- Email: [email protected]
Immuno-gold labeling for flow cytometric analysis
Journal oflmmunologicalMethods, 74 (1984) 49-57 Elsevier
49
JIM03248
Immuno-Gold Labeling for Flow Cytometric Analysis R a l p h - M . B 6 h m e r 1,, a n d N i c h o l a s J.C. K i n g 2 I Melbourne Turnout Biology Branch, Ludwig Institute for Cancer Research, P.O. Royal Melbourne Hospital Victoria 3050, and e John Curtin School of Medical Research, Australian National University, Canberra, Australia (Received 27 March 1984, accepted 17 July 1984)
Colloidal gold particles coated with goat anti-mouse immunoglobulin antibodies were used to analyse surface antigens on various cell types by flow cytometry. The gold-labeled cells showed an increasing signal amplification in the 90 o light scatter with increasing wavelength of the incident laser light, reaching a more than 10-fold amplification at 632.8 nm. This wavelength was provided by a 0.5 m W h e l i u m - n e o n laser. The magnitude of the signal amplification due to the gold label as well as the specificity of the label was sufficient for quantitative discrimination between positive and negative ceils. Cell viability was not affected by the gold label, Mouse spleen cells were labeled with various combinations of FITC- and gold-conjugated antibodies, It was found that the gold and fluorescent labels did not interfere with each other. Colloidal gold may thus be used as an additional label for multiparameter cell analysis and sorting. Biparametric cell analysis/sorting of surface antigen-labeled cells (label versus low-angle scatter) becomes possible even with a low energy h e l i u m - n e o n laser. Key words: immuno-gold label - flow cytometry - cell sorting
Introduction
The flow cytometric analysis of cell surface antigens by antibodies tagged with fluorescent dyes has become an important technique for immunological studies. Simultaneous labeling of different cell surface molecules by use of reagents labeled with different dyes has also become feasible (Loken et al., 1979). It has also been possible to combine surface labeling with specific staining of cell contents such as DNA (Loken, 1980; Steinkamp et al., 1982). While such combined labels are very useful, they are limited by the spectral overlap of the fluorescent dyes. In view of this difficulty we decided to introduce a non-fluorescent cell surface label, suitable for flow cytometry, which would provide more freedom in combining different labels. One of the possibilities of detecting a non-fluorescent label with a flow cytometer is
* To whom correspondence should be addressed. 0022-1759/84/$03.00 © 1984 Elsevier Science Publishers B.V.
50 the change in the light scatter signal (Shapiro, 1983). We found that the binding of colloidal gold particles, which are commonly used as a cell surface label for light and electron microscopic studies (Horrisberger, 1981; De M e y e t al., 1981: Bergroth et al., 1983), was able to enhance the right angle (90 °) scatter of the red laser light dramatically. This paper investigates the various factors influencing the results of the labeling technique and demonstrates its enormous potential for studies on cell surface receptors.
Materials and Methods
Cell preparation B A L B / c mice were used as donors for spleen cells and peritoneal exudate cells (PEC). Spleen cells were obtained by passing the whole organ through a fine wire mesh into RPMI 1640 and centrifuging the suspension over Ficoll to eliminate dead cells and erythrocytes. PEC were lavaged from the peritoneal cavity with cold phosphate buffered saline (PBS). Cells from the P815 line, a mastocytoma of D B A / 2 origin, were used for indirect labeling as a cell line comparison of the same d H-2 haplotype as B A L B / c mice. 6C2 is a chicken erythroblast cell line transformed by avian erythroblastosis virus. A ntisera G A M G40 (G20, G5) was obtained from Janssen Pharmaceutica, Beerse, Belgium. GAM G40 is colloidal gold of 40 nm diameter, conjugated electrostatically with affinity-purified goat anti-mouse immunoglobulin. GAM G40 was used to label the surface immunoglobulin (Ig) on murine lymphocytes directly, or as a secondary antibody (Ab) to label histocompatibility antigen ( H - 2 K / I a ) on the cell surface. Monoclonal rat anti-mouse Thyl.2-FITC and normal rat Ig were a kind gift from Dr. R. Scollay, Walter and Eliza Hall Institute, Melbourne. The anti-Thyl.2 antiserum was obtained from the supernatant fluid of the hybridoma H12 clone and conjugated to FITC after affinity purification. Anti-Thyl.2 Ab labels the Thy antigens on T cells of B A L B / c spleen and peritoneal exudate cell (PEC) populations. A monoclonal antibody against chicken erythroid cells was made by Dr. Keith Savin, Ludwig Institue for Cancer Research° Melbourne. Spleen cells from mice injected with cells from the 6C2 chicken erythroblast line were fused with NS1 mouse myeloma cells. The hybridoma produces an antibody of the IgM class which binds to cell surface glycoprotein of all chicken erythroid cells. Mouse anti-mouse H - 2 K / I a antiserum was a kind gift from Dr. R, Blanden, John Curtin School of Medical Research, Canberra. It was raised in ( A / J x DBA/1)F1 mice, injected weekly with D B A / 2 spleen cells for 4 weeks. The mice were bled and the 0.2 ffm filtered whole serum used to label the cell surface gene products of the K and Ia regions of the d haplotype. Rabbit anti-mouse Ig-FITC (RAMIg-FITC) was prepared by raising an antiserum in a New Zealand White rabbit against mouse Ig. It was passed over a Protein
51 A column and the bound portion eluted and conjugated to FITC (Goding, 1976). Cells were washed in normal rabbit serum before addition of RAMIg-FITC which was used to label Ig on B cells directly.
