Eva Jablonka
Museum of Vertebrate Zoology
3101 Valley Life Sciences Building #3160
Berkeley, CA 94720 USA

From: Eva Jablonka
To: Marcello Barbieri
Date: 13 May 2001

Dear Professor Barbieri,

It has taken me a a longer time than I anticipated to read your book, but I have recently finished it and I can now tell you what I think. I like your approach, and in general I think that what you have proposed is very important. I agree with you that there are different types of memory systems (and we do know quite a lot about cellular memory systems), as well as, possibly, different organic codes . I also agree with you that we lack, and badly need, a model of instruction-based complexification, and the type of re-construction model that you suggest is a very useful metaphor for the kind of thing that is required. It is very important that this model makes the role of memory, and of several types of memory, in development, explicit. I have however, many specific reservations and comments.

My first reservation reflects the limitations of my own education (and, I am afraid, also the limitations of many other biologists)- I am not familiar enough with 3 dimensional reconstruction models, and I did not find your description of the model in chapter 3 clear enough, so I had to work really hard to understand more than the outline of the model. I would need a more friendly explanation with a lot of examples and pictures, as well as an appendix with the formalisms which will allow me to work hard in my spare time, without stopping me and hampering my understanding while reading.

My other problems are more fundamental. You suggest that there are splicing codes, signal transduction codes, apoptosis codes, and assembly codes at the cell and tissue levels. I tend to agree with you that organic codes may exist, but I don't think that we have very good empirical reasons for this belief at present. As you have noted, in order to decide that there is a code we need more than a correspondence between two types of molecules (or "worlds'). Every allosteric enzyme forms such a correspondence. Almost every type of regulator is forming a specific link between two types of entities and imparts functional meaning on the system. Yet not every regulatory system with regulatory molecules that form specific correspondences between types of molecules (e.g. a protein forming a correspondence between DNA sequences in the regulatory regions of an operon and particular metabolites) is a code. So the correspondence and the functional meaning that it imparts is not sufficient. The genetic code is a code because there is a set of generative "system rules": every three nucleotides in a row are a unit (codon), the combinatorial possibilities are well defined as are the correspondences to amino acids, and most importantly - the codons that belong to the circumscribed set are read in a linear manner, from a defined start point up to a defined end point. I think that in order to argue that there is a splicing code or any other code, we have to show first that system-specific regularities exist, and moreover that these rules are generative, so that with a set of signals in one system we can get a corresponding set of signals or acts in another system according to the generative rules of system. So, I think that it is not even enough to have rules - they have to be generative. Have we discovered thousands of different splicing non-generative specific regularities (sort of limited "rules"), I would not think of splicing as encoded information. Again, I agree with you that there may be splicing codes and other codes as well, and that we should very actively look for them, but I think that you have to specify more clearly what it is that we should be looking for. You may disagree with my specific requirement (the need for generative rules), but I think that it is important to distinguish between a system of correspondences between two types of entities that is not a code and one that is a code.

I had problems with your treatment of epigenetic cellular memory. We know a lot more about it than you suggest. You did not mention the various epigenetic cellular memory (or inheritance ) systems that we now understand in some detail. Marion Lamb and myself have written a lot about it, and have distinguished between three types of cellular heredity systems: chromatin marking systems (e.g. the methylation system), systems of self sustaining loops, and systems of structural inheritance (as manifest in ciliates and in prions). The discussion of these systems reveals the complex relations between memory and coding. Memory does not imply encoding, although it may be related to it, of course. My understanding of cellular epigenetic memory suggests that epigenetic memory (it would be more correct to call it inheritance at the uni-cell level) is very ancient, and appeared with the first self perpetuating proto-cells. I do not agree with you that cellular memory (such as that associated with determination) was an invention of multicellular organisms, although, of course, it was a precondition for multicellularity. There was temporal differentiation in unicelles that depended on epigenetic inheritance that became spatial differentiation as well, with the advent of multicellularity. Of course epigenetic memory further evolved and became more sophisticated during the evolution of multicellular organisms, but I think we have good reasons to think that cellular memory is very ancient (all the cellular heredity systems we know in multicellular organisms are found in unicellular organisms). What has evolved are sophisticated control of "remembering" and "forgetting". A more specific point - I do not think that pattern genes are genes that contribute to cell memory - there is of course cellular memory associated with these genes (they clearly show chromatin-marking type of memory of their functional state) but they are not memory genes. If something can be called a "memory gene" (and this is a very very problematical concept) it the gene coding for methyltransferase. Other genes involved in memory function are genes coding for certain types of regulators that lead to self perpetuating properties of the system. Usually memory is not a property of a gene but a of a cellular network of genes and their products. The evolution of cellular heredity is not something we know much about but there are some speculations in the literature. Clearly the control over cellular memory in germ cells was crucial.

