A. Erythrocytes of non-mammalian vertebrates
The erythrocytes of all non-mammalian vertebrates (i.e., fish, amphibians, reptiles, birds) are nucleated, flattened, and ellipsoidal (Andrew, 1965; Rowley and Ratcliffe, 1988). Their cytoskeletal system consists principally of a membrane skeleton, containing actin and spectrin-family proteins, a marginal band of microtubules composed of tubulin plus microtubule-associated proteins, and intermediate filaments of the desmin class (Fawcett and Witebsky, 1964; Behnke, 1970; Barrett and Dawson, 1974; Cohen, 1978; Sloboda and Dickersin, 1980; Granger and Lazarides, 1982). These cytoskeletal elements are illustrated schematically in Fig. 1. Erythrocytes with a similar cytoskeletal system constitute the "primitive" erythrocyte generation in mammalian embryos (van Deurs and Behnke, 1973; Cohen et al., 1990), and are phylogenetically widespread among invertebrates as well (Cohen and Nemhauser, 1985). Marginal bands of microtubules are also a prominent component of other blood cell types, including mammalian blood platelets, vertebrate thrombocytes, and hemocytes involved in clotting and other functions in invertebrates (Cohen and Nemhauser, 1985).
Nucleated erythrocytes of non-mammalian vertebrates are available in relatively pure populations, have characteristic and consistent morphology, and exhibit greater spatial separation of cytoskeletal components than most other eukaryotic cells. In addition, their cytoskeletal elements are representative of those found in eukaryotic cells in general. For these reasons, nucleated erythrocytes have been employed by a number of laboratories in recent years as a model cell type for studying cytoskeletal structure, function, biogenesis, and molecular composition (e.g., Euteneuer et al., 1985; Centonze and Sloboda, 1986; Lazarides, 1987; Birgbauer and Solomon, 1989; Kim et al., 1987; Murphy, 1991; Winckler and Solomon, 1991; Feick et al., 1991).
B. Erythrocytes of the smooth dogfish
Erythrocytes of the smooth dogfish, Mustelus canis, are particularly valuable for experimental work. They have typical flattened ellipsoidal morphology, are large enough (~18 µm long axis) for routine monitoring of cell shape and cytoskeletal structure in phase contrast (Fig. 2), and are easily obtainable in enormous quantity without sacrificing the animal. We know of no other readily accessible non-mammalian vertebrate that can match the smooth dogfish for this combination of cell size and quantity. In addition, Mustelus canis is one of the species in which the marginal band of microtubules, a structure of primary interest, is cold-labile both in living cells and in vitro (Cohen et al., 1982). This facilitates many kinds of experiments that would not otherwise be feasible.
In this paper we report current techniques and methodology with respect to (a) availability and handling of smooth dogfish at the Marine Biological Laboratory (MBL), (b) obtaining the blood, (c) large-scale preparation of washed erythrocytes with minimal leukocyte contamination, and (d) production of erythrocyte cytoskeletons in large quantity for a variety of subsequent experimental uses. The procedures either are new or are improved versions of those previously reported, with detailed recipes and protocols provided in the Appendices.
In previous work we have used erythrocytes of the smooth dogfish for studies of cytoskeletal protein composition (Cohen et al., 1982), molecular properties of non-mammalian erythrocyte "spectrin" (a subunit = alpha-fodrin; Bartelt et al., 1982, 1984), function of the marginal band of microtubules in living cells (Joseph-Silverstein and Cohen, 1984), isolation and molecular composition of the marginal band (Cohen and Ginsburg, 1986; Sanchez et al., 1990; Sanchez and Cohen, 1994a), and the function of marginal band associated proteins in assembly of microtubule bundles in vitro (Sanchez and Cohen, 1994b).
These studies have taken advantage of many useful properties of smooth dogfish erythrocytes, as follows:
If washed dogfish erythrocytes suspended in Elasmobranch Ringer's solution at laboratory "room temperature" (~22°C, with air conditioner) are placed at 0°C (ice slurry bath) for ~1-3 hours, the marginal band disassembles. If the cells are then returned to ~22°C the band reassembles to nearly normal appearance within ~3 hours (Cohen et al., 1982).