Cell fixation Cells were washed once in PBS and suspended for 30 min at 4 ° in freshly prepared periodate-lysine-paraformaldehyde fixative (PLP) (McLean and Nakane, 1974). This solution is made up of 1 ml paraformaldehyde (4% w / v ) in water, pH 7.3, 9 mg paraperiodic acid and 3 ml 0.1 M L-lysine in 0.1 M sodium phosphate buffer, pH 7.3. PLP is a gentle fixative maintaining the serological detectability of surface H-2, Ig and Thy antigens (King and Parr, 1982; King, unpublished observations). Thereafter the cells were washed first in PBS and then in RPMI 1640 with 10% fetal calf serum (RPMI + FCS). All washes made use of the latter medium unless stated otherwise. Exept for Ficoll separation, the cells were kept strictly at 4 o C during all preparation and labeling procedures.
Flow cytometry The experiments were performed on an Ortho System 50 cell sorter, equipped with an argon ion laser providing wavelengths ranging from 451.7 nm to 528.7 nm, and a 0.5 mW helium-neon laser providing 632.8 nm red light. The red right angle (90 ° ) scatter signals were detected with a special red-sensitive photomultiplier (EMI type D160). FITC fluorescence was excited at 488 nm and measured between 510 and 530 nm. The 90 ° scatter signals as well as the green fluorescence signals were processed by use of a logarithmic amplifier. A distance of 18 units on the histogram scales was found to correspond to a factor of 2 in signal size.
Labeling procedures The cells were incubated for 60 min with anti-H-2 antiserum and for 30 min with anti-Thyl.2-FITC and RAMIg-FITC. Each incubation was followed by three washes in RPMI + FCS before addition of the secondary antiserum. Before incubation with anti-Thyl.2-FITC or RAMIg-FITC, the cells were washed for 10 rain in normal rat Ig or normal rabbit serum, respectively, in order to occupy sites that might otherwise be labeled non-specifically by the FITC reagents. For gold labeling, aliquots of 106 cells were suspended directly in 50 ~tl of the undiluted GAM G40 (concentration of reagent not given by supplier), placed in a 2-mm, flat-bottomed centrifuge tube, spun three times for 5 min at 200 x g and carefully resuspended between the spins. After the third spin, the cells were either washed twice in 10 ml RPMI + FCS before further labeling procedures, or simply resuspended in this washing medium for direct analysis in the flow cytometer. The same procedure was used for fixed and for vital cells. Results
Biparametric histograms after gold label Fig. 1 demonstrates the effect of direct GAM G40 label on the red light scatter
52
profile of murine peritoneal exudate cells (PEC). The unlabeled control in Fig. 1A shows two distinct groups of cells, the group with the lower right-angle and low-angle scatter representing the lymphocyte population and the group with high right-angle and low-angle scatter the population of macrophages. Fig. 1B shows the same cell suspension after GAM G40 label. It can be seen that the right-angle scatter signal is dramatically increased while the low-angle scatter is not significantly affected. The range of the available logarithmic scale was not sufficient to represent all clusters fully in the histogram. The centre of the clusters with high right-angle scatter is, therefore, close to the margin of the histogram. A small population of lymphocytes is seen with no label, probably representing T-cells which do not express Ig. The signal enhancement of the macrophage population was smaller than the enhancement of the B-cell population and may possibly be due to labeling of Fc receptors. The signal enhancement factors for B-cells and macrophages were approximately 16 and 6, respectively (see Materials and Methods for scale calibration). Fig. 2 shows the result of labeling 6C2 chicken erythroblasts vitally with a mouse
1@O-
108 ~ 811 ~ 60-
~ 6e-
~
~ c~ 4 e -
48.
