protein (not an enzyme) that forms complexes with other proteins (like Apo-AII) and lipids to form HDL particles. Apo-AI can self-associate, and normally forms modest amounts of dimers, trimers and quadramers as well as being in the monomeric form, and produces HDL particles in a range of sizes (7). In Apo-AIM the basic amino acid arginine (R) at position 173 has been mutated to a sulphur-containing amino acid, cysteine (C; R173C). This results in a greater ability to form stable dimers than Apo-AI as the two cysteines can form a chemical bond together via these sulphur groups (7).

Despite AiG's claim that the formation of dimers results in restricting their usefulness (this claim is not in either of the press releases), they are in fact fully functional. Cholesterol removal from cells occurs primarily through reverse cholesterol transport (RCT). The first step in RCT is the efflux of un-esterified cholesterol from cells to suitable acceptors, normally apolipoproteins (with or without lipids already bound). This occurs through two distinct mechanisms: a) nonspecific interaction of lipoprotein acceptors with the cell, and diffusion of cholesterol from the cell membrane onto the lipoprotein surface, and b) interaction of lipid-free apolipoproteins with specific acceptor sites on the cell surface. Cholesterol then diffuses into the apolipoproteins. Both of these mechanisms depend on the structure of Apo-AI.

The Apo-AIM dimers bind to the specific Apo-AI binding site (lecithin:cholesterol acyltransferase; LCAT) as efficiently as Apo-AI (5,7) and stimulates cholesterol efflux (5,8). Apo-AIM dimers form HDL particles as readily as Apo-AI monomers (7), and the HDL formed from Apo-AIM stimulates cholesterol efflux more efficiently than HDL formed from Apo-AI monomers (5,8). Apo-AIM-containing HDL is also far more efficient at inhibiting cholesterol esterification by microsomal acylCoA:cholesterol acyltransferase (ACAT), than Apo AI-containing HDL (5). This results in more cholesterol being released from the membrane for efflux into HDL particles.

Apo-AIM does form a more restricted size range of HDL, with predominant sizes of 7.8, 12.7 nm (and a rare 10.8 nm form), whereas Apo-AI forms particles of 7.8, 9.6, 12.7 and (mostly only seen at high lipid concentrations) 17.6 nm (7). This is in part because, as mentioned above, Apo-AI can form dimers and higher oligomers, while Apo-AIM can only form dimers and quadramers. Note again that normal Apo-AI can and does dimerise, and dimeric Apo-AI is incorporated into HDL particles just as dimeric Apo-AIM is. Despite the restricted size range, the HDL formed by Apo-AIM dimers are functional (5,8,7). While it was claimed in the DOE press release (2) that dimerisation is responsible for the HDL deficiency seen in people carrying the Apo-AIM gene, the dimers are actually more stable than Apo-AI monomers (9); indeed, part of their ability to reduce the risk of arteriosclerosis may be due to their stability, resulting in their hanging round for longer, mopping up more cholesterol and activating whatever they activate to inhibit cell proliferation (9). So we can see that Apo-AIM dimers are not restricted in usefulness, and do form fully functional HDL particles. These Apo-AIM dimer HDL particles are more stable than Apo-AI HDL particles (9), and are better than Apo-AI HDL particles at stimulating cholesterol efflux (5,8).

Is the antioxidant activity of Apo-AIM non-specific?

While AiG concedes that Apo-AIM has gained a new function, the antioxidant ability, they contend that this comes at the expense of specificity.

"Now in gaining an anti-oxidant activity, the enzyme has lost activity for making HDLs. So the mutant enzyme has sacrificed a lot of specificity. Since antioxidant activity is not a very specific activity (a great variety of simple chemicals will act as antioxidants), it would seem that the net result of this mutation has been a huge loss of specificity, or, in other words, information. This is exactly as we would expect with a random change."

