A monoclonal antibody equivalent to anti-rat neural antigen-1 as a marker for schwann cells

03~522/85 S3.00 + 0.00 Pergamon Press Ltd C 1985 IBRO

neuroscienceVol. 15. No. 3, pp. 877-885. 1985 Printed in Great Britain

A MONOCLONAL ANTIBODY EQUIVALENT TO ANTI-RAT NEURAL ANTIGEN-l AS A MARKER FOR SCHWANN CELLS K. L. FIELDS* and M. DAMMERMAN Departments of Neurology and Neuroscience. Albert Einstein College of Medicine,

1300 Morris Park Avenue, Bronx, NY 10461, U.S.A. Abstract-The monoclonal antibody 217c, raised by Peng ef al. [(1982) Science, Wash. 215, 1102-l 1041 in mice against the rat glioma cell line C6, can be used as a marker for normal Schwann cells. In mixed cultures of Schwann cells and fibroblasts from neonatal rat sciatic nerve, this monoclonal antibody, detected by indirect immunofluorescence, bound to the surface of cells with the same elongated morphology as those that express a previously described surface antigen, rat neural antigen-l (Ran-l), defined by polyclonal mouse antisera. In these experiments Ran-l and the antigenic determinant recognized by monoclonaI217c were both found on normal rat Schwann cells and on the rat glial tumor cell lines C6, 33B and 21A and the pheochromocytoma PC12. Neither anti-Ran-l nor the monoclonal antibody bound to neurons, fibroblasts or glial cells in newborn rat cerebellum cultures, the rat muscle cell line L6, the transformed rat fibroblast cell line Rat 1, the rat brain tumor cell line B28 or the mouse Schwannoma cell line TR6B. Thus the monoclonal217c behaved as if it were detecting Ran-l by binding to normal rat Schwann cells and to those tumor cells that have this antigen. Our data show that this monoclonal antibody is a reliable and convenient marker for rat Schwann cells in culture.

Considerable progress has been made in defining immunological markers that can be used to identify the various ceil types of the nervous system when these cells are grown in culture.‘*.“.*~ Identification of rat neural antigen-f (Ran-l),” a surface protein,* provided a means of specifically labeling Schwann cells in mixed cultures of Schwann cells and fibroblasts from rat sciatic nerve,3 in cultures from sensoryI and enteric” ganglia or in cultures from the CT%.*’ Rabbit antisera to rat tumors have consistently failed to detect Ran-I-equivalent antigens,* but one raised against a human lung tumor detected a surface antigen shared by rat and human Schwann cells2 More recently a rabbit antiserum, raised against Schwann cells isolated from neonatal rats, detected surface antigen(s) present on rat Schwann cells but not on fibroblasts or neurons.16 Monoclonal antibodies have been produced that bound to the surface of Schwann cells from mice”’ or chickens,’ but unlike Ran-l both of these bound to other normal cell types as well. Rat neural antigen-l was defined by mouse antisera raised against 33B, a rat cell line derived from an N-ethyl-N-nitrosourea-induced tumor of the nervous system.” The Ran-l antigen was found on the surface of normal Schwann cells3 and all of the rat nervous system tumor cell lines tested, except 828.’ The Ran-f-positive tumor cell lines included glioma cell line C6, other ghomas and S~hwannomas and PC12, a pheochromocytoma. Anti-Ran-l antiserum did not bind to any identified cell type in primary rat cultures *To whom all correspondence should be addressed. A&ret&rrion: Ran- 1, rat neural antigen- 1. 877

other than Schwann cells or closely related glial cells of the enteric nervous system.” Thus fibroblasts, neurons and giial cells in cultures from newborn rat cerebellum were Ran-l-negative.*’ The monoclonal antibody 217c,*Oraised against C6 cells, bound to spontaneously transformed rat astrocytes, to a rat oligodendroglioma and to a human glioma cell line, as well as to C6 cells. It bound weakly to a rodent hepatoma cell line and hepatocytes in primary cultures. The monoclonai did not label normal rat astrocytes or ohgodendrocytes, normal brain tissue sections, a non-transformed glial cell line or a non-transformed rat fibroblast line.20 The present study was undertaken to compare directly the binding specificities of anti-Ran-I antiserum and monoclonal antibody 217~. EXPERIMENTAL PROCEDURES Cells and cell lines