I liked your fifth chapter and your idea of the ribotype. I think that your idea to focus on the evolution of the cell via the code, through the mediation of riboproteins is very ingenious, it makes a lot of sense to me. However I do not think that you can establish natural conventions without natural selection. How else can you systematically change your statistical proteins? I liked very much your story at the end of chapter 5 - it is a very good way of explaining the central role of the agent in evolution (and a very good argument against the primacy of the selfish gene or the selfish meme in evolution!)

A small reservation -I disagree that replication is simpler than metabolism. Ganti's chemoton model shows that you can have a metabolic self-sustaining, growing, multiplying and even evolving unit, even when at the first stages it is lacking protein catalysts.

I had no problems with chapter 6, except the one I mentioned (about the inference of codes from specificities and correspondences ) and your reconstruction made sense to me. I had many problems with your explanation of the Cambrian explosion. First, I think that to assume that the first pre--Cambrian multicellular organisms were genetically determined and had no epigenetic mechanisms that lead to various types of plastic adaptations and to memory is wrong, in view of what we know of all extant cells and all extant organisms (unicellular and multicellular): all have cellular memory (or epigenetic inheritance) systems. Second, I do not accept that plasticity always evolves from determination - very often the opposite occurs. I think that often more stereotyped responses evolve from more plastic and fuzzy ones, and that the rigidity of the components parts, fixed by hard wired genetic mechanisms allows plasticity at a higher level of biological organization, that of the combinations among them.

I have the same intuition that you have, that in addition to neural and immune memories there must be other types of memories, such as 3D memories, probably mediated by the extra cellular matrix (Ettinger and Dolzhanski had an interesting paper developing this idea). I don't think the body plan is a memory, but it seems like the manifestation of an operating 3D structural memory system. We do not understand it at all however. So although I do not accept your stages of the Cambrian explosion and your explanation of it, I do accept that supra-cellular 3D memory systems must have been involved.

I have some reservations about universal grammar, but although I collaborate with a linguist, this is not my field and it is not crucial to your argument ( I accept that linguistic classification schemas are partially innate). I have a lot of sympathy with your ideas and as I said in the beginning I think that memories and codes are crucial to our understanding of the development and evolution of organisms. I suppose that you sent me your book because you anticipated that I will be both interested and critical about it because of my interest in cellular heredity and the evolution of information and memory. I suppose that you are aware of my work but some of the arguments that I have advanced in this letter may be more clear in view of what I have written elsewhere, so I shall enclose a short list of references that you may find useful ( as well as the Ettinger and Dolzhanski reference).

Ettinger L and Doljanski F. (1992) On the generation of form by the continuos interactions between cells and their extracellular matrix. Biological Reviews, 67, 459-489.
Jablonka E. (1994) Inheritance systems and the evolution of new levels of individuality. Journal Theoretical Biology, 170, 301-309.
Jablonka E. and Lamb M.J. (1995) Epigenetic inheritance and Evolution: The Lamarckian Dimension. OUP, Oxofrd.
Jablonka E. and Szathmáry E. (1995). The evolution of information storage and heredity. Trends in Ecology and Evolution, 10, 206-211.
Jablonka E., Oborny B, Molnár E., Kisdi E., Hofbauer J., and Czárán T. (1995) The adaptive advantage of phenotypic memory. Philosophical Transactions of the Royal Society, London B., 350, 133-141.
Lachmann M. and Jablonka E., (1996) The inheritance of phenotypes: an adaptation to fluctuating environment. Journal of Theoretical Biology, 181, 1-9.
Jablonka E. and Lamb M.J. (1998) Epigenetic inheritance in evolution (A target article). Journal of Evolutionary Biology, 11, 159-183.
Jablonka E. and Lamb M.J. (1998) Genic-Neo Darwinismis it the whole story? Journal of Evolutionary Biology, 11, 243-260.
Regev A., Lamb M.J., and Jablonka E. (1998) The role of DNA methylation in invertebrates: developmental regulation or genome defense? Molecular Biology and Evolution, 15, 880-891.
Jablonka E., Lamb M.J., and Avital E. (1998). Lamarckian mechanisms in Darwinian evolution. Trends in Ecology and Evolution, 13, 206-210.
Lachmann M., Sella G., and Jablonka E. (2000) On the advantages of Information sharing. Proc.Roy.Soc. B 267, 1287-1293.
Dor D. And Jablonka E. (2001) From cultural selection to genetic selection: a framework for the evolution of language. Selection, 1-3, pp. 33-57.

