Reprints Available Directly from the Publisher Photocopying Permitted by License Only Serum Amyloid P Component (sap)-like Protein from Botryllid Ascidians Provides a Clue to Amyloid Function

The HA-1 lectin isolated from Botrylloides leachii has an amino acid composition similar to that of mammalian serum amyloid protein (SAP). SAP is a universal component of mammalian amyloid deposits. Like SAP, HA-1 has a disc ultrastructure, and antibody to HA-1 binds both (a) to amyloidlike fibers deposited between rejected Botrylloides colonies and (b) to cerebral amyloid deposits in Alzheimer's disease brains. Deposition of protochordate amyloid within rejection sites and surrounding fouling organisms implies that these fibers function as barriers to allogeneic and infectious challenge. Similarly, mammalian amyloid may also function to contain inflammatory lesions and to limit the spread of certain infections. Pathological amyloidotic conditions in humans, such as Alzheimer's disease, may result from unregulated expression of this primitive encapsulation response.


INTRODUCTION
For some years, animals belonging to the protochordate genera Botryllus and Botrylloides have been the subjects of intensive study by immunologists concerned with the evolutionary origins of the vertebrate immune response (Weissman et al., 1990). These colonial ascidians grow on hard substrata in communities where space is a limited resource and where colonies frequently come into contact. When this happens, juxtaposed colonies undergo vascular fusion to produce a genetic chimera or initiate rejection reactions with associated necrosis at the contact site in the tissues of each partner. Fusion or rejection is controlled by a single polymorphic histocompatibility gene locus (Oka and Watanabe, 1957;; fusion occurs when one allele at this locus is shared, whereas rejection follows contact between colonies sharing no alleles. In an examination of the ultrastructure of necrotic areas in the contacted tissues between *Corresponding author. rejecting colonies of Botryllus primigenus, Watanabe (1973a, 1973b) described accumulations of stiff, hollow fibers with a characteristic diameter of around 200 angstroms. These ultrastructural features are reminiscent of polymerized serum amyloid P component (SAP), which is a universal component of mammalian amyloid deposits, and of a type of fiber observed in electron micrographs of some such deposits (Skinner et al., 1982;. Like the closely related C-reactive protein (CRP), SAP is a member of the highly conserved pentraxin family of serum and tissue proteins (Coe, 1983). Molecules belonging to this family have been identified in several representative vertebrate classes and in an invertebrate of the protostome lineage, the horseshoe crab Limulus (Robey and Liu, 1981). Until now, no pentraxin molecule has been identified from a deuterostome invertebrate.
The pentameric HA-1 lectin isolated from the hemolymph of Botrylloides leachii (Coombe et al., 1984a;Schluter and Ey, 1989) has subunit molecular weight and amino acid composition characteristics similar to those of prototypical vertebrate pentraxins (Schluter and Ey, 1989).
Because of these similarities and the suggestive 68 V.L. SCOFIELD et al.
ultrastructure of fibers deposited between rejected Botryllus colonies Watanabe, 1973a, 1973b), we set out to determine (a) whether HA-1 shows the disc ultrastructure of vertebrate pentraxins, (b) whether rejection areas and their contained fiber structures contain HA-1, and (c) whether antibodies to HA-1 identify cross-reacting determinants in mammalian amyloid deposits. Here we report (a) that HA-1 molecules show a typical pentraxin ultrastructure upon examination with the electron microscope after negative staining; (b) that antibodies to HA-1 show concentrated binding to tunicate rejection areas by light immunohistochemistry, and exhibit specific binding to rejection fibers by immunoelectron microscopy; and (c) that antibodies to HA-1 bind to mammalian amyloid in patterns identical to those of anti-SAP on the same material. The evidence presented suggests that allogeneic rejection in tunicates initiates a true amyloidotic process. Further, the concentration of amyloidlike fiber masses within rejection barriers, and the presence of HA-1 epitopes in these regions, suggests that the fiber network functions for separation of inflammatory foci from surrounding healthy tissue. These observations have interesting implications for the function of amyloid in mammals. upon sequencing. The Limulus sequences were resolved eventually by sequencing of the three Limulus pentraxin genes (Nguyen et al., 1986b). Cloning and sequencing of the Botrylloides nucleotide sequences encoding HA-1 subunits will be required for definitive assignment of the HA-1 lectin to the pentraxin family of proteins (Coe, 1983). However, comparison of the amino acid composition data in Table 1 to vertebrate pentraxins reveals marked similarities to SAP/CRP molecules from other species (Skinner et al., 1980;Liu et al., 1982;Pepys et al., 1982). In addition, the ultrastructure of the HA-1 molecule provides strong evidence that it is related to mammalian CRP and SAP (see what follows).