o~
=
~ 2e1
• 2"8
4b
low
angle
6b
" eb
" 1de
scatter
" fe low
" 4b angle
6b
eb
l@I
scatter
Fig. 1. 632.8 nm light scatter profile of peritoneal exudate cells after direct labeling with 40 nm colloidal gold conjugated with goat anti-mouse Ig (GAM G40). A: Unlabeled control; B: Labeled population. Ordinate: right-angle (90 °) scatter. Logarithmic scale, 18 units correspond to factor 2 in signal amplification• Abscissa: low-angle scatter. Linear scale.
8e-
60... o 4e-
"~ 28k--
W " •
fe
low
-q '
4b
angle
Ke
eb
' t&e
scatter
Fig. 2. Chicken erythroblasts labeled vitally with a mouse monoclonal antibody against cell surface glycoprotein, followed by GAM G40 as secondary antibody. Labeled cells were mixed approximately 1 : 1 with a control population incubated with GAM G40 without the primary antibody. Abscissa: red low-angle light scatter, linear scale. Ordinate: red right-angle scatter, logarithmic scale.
53
monoclonal antibody against surface glycoprotein, followed by GAM G40 as a secondary antibody. Since all cells of the homogeneous culture were labeled, the signal enhancement is demonstrated by mixing the sample with a non-labeled control, obtained by incubation with GAM G40 without the primary antibody. The two clusters at channels 60-80 on the abscissa represent the vital fractions of labeled and non-labeled cells, whereas the two clusters at channels 20-30 on the abscissa represent the corresponding fractions of labeled and non-labeled dead cells. The identification of these clusters with reduced low-angle scatter as dead cells was verified by use of an ethidium bromide exclusion test. For demonstrating the possibility of 2-parameter labels with FITC and gold, spleen cells were labeled in various combinations: (A) anti-Thyl.2-FITC; (B) GAM G40; (C) GAM G40 + anti-Thyl.2-FITC; (D) RAMIg-FITC + a n t i - H - 2 K / I a + GAM G40. The resulting histograms are shown in Fig. 3A-D. Fig. 2A shows a fraction of 13.7% fluorescent cells, Fig. 2B a fraction of 65.1% scatter-enhanced cells, Fig. 2C a fraction of 17.1% fluorescent cells, 67.7% scatter-enhanced cells and 15.2% cells with no label at all. The fluorescent T-cells and the non-fluorescent null cells show the same control signal level in the 90 ° scatter, which demonstrates the lack of nonspecific gold label. Fig. 2D shows a fraction of 70.7% cells labeled both with FITC and gold, the remaining cells being negative for both labels. Fig. 2C demonstrates that there was no interference between FITC and gold labels for different antigenic sites on different cells, and Fig. 2D demonstrates that there was no interference between FITC and gold for different antigenic sites on the same cell.
lee-
A
B
81i~-
6o. . 4oo 2oCD C~
!
t-
I
20
40
&l
~l
•
too
2e
"
4"e
"
6"e
"
e'e
"
t6e
"
6~
"
e'e
'
lee
iBe-
D
tOI
aa
"O
60
8o-
J
L. 40
4o-
20-
4i -.°,
• •
4e
"
6e,
"
ee
green
1 2"e
"
4~
fluorescence
Fig. 3. Biparametric histograms of spleen cells labeled with gold and FITC. Abscissa: green fluorescence, logarithmic scale. Ordinate: red right-angle scatter, logarithmic scale. A: anti-Thyl.2-FITC; B: GAM G40; C: GAM G40+anti-Thyl.2-FITC; D: RAM Ig-FITC+anti-H-2K/Ia+GAM G40. Signals from debris particles have been gated out by use of the forward light scatter parameter.
54
Cell survival after gold labefing A preliminary experiment was performed on the viability of P815 cells after applying the gold label. 10 6 cells were labeled by 4 different protocols: (1) all incubation and centrifugation steps as described in Materials and Methods in PBS with no antibody; (2) primary antibody alone (anti-H-2K/Ia); (3a, b) primary + secondary antibody (GAM G40). The successful labeling in protocol 3 was verified with the flow cytometer (not shown). All probes were suspended in 2 ml of normal growth medium (RPMI 1640 + 10% FCS) and incubated at 37 ° C. After 2 days, 5 ml of fresh medium was added, and after 4 days the total amount of vital (ethidium bromide excluding) cells was determined flow cytometrically by mixing with a defined number of fluorescent beads and staining with acridine orange and ethidium bromide (B0hmer, 1984). The cultures contained: (1) (1.71 __+0.05) × 107; (2) (1.68 _+ 0 . 0 5 ) X 107; (3a) (1.71 + 0.05) × 107; (3b) (1.54 + 0.05) × 107 vital cells. This means an approximately 4 population doublings in all protocols without significant differences and indicates that gold-labeled cells remain vital and can be subjected to proliferative assays.