I hope that you find these comments (and maybe some of the references) useful.

All the best,

Eva Jablonka.

From: Marcello Barbieri
To: Eva Jablonka
Date: 19 May 2001

Dear Eva

Thank you for accepting to correspond on a first name basis, and above all for your beautiful and thoughtful letter about The Organic Codes. Let me concentrate, for a start, only on a few points.

(1) "I did not find your description of the model in chapter 3 clear enough.....I would need a lot of examples and pictures, as well as appendix with the formalisms".
I totally agree. The 1974 model was financed for the Brookhaven Workshop, but afterwards it fell out of favour. Who in his right mind would perform a tomography from "incomplete" information? I pointed out that the problem was "theoretically" important for biology, but the mathematicians insisted that the funds should come from biologists, and the biologists from mathematicians. You may say that I did not get in touch with the right people, with those who do take seriously the complexity problem of biology. For example? For example the Santa Fe people! Yes, indeed. As a matter of fact I did write to Stuart Kauffman. The answer?....that he is too busy. See what I mean?
Anyway, do not worry. There are still modest people who do have time for new things, and soon I hope to start a nice little research project that will give you all the examples and the pictures that you are looking for.

(2) "I agree that there may be splicing codes and other codes as well, and that we should very actively looking for them, but I think you have to specify more clearly what it is that we should be looking for".
As a matter of fact, I did: we should be looking for "adaptors". It is they that make the difference between catalyzed and codified reactions, and so it they that we should dig out and bring to light.

(3) "You did not mention the various epigenetic cellular memory (or inheritance) systems that we now understand in some detail. Marion Lamb and myself have written a lot about it".
I plead guilty. I have discovered your work only recently, and I sent you my book precisely in order to start a dialogue. In my opinion, epigenetic cellular memories are more general than inheritance systems, but perhaps we should discuss this when I know more about your work (reprints would be welcome).

(4) "I disagree that replication is simpler than metabolism".
This is the conclusion that Freeman Dyson reached in his beautiful little book Origins of Life (1985,Cambridge UP), and I agree with him. I actually adopted his model as a starting point in Chapter 5 of The Organic Codes, and I would have dearly liked his opinion, but so far he has been silent (perhaps you should re-address the question to him).

(5) "I had many problems with your explanation of the Cambrian explosion.....to assume that the first pre-Cambrian multicellular organisms were genetically determined and had no epigenetic mechanisms....is wrong".
Here you are right and wrong at the same time. When Galileo addressed the problem of motion, he pretended that friction does not exist, and only in that way he could "see" that motion can go on forever. Was he right about friction? No. Was he right about inertia? Yes. So, you see, sometime you "have" to make an absurd hypothesis to see the truth.
The genetically determined pre-Cambrian animals did have epigenetic mechanisms, but if you ignore them (as if they were a sort of biological "friction") you are in a better position for seeing what did happen during the explosion.

I am looking forward to the next round. After a 20 year-long isolation, it's nice to start talking again.

Yours
Marcello

From: Eva Jablonka
To: Marcello Barbieri
Date: 20 May 2001

Dear Marcello,
I am very sorry (but not very surprised) to hear that models of reconstruction from incomplete information did not get funded. I hope that you will succeed now. It will be most important to relate such models to the cellular system and embryology.

