HYBRIDOMA Volume 10, Number 6, 1991 Mary Ann Liebert, Inc., Publishers
Can Successfully Be Prepared from Frozen/Thawn Spleen Cells
Hybridomas
ANNE BENNICK, BERIT
GR0N,
Research Institute for Internal Medicine,
and FRANK BROSSTAD Rikshospitalet, Oslo, Norway
ABSTRACT
Spleens from mice immunized with either human von Willebrand factor, albumin or fragment DD from plasminolyzed fibrin were divided into two equal parts. Subsequently, one half of the spleen cells were fused directly with X-63 mouse myeloma cells and the other half fused after freezing and thawing. The results show that hybridomas may successfully be prepared from frozen/thawn spleen cells, but that a higher cell density than with fresh cells is required. The possibility of making use of frozen spleen cell material implies that valuable time and cell material can be saved, and that fusions can be postponed and performed at the time of choice. INTRODUCTION To
our knowledge, all reports on hybridoma techniques have so far used freshly prepared spleen cell suspensions. In spite of rapid screening procedures to detect antigen-specific clones, most laboratories are unable to utilize all immunized mice. Since valuable time and cell material could be saved if spleen cell suspensions from otherwise "discarded" mice could be frozen, and thereby postponing fusion to be done at one's convenience, we investigated this possibility.
MATERIALS AND METHODS Immunization. Balb/C mice were immunized subcutaneously with 50-100 ug antigen per mouse each fortnight for 711 months. Freund's complete adjuvant (Difco Laboratories,Detroit,MI,USA)was used with the first immunization dose to stimulate the animals'immune system. At later immunizations Freund's incomplete adjuvant (Difco Laboratories,Detroit,MI,USA) was used. Adjuvant (1-2 parts) was mixed with antigen solution (1 part). The following antigens were used for the immunization: 1.Human
serum
2.Human
von
albumin (HSA) ( Kabi, Stockholm, Sweden)
Willebrand factor,
purified from cryoprecipitated plasma.
761
3.Fragment D-dimer (DD) from plasminolyzed FXIII stabilized human fibrin. Three days before sacrificing the mice, they were boosted with antigen. The mice were then sacrificed and their spleens carefully removed. The spleens were gently squeezed in a sterile mortar in the presence of Dulbecco'sMinim urn Essential Medium (MEM) (GibcoLtd,Paisley,Scotland)toobtaincell suspensions.
Freezing. After centrifugation, the cell pellets were treated with 0.83% ammonium chloride with pH=7.2 for
1 minute to remove the red blood cells. The cell suspensions were then washed with MEM before they were frozen in fetal bovine serum containing 5% dimethyl-sulfoxide (DMSO), first by -80° C and then in liquid nitrogen. After 3-20 months in liquid nitrogen, the spleen cell suspensions were thawed and fused with X-63-Ag-65 3 (Rikshospitalet, Oslo, Norway) mouse myeloma cells.
Fusion.
The cell count of the spleen cell suspensions was estimated using a Coulter Counter (Coulter Electronics Ltd,Harpenden, Herts, England) or a Burker chamber obtaining approximately 2x10 8 cells from each
spleen.
A simplified polyethylene glycol(PEG)-induced fusion (1), which is a modification of the classical method of Köhler and Milstein (2,3) was used with some smaller modifications.Polyethylene glycol 1450 from Eastman Kodak Company, Rochester,NY,USA, was used. Cells from one spleen were mixed with myeloma cells in a ratio 5:1 in MEM in a 50 ml centrifuge tube, topped with MEM and then centrifuged 10 min. at 200 x g. After having removed the supernatant and loosened the pellet, 3 ml PEG-solution (50% PEG with 5% DMSO) was added to the cells during 1 min. under constant, slight agitation. Still under slight agitation, the tube with the cells were immersed for 90 sec. in a 37°C bath. The PEG-solution was then diluted by adding 2 ml MEM over 30 sec, 6 ml MEM over the next 30 sec, 32 ml MEM over 1 min. The tube was topped up with MEM and then allowed to stand for 5 min., before it was centrifuged for 10 min. at 200 x g. The supernatant was discarded, the tube topped with standard medium without HAT (vide infra), centrifuged 10 min. at 200 x g, and after removing the supernatant once more, the cells were plated out on 96 wells culture plates in HAT-medium (vide innfra). This selective HAT-medium (Hypoxanthine, Aminopterine and Thymidine (4)) was used to select the hybridomas from the myeloma cells. Instead of using a feeder layer of mouse macrophages in the culture plates, we used 10% HECS (Human Endothelial Cell Supernatant) in the medium. Pilot experiments indicated that the best results could be obtained using cell suspensions of less dilutions than used with freshly prepared cells. Therefore, the fused cells were plated out with approximately 200, 000 spleen cells per well compared to 40, 000 cells per well when using freshly prepared cells. Pilot experiments also showed that in order to obtain the best result, it is important to use the spleen cell suspension immediately after thawing, in spite of the fact that the cells still contains some DMSO.