As we have seen, Apo-AIM has not lost the ability to make HDLs, so it has not sacrificed specificity. Indeed, as Apo-AIM HDL particles are more effective at promoting cholesterol removal from cells, one could reasonably claim that there has been an increase in specificity. However, is the antioxidant activity of Apo-AIM non-specific? Antioxidant activity is possessed by a number of small molecules, but so is protein hydrolysis (catalyzed by small molecules such as the amino acid serine), esterification and just about all important enzyme activities. What matters to an extent is if there is a specific sequence that binds a substrate to deliver it to the antioxidant amino acid (in the same way the catalytic site sequence in a protein hydrolysis enzyme delivers the substrate to catalytic serine). Is the antioxidant activity of Apo-AIM substrate and sequence specific?

The answer is yes. The antioxidant effect is sequence dependent. The Milano mutation (R173C) is much more effective at inhibiting oxidation of lipids than another mutation, the Paris mutation (R151C) (6). There is also substrate specificity. Neither mutation can quench superoxide anions in aqueous solution (6), suggesting that the R->C mutation does not generate a generic antioxidant, but rather a specific targeted antioxidant that will only work in a specific way. Neither mutation is capable of preventing oxidation of a control protein (cytochrome c) (6) so the mutation is specific for lipid substrates. Further studies (10) have indicated that small peptides derived from these regions (amino acids 167-R173C-184 and 145-R151C-162) also retain the specificity for lipids, indicating that it is not simply the presence of the cysteine, but rather the position of the cysteine within the structural constraints of the protein, that confer the healthy antioxidant properties. Thus we can see that the antioxidant properties are specific, in the sense that they are substrate and sequence dependent.

The AiG response also implies that the dimerisation of Apo-AIM is irreversible. In fact, cysteine dimerisation is a readily reversible reaction, and is used as a control mechanism in several proteins (for example HSP33 (11)). Dimer formation is reversible by exposure to reducing environments (6), and while binding to arteriosclerotic plaques puts the Apo-AIM HDL in an oxidizing environment, the Apo-AI HDL also binds to a number of other cellular environments where there are coupled reduction-oxidization reactions which specifically reverse cysteine dimerisation. Furthermore, disulphides (R-S-S-R, where R is an organic compound or protein) are also antioxidants in certain circumstances. R-S-S-R compounds are quite capable of being sinks for electron oxidation (12).

Does the Apo-AIM mutation represent a loss of information?

The AiG page claims that information-increasing mutations are required for evolution. The concept of "information" is a problematic one in biology, as most measures only imperfectly capture key aspects of genetic change. Biologists prefer to think in terms of gene number and gene/protein function. While an increase in gene number and function is not required (parasites which have lost genes do quite nicely), increase in gene number has occurred. Certainly the average vertebrate has more protein coding genes than worms or insects, and these in turn have more protein coding genes than unicellular yeast. The basis of this increase is largely via duplication of pre-existing genes. For example, one of the major differences between vertebrates and worms, and worms and yeasts, is an increase in the numbers of modified copies of a class of enzymes called tyrosine kinases. By most measures of "information" a vertebrate with 30,000 genes in its genome has more information than a yeast with a mere 6,000 or so genes, and the role of gene duplication in this rise is well understood (13).

AiG uses another definition of "information," equating it with "specificity." This was originally coined by Dr . L. Spetner, and is related to the number of substrates an enzyme binds (the fewer substrates, the more specific the enzyme is and the more "information" it has (14)). Applying this measure to non-enzymes is not entirely straightforward. With this measure, it is claimed that random mutations do not increase "information" in a protein. In the case of Apo-AIM, AiG claims that the mutant apolipoprotein has lost specificity as it has lost (or restricted) the ability to form HDL particles, and the antioxidant ability of Apo-AIM is "non-specific". We have seen that in fact Apo-AIM has not lost the ability to form HDL particles, and that these HDL particles that are formed bind to specific acceptor sites and are more effective at promoting cholesterol efflux than normal HDL particles. We have also seen that the Apo-AIM antioxidant ability is both sequence and substrate specific. Thus Apo-AIM has not lost "information" by AiG's own measures. If anything it has gained Spetner "information".