It is important to note that the upper end of the natural physiological temperature range for smooth dogfish extends into the range at which researchers work comfortably in the laboratory. This is advantageous in permitting use of the living erythrocytes on the lab bench at a temperature approximately physiological for the cells. In contrast, chickens have a body temperature of ~38°C, so that their erythrocytes may be as much as15°C below their normal temperature in routine lab use, with possible negative effects on experiments.
Details of the erythrocyte wash procedure (with removal of leucocytes) are provided in Appendix 1.
2. Inhibition of marginal band disassembly and reassembly
In cells maintained in Elasmobranch Ringer's solution at ~22°C, the band remains stable for at least several hours. Exposure to microtubule inhibitors such as colchicine or colcemid (0.1-1 mM) at that temperature does not produce marginal band disassembly. However, if the band is then disassembled by incubation of cells at low temperature (as above), these agents, as well as nocodazole (10 µg/ml), will block reassembly of the marginal band upon rewarming (Cohen et al., 1982; Joseph-Silverstein and Cohen, 1984). This has permitted production of living cell populations that either contain or lack marginal bands at physiological temperature, and these populations can be compared with respect to cytoskeletal molecular composition, osmotic responses, and mechanical behavior. It can then be demonstrated that the band maintains mature cell shape against experimental deformation. Similar experiments are possible using cells at low temperature, because taxol (10 µg/ml) blocks marginal band disassembly in living cells even at 0°C (Joseph-Silverstein and Cohen, 1984).
3. Natural marginal band breakage, generating doubly pointed cells
Although morphologically stable for several hours in Elasmobranch Ringer's solution at lab temperature (~22°C), cells that are pointed rather than curved at the ends of their long axis begin to appear after about 4 hours . The number of such doubly pointed erythrocytes increases with time. These cells contain correspondingly pointed marginal bands of microtubules, in which the band is "broken" at the two ends of the ellipse (Fig. 3). We do not know the cause of breakage, but it is quite useful experimentally. Studies using microtubule assembly inhibitors show that pointed cell morphology does not occur in cells that lack marginal bands, i.e., it is the band that generates pointed cell shape, not the reverse (Cohen et al., 1982; Cohen, 1991).
Marginal band thickness is positively correlated with cell size (long axis of ellipse) in diverse species (Goniakowska-Witalinska and Witalinski, 1976). In species with relatively small nucleated erythrocytes, such as most birds (e.g., chickens) and bony fish, the erythrocyte marginal band is so thin that it can be seen in phase contrast only with difficulty. It becomes necessary to use oil immersion objectives and sometimes video-enhanced imaging, and even then the band is not easily isolated. In contrast, the dogfish erythrocyte is large (~18µm long axis). Its marginal band is thick enough (~0.2-0.3µm) to be seen easily under a 40X phase contrast objective, and it has sufficient stability for isolation from cytoskeletons. Although cell lysis using the detergent Brij-58 is preferred for mass preparation of cytoskeletons, their further extraction with Triton X-100 provides improved views of cytoskeletal structure for light and electron microscopy (Fig. 2 ).
Our original isolation methods for isolating dogfish erythrocyte marginal bands involved the ue of proteases to remove the membrane skeleton (Cohen and Ginsburg, 1986). More recently we have developed a non-proteolytic detergent-based method (Sanchez et al., 1990). Isolation is facilitated by the use of taxol (Vallee, 1982) to stabilize cytoskeleton marginal bands during long-term storage (Appendix 2). Marginal band isolation provides one approach to the localization of cytoskeletal proteins, particularly microtubule-associated proteins (MAPs). Tau, a MAP best known as a neuronal component with possible involvement in Alzheimer's disease (e.g., Kosik et al., 1988; Wille et al., 1992), is found in isolated dogfish marginal bands (Sanchez and Cohen, 1994a). It was identified previously in chicken erythrocyte microtubule protein preparations (Murphy and Wallis, 1985; Lichtenberg-Kraag and Mandelkow, 1990; Murphy, 1991). Isolated marginal bands are also excellent material for studies of the mechanical properties of cytoskeletal elements (Cohen, 1978; Waugh and Erwin, 1989). Although such work is most readily done with the extremely large bands (>40 µm long axis) of certain amphibians, the isolation methods employed have been developed primarily through use of dogfish erythrocytes.