Freshly dissected sciatic nerves from newborn Wistar rats were incubated with trypsin and collagenase and dissociated by a procedure described in detail previously.’ Debris was removed by ~t~fugation through a 4% albumin solution and cells were plated on 13 mm glass cover slips in multiwell plates. Dissociated cell cultures were prepared from adult rat sciatic nerves essentially according to Scott.‘.*’ All these peripheral nerve or ganglionic cells were grown in minimum essential medium (Eagle) (G&co) plus 10% fetal calf serum and 10% horse serum (KC Biologicals). Dissociated cell cultures from newborn rat cerebelfum or olfactory bulb were prepared following the procedure of Currie and Dutton sd The rat cell lines 33B, 21A, B28, RN2, Rat 1 and L6, the mouse cell line TR6B and the human cell line HeLa were taken from frozen stocks, grown in Dulbecco’s modified Eagle medium plus 10 or 15% fetal calf serum, and plated onto glass cover slips for indirect immunofluorescence as-

K. L. Fields and M. Dammerman

878

says. The cell line B2SL”was a gift of Dr. D. Schubert (Salk Institute for Biological Studies. La Jolla, CA), while C6 and RN2 were kindly provided by Dr. S. Pfeiffer (University of Connecticut Health Center. Farmineton. CT). %La cells were a gift of Dr. L. Reid and Rat- 1 was obtained from Dr. S. I. Shin (both at the Albert Einstein College of Medicine). The PC12 pheochromocytoma cell line” was obtained from Dr. L. Greene (New York Universitv Medical School) and from Dr. L. Reichardt (University- of California a; San Francisco) and grown according to Greene and Tischler’4 in RPM1 1640 (GIBCO) with 107; horse serum and 5% fetal calf serum. Anribodr staining methods Anti-Ran-l antiserum (A-strain mouse anti-33Bf was adsorbed three times with rat liver, three times with spleen and thymus, and twice with rat muscle fibroblasts, as It was used for indirect previously described.3,” immunofluorescence at a dilution of 1:25 or 1:50. Monoclonal antibody 217~. described by Peng et 01.,” in the form of culture su~matant was the generous gift of Dr. J. de Vellis (University of California at Los Angeles) and it was used undiluted in some experiments, but worked equally well at I:50 dilution. Goat anti-mouse immunoglobulin G-rhodamine (Cappel) was used at a I : 100dilution. Staining of live cells was performed as previously described.3 CNS cultures were double-labeled with anti-Ran-l or monoclonal 2I7c and with tetanus or cholera toxin, followed by rabbit anti-tetanus or anti-cholera antiserum and then by goat anti-mouse immunoglobulin G-rhodamine plus goat anti-rabbit immunoglobulin G-fluorescein in a procedure described elsewhere.‘2,” Double labeling with a surface antigen and anti-&al filament serum was done as described previously.?’ Antibody controls included several other unrelated monocionaf antibodies, normal mouse serum and normal mouse ascites fluid. In several experiments cells were prefixed with 3.73: p-formaldehyde in phosphate-buffered saline for 10 min at 20’ or with 5”; acetic acid in 95% ethanol (acid alcohol) for IO min at 0 or 20 After incubation with all of the antisera, the cells were fixed again with acid alcohol and then mounted in a permanent mounting medium.” Coverslips were examined and photographed using a Zeiss standa-d microscope equipped for phase contrast and epifluorescence and a 63 x objective.

RESULTS Sciatic neroe cultures In rat sciatic nerve cultures, all cells with the morphology typical of Schwann cells were unambiguously positive for binding either anti-Ran-l serum or the monoclonal 217c, while fibroblasts in the same cultures bound neither antibody (Fig. 1). Cultures of adult rat sciatic nerve also contained cells, sometimes round but usually bearing processes, that bound either Ran-l or the monocional 217~ antibody (Fig. 2). The morphology of these cells was better preserved when the cells were fixed with formaldehyde or acid alcohol before staining. Structures resembling growth cones were then visualized by antibody binding, and flaps of plasma membrane were seen extending from the ceil body or from the sides of processes (Fig, 3). None of these structures were seer. clearly by phase contrast microscopy alone, and they retracted during the procedure used for staining live cells where, in addition, the distribution of antigen became patchy (Fig. 2). bipolar