After this page was made public [20/4/03] AiG added a paragraph that expands on the concept of specificity. Here they introduce the analogy of "fixing cars" and claim that a statement "fix the porsche" has more information than the statement "fix the car and the truck" as the latter, although having two "functions" is less specific than the first. They then refer readers to Werner Gitt's page.

There are two fundamental problems with this argument. First, as we have shown above, Apo-AIM is not less specific than Apo-AI, and can be reasonably considered more specific by AiG's very own criteria. Apo-AIM produces a more specific range of HDL partical sizes, rather than a broader non-specific distribution, it is more specific in activating ACAT, and it specifically reverses oxidation of lipids binding to a specific recognition sequence - all of which increase the amount of information in Apo-AIM according to AiG's logic. Secondly, they have chosen the wrong analogy to illustrate Apo-AIM's "function", and when applying information theory, one must be very careful of the analogy being used. In this case, the functions are (1) increased cholesterol efflux and (2) repair of damaged proteins, very different to the general "fix the car" scenario AiG discuss. These functions are more akin to the sentence "fill the car with fuel and if the car has a flat tyre, fix it". Does this sentence really have less information than "fix the porsche"? We will address this next, but it is obvious that AiG's conclusion is invalid because of these two fundamental problems.

With this more appropriate analogy we can now approach the question of whether Apo-AIM has more or less information than Apo-AI. We will use a formal approach to information theory called algorithmic information theory which is appropriate for comparing systems with multiple functions and will illustrate the case more clearly. We can formally represent an enzymic function as a computer program which performs a given set of functions. In algorithmic information theory, program size is directly proportional to information content, provided that the programs are written in the same language and written equally efficiently. An enzyme with multiple functions must have a longer program to code for its functions, relative to an enzyme which performs a smaller subset of these functions. Hence, an enzyme performing both functions A and B must necessarily have more information than an enzyme which performs only function A or only function B.

We can represent our more reasonable analogy in algorithmic information theory by envisaging wild-type Apo-AI as a robot which has been programmed with the task of filling a car with fuel (in the spirit of AiG's car analogy, as activation of ACAT fills HDLs with cholesterol). To do this task, a program is required to be a certain number of bits long. Now Apo-AIM would be like a robot that has been programmed to fill cars with fuel and recognise and replace flat tyres (an analogy for the anti-oxidant "repair" activity). It is clear this program must be longer than the program that simply fill the cars with fuel. Therefore, according to algorithmic information theory, the longer program has more bits of information.

More technical information can be found on this page on algorithmic information theory, and this page has a critique of Werner Gitt's ideas. Once again we refer readers to this page on Spetner 's formulation of information for specific critiques of Spetner's ideas. It is important to note that Werner Gitt's information theory formulations and Lee Spetner's applications of information theory have not been published in scientific journals and are not peer-reviewed. Currently, they are simply pseudo-scientific concepts, completely unused by professional research scientists. More importantly, however, even using these formulations we can see that Apo-AIM has more AiG-style "information."

Are homozygous Apo-AIM mutations lethal?

In addition, AiG claims that since only heterozygotes of the Apo-AIM mutation have been identified so far,

"This may suggest that the homozygote (both genes the same) A-I Milano mutation is lethal."

Contrary to AiG's claim, this finding is unsurprising given the present rarity of the Milano mutation. The early work studying the heterozygous carriers of this mutant allele identified only 33 individuals, after genetic testing of all inhabitants in an isolated northern Italian village (about 1000 people, 15). In the gene pool of this village, where the mutant allele originated and which has an extremely high concentration of Apo-AIM mutants relative to other human populations, the Milano allele has a frequency of only 1.65% (33 mutant alleles out of 2000 total alleles of this gene). Assuming that those individuals are mating randomly, it would be somewhat surprising to find a homozygous individual in a population of 1000 since we expect to find a homozygote at a frequency of about 1/3700 (the chance of a homozygote is equivalent to 0.0165 squared). However, all 33 of these individuals are known to be descendents of one original 18th century couple carrying the Milano mutation. Humans mate with relatives much less than random, and therefore finding a homozygote in this population of carriers is highly improbable. Thus, based upon our current knowledge of the allelic distribution, there is no reason to suspect that homozygous Milano mutations are lethal [see Note].

