Immunoglobulin genes. pages 165 - 194

    A quick look through the text (pages  ) will convince you that immunoglobulin genes are quite unusual. In fact it challenges you to decide what a gene is. As a part of the problem of how to make a cell functionally haploid in order to allow evolution to efficiently select on cells instead of whole organisms (recall that evolution proceeds faster but with more waste in haploids), the Ig gene is split into several segments (I think that the term gene segments best describes the situation), which are fused into a single functional transcription unit.

The gene segments are fused such that the joints fall within coding exons rather than in non coding introns. The joints are not able to maintain the reading frame and given a triplet code, only 1/3 of joints will be in-frame and 2/3 will be out of frame.

Applying this simple case to L and H chains, the probability that a cell would have a functional L and H chain is 1/3 + 2/9 (i.e., 1/3x2/3) = 5/9 for each chain, and 25/81 for both chains. In other words, about 31% of cells attempting to produce an Ig will succeed. The remaining ~70% are missing one or both chains and will become dead cells. Note that the genetic cost of loosing 70% of cells that start out wanting to be B cells is not all that great, especially considering how many sperm are wasted. In other words, if the payoff is sufficiently large, evolution is prepared to tolerate considerable waste. After all, better to loose 70% of B cells instead of 70% of organisms!

The story is actually a little more complicated, and the  reason is that among that 30% that are Ig⁺ (i.e., Ig-expressing) a considerable number will produce two L and one H, or two H and one L, even two L and two H chains and will not express a single specificity of Ig. Since selection has to operate at the level of the Ig, cells producing multiple specificities are potential mixtures of anti-S and anti-NS. Such mixtures are potential sources of anti-S and possible autoimmune self-destruction.  Recall that for antibody to function in ridding antigen it has to cross-link the antigen and this requires antibody to have two identical binding sites. Thus mixed molecules with differing paratopes would inhibit rather than induce antigen aggregation.

One way to reduce some of the doubles is to introduce even more errors into the gene fusion process. Adding one more joint adds another ~1/2 (4/9)  to the losses and decreases the doubles a little further. Adding this to both L and H chains brings the yield of functional cells below 10% and ends up not solving all of the problems. As we will see, evolution has to keep track of all of the consequences of the choices for making cells functionally hapliod (hence the name haplotype exclusion for this tweaking of the Ig-genes), and we have not yet gone far enough to see those pitfalls yet. Thus, for Ig H chains there are two gene fusion events and for L chains one gene fusion event per locus.

We need to return to the structure of the Ig molecule and go over more carefully some of the functions of these various gene segments. First, we recall that one gene segment is concerned with Ig specificity and is so call the V-gene segment. There are roughly 300 total V gene segments (the entire regions of human and mouse have now been sequenced) for L and H. Due to random mutations in the germline a number of V genes have stop codons and frameshifts that make them obviously non functional (there must also be cases of muations that lead to a protein, but it cannot pair with it's partner and make an Ig). Not surprisingly a reasonable estimate of the number of functional V gene segments is around 100 - why is this not surprising? Because the gene fusion process results in 1/3  functional joints per chromosome and there would be no way for evolution to select for a higher proportion of functional gene segments if selection was based on the need to have a certain frequency of non functional chains. Thus, whether the V gene is defective or the joint is defective the result is the same. This is a vastly under appreciated fact.

Next we find that between the V gene segment and what is called the C gene (i.e., the relatively constant part of the L or H chain) is another short segment called J (for joining). There are several J segments when there are several V segments (as we will see there are some cases where there is a single V and a single J). We are not at all sure what J segments do for the process. The J segment inserts an intron between J and C and this allows for the further joining to a different C gene without interfering with the V gene segment. The arrangement of gene segments for the H chain, which we recall has several different isotypes (e.g., for IgM, IgG, IgE, IgA, etc.).
 