Screening.
To select clones producing monoclonal antibodies, ELISA procedures were used. NUNC immunoplates(Nunc A/S, Roskilde, Denmark) were coated with 5-10 pg antigen/ml Buffer 1 (vide infra) and shaken overnight at room temperature. After 3 x wash with Buffer 2(vide infra), the wells were blocked with Buffer 3 withl% bovine serum albumin (BSA)(Sigma Chemical Company, St.Louis, MO, USA) for 1/2-1 h under slight agitation at room temperature and finally washed 3 X in Buffer 2. Supernatants from wells containing clones with a sufficient number of cells to turn the media yellow were then added, and incubated under agitation for 1 h. at room temperature.
762
After washing 3x with Buffer 2, specific alkaline phosphatase conjugated goat-anti-mouse antibodies to Ig-light chains (Jackson Immunoresearch Lab Inc. Avondle,PA,US A) diluted 1/2000 in Buffer 3with 1 % BSA was used to detect wells binding primary monoclonal antibodies. Subsequently, the wells were incubated with the conjugate for 2-3 h under agitation at room temperature, and then washed 3 x with Buffer 2 and once with Buffer 4. To visualize the binding of the monoclonal antibody to the antigen p-nitrophenyl disodium phosphate (Sigma Chemical Company, St.Louis, MO, USA) was used. This substrate was dissolved in Buffer 4 to 1 mg/ml. The color-reaction was stopped with IM NaOH and read at 405nm in a Titertek Multiscan ELISA-reader (Flow Laboratories, Irvine, Scotland). Determination of the immunoglobuline classes and sub-classes was carried out using Mouse-Typer subisotyping kit (Bio-Rad Laboratories, Richmond, CA,USA).
Media.
(1) DMEM, standard medium: Dulbecco's MEM containing 20% fetal bovine serum, 2mM streptomycin, ImM Sodium pyruvate, 2.5 mg Fungizone
1-glutamine, 50 IU/ml penicillin, 50 ug/ml
(2) HAT-medium: DMEM, standard medium (1) containing 10% HECS, 100 uM hypoxanthine, 0.4 uM aminopterine, 16 uM
thymidine
(3) HT-medium: HAT-medium (2) without
aminopterine
Buffers.
(1) Phosphate buffered saline (PBS) 0.015 M pH=7.4: 0.14 M NaCl, 2.7 mM KC1, 1.5 mM KH2P04, 8.1 mM
Na2HP04, 0.01% thimerosal
(2) PBS-tween: Buffer 1 with 0.05% Tween 20 (3) Buffer 2 with 0.1% or 1% BSA (4) Diethanolamine buffer, 1 M pH 9.8: IM diethanolamine, 0.5 mM MgCL/6 H20, 3 mM NaN3 RESULTS AND DISCUSSION
Previous work in our laboratory with immunization of mice with the same antigens as used in this presentation have shown that these antigens tend to give a high IgM immunoglobulin response. The immunization periods were, therefore, considerably extended (7-11 months), hopefully to obtain a maximum number of IgG responding animals. Moreover, pilot experiments clearly demonstrated that in spite of the fact that the thawn spleen cells contained small amounts of DMSO from the freezing-media, the best results were obtained when the spleen cells were used in fusion immediately after thawing. If the spleen cells were cultured for a short period before the fusion to remove the DMSO, the yield from the fusion was very low. This is probably due to the fact that the spleen cells are very short-lived in culture, and maybe even more sensitive after
763
freezing/thawing. It is also possible that the usually is added to the fusion agent, PEG.