The disc structure of HA-1 is pentraxinlike. Despite the wide phylogenetic separation of mammals, fish, and ancient arthropods, a unique property of vertebrate pentraxins that has been retained is the disclike form of their molecules in the electron microscope after negative staining (Fernandez-Moran et al., 1968;Pepys et al., 1978;Coe, 1983). Figure 1 shows that the HA-1 molecule has this characteristic. Both the wide-field and enlarged images of HA-1 in this figure show structures with subunits arranged as discs with a RESULTS HA-1 Has Pentraxin Characteristics The amino acid composition of Botrylloides HA-1 molecules is similar to those of other ver-Asp tebrate pentraxins. HA-1 is a pentameric mol-Glu ecule whose identical disulfide-bonded subunits Ser Gly have a molecular weight similar to that of a His prototypical pentraxin subunit (around 28,000; Arg Coe, 1983;Coombe et al., 1984a;Schluter and Ey, Thr Ala 1989). In the present study, the blocked N-ter-Pro minus of HA-1 necessitated cleavage and separ-Tyr ation of the cleaved fragments prior to attempts Val Met at N-terminal amino acid sequencing. Even after Ile these steps were taken, however, single amino Leu acids could not be assigned confidently to each Phe position (not shown). It is noteworthy that this Lys result is reminiscent of similar findings with the Limulus pentraxin (Nguyen et al., 1986a(Nguyen et al., , 1986b, which exhibits microheterogeneity between subunits giving rise to multiple assignments aAn aliquot of HA-1 evaporated to dryness, hydrolyzed in M HCI, and derivatized for amino acid analysis by reverse-phase HPLC (Bidlingmeyer et al., 1984). bHA-1 has 259 amino acid residues.
PhCAP: plaice SAP composition (Pepys et al., 1982). SAP: human SAP composition (Skinner et al., 1980). eICRP: Limulus CRP subunit B (Liu et al., 1982). 69 hole in the center. Like mammalian and fish SAP, some of the HA-1 molecules are loosely arrayed in chains or stacks. Two characteristics of the molecules in Fig. 1 Fig. 1 appear to display the 100-angstrom diameter typical of vertebrate pentraxins (Pepys et al., 1978), the other half appears to have a diameter of 220 angstroms ( Fig. 1 and Fig. 1 inset), which is the diameter of the amyloidlike fibers deposited in rejection regions between Botryllus and Botrylloides oozooids (see what follows). The HA-1 preparation used in these studies is electrophoretically homogeneous under reducing conditions (Schluter and Ey, 1989), and thus the observed differences in subunit and molecular diameter are not likely to be due to the presence of a second subunit of a different molecular weight. Mammalian SAP occurs in the serum as paired pentameric discs that, like the HA-1 molecules in Fig. 1, tend to form stacks (Pepys et al., 1978;Coe, 1983). The twofold size difference between HA-1 images in Fig. 1 Figs. 2(a) and 2(b). After the initial fusion of tunics between contacted oozooids, the interactants either complete the establishment of shared blood vessels leading to fusion (Katow and Watanabe, 1980), Fig. 2(a), or initiate rejection reactions that lead to reseparation Watanabe, 1973a, 1973b;Katow and Watanabe, 1980;Scofield and Nagashima, 1982), Fig. 2(b). As in Botryllus, both the fusion and rejection reactions between juxtaposed Botrylloides oozooids are completed within 12 hr of first contact. Reseparation of rejected oozooids is accompanied by appearance of an opalescent mass of birefringent fibers in the tunic throughout the contact zone Watanabe, 1973a, 1973b), Fig. 2 The fibers are stained by aniline dyes. To determine whether ascidian amyloidlike fibers exhibit the histochemical characteristics of vertebrate amyloid, fiber deposits in sections from methacrylate-embedded oozooid pairs were subjected to standard histological tests for amyloid. Certain dyes, particularly fuchsin and Congo FIGURE 1. Electron micrograph of negatively stained Botrylloides leachii HA-1 molecules (Coombe et al., 1984a;Schluter and Ey, 1989).
