Evolution of vascular systems

Once upon a time, in Paris in 1830, Etienne Geoffroy St. Hilaire debated Georges Léopole Chrétien Frédéric Dagobert,
Baron Cuvier on the subject of the unity of organismal form. Geoffroy favored the idea of a deep homology, that all animals shared a common archetype: invertebrates with their ventral nerve cord and dorsal hearts were inverted vertebrates, which have a dorsal nerve cord and ventral hearts, and that both were built around or within an idealized vertebra. While a thought-provoking idea, Geoffroy lacked the substantial evidence to make a persuasive case—he had to rely on fairly superficial similarities to argue for something that, to those familiar with the details, appeared contrary to reason and was therefore unconvincing. Evolutionary biology has changed that — the identification of relationships and the theory of common descent has made it unreasonable to argue against origins in a common ancestor — but that difficult problem of homology remains. How does one argue that particular structures in organisms divided by 600 million years of change are, in some way, based on the same ancient organ?

One way is sheer brute force. Characterize every single element of the structures, right down to the molecules of which they are made, and make a quantitative argument that the weight of the evidence makes the conclusion that they are not related highly improbable. I’ll summarize here a recent paper that strongly supports the idea of homology of the vertebrate and arthropod heart and vascular systems.

First, though, some of the differences. Diagram A, below, illustrates some of the major differences between the two kinds of heart, besides whether they are dorsal or ventral. We’re familiar with how our hearts are organized: we have a large muscular pump that drives blood through a closed, high pressure vascular system. Blood is confined entirely to the arteries and veins, and nutrients filter through the epithelial walls of the capillaries to reach other tissues. Insects have an open circulatory system. The ‘blood’, called hemolymph, simply saturates everything, and the heart is a muscular tube that makes peristaltic motions to keep the fluid churning and flowing. It sounds crude and primitive, but it is very efficient.

Blood cells, vascular cells and nephrocytes.
A: Blood cells in closed (top) and open (bottom) vascular
systems. B: Close ontogenetic relationship between cells of the
vascular system, blood and excretory system.

The bottom diagram, B, illustrates a significant closed ontogentic relationship in development. All this means is that a certain linked set of specialized cells are all drawn from the same pool of cells early in development—they are all related by developmental ancestry. In this case the cell types that are all linked are 1) the hemocytes, or various blood cells; 2) the vascular system walls, the endothelia and mesothelia that build blood vessels, and 3) the nephrocytes, or cells that make the linings of the excretory system.

You might be thinking that that third one sounds like it doesn’t quite belong, but it actually makes perfect sense. All of these cells contribute to the maintenance of the coelomic space, the fluid filled cavity lined by mesoderm in which your organs are sloshing around right now. Embryonically, the excretory is a system of tubes through which coelomic fluid flows and is filtered before being lost; our blood vessel system is similarly a system of tubes through which fluid flows, allowing other tissues to extract nutrients. It’s actually not at all surprising that these similar functional elements have a similar developmental origin.

In vertebrates these three tissues are initially derived from the same pool of cells, and in Drosophila a similar phenomenon is observed: the progenitors of the heart (cardioblasts), blood cells (hemangioblasts), and the nephrocytes are drawn from the same pool of mesodermal tissue. That the same developmental relationships between this triad of cell types exists in both lineages is suggestive that we’re also seeing a preserved evolutionary relationship. One could argue, though, that the similarity in embryonic function and their relationship to the coelom imposes a functional constraint, and that’s why vertebrate and invertebrate systems resemble one another in this regard. It’s simply a likely outcome of building plumbing.

It’s when we get into the details of that plumbing that we start to see an accumulation of similarities that cannot be accounted for by mere coincidence, or by the necessity of convergence. Below is a complicated table, but here’s the basic explanation: it’s a parts catalog. It’s addressing the question of how many parts are shared in the construction of the hearts of insects and mammals.

The way it was made was to start with the Berkeley Drosophila Genome Project, the BDGP database. This database allows one to search for genes by their embryonic expression pattern; in this case, all the fly genes that are expressed in the dorsal vessel or heart of the fly, but not in other muscles, were extracted. 62 genes were found in this way that have functions specifically in the heart, but not in other more mundane cellular functions. Of these 62, 53% have vertebrate orthologs, and those genes were pulled out and are listed below.

The left column, CG, contains an identification number which makes it easy to look up in the fly database. All of these genes are expressed in the fly heart, remember. The “Mouse/Human” column identifies the vertebrate ortholog of that gene. The remaining columns list where the gene is expressed in the mouse or human, and the question being asked is whether these fly heart genes also play a role in the formation and function of the vertebrate heart (or blood vessels or kidney; recall that those three tissues are all developmentally related!) If they are, you’d expect many of these genes to be active in vertebrate heart formation, sharing roles in both flies and people. I think you can see the result: over half are found in the vertebrate heart, and others are in those related tissues, blood vessels and kidneys.