intracellular DMSO
can
be
advantageous,
since DMSO
As seen from Table 1, when fresh and frozen fused cells were plated out with a cell density of 40,000 per well, the number of clones generated from frozen material was significantly lower than when fresh material was used. If, however, the cell density stemming from frozen material was increased to about 180,000 cells per well, a satisfactory amount of clones resulted. Thus, three fusions with fresh spleen cells (fusion no. 1,4,7) gave hybrid clones in 3-42 % (mean 23.3 %) of the wells, and 7-29 % (mean 18.6%) of the clones were specific clones to the antigens. As is seen,the percentage varies with the antigen used. Three fusions with frozen spleen cells (fusion no.3,6,9) gave hybrid clones in 17-100 % (mean 40.4 %) of the wells, and 2-11 % (mean 8.6 %) of the clones were specific.
Number and
Specificity
TABLE 1 of Hybdridoma Clones generated from fresh
or
frozen/thawn Spleen Cells.
Hybridoma clones generated from spleen cells from 1): mouse immunized with human serum albumin (HSA), 2): mouse immunized with human von Willebrand factor (vWF) or 3):mouse immunized with fragment D-dimer from plasminolyzed human fibrinogen (DD).The spleen cells were either fused directly (=a) or subsquent to freezing/thawing (=b). Cell concentration per well was either 40,000 (fresh and frozen cells) or 180,000 (frozen cells). Number of generated clones is given as percent (i.e. no. of wells with clones relative to total no. of wells). Number of specific clones is given as percent (i.e. no.of specific clones relative to total no. of clones). FUSION NO.
ANTIGEN USED
HSA HSA HSA vWf vWf vWf DD DD DD
The
SPLEEN
FRESH(a) FROZEN(b)
SPLEEN CELLS PER WELL
40,000 40,000 180,000 40,000 40,000 180,000 40,000 40,000 180,000
%CLONE % SPECIFIC GENECLONES RATED
25 0.4 30 3 0.7 100 42 2 17
IgG:IgM RATIO
1:34
29 0 25 20 0 11 7 10
2:1
11
1:6
1:30 1:5 1:16 1:70
tendency seems to be that a fusion with frozen spleen cells gives a higher percentage of generated
clones, and a lower percentage of specific clones compared with a fusion with fresh spleen cells. It is, however, reason to believe that this will vary with the antigen used. The IgG/IgM ratio of the
specific clones based on immunotyping varied from 1:34 to 2:1 (Table 1).
When the fusions with fresh spleen cells and with frozen spleen cells are regarded separate, the ratios will be from 1:5 to 1:70 for fresh spleen cells and from 2:1 to 1:30 for frozen spleen cells. This may partly be due to the nature of the antigens used, but may also indicate that even though the freezing of the spleen cells influence the percentage of specific clones, the IgG/IgM ratio is not affected. One possibility may also be that the IgG producing cells are more likely to survive the freezing/thawing process as compared to the IgM producing cells. One can also speculate if the freezing/thawing process induces a Ig-class-shift on the DNA-level.
764
CONCLUSIONS
Spleen cell suspensions from immunized mice may be frozen and thawn, preserving their viability. a consequence, specific hybridomas may successfully be produced from such cells.
As
The present technique offers the possibility to save time, labor and valuable material and makes it possible to perform a fusion at short notice. Previous observations indicate that the type of antigen used for immunization influences the resulting IgG/IgM ratio of the producing hybridomas. The large proportion of IgM-producing hybridomas observed in the present study may be related to the antigens selected.
REFERENCES 1.
Fazekas, D. S., Groth, S. and Scheidegger, D.(1980) Production of monoclonal antibody: strategy and tactics. J. Immunol. Meth. 35: 1-21.
2.
Köhler, G and Milstein, C.(1975) Continuous cultures of fused cells secreting antibody of
predifined specificity. Nature 256: 495-497. 3.
Köhler, G. and Milstein ,C.(1976) Derivation of specific antibody-producing tissue culture and tumor
4.
lines
by
cell fusion. Eur. J. Immunol. 6: 340-347.
Littlefield, J. W.(1964) Selection of hybrids from matingsof fibroblasts in vitro and theirpresumed recombinants. Science, 145: 709-710.
Requests for reprints to: A.Bennick,Ph.D. Res.Inst. for Int.Med.