Dependence of signal enhancement on wavelength By use of a probe of gold-labeled spleen cells the signal amplification was measured as a function of incident laser wavelength. In Fig. 4 the factor of signal enhancement between the peaks of labeled B-cells and non-labeled T- and null cells (as shown in Fig. 2B for 632.8 nm) is plotted versus wavelength. It can be seen that the signal amplification increases from a factor of 3 at 480 nm to a factor of 11 at 632 nm.
Stability of the gold label PLP-fixed cells were labeled with gold, stored at 4 ° C in RPMI 1640 + FCS and
o
1C
E E Q)
~5 t-
'
. . . .
. . . .
660'
'
'
wavelength (nm) Fig. 4. Gold-induced scatter signal enhancement as a function of laser wavelength, determined with GAM G40 label of murine spleen cells. Signal enhancement factor between the peaks of positive and negative cells (ordinate) versus wavelength (abscissa). Vertical errors estimated as maximum errors in determination of peak locations.
55 renanalysed every day. During a period of 5 days no qualitative or quantitative change in the scatter distributions was detected, though after about 4 days a slight increase in cell debris was noted, indicating the onset of cell disintegration equally from the clusters of labeled and non-labeled cells. Gold-labeled cells were fixed overnight with 2.5% glutaraldehyde in 0.1 M sodium cacodylate. After this fixation, which is reputed to stabilize cells almost indefinitely, the scatter profiles were not different from those before fixation.
Smaller size of colloidal gold particles Cells were also labeled with GAM G20 and GAM G5. The 20 nm gold produced about 20% weaker signal enhancement than the 40 nm gold, and the 5 nm gold showed almost no signal enhancement. Higher centrifugation speeds and longer incubation times during the labeling procedure did not help. Electron microscopic inspection showed that the cells were labeled much more densely with the 5 nm gold than with the 40 nm gold. Therefore, it would appear that it is colloidal particle size which is critical for the enhancement of the right-angle scatter signal at longer wavelengths.
Other light scatter angles A preliminary study was performed (data not shown) in order to find whether the 90° light scatter analysis as provided by the commercial flow cytometer represents the optimum for detection of gold label. For this purpose an optical fibre (light conductor) was placed close to the flow chamber with a solid angle of light acceptance similar to the one of the standard detection system, and the mean detection angle then varied between approximately 60 ° and 120 °. The optimum discrimination between labeled and non-labeled cells was found in the 90 o position. In a further study, the light acceptance angle of the standard system was narrowed by placing either an iris or narrow vertical slots in front of the 90 o light collection lens. No improvement in positive/negative signal discrimination could be found as compared to the standard optical setting.
Discussion
It is shown in this paper that immuno-gold labeling of cells resulted in a drastic enhancement of the 90° red light scatter (factor 10-12 for spleen B-lymphocytes and factor 16 for peritoneal exudate B-lymphocytes) which was sufficient for the quantitative distinction between positive and negative cells. Furthermore, gold and FITC label could be applied together in a two-parametric label analysis. In this work we did not attempt to compare quantitatively the sensitivities of gold and FITC labels, which needs to be done by optimizing both labels for the same antigenic determinants, because our flow cytometer was not yet equipped for fully exploiting the potentials of the gold technique (see below). Thus we have not as yet been able to determine whether the new immuno-gold technique will be superior to FITC staining for detecting low densities of antigenic determinants, or whether it will simply provide a useful additional label for multi-parameter flow cytometry.