Most molecules are oriented face up (arrows), and many are aligned in rows (top arrow). Individual molecules appear to be pentameric Red, show remarkable affinity for amyloid (Skinner et al., 1982). Congo Red binds to amyloid fibers via hydrophobic interactions with proteins folded as beta-pleated sheets, and fuchsin dyes similarly show specific binding to amyloid fibers, but by an unknown mechanism (Puchtler et al., 1982). The regular alignment of dye molecules along the fibrils amplifies their intrinsic birefringence (Skinner et al., 1982). Figure 3 shows a section through a rejected Botryllus oozooid pair that has been stained with basic fuchsin and photographed under cross-polarized light. A wide birefringent wall of fibers cleanly bisects the area of joined tunic between the oozooids; see Fig. 3(a). A Congo Red-stained section is shown under higher magnification in Figs. 3(b) and 3(c) which are of the same field photographed under two different polarizing prism positions. When illuminated by cross-polarized light, the fiber masses composing the rejection "barrier" in these photographs, and in the inset showing a similar field in a replicate section, show the "apple-green" color and birefringence characteristics of Congo Red-stained amyloid deposits in mammals.
The ultrastructure of fibers in rejection areas is similar to that of stacked SAP. Reports detailing the ultrastructure of mammalian amyloid deposits frequently include descriptions of stiff, hollow rodlike fibers whose cross-sectional morphology is suggestive of stacked SAP molecules (Holck et al., 1979;Miyakawa, 1988). A high-magnification transmission electron micrograph of fibers in rejection areas, like those in Figs. 3(a) to 3(c), is presented in Fig. 4. The 200-angstrom-wide fibers appearing in this electron micrograph are similar in ultrastructure and diameter and to those described by Watanabe (1973a, 1973b) in rejecting Botryllus primigenus colonies.
Where bundles of fibers criss-cross in and out of the plane of section, those appearing in cross-section are hollow and appear to be pentameric or hexameric, and to contain a central hole. Structures of similar size and appearance have been described in the kidney mesonephros of lampreys (Ellis and Youson, 1989) and have also reported to be present in mammalian connective tissue (Inoue and Le Blond, 1986). Where they are described in mice, the fibers are immunoreactive with anti-SAP antibody .
Antibody to HA-1 Binds to Protochordate Amyloidlike Fibers HA-1 determinants are concentrated in rejection areas.
The ground substance of the tunicate tunic is composed of tunicin, a celluloselike polymer (Deck et al., 1966). Birefringence patterns in tunicate tissues stained with Congo Red or fuchsin, both of which are cellulose dyes (Puchtler et al., 1962), may not be specific markers for amyloidlike material in ascidian tissues. Instead, the presence of immunoreactive SAP is now considered to be a better diagnostic indicator for human amyloid (Pepys, 1988). If ascidian rejection fibers are structurally similar to mammalian amyloid, and if the HA-1 lectin is SAP-like in function as well as morphology, the HA-1 protein should be concentrated in the rejection area and present as a structural component of the double-tracked fibers in these locations. We therefore used the rabbit antibody to HA-1 (Coombe et al., 1984b) to test for the presence and location of HA-1 epitopes in oozooid rejection areas and rejection fibers, and to identify HA-1 epitopes in other body tissues. Figure 5 shows replicate methacrylate sections taken through a rejected Botrylloides oozooid pair, incubated with normal rabbit serum, anti-SAP or anti-HA-1. Whereas neither the normal rabbit serum nor the antihuman SAP displayed significant binding to these sections (as evidenced by failure of goat antirabbit Ig and the Avidin-D and Protein A-labeled Fluoresbrite beads to identify sites of bound antibody on the sections in Figs. 5 (a) to 5 (d)), the anti-HA-1 bound to all the zooid tissues, with the heaviest labeling in the contact zone (Figs. 5(e) and 5(f)).
