Sunday, November 28, 2010

Golden oldies: Sarich and Wilson

I introduced the topic of molecular phylogenetics to students by looking at some classic papers in the field. I want to summarize a few of those (by Vincent Sarich and Allan Wilson) here. This will give me a chance to show some nice pictures like the montage above.

The images are from wikipedia. Top-left is Colobus, an "old world" monkey. On the right is Cebus, a "new world" monkey. And bottom left is a Lemur. Just for fun, here is Vincent Sarich himself.

I found the image here.

I'm sure you know the story, but just to remind everyone, there is a huge amount of geological evidence that the American continents have moved with respect to (particularly) the African continent. I'm not sure of current estimates as to the date of separation but one reference says about the end of the Cretaceous period (65 Mya). (There is apparently an argument about whether a land bridge remained for an extended period of time). Here is what the world is thought to have looked like a little while before that:

found it here, with a USGS credit

Separation enforced separate evolution of lineages giving rise to modern old and new world monkeys, with the hominid lineage separating from that of the old world monkeys at a substantially later date. Here is a figure illustrating the point from the 1967 PNAS paper (Sarich and Wilson 1967 PMID 4962458).

It's a classic example (perhaps the classic example) of allopatric speciation, with speciation happening after geographical isolation by barrier formation (the Atlantic Ocean). Image from here:

In a series of papers, Sarich and Wilson described one of the first "molecular" phylogenetic studies. To be clear, it is primarily based on a relatively complex immunological test, but antibodies are molecules even if this is not a sequence-based method. Our goal for today is to understand the test, and look at some of the data from the first paper (Sarich and Wilson 1966 PMID 4958934).

The assay is a micro complement-fixation assay, which is slightly funny since the assay volume is 7 ml---massive by modern standards. It's an immunological assay involving antibody and antigens. The antibody in this case is rabbit antiserum from animals immunized with purified human serum albumin. The antiserum is diluted (between 1:1000 and and 1:11000) and incubated with varying amounts of antigen---serum albumin. In the first part of the paper they use purified human protein, and then later they use serum containing albumin from a variety of primate species.

It isn't clear to me what fraction of the 7 ml is antiserum, but they say a curve (8 points) requires 1 ul of antiserum, so it isn't very much.

The test used here is a complement fixation assay. The readout at the end of the test is lysis of sheep red blood cells (RBCs) by the membrane-busting action of complement, after the RBCs have been "sensitized" by pre-treatment with Ab (a different Ab, directed against surface antigens of the RBCs). Lysis releases soluble hemoglobin into the test tube, which can be quantified using a spectrophotometer.

The more lysis observed, the more active complement was present. If the rabbit antibody to human serum albumin finds available antigen and binds to it, that activated antibody can bind to complement, which removes it from the system, so that it will no longer promote lysis of the RBCs. In contrast, antibody molecules alone, or antigen (human serum albumin) molecules alone, are not able to bind to complement. Complement "fixation" (and thus, antibody-antigen complex formation) is measured by a decrease in the observed lysis of RBCs. So far, so good.

Now, in typical immunological fashion, the reaction between antibody and antigen depends on both the concentration and the relative amounts of the two components. The dependence on relative amounts arises because antibodies are multivalent (divalent for IgG), and the antigen used here has multiple epitopes to which the many different kinds of antibodies in the antiserum may bind as well. In antibody-excess (left part of the curve; top of the figure below) most complexes are of single antigen molecules bound to multiple antibodies; in antigen-excess (right part of the curve, bottom panel of figure) most complexes are of single antibodies bound to multiple antigen molecules.

Aggregates containing many molecules of each type are formed when just the right ratio of antibody to antigen is present. In a "macro" assay, this would be observed as a precipitation, observable by eye. Here, for some reason only the networks are efficient at "fixing" complement, that is, removing it from the system. The reason seems to be that multiple Fc chains (the magenta rods in the figure) need to cooperate to "fix" the complement. I've shown the free complement as red triangles, and the bound complement as red squares, because a conformation change occurs upon binding, and the complement component (C1q according to this) is sequestered (nice animation here). I'm a little uncertain as why this disables the complement activity against the sheep RBCs, since the point of binding to the antibody is to do something, but that seems to be the story.

In any case, the use of complement gives a very large amplification of the signal from each antibody-antigen complex, because loss of each "bit" of complement by binding to our complexes causes a large loss of the RBC lysis signal. (Sound complicated: it is! Is it linear with evolutionary distance? Who knows?)

The sharpness of the peaks has been accentuated by a logarithmic transformation of the x-axis.

Not surprisingly, the reaction of these rabbit antisera to human serum albumin is very strong, and can be observed at a substantial dilution (1:11000). At this highest dilution only human and chimp proteins give a visible signal in the assay. Quite small adjustments (panel b antiserum is 2.5 x concentrated, panel c is 4.4 x), result in observation of reactivity for the purified serum albumins from other species.

My inclination might have been to compare the position of the peak on the x-axis with a purified standard antigen to that observed with the serum samples (below), which seems like it would be subject to a fair amount of error. But what these authors do is focus on the height of the peak. The difference between peak heights for human and chimp, say, is easily measured.
The chimpanzee curve has a peak whose height is 67 percent as great as that given by HSA.

The analysis proceeds by plotting the peak height observed for different dilutions of antiserum. (Log-transformed) values give a straight line. The difference between antigens is summarized by taking the difference in dilution which yields 50% fixation. Guesstimating from the figure 50% fixation occurs for human at about 1:9500, for Rhesus at about 1:4000, so the "index of dissimilarity or immunological distance" between them is 9500/4000, approximately 2.4.

They also showed that reciprocal tests (anti-human SA with chimp SA compared to anti-chimp SA with human SA) gave comparable results. Then, the real interest was to extend the test to a number of different species, which was done by using not purified antigen, but antigen in serum samples.

The ID observed using rabbit antiserum to human serum albumin combined with serum from various species was:

human              1.0
chimp 1.14
gorilla 1.09
old world monkeys 2.23-2.65
new world monkeys 4.2-5.0 *
prosimians 7.6-18
* one exception

The species studied were:

Homo sapiens
Pan troglodytes
Gorilla gorilla
Pongo pygmaeus
Hylobates lar
Symphalangus syndactylus

Cercopithecoidea (Old World monkeys):
Macaca mulatta (Rhesus)
Papio papio
Cercocebus galeritus
Cercopithecus aethiops
Colobus polykomos
Presbytis entellus

Ceboidea (New World monkeys):
Aotes trivirgatus
Ateles geoffroyi (spider monkey)
Saimiri capucinus
Callicebus torquatus

Prosimii (Prosimians):
Tarsius spectrum
Galago crassicaudatus
Nycticebus coucang
Lemur fulvus
Tupaia glis

Bos taurus
Sus scrofa