56 Preliminary results obtained with non-fixed cells from a permanent cell culture line indicated that gold-labeled cells retained their ability to proliferate in culture. In the case of vital cell labeling the red low-angle scatter signal offered the additional feature of a much better discrimination between vital and dead cells than is provided by the blue or green low-angle light scatter. The usual ethidium bromide counter-stain for dead cells, which would interfere with the red 90 ° scatter signal, is, therefore, not required. Further studies upon the correlation between dye exclusion and low-angle red light scatter are in progress. It seems to us that this non-fluorescent cell surface label offers a variety of possible combinations of simultaneous cell labels for multi-parametric flow cytometry. Colloidal gold particles of defined size may be prepared easily in the laboratory and coupled to a broad range of antibodies, growth factors, lectins, etc. (Geoghegan and Ackerman, 1977; Horrisberger, 1981). Our studies on the basic technical factors influencing the results of this technique were limited by the lack of a tunable red laser. Fig. 4 seems to indicate that an even better than 12-fold signal enhancement might have been achieved with wavelengths beyond 632.8 nm. Furthermore, since the red light angle signals are weak as compared to scatter signals at other angles, and due to the weak output of our laser (0.5 mW), we were analysing extremely weak signals. At the standard setting of the machine, most fluorescent dyes specific for D N A or other cell contents would significantly add to the weak scatter signals at this wavelength. There are three ways of reducing the relative contribution of fluorescence light to the measured red scatter signal: (1) Use of very narrow band-pass filter for the light scatter signal. (2) Separate the two incident laser beams, so that the scatter signal and the corresponding fluorescence signal become separated in time, and correct electronically for this delay. (3) Increase the power of the red laser. Higher power would be provided by krypton or dye lasers, which would also provide the possibility of screening different red wavelengths for optimizing signal discrimination.
Acknowledgement We wish to thank Dr. A.W. Burgess, Dr. J.D. Gardner and Dr. R.G. Ashcroft for their support during the experiments and preparation of the manuscript, and Mr. J. Papaioannou for skilfully running the flow cytometer.
References Bergroth, V., Y.T. Konttinen and S. Reitamo, 1983, J. Histochem. Cytochem. 31, 837. B6hmer, R.-M., 1984, Cell Tissue Kinet. in press. De Mey, J., M. Moeremans, G, Geuens, R. Nuydens and M. De Brabander, 1981, Cell. Biol. Int. Rep. 5, 889.
57 Geoghegan, W.D. and G.A. Ackerman, 1977, J. Histochem. Cytochem. 25, 1187. Goding, J., 1976, J. Immunol. Methods 13, 215. Horrisberger, M., 1981, Scanning Electron Microscopy, (Sem Inc., AMF O'Hare, Chicago) pp. 9-31. King, N.J.C. and E.L. Parr, 1982, Aust. J. Exp. Biol. Med. Sci. 60, 655. Loken, M.R., 1980, Cytometry 1,136. Loken, M.R., R.D. Stout and L.A. Herzenberg, 1979, Flow Cytometry and Sorting, eds. Melamed, Mullaney and Mendelsohn (Wiley, New York). McLean, I.W. and P.K. Nakane, 1974, J. Histochem. Cytochem. 22, 1077. Shapiro, H.M., 1983, Cytometry 3, 227. Steinkamp, J.A., C.C. Stewart and H.A. Crissman, 1982, Cytometry 2, 226.
49
JIM03248
Immuno-Gold Labeling for Flow Cytometric Analysis R a l p h - M . B 6 h m e r 1,, a n d N i c h o l a s J.C. K i n g 2 I Melbourne Turnout Biology Branch, Ludwig Institute for Cancer Research, P.O. Royal Melbourne Hospital Victoria 3050, and e John Curtin School of Medical Research, Australian National University, Canberra, Australia (Received 27 March 1984, accepted 17 July 1984)
Colloidal gold particles coated with goat anti-mouse immunoglobulin antibodies were used to analyse surface antigens on various cell types by flow cytometry. The gold-labeled cells showed an increasing signal amplification in the 90 o light scatter with increasing wavelength of the incident laser light, reaching a more than 10-fold amplification at 632.8 nm. This wavelength was provided by a 0.5 m W h e l i u m - n e o n laser. The magnitude of the signal amplification due to the gold label as well as the specificity of the label was sufficient for quantitative discrimination between positive and negative ceils. Cell viability was not affected by the gold label, Mouse spleen cells were labeled with various combinations of FITC- and gold-conjugated antibodies, It was found that the gold and fluorescent labels did not interfere with each other. Colloidal gold may thus be used as an additional label for multiparameter cell analysis and sorting. Biparametric cell analysis/sorting of surface antigen-labeled cells (label versus low-angle scatter) becomes possible even with a low energy h e l i u m - n e o n laser. Key words: immuno-gold label - flow cytometry - cell sorting
Introduction
The flow cytometric analysis of cell surface antigens by antibodies tagged with fluorescent dyes has become an important technique for immunological studies. Simultaneous labeling of different cell surface molecules by use of reagents labeled with different dyes has also become feasible (Loken et al., 1979). It has also been possible to combine surface labeling with specific staining of cell contents such as DNA (Loken, 1980; Steinkamp et al., 1982). While such combined labels are very useful, they are limited by the spectral overlap of the fluorescent dyes. In view of this difficulty we decided to introduce a non-fluorescent cell surface label, suitable for flow cytometry, which would provide more freedom in combining different labels. One of the possibilities of detecting a non-fluorescent label with a flow cytometer is
* To whom correspondence should be addressed. 0022-1759/84/$03.00 © 1984 Elsevier Science Publishers B.V.