Central nervous q~srem culture.5 In newborn

rat cerebellum

cultures.

neither

nru-

rons, which were identified by their binding of tetanus toxin.‘* nor 9976 of the non-neuronai cells were stained by either antibody. Both Ran-l and 217~ antibodies stained a very small number r)T nonneuronal cells (less than 1”;) that were not seen when normal serum or monoclonal controls were tested. In cultures of the olfactory bulb, tetanus toxin-binding neurons were all negative for Ran-l or monoclonal 217c antibody binding. Astrocytes. defined by the binding of rabbit anti-glial filament antiserum. were present in these cultures but bound neither anti-Ran1 nor 217~ antibody. However, both anti-Ran-i and monoclonal217c labeled S-lOoi/, of the non-neuronal cells (Fig. 4). Cultures exposed to a mixture of anti-Ran-l and 217~ monoclonal antibody stilt showed only 8-109; positive cells. Some of the positive cells were of bipolar morphology while others were flatter and broader than typical Schwann cells. Cell lines ~-Ethyl-~-nitrosourea-induced neural tumor cell Iines 33B and C6 were labeled by both anti-Ran-f and monoclonal antibody 217c, with similar surface staining patterns and intensities, whereas line 2lA showed less intense staining by both antibodies. PC1 2 cells, derived from a rat pheochromocytoma tumor, were positive for Ran-l and monoclonal antibody 21%~ binding. In both cases (Fig. 5) staining intensities were variable within the PC12 cell population. RN2, a rat Schwannoma line, was stained faintly by anti-Ran-l or the monoclonal. The following were not labeled by either antibody: L6, a rat muscle cell line; Rat-l. a transformed ~broblast cell line; TR6B, a mouse Schwannoma line; and HeLa, a human carcinoma cell line. The brain tumor cell line B28, the only one of the 12 Salk Institute lines2J~26 tested previously that was Ran-l -negative,‘.” also

lacked the determinant ctonal.

recognized

by the mono-

Correspondence qf mon~clonal 2 17~ und anti- Ran - I binding All of the Ran-l-positive cell types examined in this study also expressed the antigen recognized by the monoclonal antibody 217~. The monoclonal 217~ failed to bind to any of the Ran-l-negative cell types tested. These results are summarized in Table 1. DISCUSSION

Utility of the Ran- 1 marker Of the known Schwann cell surface antigens, Ran1 has been most thorou~iy investigated for ceit-type specificity. It is expressed by cultured Schwann cells from rats of all ages tested, including adults.3.9 It has been used to evaluate a method for obtaining cultures of pure Schwann cells: to identify Schwann cells by electron microscopy,” to study Schwann cell mito-

Fig. I. Binding of monoclonal antibody 217~ or anti-Ran-l to elongated cells in cultures of neonatal rat sciatic nerve. Binding was detected with goat anti-mouse immunoglobulin-rhodamine using live cells, which were fixed after staining with acid alcohol. In (A) and (B) the primary antibody was the monoclonal 217~ (diluted 1: SO). (A) Phase contrast; (B) fluorescence of the same field. In (C) and (D) anti-Ran-l (diluted 1:25) was used. (C) Phase contrast; (D) fluorescence of the same field. Cells were 4 days in culture. Flat fibroblastic cells did not bind either antibody. Scale bar, IOpm.

a79

Fig. 2. Binding of monoclonal antibody 217~ and anti-Ran-l to cells cultured from adult rat sciatic nerve. (A) Phase contrast; (B) fluorescence of the same field. Monoclonal antibody 217c, diluted 1: 50, was used. (C) Phase contrast; (D) fluorescence, using anti-Ran-l diluted 1:25. Cells, 6 days in oifro, were stained live. Scale bar, 10pm.

880

Fig. 3. minding of mon~~onal antibody 21%~and anti-Ran-l to fixed cells from adult rat sciatic nerve. Cells, after 6 days in v&w, were fixed with ~-fo~a~dehyde and then exposed to antibodies. (A) Phase contrast; (B) fluorescence of the same field. Nonoclonal antibody 21?c, diluted 1: 50.(C) Phase contrast; (13) Auorescence, using anti-Ran-I diluted I : SO. Scale bar, 10 pm.