Rikshospitalet Oslo, Norway Received for publication June 3, 1991, accepted after revision July 1, 1991
765
Can Successfully Be Prepared from Frozen/Thawn Spleen Cells
Hybridomas
ANNE BENNICK, BERIT
GR0N,
Research Institute for Internal Medicine,
and FRANK BROSSTAD Rikshospitalet, Oslo, Norway
ABSTRACT
Spleens from mice immunized with either human von Willebrand factor, albumin or fragment DD from plasminolyzed fibrin were divided into two equal parts. Subsequently, one half of the spleen cells were fused directly with X-63 mouse myeloma cells and the other half fused after freezing and thawing. The results show that hybridomas may successfully be prepared from frozen/thawn spleen cells, but that a higher cell density than with fresh cells is required. The possibility of making use of frozen spleen cell material implies that valuable time and cell material can be saved, and that fusions can be postponed and performed at the time of choice. INTRODUCTION To
our knowledge, all reports on hybridoma techniques have so far used freshly prepared spleen cell suspensions. In spite of rapid screening procedures to detect antigen-specific clones, most laboratories are unable to utilize all immunized mice. Since valuable time and cell material could be saved if spleen cell suspensions from otherwise "discarded" mice could be frozen, and thereby postponing fusion to be done at one's convenience, we investigated this possibility.
MATERIALS AND METHODS Immunization. Balb/C mice were immunized subcutaneously with 50-100 ug antigen per mouse each fortnight for 711 months. Freund's complete adjuvant (Difco Laboratories,Detroit,MI,USA)was used with the first immunization dose to stimulate the animals'immune system. At later immunizations Freund's incomplete adjuvant (Difco Laboratories,Detroit,MI,USA) was used. Adjuvant (1-2 parts) was mixed with antigen solution (1 part). The following antigens were used for the immunization: 1.Human
serum
2.Human
von
albumin (HSA) ( Kabi, Stockholm, Sweden)
Willebrand factor,
purified from cryoprecipitated plasma.
761
3.Fragment D-dimer (DD) from plasminolyzed FXIII stabilized human fibrin. Three days before sacrificing the mice, they were boosted with antigen. The mice were then sacrificed and their spleens carefully removed. The spleens were gently squeezed in a sterile mortar in the presence of Dulbecco'sMinim urn Essential Medium (MEM) (GibcoLtd,Paisley,Scotland)toobtaincell suspensions.
Freezing. After centrifugation, the cell pellets were treated with 0.83% ammonium chloride with pH=7.2 for
1 minute to remove the red blood cells. The cell suspensions were then washed with MEM before they were frozen in fetal bovine serum containing 5% dimethyl-sulfoxide (DMSO), first by -80° C and then in liquid nitrogen. After 3-20 months in liquid nitrogen, the spleen cell suspensions were thawed and fused with X-63-Ag-65 3 (Rikshospitalet, Oslo, Norway) mouse myeloma cells.
Fusion.
The cell count of the spleen cell suspensions was estimated using a Coulter Counter (Coulter Electronics Ltd,Harpenden, Herts, England) or a Burker chamber obtaining approximately 2x10 8 cells from each
spleen.
A simplified polyethylene glycol(PEG)-induced fusion (1), which is a modification of the classical method of Köhler and Milstein (2,3) was used with some smaller modifications.Polyethylene glycol 1450 from Eastman Kodak Company, Rochester,NY,USA, was used. Cells from one spleen were mixed with myeloma cells in a ratio 5:1 in MEM in a 50 ml centrifuge tube, topped with MEM and then centrifuged 10 min. at 200 x g. After having removed the supernatant and loosened the pellet, 3 ml PEG-solution (50% PEG with 5% DMSO) was added to the cells during 1 min. under constant, slight agitation. Still under slight agitation, the tube with the cells were immersed for 90 sec. in a 37°C bath. The PEG-solution was then diluted by adding 2 ml MEM over 30 sec, 6 ml MEM over the next 30 sec, 32 ml MEM over 1 min. The tube was topped up with MEM and then allowed to stand for 5 min., before it was centrifuged for 10 min. at 200 x g. The supernatant was discarded, the tube topped with standard medium without HAT (vide infra), centrifuged 10 min. at 200 x g, and after removing the supernatant once more, the cells were plated out on 96 wells culture plates in HAT-medium (vide innfra). This selective HAT-medium (Hypoxanthine, Aminopterine and Thymidine (4)) was used to select the hybridomas from the myeloma cells. Instead of using a feeder layer of mouse macrophages in the culture plates, we used 10% HECS (Human Endothelial Cell Supernatant) in the medium. Pilot experiments indicated that the best results could be obtained using cell suspensions of less dilutions than used with freshly prepared cells. Therefore, the fused cells were plated out with approximately 200, 000 spleen cells per well compared to 40, 000 cells per well when using freshly prepared cells. Pilot experiments also showed that in order to obtain the best result, it is important to use the spleen cell suspension immediately after thawing, in spite of the fact that the cells still contains some DMSO.