2. Production of anucleate ghosts and anucleate cytoskeletons from living erythrocytes
For certain purposes, particularly protein analysis using SDS-PAGE and other techniques, it is advantageous to eliminate nuclei from erythrocyte cytoskeletons. This is extremely difficult to do after the cytoskeletons have been prepared using detergents. However, living dogfish erythrocytes are remarkably cooperative in expelling their nuclei upon exposure to hypo-osmotic microtubule-stabilizing media and subsequent fluxing through syringe needles. This produces anucleate ghosts, from which (anucleate) cytoskeletons can be prepared by subsequent detergent extraction (Cohen et al., 1982). These anucleate cytoskeletons consist principally of the marginal band plus the membrane skeleton. It is worth noting that pellets of quite cleanly isolated erythrocyte nuclei are a preparation by-product that could be of value for other types of studies.
3. Binding of calmodulin to membrane skeleton alpha-fodrin
The 245kD protein of the dogfish erythrocyte cytoskeleton (a membrane skeleton component) binds calmodulin strongly. Thus, dogfish erythrocytes were one of the first systems in which the similarity between non-mammalian erythrocyte alpha-spectrin and brain alpha-fodrin was readily demonstrated (Bartelt et al., 1982). Subsequent work documented this by immunoblotting with antibodies to alpha-fodrin (= a subunit of "brain spectrin"; Bartelt et al., 1984). For studies of membrane skeleton components, anucleate cytoskeletons (described above) are particularly useful.
4. Low-temperature marginal band disassembly in vitro, yielding protein that reassembles into microtubule bundles
Low-temperature disassembly of the marginal band is observed not only in living cells, but also in whole cytoskeletons and in isolated marginal bands in vitro. With "Brij cytoskeletons" (prepared using Brij detergent; details in Appendix 2) suspended in microtubule disassembly/reassembly media at 0°C, the band disassembles completely in ~3 hours. The cytoskeletons (minus bands) can be sedimented by centrifugation, producing a supernate containing microtubule protein. Upon rewarming of this supernate to temperatures within or close to the physiological range (for dogfish! ~22°C), microtubule assembly and microtubule bundle formation occur. Analysis of these bundles indicates that tau protein is present and functioning as a co-assembly factor in microtubule bundling in vitro (Sanchez and Cohen, 1994b).
Smooth dogfish are typically brought in by the MBL collecting boats from late May through mid September. They are maintained in large circular holding tanks to reduce the likelihood of bumping into tank walls. Bumping is much more frequent in rectangular tanks, and damage to the "nose" often leads to infection and ultimately death.
M. canis of the Woods Hole/Northeast coast region migrate to warmer waters in September and October, and are then no longer routinely collected. Although experimental material can be stored long-term (see below), freshly prepared material is often desirable. Lack of availability of smooth dogfish at the MBL during colder months has thus been a disadvantage in the past. However, pilot projects conducted during 1992-94 have shown that smooth dogfish collected during the summer months can be maintained in the MBL Marine Resources Center in useful numbers for most of the year (e.g., 10-20 fish maintained for 6-8 months).
Although they lack sharp teeth and almost never try to bite when handled, smooth dogfish are still members of the shark family. As such they are quite powerful, and even a small dogfish can inflict scratches and scrapes on the skin of one's arm with a slap of its body while attempting swimming-like escape movements during handling. The fish feels very smooth when stroked from anterior to posterior, but quite rough in the opposite direction.