CG Fly gene
Mouse/Human Heart Blood vessels Kidney Others
CG1049 Cct1 Pcyt1a/PCYT1A Heart/Skeletal muscles no no others
CG1242 Hsp83 HSPCA/human Heart lymphocyte no others
CG1429 Mef2 Mef2c/mouse
Heart/Skeletal muscles
Aorta, cardiovascular
CG2969 Atet Abcg1/mouse Heart Blood vessels Meso/meta others
CG3132 Ect3 Glb1/mouse
no Bone marrow leukocytes others
CG3171 Tre1 GPR84/human Heart Neutrophils eosinophils kidney others
CG3268 phtf Phtf2/mouse Heart Bone marrow
CG3365 drongo Hrb/mouse
Tumor, placenta
CG3722 Shg*
CG4058 Nep4 Mme/mouse
Heart/Skeletal muscles Leukemic cells Kidney
CG4262 elav Elavl1/mouse
Heart/Skeletal muscles
Breast carcinoma
CG4451 Hs6st Hs6st1/mouse Heart others
CG5408 trbl Trib2/mouse Heart Aorta, lymphocyte kidney others
CG5661 Sema-5C Sema-5A Heart/Skeletal muscles kidney
CG5772 Sur Abcc9/mouse Heart kidney others
CG6281 Timp Timp2/mouse Heart/Skeletal muscles kidney others
CG6605 BicD Bicd1/BICD1 Heart/Skeletal muscles kidney others
CG6811 RhoGap68F arhGap1/mouse
Heart Mammary gland tumor
Platelets, fibrosarcoma
kidney others
Cg7033 Cct2/mouse
B cell lymphoma, platelets others
CG7223 htl Fgfr2/mouse Heart kidney others
CG7524 Src64B Src/mouse Heart/Skeletal muscles Placenta eosinophils kidney others
CG7867 nuf Rab11fip4/mouse Heart Mammary gland tumor others
Cg7895 tin Nkx2-5/mouse
Heart/cardiac muscles
kidney others
CG8049 Btk29A Tec/mouse
Heart placenta
CG9256 Nhe2 Slc9a1/mouse Heart kidney others
CG9579 Annx Anxa6/mouse
Heart Blood progenitors, T lymphocytes
CG10275 Cspg4/CSPG4 no Blood vessels melanoma metanephros
CG11331 Spn27A SERPIN B3 Carcinoma antigen
CG13521 robo-1 Robo-1/mouse Blood vessels others
CG17334 lin-28 Lin-28/mouse Cardiac muscles metanephros others
CG17927 mhc Myh7/mouse Heart/Skeletal muscles others
CG31043 gurh NHS/human no Platelets, lymphocytes kidney others
CG31175 Dys/DmDys Dmd/mouse Cardiac muscles kidney others
CG32858 sn Fscn1/mouse Heart/Skeletal muscles kidney others
*Drosophila E- cadherin, shotgun(shg), is closely related to vertebrate VE-cadherin or paralogous cadherins, which are present in all possible tissues including heart, blood vessels, epithelial cells etc.

I am not entirely convinced by these data, though. One thing lacking is any indication with what frequency any random fly ortholog, one not associated with the dorsal vessel, might also be expressed in the vertebrate heart. Pleiotropy is the rule, as you can see by the “others” column (these genes are also active in many other vertebrate tissues), so I’d like to see some other comparative expression values. I’m more impressed by the correspondence of key regulatory genes, like tinman/Nkx2-5.

The totality of the similarity does add up to a good case for homology between the vertebrate and invertebrate heart, and the authors make a case for the evolutionary scenario illustrated here. It’s interesting that acoelomate flatworms lack a discrete circulatory system, but they do have excretory epithelia—it implies that maybe the heart and blood vessels were cobbled together out of genetic pathways that were first pioneered in the building of the organism’s sewage treatment system.

In the ancestral form, represented by a polychaete worm, a subset of mesothelial cells lining the coelom are specialized to form channels for the transport of fluids. This is where the homology lies, and where the homology is actually a little bit complicated. What’s homologous between flies and vertebrates isn’t the heart, precisely, but the developmental program that sets aside a portion of the mesodermal cells associated with the lining of the coelom and dedicates them to a vascular function.