 

 Notice first the string of C gene segments and recall that any given V-J set of rearrangements give a particular specificity can be associated with any of the C gene segments. Recall that a single messenger RNA encodes the V-J Cm+Cd, and at the time of RNA processing either the product is IgM or IgD. In the case of other C gene segments there is a further gene fusion even that links a "switch" site in the J-C intron to switch sites in front of the more distal C gene segments. Note especially that the fusion that links these switch sites occurs within introns and so causes no non functional joints.  Note also the delta (d) segment is produced on the a new gene segment termed D (it was called D for diversity), which sits between V and J segments. This is the site of the second gene fusion event that we mentioned above as a means of further reducing the number of functional H chains in a cell. There is some controversy over the role of the D gene segment, but one role can be almost certainly eliminated, namely diversity of paratopes. There is enormous diversity of amino acid sequence, but very little of this is found in antibody specificity differences. It seems as if this is a highly promiscuous region of the protein that is important for some other reason than antibody diversity. One further oddity about the D gene segment is that the gene fusion events that link D to J (the first to occur in B cell differentiation) and V to D (the second event), both allow terminal transferase to add random (not quite) nucleotides at the two joint interfaces during the gene fusion event. This additional amino acid diversity in this region is so large as to be unselectable in terms of antibody specificity. This does not mean that most immunologists believe that there is a nearly one-to-one correspondence in amino acid diversity and antibody specificity. Later we will see where the argument is flawed and why. One additional point to make about these H chain gene fusion events is that each gene segment has a characteristic boundary sequence that determines which segments can fuse to which. For example Vh can only fuse to Dh and Dh can fuse to Jh but Jh, but Vh to Jh is forbidden, making the whole gene fusion process highly regulated, but rarely is there any discussion of why.

Recall that there were two light chains (kappa and lambda) that functioned equivalently in antibody. In mouse there is essentially only one V-lambda segment that can link with either of two J-C pairs as illustrated below. In the vast majority of cases V1 links to C1. The kappa locus is almost as complex as the H chain with 300 or so V segments, several J's and a single C. In humans there is roughly a 60:40 split lambda to kappa, and in cow it is the reverse of mouse with almost 100 lambda to 1 kappa V. The wild fluctuation of lambda and kappa light chains in various species is fact, but rarely explained.
 

    In the beginning we talked about L chains as if there were only one when calculating the probability of getting a functionally haploid cell. The presence of two L chains means that are possibly two chains in every cell even if each locus is properly haplotype excluded. The question is how to make sure only one of the two possible loci is engaged in making the L chain to be used. Even if in the mouse kappa always fused before lambda, both would produce a light chain and with one H chain the cell would always produce two specificities.

    There is a strictish order to the gene fusion events leading up to the synthesis of a complete LH immunoglobulin. The first gene fusion is Dh to Jh, then Vh to Dh, and there is probably some kind of "stop" that pauses fusion events if there is a full length H chain that can pair with a pseudo light chain (some times called lambda 5). Next the kappa locus attempts to produce a light chain, and then if that fails, the lambda locus. Since the simple probability of a kappa producing a functional light chain is low, there is a roughly equal chance of the productive light chain coming from lambda or kappa. However, there are many more kappa genes that can be eventually selected by antigen and so the ratio of antbobodies expressing kappa and lambda in serum is close to the ratio of the number of V genes. Some controversy surrounds the gene fusions at kappa as some have claimed that if one V-J fusion fails, then another can occur, and if the J segments are used in order from 5' to 3' then there would be a much higher probability of producing a kappa chain than a lambda chain. One piece of evidence that makes the multiple fusions at kappa believable is that about half of the kappa V segments are inverted so that the looping out normally seen with the H chain genes becomes an enlarged inversion, with no loss of the looped out genes. I would have made little sense to have multiple V-J fusions if the looped out material (all the V segements between the used V and J) was lost. It is still not quite clear why multiple gene fusions at kappa is so advantageous if it does indeed occur at high frequency (one of the dangers in some of these kinds of experiments is that rare events can be amplified out of the background noise because they are being looked for and because they are novel, and they are soon treated as high frequency events in the animal without a shred of direct evidence). Time will tell how this works out.

    Because there are two light chain loci and each could produce a L chain, in order to stop this being a common occurence there is a special "STOP" signal produced when an LH dimer can be inserted into the memebrane. The stop signal shuts down all of the gene fusion system. For this to work, the stop signal has to he fast relative to the gene fusions, which have to be rather slow to ensure that both allelles are not fused before the stop could act.

    All of this tends to make one wonder why there is a kappa and lambda locus that in most species fall into sharply diffferent groups such that there is one locus where the vast majority of V segements are housed and another locus where few V segements are maintained. Human is a rare exception where the numbers of kappa and lambda V segments are roughtly the same.