Prior to obtaining blood, one or more dogfish are transferred from their holding tank into a large container of seawater on a dolly, and transported to the site designated for animal procedures. The container must be kept covered to prevent the fish from jumping out. We handle the fish during removal from tank or container in either of two ways: by grasping them tightly with one hand just in front of the tail, or by grasping the body dorsally with two hands, one in front of the dorsal fin and the other behind it (Figs. 4a,b). The first method typically initiates rapid and powerful swimming-type movements, making the fish difficult to hold. Allowing one's arm to swing loosely helps greatly. The second method, if done properly, is the more gentle in terms of immediate fish response. For our purposes, however, the first method appears to be preferable. Though temporarily stressful, it tires the animal to some extent, reducing subsequent squirming during blood withdrawal and increasing our efficiency. We believe that stress is thus reduced because the fish spends less time out of water.
Investigators new to dogfish must take care to distinguish between smooth and spiny species if both are available (as at the MBL). The spiny dogfish, Squalus acanthias, has a sharp spine just in front of both dorsal fins, and dorsal rows of small light-colored spots as well as a somewhat different body shape (Fig. 5). It prefers colder waters than the more uniformly gray-colored Mustelus. In our experience, blood is not readily obtained from the spiny dogfish. The syringe procedure described below for Mustelus does not work with Squalus acanthias, and thus it cannot easily be substituted for the smooth dogfish. We have not determined whether its erythrocytes or the cytoskeletal fractions prepared from them behave similarly to those of the smooth dogfish.
We use a V-shaped wooden restraining device with attached straps (actually, belts; Fig. 6) to hold the fish relatively still while taking blood. This prevents them (and us!) from possible movement-induced injury. Tilting the restrainer appears to enhance blood flow to the posterior end of the animal, where blood is drawn (Fig. 7). To avoid being scratched and to keep the animal wet, we routinely wrap the fish in a large towel soaked in seawater before strapping it in. Covering the eyes also protects them and makes the animal more tranquil. With seawater poured over the head and gills periodically, the dogfish easily tolerates being out of the water for the 5-10 minutes required to take blood.
Blood volumes in the 20-50 ml range are readily obtained from the animals, depending on their size. After bleeding, the dogfish are tagged on the dorsal fin (Fig. 8) and returned to the holding tank. With appropriate care, the animals can be maintained long-term for repeated use. Details of the procedures for blood withdrawal, a guide to fish size vs. blood volumes taken, and information on repeated use of animals are provided in Appendix 3 .
The major steps are summarized here; details are provided in Appendix 1. The bulk of the white cells are first removed by low speed centrifugation of the initial blood cell suspension in heparinized Elasmobranch Ringer's. The white cells are less dense than the erythrocytes; most of them sediment into a layer above the red cells (the "buffy coat"; Fig. 9a,b), which is then resuspended and removed by aspiration. The red cell pellet is resuspended in heparinized Ringer's, and most remaining contaminating white cells are removed by centrifugation through a sucrose-Ringer's solution in which increased solution density retards white cell sedimentation. The supernate is discarded, and the erythrocytes washed by two rounds of resuspension and centrifugation in Ringer's (no heparin needed now). The erythrocytes are then ready for use. If they are not to be used immediately we keep them as a suspension in Ringer's at ~22°C, a more physiological condition than a pellet. This procedure, involving use of a sucrose-Ringer's density step, typically yields erythrocyte suspensions with white cell contamination of less than 1 in 1000 cells. Similar low white cell contamination levels can be achieved without the density step, but a great many rounds of centrifugation and removal of the upper white layer are then required.
Erythrocyte cytoskeletons are produced by lysis with detergents and subsequent washes in stabilizing media. Soluble cytosolic components, especially hemoglobin, are removed leaving a white "skeletal" structural residue. The cytoskeletons are excellent starting material for a variety of experimental purposes, such as localization of molecular components, marginal band isolation, or direct extraction of microtubule protein. We summarize here the major steps in mass preparation and storage of dogfish erythrocyte cytoskeletons; details are provided in Appendix 2. The procedure described represents a considerable improvement over its predecessor (Cohen and Ginsburg, 1986) with respect to increased production in less time, and reduction of mechanical stress to the cytoskeletons.