(click for larger image)

Vascular structure and development. A: Acoelomate (e.g. platyhelminthes). Schematic cross section shows parenchyma, lined by
musculature and intestine. The parenchyma contains various types of freely moving cells, among them the stem-cell-like neoblasts.
Nephrocytes are integrated in the walls of mesodermally derived tubules, called protonephridia, that open to the outside. Drawing on left shows
longitudinal section of protonephridium. B: Coelomate invertebrate (e.g. polychaete annelid). Schematic cross sections of embryo (bottom
left) and adult (bottom right). Blood/vascular cells and nephrocytes are derivatives of the mesothelial coelomic wall, in particular its inner leaf
(splanchnopleura). In the adult the coelom has expanded into the secondary body cavities filling the interior of the animal. Mesothelia line the
inner side of the bodywall and the outside of the intestine. Blood vessels, shown at higher magnification at the top, are formed as clefts between
the basal surfaces of the mesothelia. Blood vessel walls contain contractile myofibrils; nephrocytes are integrated in the vessel wall.
C: Myxocoelomate (open coelom; e.g., insect). Schematic cross section of embryo (bottom left) and adult (bottom right). The embryonic
mesoderm transiently forms coelomata with an inner layer (splanchnopleura) and outer layer. Blood/vascular cells originate from the
cardiogenic mesoderm, located at the junction between the two layers. Vascular progenitors (cardioblasts) migrate dorsally, meet in the midline
and form the myoepithelial dorsal vessel (top). Pericardial nephrocytes align themselves beside the dorsal vessel. D: Vertebrate. Schematic
cross section of embryo (bottom left) and adult (bottom right). The lateral plate mesoderm of the embryo cavitates and forms the mesothelial
walls of the coelomic cavities. Progenitors of the vascular system and blood derive from the lateral plate. The vascular system initially
comprises endothelial cells (yellow). These cells recruit muscle cells (vascular smooth muscle cells; bright green) from the surrounding
mesoderm. The cardiac mesoderm that produces the endothelial layer (endocardium) and muscular layer (myocardium) of the heart forms an
anterior domain of the lateral plate. Excretory nephrocytes derive from the intermediate mesoderm located adjacent to the lateral plate.

Homologies can be very difficult to interpret, and we do have to be cautious about implying too much with a statement of homology, and I think the authors draw the line carefully enough:

In conclusion, comparative morphological and molecular data suggest that many similarities in the development and molecular control of blood and cardiovascular cells in ver tebrates and Drosophila are likely to reflect true homologies: the finding that, for example, blood cells and blood vessel cells in both systems derive from a common pool of progenitors (hemangioblasts) can be easily explained if one assumes that in the last common ancestor, mesothelial cells lining the coelom differentiated into both vascular cells and hemocytes. However, tentative homologies have to be carefully stated; there is no evidence, for example, that the Drosophila heart is any more “homologous” to the vertebrate myocardium, endocardium, or endothelia in general.

I think this is also where Geoffroy erred in his attempted synthesis. He tried to homologize detailed structures in the morphology of vertebrates and invertebrates, basing his ideas on a common root in the structure of the vertebra and trying to shoehorn the invertebrate pattern into a derived vertebrate structure. What we think is going on, though, is that the morphology of both groups is rooted in a generalized ancestral form that lacked anything resembling a vertebra and expressing only the most marginally recognizable rudiment of a heart, and that later forms have modified and elaborated upon that primitive form in radically different ways.

Hartenstein V, Mandal L (2006) The blood/vasculare system in a phylogenetic perspective. BioEssays 28:1203-1210.


  1. #1 Epistaxis
    December 19, 2006

    Etienne Geoffroy St. Hilaire debated Georges Lèopole Chrètien Frèdèric Dagobert, Baron Cuvier

    FYI, your accent marks are all backwards: it should be “Georges Léopold Chrétien Frédéric Dagobert, Baron (de) Cuvier” (“de” is optional in English, but as long as you’re going to the trouble of listing all his names…). Both grave accents (the kind you had) and acute accents (the kind you should have had) are used in French, and they have different meanings.

  2. #2 PZ Myers
    December 19, 2006

    Dang French. Fixed now.

  3. #3 wright
    December 19, 2006

    Fascinating stuff. A great example of how theory is reworked over long periods (in this case, the better part of two centuries) as evidence challenging the original ideas accumulates.

  4. #4 truth machine
    December 20, 2006

    I’m a bit confused here, doesn’t the theory of common descent basically assume that there is a common ancestor?

    Science is based on evidence, not assumption. What would be the value of making such assumption? The theory of common descent is the body of evidence and reasoning that supports the thesis of common descent.

    Do you mean that the theory of common descent has made good predictions that were verified?

    The theory of common descent has produced extensive verified predictions. See, e.g.,

  5. #5 PZ Myers
    December 20, 2006

    All of the above, but what I meant by it was that theory is the lens through which we view data. Scientists in the 1830s were smart and well-educated, but they lacked the particular perspective we now have that allows Geoffroy’s hypothesis to have great power.

  6. #6 David Marjanovi?
    October 2, 2007

    Dang French. Fixed now.

    Léopold is still not fixed, and the de is still missing.

    Very nice post, though! :-)