Anti-HA-1 binds to fibers in rejection regions. Figure 6 shows different areas from thin sections taken through birefringent areas of Epon-embedded rejected Botrylloides oozooids.
Figure 6(a) shows both the tunic cuticle and an area underlying it where debris from degenerating test cells is surrounded by double-tracked fibers, as described by Watanabe (1973a, 1973b). Binding of anti-HA-1 to fibers within these areas is clearly evident in Fig. 6(a) and in the magnified area from this photograph shown in 6(b). In Fig,. 6(c), heavy gold labeling marks the area underlying a cnidarian polyp embedded in the tunic. As in Fig. 5, neither normal rabbit serum nor anti-SAP showed significant binding to tunicate tissues in similar sections (Figures 7 and 8).
HA-1 Tissue Distribution Is Similar to That of CRP and SAP in Mammals HA-1 determinants are present on blood cells, in basement membranes, and in endostylar tissues. By immunoelectron microscopy, it was apparent that anti-HA-1 bound to Botrylloides cells and tissues in a pattern suggesting the blood cell and tissue distribution of SAP in mammals (Fig. 7). Previous studies have shown that anti-HA-1 binds to the macrophagelike granular amoebocytes from Botrylloides hemolymph (Coombe, 1983). In mice and humans, monomeric SAP and CRP are present on B-cell and NK-cell surfaces (Bray et al., 1988;James et al., 1983) and SAP occurs as a receptor-bound surface molecule on macrophages (Siripont et al., 1988). In addition, SAP is present in endothelial basement membranes (Dyck et al., 1980), and is a component of elastic fibers in the skin and connective tissues of lower chordates and mammals alike Ellis and Youson, 1989). Figure 7 shows that HA-l, like mammalian SAP, is apparently present on (and in) some blood cells, and on and in both the  Fig.1 and to stacked SAP (Pepys et al., 1978). The bar spans 1500 nm. epithelial cells of blood vessel walls, with epitopes also extending into the surrounding microfibrillar network. In addition, it appears that HA-1, like mammalian CRP (Braun et al., 1986), may be present both in the blood and in seromucous fluids.
Tunicates are filter-feeding organisms, and have a ciliated epithelial structure called the branchial basket that both supplies particle-trapping mucus and rolls it into a tube for digestion in the gut (Goodbody, 1974). The rolling is done by specialized elongated languet cilia inside the thyroidlike endostyle (a structure that is also present in larval lampreys; Ellis and Youson, 1989).

Figures 8(a) to 8(c)
show that HA-1 determinants are concentrated heavily on surfaces of the languet cilia, where they may be components either of the ciliary membrane or of the mucus that bathes these structures in the living animal. That the latter possibility is more likely is suggested by the presence of concentrated HA-1 determinants in the membranes of endostylar cell   Maudsley and Pepys, 1987;Bray et al., 1988). To test mammalian amyloid for the presence of HA-1 cross-reactive determinants, tissue was obtained from the brains of patients who had 'died with Alzheimer's disease.
Unlike amyloidotic tissue from other locations, the cerebral amyloid of Alzheimer's disease brains has a finely specialized pattern of deposition into three sites: the walls of microvessels, focal plaque cores, and halos of degenerating neurites surrounding the cores (Vinters and Gilbert, 1983;Coria et al., 1988) (Fig. 9). Incubation of Alzheimer's brain tissue with anti-SAP followed by fluorescent detection reagents reveals  antibody binding to amyloid deposits in all three of the locations identified in Fig. 9 (Liu et al., 1982); Figs. 10(a) to 10(c). When the anti-SAP is replaced by anti-HA-I, an identical binding pattern is seen; Figs. 10(d) and 10(e). Neither anti-SAP nor anti-HA-1 bound to similar tissue sections from two age-matched normal control brains (not shown). Similarly, sections incubated with the second and third reagents (alone or together) displayed no binding (not shown).