50 the change in the light scatter signal (Shapiro, 1983). We found that the binding of colloidal gold particles, which are commonly used as a cell surface label for light and electron microscopic studies (Horrisberger, 1981; De M e y e t al., 1981: Bergroth et al., 1983), was able to enhance the right angle (90 °) scatter of the red laser light dramatically. This paper investigates the various factors influencing the results of the labeling technique and demonstrates its enormous potential for studies on cell surface receptors.
Materials and Methods
Cell preparation B A L B / c mice were used as donors for spleen cells and peritoneal exudate cells (PEC). Spleen cells were obtained by passing the whole organ through a fine wire mesh into RPMI 1640 and centrifuging the suspension over Ficoll to eliminate dead cells and erythrocytes. PEC were lavaged from the peritoneal cavity with cold phosphate buffered saline (PBS). Cells from the P815 line, a mastocytoma of D B A / 2 origin, were used for indirect labeling as a cell line comparison of the same d H-2 haplotype as B A L B / c mice. 6C2 is a chicken erythroblast cell line transformed by avian erythroblastosis virus. A ntisera G A M G40 (G20, G5) was obtained from Janssen Pharmaceutica, Beerse, Belgium. GAM G40 is colloidal gold of 40 nm diameter, conjugated electrostatically with affinity-purified goat anti-mouse immunoglobulin. GAM G40 was used to label the surface immunoglobulin (Ig) on murine lymphocytes directly, or as a secondary antibody (Ab) to label histocompatibility antigen ( H - 2 K / I a ) on the cell surface. Monoclonal rat anti-mouse Thyl.2-FITC and normal rat Ig were a kind gift from Dr. R. Scollay, Walter and Eliza Hall Institute, Melbourne. The anti-Thyl.2 antiserum was obtained from the supernatant fluid of the hybridoma H12 clone and conjugated to FITC after affinity purification. Anti-Thyl.2 Ab labels the Thy antigens on T cells of B A L B / c spleen and peritoneal exudate cell (PEC) populations. A monoclonal antibody against chicken erythroid cells was made by Dr. Keith Savin, Ludwig Institue for Cancer Research° Melbourne. Spleen cells from mice injected with cells from the 6C2 chicken erythroblast line were fused with NS1 mouse myeloma cells. The hybridoma produces an antibody of the IgM class which binds to cell surface glycoprotein of all chicken erythroid cells. Mouse anti-mouse H - 2 K / I a antiserum was a kind gift from Dr. R, Blanden, John Curtin School of Medical Research, Canberra. It was raised in ( A / J x DBA/1)F1 mice, injected weekly with D B A / 2 spleen cells for 4 weeks. The mice were bled and the 0.2 ffm filtered whole serum used to label the cell surface gene products of the K and Ia regions of the d haplotype. Rabbit anti-mouse Ig-FITC (RAMIg-FITC) was prepared by raising an antiserum in a New Zealand White rabbit against mouse Ig. It was passed over a Protein
51 A column and the bound portion eluted and conjugated to FITC (Goding, 1976). Cells were washed in normal rabbit serum before addition of RAMIg-FITC which was used to label Ig on B cells directly.
Cell fixation Cells were washed once in PBS and suspended for 30 min at 4 ° in freshly prepared periodate-lysine-paraformaldehyde fixative (PLP) (McLean and Nakane, 1974). This solution is made up of 1 ml paraformaldehyde (4% w / v ) in water, pH 7.3, 9 mg paraperiodic acid and 3 ml 0.1 M L-lysine in 0.1 M sodium phosphate buffer, pH 7.3. PLP is a gentle fixative maintaining the serological detectability of surface H-2, Ig and Thy antigens (King and Parr, 1982; King, unpublished observations). Thereafter the cells were washed first in PBS and then in RPMI 1640 with 10% fetal calf serum (RPMI + FCS). All washes made use of the latter medium unless stated otherwise. Exept for Ficoll separation, the cells were kept strictly at 4 o C during all preparation and labeling procedures.