881

Fig. 4. A minority of olfactory bulb non-neuronal cells bind monoclonal antibody 217~. Cells were IO days in oitro and stained live using the monoclonal antibody diluted 1: 50. (A) Phase contrast; (B) fluorescence of the same field. Note that neurons (n) and astrocytes (a) did not bind the monoclonal antibody. Scale bar, 1Okm.

Fig. 5. Pheochromocytoma PC12 cells bound monoclonal antibody 217~. (A) Phase contrast; (B) fluorescence. Not all cells were equally positive; arrowhead marks a faintly outlined cell. Scale bar, 10 pm.

882

A monoclonal antibody to a Schwann cell surface antigen

883

Table I. Binding specificities of anti-Ran-l antiserum and monoclonal antibody 217~ as determined by indirect immunofluorescence* Cells Primary cultures of rat PNS (neonatal) Schwann cells (adult) (neonatal or adult) Fibroblasts Primary cultures of rat CNS Cerebellum Neurons Astrocytes Oligodendroglia Olfactory bulb Neurons Non-neuronal cells Rat neural tumor cell lines C6 33B PC12 21A RN2 B28 Non-neural rat cell lines Rat 1 (fibroblast) L6 (muscle)

Tumor cell lines (other species) TR6B (mouse Schwannoma) HeLa (human carcinoma)

Anti-Ran-l

Antibody Monoclonal 217~

++ ++

++ ++

0

0

0

0 0 0

0

0 0 + + (10%)

++ ++ + +/+ + + /o

0

+ + (10%)

++ ++ + +I+

0

+ IO 0

0 0

0 0

0

0

0

0

* + + , very positive; + , definite positive; + + / +, positive but of variable intensity; + /O, faint; 0, not detectable.

gens,’ to characterize the Schwann cell-like glial cells of the enteric nervous system” and to investigate myelin-specific glycolipids and basic proteins expressed by cultured Schwann cells.” It would be advantageous in some studies to immunoselect the antigen-positive Schwann cells but this has been precluded by limited supplies of Ran-l antiserum. The monoclonal antibody 217~ should be a superior reagent for all such applications, for in our experience it has a higher titer and is a cleaner reagent than any of the adsorbed Ran-l antisera. Rat neural antigen-l has not been localized in tissue sections and the monoclonal immunoglobulin G antibody should be a better reagent for further attempts to do this. Monoclonal antibodies offer distinct advantages over conventional antisera for use as cell type-specific markers. There is no need for the repeated adsorptions that are required to remove non-cell typespecific antibodies from polyclonal xenogeneic sera. In addition monoclonals are of constant specificity and may be available in unlimited quantities. Biochemistry of Ran - 1 The Ran-l antigen was shown to be greatly enriched in cell membrane preparations. However, a more complete characterization has been prevented by a loss of antigenic activity following detergentinduced release from cell membranes.’ The antigen was sensitive to heat inactivation, with substantial loss of activity at temperatures as low as 52” and with

95% of the antigen destroyed at 80”. These findings, together with its sensitivity to digestion by proteolytic enzymes, make it likely that the Ran-l antigen is a cell-surface protein.* Its function is not known. At the present time, the evidence for Ran-l and the 217~ antigens being identical rests entirely on similarities in distribution, method of immunizing, and stability to various fixation procedures. The possibility that Ran-l and the 217~ antigens are lipids is ruled out by their insensitivity to methanol or acidalcohol extractions. In contrast, alcohols remove the receptors for cholera toxin, tetanus toxin and R24 monoclonal antibody (K. L. Fields, unpablished observations). There is good evidence, reviewed recently,” that the toxins, R24 and similar monoclonal antibodies detect gangliosides. Other lipids, such as galactocerebroside, sulfatide, or phosphorylcholine, are also ruled out by the alcohol insensitivity of Ran- 1 and 217~ antigen(s). Unlike several antibodies, some directed at lipids (A2B5, galactocerebroside, phosphorylcholine and several oligodendrocyte antigens) and some that recognize glycoproteins (UC45 and two anti-Thy- 1.1 monoclonal antibodies),” cross-reactions with cytoplasmic, mitochondrial or cytoskeletal components were not detected for Ran- 1 antisera or the monoclonal antibody 217~. Further experiments are required to determine whether Ran- 1 and 217~ recognize the same antigenic determinant or different proteins. Immunoprecipitation or immunoblotting may establish the