Screening.
To select clones producing monoclonal antibodies, ELISA procedures were used. NUNC immunoplates(Nunc A/S, Roskilde, Denmark) were coated with 5-10 pg antigen/ml Buffer 1 (vide infra) and shaken overnight at room temperature. After 3 x wash with Buffer 2(vide infra), the wells were blocked with Buffer 3 withl% bovine serum albumin (BSA)(Sigma Chemical Company, St.Louis, MO, USA) for 1/2-1 h under slight agitation at room temperature and finally washed 3 X in Buffer 2. Supernatants from wells containing clones with a sufficient number of cells to turn the media yellow were then added, and incubated under agitation for 1 h. at room temperature.
762
After washing 3x with Buffer 2, specific alkaline phosphatase conjugated goat-anti-mouse antibodies to Ig-light chains (Jackson Immunoresearch Lab Inc. Avondle,PA,US A) diluted 1/2000 in Buffer 3with 1 % BSA was used to detect wells binding primary monoclonal antibodies. Subsequently, the wells were incubated with the conjugate for 2-3 h under agitation at room temperature, and then washed 3 x with Buffer 2 and once with Buffer 4. To visualize the binding of the monoclonal antibody to the antigen p-nitrophenyl disodium phosphate (Sigma Chemical Company, St.Louis, MO, USA) was used. This substrate was dissolved in Buffer 4 to 1 mg/ml. The color-reaction was stopped with IM NaOH and read at 405nm in a Titertek Multiscan ELISA-reader (Flow Laboratories, Irvine, Scotland). Determination of the immunoglobuline classes and sub-classes was carried out using Mouse-Typer subisotyping kit (Bio-Rad Laboratories, Richmond, CA,USA).
Media.
(1) DMEM, standard medium: Dulbecco's MEM containing 20% fetal bovine serum, 2mM streptomycin, ImM Sodium pyruvate, 2.5 mg Fungizone
1-glutamine, 50 IU/ml penicillin, 50 ug/ml
(2) HAT-medium: DMEM, standard medium (1) containing 10% HECS, 100 uM hypoxanthine, 0.4 uM aminopterine, 16 uM
thymidine
(3) HT-medium: HAT-medium (2) without
aminopterine
Buffers.
(1) Phosphate buffered saline (PBS) 0.015 M pH=7.4: 0.14 M NaCl, 2.7 mM KC1, 1.5 mM KH2P04, 8.1 mM
Na2HP04, 0.01% thimerosal
(2) PBS-tween: Buffer 1 with 0.05% Tween 20 (3) Buffer 2 with 0.1% or 1% BSA (4) Diethanolamine buffer, 1 M pH 9.8: IM diethanolamine, 0.5 mM MgCL/6 H20, 3 mM NaN3 RESULTS AND DISCUSSION
Previous work in our laboratory with immunization of mice with the same antigens as used in this presentation have shown that these antigens tend to give a high IgM immunoglobulin response. The immunization periods were, therefore, considerably extended (7-11 months), hopefully to obtain a maximum number of IgG responding animals. Moreover, pilot experiments clearly demonstrated that in spite of the fact that the thawn spleen cells contained small amounts of DMSO from the freezing-media, the best results were obtained when the spleen cells were used in fusion immediately after thawing. If the spleen cells were cultured for a short period before the fusion to remove the DMSO, the yield from the fusion was very low. This is probably due to the fact that the spleen cells are very short-lived in culture, and maybe even more sensitive after
763
freezing/thawing. It is also possible that the usually is added to the fusion agent, PEG.