As to your answers to my comments: First the finding of an adaptor is not sufficient in my opinion. You have to specify in what way an adaptor is different from any allosteric enzyme which can be considered an adaptor too (an allosteric enzyme is not a simple catalyst). But let me clarify what I mean by imagining an adaptor t-RNA molecule but no code. Let us say that there is a t-RNA that is forming a correspondence between a particular amino acid and a particular triplet, but it is doing so only under very special conditions. In other words, there is a very very high degree of context dependence in the system. Let us even assume that you have a system of 4 triplets and 4 amino acids that are specifically related by four specific t-RNA's. Yet the high context dependence will lead to the 4 codons being rarely (or never !) decoded together. Calling such a system a code, even when it has molecules that can be called "adaptors" is stretching the concept. Coding must involve a certain level of context independence in the encoding and decoding processes, a situation that is not simple to evolve in highly functional systems that are very context-sensitive. But that is not all. A single t-RNA adaptor that forms a correspondence between an amino acid and a triplet even in a context-independent manner is not a code, for we need a system of such adaptors. How many then? What is the threshold that makes a regulatory system of such a type into a code? Do we have to assume that encoded information is not displayed or expressed? At what level? I think that you have to address these questions.

The relationship between epigenetic inheritance systems and cellular memory is not simple, and you can have cellular memory in non-dividing cells so cellular multiplication is not mandatory for cell memory although it is for epigenetic inheritance. The relations between inheritance and memory are very important and very interesting, and crucial to the understanding of many aspects of evolution, for example the evolution of soma-germ line distinction, and of course for understanding higher level systems, such as the immune system, neural memory and cultural transmission. As I pointed out in my previous letter, there is good evidence for ample cellular memory and epigenetic inheritance being very ancient. So back to the Cambrian explosion - I do not see how you can ignore cellular memory and plasticity when you discuss the Cambrian explosion, and then argue that it as the evolution of memory and plasticity that drove this explosion! What you are suggesting is therefore not a fruitful idealization. I will be very interested to hear your opinion about the ideas Marion Lamb and I have been developing. Unfortunately I am now in the USA and do not have reprints of my paper (they were left in Israel), but most of the journals in which they were published are easy to access, so I hope it will not be a problem (there is a book too that we wrote that maybe in your library).

I read Dyson's delightful little book, and I see his point, but his model does not involve the kind of interrelated processes that Ganti is talking about, processes that result in a basic protocell. I am not sure that the Ganti type of system will not be stable over time and would not be able of some (limited) evolution. But more important - in what sense is the Dyson system more complex than replication? It is more complex in the sense that it involves many types of molecules and many interactions. But as Dyson and as you have argued it may be easier to generate than the replication system, and may be the precondition for replication. If the conditions leading to replication are more demanding than those for metabolism and if replication require a complex metabolic system, the simplicity of replication is an illusion. But I think that we may be arguing about words here, since like you, I agree with Dyson's arguments. By the way, did you read Iris Fry's new book on the origin of life? It is very good and I think you may like it.

All the best,
Eva

From: Marcello Barbieri
To: Eva Jablonka
Date: 21 May 2001

Dear Eva Thank you for your solidarity about the importance of funding the new mathematics. On June the 6th I will present a project to the Lausanne Polytechnic, where there seem to be some interest. Fingers cross. And now for your points.

(1) "The finding of an adaptor is not sufficient in my opinion. You have to specify in what way an adaptor is different from any allosteric enzyme".
I did not speak of "an adaptor" but of "adaptors". I stated clearly that adaptors have "collective" properties, not individual ones. The difference with an allosteric enzyme is that any adaptor can make a vast plurality of couplings (this can be tested). It is the choice of a whole set of adaptors what gives in fact origin to a code (the "logic" of these multiple choices is an issue that has not been studied yet, and represents an entirely new field of research).

(2) "As I pointed out, there is good evidence for ample cellular memory and epigenetic inheritance being very ancient".
I went even further when I proposed that epigenesis and organic memories are intrinsic characteristics of every single cell, and so must have been present from the very beginning (the phenotype is always more complex than the genotype, and so every cell is a system that is reconstructed from incomplete information). We may however have different ideas about organic memories. For this I will need to get more familiar with your work.

(3) "If the conditions leading to replication are more demanding than those for metabolism, the simplicity of replication is an illusion".
I accept, with Dyson, that replicators could have formed in primordial "test tubes", just as Dawkins and Dennett are saying, and then notice that those systems could not go anywhere. Replicators did exist, but had no future: this is what the laws of error catastrophes tell us. The simplicity of replication is not an illusion, because it can be proved in a test tube. It is simply worthless, because it is "thermodymically" incapable of producing a cell.

Thank you for the tip about Iris Fry's new book. You cannot imagine how difficult it is for me to get hold of books and papers. I do envy your privileges.
Take care

Marcello