Amyloidlike Fibers Form Barriers Between Allogeneic Colonies
Double-tracked fibers are concentrated in the contacted tunic tissues of rejecting oozooids, where they appear to surround degenerated test cells Watanabe, 1973a, 1973b); Figs. 5 to 7. The presence of HA-1 epitopes on the FIGURE 7. Binding by anti-HA-1 to Botrylloides blood cells, vascular endothelial cells, and endothelial cell basement membrane. (a) and (b) Thin sections through ampullae between rejected Botrylloides oozooids, incubated with normal rabbit serum (a), anti SAP (a inset) or anti-HA-1 (b). No gold particles mark the blood cells or ampullar wall after incubation of sections in normal rabbit serum (a), but two of the three blood cells and the ampullar epithelium are heavily labeled after incubation in anti-HA-1 (b). Magnification: x5000. (a inset) Ampullar epithelium (arrow) from a section incubated with anti-SAP. As with the anti-SAP-incubated methacrylate section in Fig. 5, some antibody binding is evident, but no distinct pattern is seen. Magnification: x5000.  underlying a fouling cnidarian, Figs. 7(a) to 7(c), suggests that HA-1 is a physical component of these structures, and that their deposition is a true amyloidotic process. It is interesting that the blood-cell and epithelial HA-1 epitopes vis-ualized by immunelectron microscopy in Figs. 7 and 8 were accessible to antibody without prior protease treatment of the sections, whereas those in the rejection area in Fig. 5 and in the Alzheimer's disease brain tissue in Fig. 10 were not. Likewise, although double-tracked fibers are  Fig. 9, incubated with normal rabbit serum (a), anti-SAP (b and c), and anti-HA-1 (d and e). As shown previously by others, anti-SAP labels microvessel, plaque, and neurite amyloid (Coria et al., 1988)  binding by anti-HA-1 occurred only in areas where a defense response had occurred (arrows in Figs. 6(a) to 6(c)). It is possible that the inflammatory reactions accompanying fiber deposition at these sites included local production of proteases capable of unmasking HA-1 epitopes; alternatively, fibers deposited during such reactions may have compositions or secondary structures distinct from those present in normal tissue (Figs. 7 and 8).
The locations of birefringent fiber masses and areas of HA-1 epitope concentration in Figs. 2 to 6 imply that amyloid in tunicates operates for encapsulation of inflammatory foci and for barrier formation against allogeneic challenge. The association of human amyloid deposits with skin lesions in leprosy, with lung granulomas in tuberculosis and in joints in rheumatoid arthritis (Jayalakshmi et al., 1987;Breedveld et al., 1989;Mazur, 1989) suggests that mammalian amyloid serves a similar containment function.
HA-1 Epitopes Are Present in Protochordate and Mammalian Amyloid Table 1 and Fig. 1 show that the amino acid composition and three-dimensional structure of the Botrylloides HA-1 molecule resemble those of other vertebrate pentraxins (Coe, 1983;Schluter and Ey, 1989). That HA-1 may have functions similar to those of mammalian SAP is suggested by Figs. 5 and 6, which show that HA-1 epitopes are concentrated in areas between rejected oozooids and on fibers deposited near degenerating test cells in these locations, and by Figs. 7 and 8, which show that HA-1 epitopes have a tissue distribution similar to that of CRP and SAP in mammals. Figures 9 and 10 show that anti-HA-I, like anti-SAP (Coria et al., 1988), binds to amyloid fibers within the microvessels of Alzheimer's disease brains, and identifies both the core and neurite amyloid deposits within the senile plaques.
In mammals, both CRP and SAP can serve as acute-phase serum proteins (Coe, 1983) whose concentrations increase upon injury or immunological challenge. CRP opsonizes bacteria (Coe, 1983) and fixes complement (Volanakis and Kaplan, 1974;Kilpatrick and Volanakis, 1985). Both SAP and CRP participate in regulation of blood clotting and platelet function (Meyer et al., 1987), and the proteolytic fragments of CRP are potent immunomodulators (Robey et al., 1987). Human and mouse NK and B cells have monomeric surface CRP and SAP that comodulates with Fc receptors for immunoglobulins Bray et al., 1988). NK cell surface CRP appears to participate in target cell killing, by mechanisms that remain to be defined . Finally, mammalian SAP, like HA-1 (Coombe et al., 1984a), is a serum lectin that mediates calcium-dependent hemagglutination via carbohydrates on red cell surfaces (Hamazaki, 1988). The host defense functions of pentraxins intersect with those of adaptive immunity on several levels, and comparisons of pentraxins and immunoglobulins suggest a limited structural similarity between their molecules as well (Vasta et al., 1986;Schluter and Marchalonis, 1989). Isolation and sequencing of the HA-1 genes of tunicates will allow positive identification of the HA-1 lectin as a pentraxin. Identification of cell surface molecules associated with it then may offer clues to relationships between the pentraxin/immunoglobulin superfamilies in the early evolution of adaptive immunity.