Furthermore, transgenic mice have been made that are homozygous for the human Apo-AI Milano gene, and they are healthy (16). In fact, the results indicate that there is a dose responsive benefit for the allele - one mutant allele is better than none, but two are best. The physiology and biochemistry of these transgenic mice (both homo- and heterozygotes) are extremely similar to that found in humans. These facts prompted the authors to conclude that the transgenic Apo-AIM mice are an excellent experimental model system for studying the effects of this beneficial mutation.

The Apo-AIM mutation and lifestyle

AiG further suggests that even if the Apo-AIM mutation is benefical, changes in lifestyle to reduce the risk of heart disease will render the Apo-AIM mutation pointless.

"Needless to say, if someone follows a healthy lifestyle, eats the right things (something like the food pyramid as recently revised by Harvard Medical School, although this could be improved further), exercises, maintains a healthy weight and does not abuse their body by smoking, the A-I Milano mutation will likely be of no use. Epidemiological studies show that heart disease can probably be avoided. "

While it is indeed true that life style changes can reduce heart disease, the original Apo-AIM population was found in Italy. This is the the country that typifies the heart-healthy Mediterranean diet, a diet recommended to lower heart disease (17). Italy has less than half the heart disease found in places like North America. Furthermore, the Apo-AIM bearing population is in a community that has even lower than average rates of heart disease for Italy itself (1, 15). Thus the Apo-AIM mutation is still beneficial even in healthy populations with low risk factors for cardiovascular disease.


AiG claims that the Apo-AIM mutation, which produces a reduction in risk from heart attack and stroke, results in a loss of specificity. However, these claims are incorrect. Instead, Apo-AIM is 1) of a more complex tertiary structure 2) more stable and 3) activates cholesterol efflux more effectively than Apo-AI. Furthermore, Apo-AIM has an antioxidant activity not present in Apo-AI that is sequence and substrate specific. Thus, far from a loss of specificity, Apo-AIM represents a gain of specificity and "information" by AiG's own measures. Contrary to AiG's suggestion, all current evidence indicates that the Apo-AIM mutation is beneficial for its carriers, whether heterozygous or homozygous.


1. Franceschini G, et al. (1980) "A-IMilano apoprotein. Decreased high density lipoprotein cholesterol levels with significant lipoprotein modifications and without clinical atherosclerosis in an Italian family." J Clin Invest. 66, 892-900. [PubMed]



4. Soma MR, et al. (1995) "Recombinant Apolipoprotein A-IMilano Dimer Inhibits Carotid Intimal Thickening Induced by Perivascular Manipulation in Rabbits." Circ Res. 76, 405-11.

5. Calabresi L, et al. (1999) "Cell cholesterol efflux to reconstituted high-density lipoproteins containing the apolipoprotein A-IMilano dimer." Biochemistry 38, 16307-14. [PubMed]

6. Bielicki and Oda (2002) "Apolipoprotein A-IMilano and apolipoprotein A-IParis exhibit an antioxidant activity distinct from that of wild-type apolipoprotein A-I." Biochemistry 41, 2089-2096. [PubMed]

7. Calabresi L, et al. (1997) "Reconstituted high-density lipoproteins with a disulfide-linked apolipoprotein A-I dimer: evidence for restricted particle size heterogeneity." Biochemistry 36, 12428-33. [PubMed]

8. Calabresi L, et al. (1997) "Activation of lecithin cholesterol acyltransferase by a disulfide-linked apolipoprotein A-I dimer." Biochem Biophys Res Commun. 232, 345-9. [PubMed]

9. Franceschini G, et al. (1990) "Apolipoprotein AIMilano. Disulfide-linked dimers increase high density lipoprotein stability and hinder particle interconversion in carrier plasma." J Biol Chem 265, 12224-31. [PubMed]