884

K

L. Fields and M. Dammerman

molecular weight of one or both antigens, while blocking experiments with biot~nylated or radiolabelled 217~ immunoglobulin could show whether the antigens are on the same molecule. Antigen-positive

cells

in

centrul

nervous

s.vstem

cultures

Less than l”/‘, of non-neuronal cells in cultures of newborn rat cerebellum were stained by either antiRan-l or the monodonal antibody 21%. In contrast, in newborn rat olfactory bulb cultures, S-IOO~ of the non-neuronal cells were stained by either reagent. Some of these cells were bipolar, while others were broader and flatter. Since this percentage did not increase in cuitures exposed to a mixture of the two antibodies, we conclude that by and large the same cells were labeled by anti-Ran- 1 and 2 17~. The cells stained by these antibodies did not have tetanus toxin receptors or glial-type intermediate filaments and were therefore unlikely to be typical neurons or astrocytes. Barber and Lindsay’ also found such cells in olfactory bulb cultures and argued that they must accompany the olfactory nerve fibers into the CNS. Monoclonal antiboc+ 217~ as a Schwann cell marker

It must not be assumed from the presented work that Ran-l or monocIona1 antibody 217~ may not bind to other ceil types besides Schwann cells and

neural tumors. Preliminary experiments with super~c’r cervical ganglion cultures. in fact. strongly iuggesr that these neurons are labelled iti vitro lw both reagents, as are the majority of ganglioni; nonneuronal cells (K. L. Fields and J Kessler. unpublished observations). Antigen expression by sensory (dorsal root ganglion) neurons. previously reported to be negative for Ran-l antigen.” is currently under investigation. Since the 217~ monoclonal antibody was reported to bind to one human glioma,“’ it will be of interest to determine whether it binds to human Schwann cells. In one sense there is a precedent for this, since a human lung tumor antigen was demonstrated on rat Schwann cells in culture’. The results of the present study indicate that the antigen recognized by monoclonal 217~ is equivalent in its distribution on normal and tumor cells to the antigen Ran-t The monoclonal therefore constitutes a convenient and effective immunochemical marker for rat Schwann cells in culture. Acknou,fedgemenrs-The authors wouid like to thank L. Stein for skillful technical assistance and M. Levine and R. Sasso for typing the manuscript. This study would not have been possible without the generous gift of monocional antibody 217~ from Dr. J. de Veilis. We were supported by NIH Grant NS14580 and the National Science Foundation Graduate Fellowship Program.