Animals
Colonies of Botryllus schlosseri and Botrylloides violaceous were collected from local harbors and maintained in the laboratory in aquaria under aeration. Embryos and mature tadpole larvae were removed by dissection and allowed to metamorphose, after which the oozooids were removed carefully and reattached in pairs to glass or plastic surfaces, as described previously . After the transplantation reactions (fusion or rejection) were completed, the rejected pairs were selected, removed with razor blades, and embedded in methacrylate for histochemistry and immunofluorescence studies, interfaces between oozooids, sections were stained with alcoholic basic fuchsin (Puchtler et al., 1962) (tunicate tissue) or Congo Red (Skinner et al., 1982) (tunicate and human tissue), and photographed under cross-polarized light (Puchtler et al., 1962;Skinner et al., 1982).

Proteins and Antibodies
HA-1 was prepared from Botrylloides hemolymph, as described previously (Coombe et al., 1984a;Schluter and Ey, 1989), and affinity-purified anti-HA-1 prepared from serum of a rabbit immunized with HA-1 in complete Freund's adjuvant (Coombe et al., 1984b). The blood cellbinding characteristics of this reagent have been described (Coombe, 1983). Affinity-purified rabbit antihuman SAP was from Accurate Chemical Corporation. For visualization of antibody binding to sections, biotinylated goat antirabbit immunoglobulin and avidin-D fluorescein isothiocyanate were obtained from Vector Laboratories, Fluoresbrite beads purchased from Polysciences, and Staph Protein A bought from Sigma. The Protein A-Gold used was Auroprobe G10 from Janssen Life Sciences.

Histochemistry and Immunohistochemistry with Tunicate Tissues
Tunicate oozooid pairs were fixed briefly in 1% formalin and embedded in methacrylate (Immuno-Bed; Polysciences). Sections were cut in a plane that incorporated both oozooid bodies and the rejection area between them. Congo Red and basic fuchsin staining were done with unfixed sections after a brief hydration step in distilled water. For assays of antibody binding to tunicate tissues embedded in methacrylate, all sections were incubated for 1 hr in anti-HA-I, anti-SAP, or normal rabbit serum at a 1:20 dilution. Areas of antibody binding were visualized by their fluorescence under ultraviolet illumination after sequential incubation in biotinylated goat antirabbit Ig (1:50) and mixed Protein Aand avidin D-conjugated Fluoresbrite beads (Polysciences). All sections in Fig. 5 had been preincubated fol 30min in a trypsin solution prior to addition of normal rabbit serum or specific antibodies.

Human Tissues
The test tissue for light microscopic detection of antibody binding to human amyloid was human brain removed at autopsy and fixed in 10% neutral buffered formalin. Blocks were embedded in paraffin by methods used in routine pathological work. Confirmation of Alzheimer's disease was by standard pathological criteria, including identification of Congophilic microvascular and neuritic plaques (Vinters and Gilbert, 1983). As with the tunicate tissues, all sections were first incubated briefly in a dilute protease solution (in this case, with pepsin; Coria et al., 1988), followed by 1:50 dilutions of the test antibodies or normal rabbit serum. This was followed sequentially by biotinylated goat antirabbit Ig and avidin-D fluorescein.

Electron Microscopy
For transmission electron microscopy, rejected oozooids were fixed in 2% glutaraldehyde, embedded in Epon, stained with osmium tetroxide and uranyl acetate, and sectioned in areas previously selected by light microscopy for birefringence. HA-1 molecules were visualized on a Formvar-coated grid after negative staining with phosphotungstic acid, as described previously for vertebrate CRP and SAP preparations (Pepys et al., 1978). Observation and photography were done with a JOEL-100X electron microscope.

Amino Acid Composition
Amino acid analysis of HA-1 was by reversephase HPLC using the Pico-Tag method (Bidlingmeyer et al, 1984).