To return to the home page, click here
*Aneurysm Paper of the Month - December 1996*
I'm doing something different this month, as I have not
previously used the Papers of the Month feature to make
a preprint available; but as I am on Holiday and left my
December journals at the office, I'm giving myself a
shortcut. My friend and colleague at Cornell, Gary
Fantini, recently asked me to contribute a chapter for an
upcoming SURGICAL BIOTECHNOLOGY INTERNATIONAL which he is
editing. I have written up the following manuscript for
him. It is not complete in two respects, but I think
you'll get the idea. Figures 4 and 5 are not available,
as I don't have a color scanner at home. Also, I work in
WP51, which will not convert the references to DOS text.
For the complete paper with references and all figures,
please watch for the "Aneurysm" issue of SBI.
THREE-DIMENSIONAL MODELING AND MOLECULAR
EVOLUTION OF MATRIX CELL ADHESION
MOLECULE #2: "MAT-CAM 2"
December 27, 1996
M. David Tilson, MD
Shichao Xia, MD
Hitoshi Hirose, MD
Kathleen J. Ozsvath, MD
James Knoetgen, III, MD
David K. W. Chew, MD
Chenzhong Fu, MD
From Columbia University and the Department of Surgery
of the St. Luke's / Roosevelt Hospital Center, New
York, NY
Corresponding author:
Dr. M. David Tilson
St. Luke's / Roosevelt Hospital Center
1000 Tenth Avenue, New York, NY 10019
Telephone: 212-523-7779
FAX: 212-523-6495
Supported by the Special Fund for Resident Research,
Department of Surgery, St. Luke's /Roosevelt Hospital
Center
THE DISCOVERY OF AAAP-40
Our research group has recently reported the
partial amino acid sequence of an aortic adventitial
elastin-associated fibrillar protein, which may be the
target of an autoimmune response in patients with
abdominal aortic aneurysms (AAA). Its apparent
molecular weight is approximately 40 kDa, so we have
named it aneurysm-associated antigenic protein - 40 kDa
(AAAP-40). It has similarities to microfibril-
associated glycoprotein-36 kDa (MAGP-36), which was
reported by Kobayashi et al to have a tissue
distribution limited to the aorta in pig; unlike other
microfibrillar proteins which appear to distribute
ubiquitously throughout the body with elastin.
Accordingly, if AAAP-40 is the human homolog of MAGP-
36, it would explain how an autoimmune reaction against
this protein might have consequences more or less
limited to the aorta and its branches.
CLONES 1 AND 5 HAVE IMMUNOGLOBULIN DOMAINS
An expression library was made from mRNA of the
human aortic adventitia and screened in an effort to
clone the cDNA for AAAP-40. Two of the first five
clones (#1 and #5) have novel features and strongly
resemble each other in domain structure. While both
have some features that occur in the families of
microfibril-associated glycoproteins and other matrix
proteins (like collagen), their most remarkable
property is that each begins with a lengthy amino acid
sequence (~100 residues) that is highly homologous to
members of the immunoglobulin kappa family. The
probability that this high degree of similarity is due
to chance alone (as calculated by BlastP) is in the
range of 1 x 10-50.
These are novel hypothetical proteins, as potentially
tissue-specific, elastin-fibril associated matrix
molecules with immunoglobulin domains have not been
previously described. By analogy to the membrane-bound
cell adhesion molecules (CAMs), we propose the name
Matrix Cell Adhesion Molecules (MAT-CAMs) for these
unique clones. Hereafter, Clone 5 will be referred to
as MAT-CAM 2; since it is the second member of this new
family to be discovered.
THREE-DIMENSIONAL MODELING OF MAT-CAM 2 (CLONE 5)
In a separate communication (presently under
editorial consideration) we are reporting the three
dimensional modeling of Clone 1 (MAT-CAM 1).
Interestingly, the N-terminal 124 residues form a
sandwich of two antiparallel beta sheets,
characteristic of the immunoglobulin superfamily. The
structure of the N-terminal 124 residues of MAT-CAM 2
was modeled by SwissModel on 11MIG.PDB, which was the
most similar immunoglobulin light chain for which a
template was available (Blastp p = 1.3 x 10-31) and
displayed in three-dimensions by RasMol. The
antiparallel beta sheet sandwich of MAT-CAM 2 is
illustrated in
Figure 1.