A. Improved cytoskeleton preparation
Previously, cytoskeletons were washed by rounds of centrifugal packing and resuspension, breaking many marginal bands and distorting cytoskeletal structure in the process. The new method eliminates such washes by substituting a glycerol step gradient (25-40-50%) onto which a Brij-lysed cell suspension is layered (Fig. 10a). Hemoglobin and other soluble components are left in upper layers during centrifugation. The cytoskeletons are transferred through detergent lysis and wash solutions stepwise in the gradient, ending with microtubule-stabilizing medum containing 50% glycerol (Fig. 10b). The cytoskeletons are never tightly packed, and resuspend with ease for subsequent storage or direct use. This procedure for dogfish erythrocytes is based on one first developed for preparing chicken erythrocyte cytoskeletons prior to isolation of their extremely thin marginal bands (Sanchez and Cohen, 1994a).
Brij-58 remains our detergent of choice for mass preparations of dogfish erythrocyte cytoskeletons. Cells lyse more slowly than in Triton X-100, but the cytoskeletons are far less "sticky" and resuspend readily after centrifugation. As described previously, a protease inhibitor "cocktail" is used as a precaution against possible endogenous protease release during cell lysis. However, the current procedure substitutes "Pefablock" (Boehringer-Mannheim Corp.) for phenylmethylsulfonyl fluoride (PMSF). Unlike PMSF, which loses potency rapidly in solutions, Pefablock can be added to solutions in advance. Although Pefablock is much more expensive than PMSF, the cost is offset by the time saved and the reduced amount of PIPES buffer needed for the step gradient vs. repeated washes (Cohen and Ginsburg, 1986).
B. Cytoskeleton storage conditions
Brij cytoskeletons can be stored for long periods (months) in the lab freezer (-20°C) in PEM containing 50% glycerol, but the marginal band appears to get thinner gradually with time unless taxol (10 µM) is included in the storage medium. Stored taxol-treated dogfish erythrocyte cytoskeletons are excellent starting material for marginal band isolation (Sanchez and Cohen, 1994a). However, once the marginal bands have been stabilized by taxol, they can no longer be used for extraction of microtubule protein by marginal band disassembly at low temperature.
We thank Edward Enos, Superintendent of the MBL Department of Aquatic Resources, for testing long-term maintenance of dogfish for experimental purposes, and both Ed and his predecessor, John Valois, for helpful suggestions during the course of our work. We are indebted to Robert Bullis, D.V.M., Microbiologist at the Laboratory of Aquatic Animal Medicine and Pathology, for critically reviewing part of the manuscript, and to Linda Golder of the MBL Photo Lab/Graphics Department for providing the color photographs. Research support from the National Science Foundation (NSF MCB-9118773) and from PSC-CUNY 663151 and NIH-NIGMS/MBRS GM08176 grants is gratefully acknowledged.
| Appendix 1. Erythrocyte washes; removal of leucocytes (recipes and protocols) |
| Appendix 2. Preparation of erythrocyte cytoskeletons (recipes and protocols) |
| Appendix 3. Blood withdrawal and volume; repeated use of animals (procedural details) |
Fig. 1: Diagrammatic representation of the cytoskeletal system of nucleated erythrocytes in all non-mammalian vertebrates.
Fig. 2: Dogfish erythrocytes and cytoskeletons.
Fig. 3: Pointed erythrocytes and cytoskeletons.
Fig. 4: Handling smooth dogfish.
Fig. 5: The spiny dogfish, Squalus acanthias, compared with the smooth dogfish.
Fig. 6: A smooth dogfish in a restrainer.
Fig. 7: Blood withdrawal from caudal blood vessels.
Fig. 8: Tagging dogfish.
Fig. 9: Erythrocyte preparation.
Fig. 10: Cytoskeleton preparation.