intracellular DMSO
can
be
advantageous,
since DMSO
As seen from Table 1, when fresh and frozen fused cells were plated out with a cell density of 40,000 per well, the number of clones generated from frozen material was significantly lower than when fresh material was used. If, however, the cell density stemming from frozen material was increased to about 180,000 cells per well, a satisfactory amount of clones resulted. Thus, three fusions with fresh spleen cells (fusion no. 1,4,7) gave hybrid clones in 3-42 % (mean 23.3 %) of the wells, and 7-29 % (mean 18.6%) of the clones were specific clones to the antigens. As is seen,the percentage varies with the antigen used. Three fusions with frozen spleen cells (fusion no.3,6,9) gave hybrid clones in 17-100 % (mean 40.4 %) of the wells, and 2-11 % (mean 8.6 %) of the clones were specific.
Number and
Specificity
TABLE 1 of Hybdridoma Clones generated from fresh
or
frozen/thawn Spleen Cells.
Hybridoma clones generated from spleen cells from 1): mouse immunized with human serum albumin (HSA), 2): mouse immunized with human von Willebrand factor (vWF) or 3):mouse immunized with fragment D-dimer from plasminolyzed human fibrinogen (DD).The spleen cells were either fused directly (=a) or subsquent to freezing/thawing (=b). Cell concentration per well was either 40,000 (fresh and frozen cells) or 180,000 (frozen cells). Number of generated clones is given as percent (i.e. no. of wells with clones relative to total no. of wells). Number of specific clones is given as percent (i.e. no.of specific clones relative to total no. of clones). FUSION NO.
ANTIGEN USED
HSA HSA HSA vWf vWf vWf DD DD DD
The
SPLEEN
FRESH(a) FROZEN(b)
SPLEEN CELLS PER WELL
40,000 40,000 180,000 40,000 40,000 180,000 40,000 40,000 180,000
%CLONE % SPECIFIC GENECLONES RATED
25 0.4 30 3 0.7 100 42 2 17
IgG:IgM RATIO
1:34
29 0 25 20 0 11 7 10
2:1
11
1:6
1:30 1:5 1:16 1:70
tendency seems to be that a fusion with frozen spleen cells gives a higher percentage of generated
clones, and a lower percentage of specific clones compared with a fusion with fresh spleen cells. It is, however, reason to believe that this will vary with the antigen used. The IgG/IgM ratio of the
specific clones based on immunotyping varied from 1:34 to 2:1 (Table 1).
When the fusions with fresh spleen cells and with frozen spleen cells are regarded separate, the ratios will be from 1:5 to 1:70 for fresh spleen cells and from 2:1 to 1:30 for frozen spleen cells. This may partly be due to the nature of the antigens used, but may also indicate that even though the freezing of the spleen cells influence the percentage of specific clones, the IgG/IgM ratio is not affected. One possibility may also be that the IgG producing cells are more likely to survive the freezing/thawing process as compared to the IgM producing cells. One can also speculate if the freezing/thawing process induces a Ig-class-shift on the DNA-level.
764
CONCLUSIONS
Spleen cell suspensions from immunized mice may be frozen and thawn, preserving their viability. a consequence, specific hybridomas may successfully be produced from such cells.
As
The present technique offers the possibility to save time, labor and valuable material and makes it possible to perform a fusion at short notice. Previous observations indicate that the type of antigen used for immunization influences the resulting IgG/IgM ratio of the producing hybridomas. The large proportion of IgM-producing hybridomas observed in the present study may be related to the antigens selected.
REFERENCES 1.
Fazekas, D. S., Groth, S. and Scheidegger, D.(1980) Production of monoclonal antibody: strategy and tactics. J. Immunol. Meth. 35: 1-21.
2.
Köhler, G and Milstein, C.(1975) Continuous cultures of fused cells secreting antibody of
predifined specificity. Nature 256: 495-497. 3.
Köhler, G. and Milstein ,C.(1976) Derivation of specific antibody-producing tissue culture and tumor
4.
lines
by
cell fusion. Eur. J. Immunol. 6: 340-347.
Littlefield, J. W.(1964) Selection of hybrids from matingsof fibroblasts in vitro and theirpresumed recombinants. Science, 145: 709-710.
Requests for reprints to: A.Bennick,Ph.D. Res.Inst. for Int.Med.
Rikshospitalet Oslo, Norway Received for publication June 3, 1991, accepted after revision July 1, 1991
765