10. Jia et al (2002) "Thiol-bearing synthetic peptides retain the antioxidant activity of apolipoproteinA-IMilano." Biochem Biophys Res Commun 297, 206-213. [PubMed]

11. Kim SJ, et al., (2001) "Crystal structure of proteolytic fragments of the redox-sensitive Hsp33 with constitutive chaperone activity." Nat Struct Biol. 8, 459-66. [PubMed]

12. Thomas JA and Mallis RJ (2001) "ging and oxidation of reactive protein sulfhydryls." Exp Geront., 36, 1519-1526. [PubMed]

13. Long M (2001) "Evolution of novel genes." Curr Opin Genet Dev. 11, 673-80. [PubMed]

14. Spetner, L. M. (1998) NOT BY CHANCE! Shattering the Modern Theory of Evolution, Judaica Press, New York.

15. Gualandri V, Franceschini G, Sirtori CR, Gianfranceschi G, Orsini GB, Cerrone A, Menotti A (1985) "AIMilano apoprotein identification of the complete kindred and evidence of a dominant genetic transmission." Am J Hum Genet. 37, 1083-97. [PubMed]

16. Franceschini G, Calabresi L, Chiesa G, Parolini C, Sirtori CR, Canavesi M, Bernini F. (1999) "Increased Cholesterol Efflux Potential of Sera From ApoA-IMilano Carriers and Transgenic Mice." Arterioscler Thromb Vasc Biol. 19, 1257-62.

17. Curtis BM, O'Keefe JH Jr. (2002) "Understanding the Mediterranean diet. Could this be the new "gold standard" for heart disease prevention?." Postgrad Med. 112, 35-8. [Pubmed]


AiG has been made aware of some of the errors in its feedback page (as accessed on 17/3/03); after correspondence with Dr. Pirie-Shepherd they made several modifications (accessed on 23/3/03) but did not acknowledge Dr. Pirie-Shepherd. At that time they did not make it clear that the enthusiastic endorsement of the feedback item was made to the previous version. AiG were notified when this page was made public [20/4/03] . Since then they have acknowledged that they have modified the page, that the endorsment was for an earlier version and credited Dr. Pirie-Shepherd. Significant errors still remain.

70% of the enzymes manufactured bind together in pairs (called dimers) and are useless. has become
70% of the enzymes manufactured bind together in pairs (called dimers), restricting their usefulness.

..[T]he enzyme has lost activity for making HDLs. So the mutant enzyme has sacrificed a lot of specificity. has become
..[T]he enzyme has lost activity for making HDLs. So the mutant enzyme has sacrificed specificity.

A paragraph has been added after our page was made public [20/4/03] which incorportates our point about the reason for the lack of human heterozygotes, but curiously omits the information that mice homozygous for the human Apo-AIM gene are perfectly healthy.

As we have seen these changes do not substantially ameliorate the errors, or address the the criticisms in Dr. Pirie-Shepherd's original emails, nor the criticisms in this essay.

AiG also removed Dr. Pirie-Shepherd's email to the original feedback response, which pointed out some errors in assumptions about the reversibility of the dimerisation process. AiG now acknowledge that the correspondence took place, but misrepresent the correspondence. AiG also take exception to Dr. Pirie-Shepherd being a signatory to Project Steve, a tongue-in-cheek parody of a creationist tradition of amassing lists of "scientists who doubt evolution" or "scientists who dissent from Darwinism".


Many thanks to Chris Ho-Stuart, Adam Marczyk, and Michael Hopkins for helpful suggestions, proof reading, and XHTML coding.

Access statistics of these pages counted using WebCounter.

Home Browse Search Feedback Other Links The FAQ Must-Read Files Index Evolution Creationism Age of the Earth Flood Geology Catastrophism Debates
Home Page | Browse | Search | Feedback | Links
The FAQ | Must-Read Files | Index | Creationism | Evolution | Age of the Earth | Flood Geology | Catastrophism | Debates