REFERENCES I. Barber P. C. and Lindsay R. M. (1982) Schwann cells of the olfactory nerves contain gliai fibrillary acidic protein and resemble astrocytes. Neuroscience 7, 3077-3090. 2. Bell C. E. and Seetharam S. (1977) Identification of the Schwann ceil as a peripherai nervous system ceil possessing a djffe~tiation antigen expressed by a human lung tumor. f. fmmun. l%S, 826831. 3. Brockes J. P., Fields K. L. and Raff M. C. (1977) A surface antigenic marker for rat Schwann ceils. Nature 266,364-366. 4. Brockes J. P., Fields K. L. and RafT M. C. (1979) Studies on cultured rat Schwann ceils. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res. 165, 105-i 18. 5. Currie D. N. and Dutton G. R. (1980) f’H]GABA uptake as a marker for ceil type in primary cultures of cerebellum and olfactory bulb. Brain Res. 199, 473-481. 6. Dutton G. R.. Currie D. N. and Tear K. (1981) An improved method for the bulk isolation of viable perikarya from postnatal cerebellum, J. Neurosci. Meth. 3, 421-427. 7. Fambrough D. M., Bayne E. K., Gardner J. M., Anderson M. J.. Wakshuli E. and Rotundo R. L. (1982) Monoclonai antibodies to skeletal muscle ceil surface. In Neurojmmunafog~ (ed. Brockes J.). Cnrrenr Topics in N~robfolog~, Vol. 5, pp. 49-89. Plenum Press, New York. 8. Fields K. L. (1977) Biochemical studies of the common and restricted antigens. two neural ceil surface antigens. Prug. clin. biol. Res. 15, 179-190. 9. Fields K. L. (1983) Differentiated Schwann cells cultured from adult sciatic nerves contain astrocyte-type intermediate filaments. Sot. Neurosci. A&r. 9, 5. 10. Fields K. L. (1985) Neuronai and giiai surface antigens on ceils in cuiture. In Ceil Culture in the ~eu~oseie~ce~ (eds Bottenstein J. and Sato G.). Plenum Press, New York. In press. ii. Fields K. L., Gosling C., Mtgson M. and Stem P. L. (1975) New cell surface antigens in rat defined by tumors of the nervous system. froc. narn. Acad. Sri. U.S.A. 72, 1296-I 300. 12. Fields K. L., Brockes J. P., Mirsky R. and Wendon L. M. B. (I 978) Cell surface markers for distinguishing different types of rat dorsal root ganglion ceils in culture. Cell 14, 43-5 1. 13. Fields K. L. and Raine C. S. (1982) Witrastructure and immunocytochemistry of rat Schwann ceils and fibrobiasts in vitro, J. Neuroimmun. 2, 155-166. 14. Greene L. A. and Tischier A. S. (I 976) Establishment of a noradrenergic cional line of rat adrenal pheochromocytoma ceils which respond to nerve growth factor. Proc. mtn. Acad. Sci. U.S.A. 73, 2424-2428. 15, Jessen K. R. and Mirskv R. (1983) Astrocyte-like giia in the peripheral nervous system: an imm~nohist~hemicai study of enterlc giia. J. Neur&ci. 3, 2266-2218 16. Knider B. Q., Messing A., Doan H., Kim S. U., Lisak R. P. and Pleasure D. E. (1981) Enrichment of Schwann cell cultures from neonatal rat sciatic nerve by differential adhesion. Bruin Res. 207, 433-444. 17. Mirsky R. (1982) The use of antibodies to defme and study major cell types in the central and peripheral nervous system. In Neuroimmunoiog~ (ed. Brockes J.). Current Topics in ~eurff~~ofo~~. Vol. 5, pp. 141-181. Plenum Press, New York.

A monoclonal anti~dy

to a Schwann cell surface antigen

885

18. Mirsky R., Wendon L.. Black P., Stolkin C. and Bray D. (1978) Tetanus toxin: a cell surface marker for neurones in culture. Brain Res. 148, 251-259. 19. Mirsky R., Winter J.. Abney E. R., Pruss R. M., Gavrilovic J. and Raff M. C. (1980) Myelin-specific proteins and glycolipids in rat Schwann cells and oligodendrocytes in culture. J. Ceil Biol. 84, 48H94. 20. Peng W. W., Bressler J. P., Tiffany-Castiglioni E. and de Vellis J. (1982) Development of a monoclonal antibody against a tumor-associated antigen. Science, Wash. 215, 1102-l 104. 21. Raff M. C., Fields K. L., Hakomori S.-i., Mirsky R.. Pruss R. M. and Winter J. (1979) Cell-type-specific markers for distinguishing and studying neurons and the major classes of glial cells in culture. Bruin Res. 174, 283-308. 22. Rodriquez J. and Deinhardt F. (1960) Preparation of a semipermanent mounting medium for fluorescent antibody studies. Virolog_v 12, 316317. 23. Schachner M. (1982) Immunological analysis of cellular heterogeneity in the cerebellum. In Neuroimmunology (ed. Brockes J.). Current Topics in Neurobiology, Vol. 5, pp. 215-250. Plenum Press, New York. 24. Schachner M., Kim S. U. and Zehnle R. (1981) Developmental expression in central and peripheral nervous system of oligodendrocyte cell surface antigens (0 antigens) recognized by monoclonal antibodies. Derf Viol. s3, 328-338. 25. Schubert D., Heinemann S., Carlisle W., Tarikas H., Kimes B., Patrick J., Steinbach J. H., Culp W. and Brandt B. L. (1974) Clonal cell lines from the rat central nervous system. Nurure 249, 224-227. 26. Schubert D., Tarikas H. and LaCorbiere M. (1976) Neurotransmitter regulation of adenosine 3’,5’-monophosphate in clonal nerve, glia and muscle cell lines. Science, Wmh. 192, 471-472. 27. Scott B. S. (1977) Adult mouse dorsal root ganglion neurons in cell culture. J. Neurobiol. 8, 417-427,

NSC t.5 3-a