Flow cytometry The experiments were performed on an Ortho System 50 cell sorter, equipped with an argon ion laser providing wavelengths ranging from 451.7 nm to 528.7 nm, and a 0.5 mW helium-neon laser providing 632.8 nm red light. The red right angle (90 ° ) scatter signals were detected with a special red-sensitive photomultiplier (EMI type D160). FITC fluorescence was excited at 488 nm and measured between 510 and 530 nm. The 90 ° scatter signals as well as the green fluorescence signals were processed by use of a logarithmic amplifier. A distance of 18 units on the histogram scales was found to correspond to a factor of 2 in signal size.
Labeling procedures The cells were incubated for 60 min with anti-H-2 antiserum and for 30 min with anti-Thyl.2-FITC and RAMIg-FITC. Each incubation was followed by three washes in RPMI + FCS before addition of the secondary antiserum. Before incubation with anti-Thyl.2-FITC or RAMIg-FITC, the cells were washed for 10 rain in normal rat Ig or normal rabbit serum, respectively, in order to occupy sites that might otherwise be labeled non-specifically by the FITC reagents. For gold labeling, aliquots of 106 cells were suspended directly in 50 ~tl of the undiluted GAM G40 (concentration of reagent not given by supplier), placed in a 2-mm, flat-bottomed centrifuge tube, spun three times for 5 min at 200 x g and carefully resuspended between the spins. After the third spin, the cells were either washed twice in 10 ml RPMI + FCS before further labeling procedures, or simply resuspended in this washing medium for direct analysis in the flow cytometer. The same procedure was used for fixed and for vital cells. Results
Biparametric histograms after gold label Fig. 1 demonstrates the effect of direct GAM G40 label on the red light scatter
52
profile of murine peritoneal exudate cells (PEC). The unlabeled control in Fig. 1A shows two distinct groups of cells, the group with the lower right-angle and low-angle scatter representing the lymphocyte population and the group with high right-angle and low-angle scatter the population of macrophages. Fig. 1B shows the same cell suspension after GAM G40 label. It can be seen that the right-angle scatter signal is dramatically increased while the low-angle scatter is not significantly affected. The range of the available logarithmic scale was not sufficient to represent all clusters fully in the histogram. The centre of the clusters with high right-angle scatter is, therefore, close to the margin of the histogram. A small population of lymphocytes is seen with no label, probably representing T-cells which do not express Ig. The signal enhancement of the macrophage population was smaller than the enhancement of the B-cell population and may possibly be due to labeling of Fc receptors. The signal enhancement factors for B-cells and macrophages were approximately 16 and 6, respectively (see Materials and Methods for scale calibration). Fig. 2 shows the result of labeling 6C2 chicken erythroblasts vitally with a mouse
1@O-
108 ~ 811 ~ 60-
~ 6e-
~
~ c~ 4 e -
48.
o~
=
~ 2e1
• 2"8
4b
low
angle
6b
" eb
" 1de
scatter
" fe low
" 4b angle
6b
eb
l@I
scatter
Fig. 1. 632.8 nm light scatter profile of peritoneal exudate cells after direct labeling with 40 nm colloidal gold conjugated with goat anti-mouse Ig (GAM G40). A: Unlabeled control; B: Labeled population. Ordinate: right-angle (90 °) scatter. Logarithmic scale, 18 units correspond to factor 2 in signal amplification• Abscissa: low-angle scatter. Linear scale.
8e-
60... o 4e-
"~ 28k--
W " •
fe
low
-q '
4b
angle
Ke
eb
' t&e
scatter
Fig. 2. Chicken erythroblasts labeled vitally with a mouse monoclonal antibody against cell surface glycoprotein, followed by GAM G40 as secondary antibody. Labeled cells were mixed approximately 1 : 1 with a control population incubated with GAM G40 without the primary antibody. Abscissa: red low-angle light scatter, linear scale. Ordinate: red right-angle scatter, logarithmic scale.