Figure 2 shows a pair-wise comparison of the
modeled sequence of MAT-CAMs 1 and 2 (as computed at
the Biologist's Control Panel) and displayed by
LalnView. There has obviously been substantial
divergence in sequence, although the structure of the
immunoglobulin sandwich is preserved. This result
confirms again what has come to be seen as a common
theme in molecular evolution; namely, that structure is
much more highly conserved than sequence.
MOLECULAR EVOLUTION OF THE MAT-CAMS
In communications pending editorial review
elsewhere, we have calculated the PAM distances (point
accepted mutations per 100 amino acids) with AllAll for
the MAT-CAMs, the most similar immunoglobulins, the
microfibril-associated glycoproteins, and their
presumed fibrinogen-like ancestor: fibrinogen-related
protein-A precursor (SP|P19477) of the invertebrate
Echinoderm sea cucumber. For fibrinopeptide-B related
proteins, a PAM unit has been estimated to be about 1
million years.
Figure 3 which is drawn roughly to scale,
illustrates some of these distances. One might surmise
that there was a cluster of evolutionary activity among
the matrix molecules around the time of the Cambrian
radiation, followed much more recently by a cluster of
activity around the time of appearance of soluble
immunity in the cartilagenous sharks.
ARE THESE PROTEINS FOR REAL?
A concern as this story began to unfold was that
what we are calling the MAT-CAMs might be an artifact,
based on a highly improbable but not impossible
laboratory curiosity. If separate immunoglobulin
cDNA's and matrix cDNA's somehow got scrambled and
fused during construction of the library, the results
would of course be trivial. Two recent findings
suggest that the hypothetical proteins may be authentic
gene products. First, Figure 4 shows the result of an
immunohistochemical experiment which compared the
binding of antibody against Ig-kappa to the elastin-
associated fibril of an aneurysmal aorta (as compared
with the binding of an antibody against Ig-heavy
chain). The binding of anti-Ig kappa to the matrix
suggests that the putative MAT-CAMs are authentic
connective tissue proteins. Second, Figure 5
illustrates an in-situ hybridization of labeled clone 1
with a section of aortic wall. The observation that
binding occurs to a mesenchymal cell instead of a
lymphocyte buttresses the argument for authenticity.
Preliminary studies with a limited number of AAA
patients and normals suggest that MAT-CAMs 1 and 2 may
also be autoantigens like AAAP-40, but substantially
more work must be done prior to considering that
conclusion to be secure. However, it is tantalizing at
the moment to speculate that there may be a whole
family of aortic antigenic proteins related to the
development of aneurysmal diseases.
LEGENDS FOR FIGURES
Figure 1 - Swiss-Model of the three-dimensional
structure of MAT-CAM 2 revealing it to have an
immunoglobulin-like sandwich, formed by two
antiparallel beta sheets, connected by a helix
between the e and f strands.
Figure 2 - Pair-wise comparison of MAT-CAMs 1 (Clone 1)
and 2 (Clone 2), as displayed by LalnView, with color
coding for the regions of greatest similarity in amino
acid sequence. The three dimensional structures have
been conserved despite substantial divergence in
sequence.
Figure 3 - An evolutionary tree, roughly scaled in PAM
units (point accepted mutations / 100 amino acid
residues), as computed by AllAll. For proteins of the
fibrinogen family, a PAM unit has been estimated to be
about 1.1 million years.
Figure 4 - Histological section of aneurysmal aorta
probed with antibodies against immunoglobulin kappa and
against immunoglobulin heavy chain. Anti-Ig kappa
localizes to the adventitial elastin-associated fibril.
Figure 5 - In-situ hybridization of clone 1 against a
histological section of normal human aorta. The
predominant cell type positive for hybridization in
this section is mesenchymal, although further studies
would be required to differentiate whether the positive
cells are fibroblasts or smooth muscle cells.
REFERENCES