53
monoclonal antibody against surface glycoprotein, followed by GAM G40 as a secondary antibody. Since all cells of the homogeneous culture were labeled, the signal enhancement is demonstrated by mixing the sample with a non-labeled control, obtained by incubation with GAM G40 without the primary antibody. The two clusters at channels 60-80 on the abscissa represent the vital fractions of labeled and non-labeled cells, whereas the two clusters at channels 20-30 on the abscissa represent the corresponding fractions of labeled and non-labeled dead cells. The identification of these clusters with reduced low-angle scatter as dead cells was verified by use of an ethidium bromide exclusion test. For demonstrating the possibility of 2-parameter labels with FITC and gold, spleen cells were labeled in various combinations: (A) anti-Thyl.2-FITC; (B) GAM G40; (C) GAM G40 + anti-Thyl.2-FITC; (D) RAMIg-FITC + a n t i - H - 2 K / I a + GAM G40. The resulting histograms are shown in Fig. 3A-D. Fig. 2A shows a fraction of 13.7% fluorescent cells, Fig. 2B a fraction of 65.1% scatter-enhanced cells, Fig. 2C a fraction of 17.1% fluorescent cells, 67.7% scatter-enhanced cells and 15.2% cells with no label at all. The fluorescent T-cells and the non-fluorescent null cells show the same control signal level in the 90 ° scatter, which demonstrates the lack of nonspecific gold label. Fig. 2D shows a fraction of 70.7% cells labeled both with FITC and gold, the remaining cells being negative for both labels. Fig. 2C demonstrates that there was no interference between FITC and gold labels for different antigenic sites on different cells, and Fig. 2D demonstrates that there was no interference between FITC and gold for different antigenic sites on the same cell.
lee-
A
B
81i~-
6o. . 4oo 2oCD C~
!
t-
I
20
40
&l
~l
•
too
2e
"
4"e
"
6"e
"
e'e
"
t6e
"
6~
"
e'e
'
lee
iBe-
D
tOI
aa
"O
60
8o-
J
L. 40
4o-
20-
4i -.°,
• •
4e
"
6e,
"
ee
green
1 2"e
"
4~
fluorescence
Fig. 3. Biparametric histograms of spleen cells labeled with gold and FITC. Abscissa: green fluorescence, logarithmic scale. Ordinate: red right-angle scatter, logarithmic scale. A: anti-Thyl.2-FITC; B: GAM G40; C: GAM G40+anti-Thyl.2-FITC; D: RAM Ig-FITC+anti-H-2K/Ia+GAM G40. Signals from debris particles have been gated out by use of the forward light scatter parameter.
54
Cell survival after gold labefing A preliminary experiment was performed on the viability of P815 cells after applying the gold label. 10 6 cells were labeled by 4 different protocols: (1) all incubation and centrifugation steps as described in Materials and Methods in PBS with no antibody; (2) primary antibody alone (anti-H-2K/Ia); (3a, b) primary + secondary antibody (GAM G40). The successful labeling in protocol 3 was verified with the flow cytometer (not shown). All probes were suspended in 2 ml of normal growth medium (RPMI 1640 + 10% FCS) and incubated at 37 ° C. After 2 days, 5 ml of fresh medium was added, and after 4 days the total amount of vital (ethidium bromide excluding) cells was determined flow cytometrically by mixing with a defined number of fluorescent beads and staining with acridine orange and ethidium bromide (B0hmer, 1984). The cultures contained: (1) (1.71 __+0.05) × 107; (2) (1.68 _+ 0 . 0 5 ) X 107; (3a) (1.71 + 0.05) × 107; (3b) (1.54 + 0.05) × 107 vital cells. This means an approximately 4 population doublings in all protocols without significant differences and indicates that gold-labeled cells remain vital and can be subjected to proliferative assays.
Dependence of signal enhancement on wavelength By use of a probe of gold-labeled spleen cells the signal amplification was measured as a function of incident laser wavelength. In Fig. 4 the factor of signal enhancement between the peaks of labeled B-cells and non-labeled T- and null cells (as shown in Fig. 2B for 632.8 nm) is plotted versus wavelength. It can be seen that the signal amplification increases from a factor of 3 at 480 nm to a factor of 11 at 632 nm.
Stability of the gold label PLP-fixed cells were labeled with gold, stored at 4 ° C in RPMI 1640 + FCS and
o
1C
E E Q)
~5 t-
'
. . . .
. . . .
660'
'
'
wavelength (nm) Fig. 4. Gold-induced scatter signal enhancement as a function of laser wavelength, determined with GAM G40 label of murine spleen cells. Signal enhancement factor between the peaks of positive and negative cells (ordinate) versus wavelength (abscissa). Vertical errors estimated as maximum errors in determination of peak locations.
55 renanalysed every day. During a period of 5 days no qualitative or quantitative change in the scatter distributions was detected, though after about 4 days a slight increase in cell debris was noted, indicating the onset of cell disintegration equally from the clusters of labeled and non-labeled cells. Gold-labeled cells were fixed overnight with 2.5% glutaraldehyde in 0.1 M sodium cacodylate. After this fixation, which is reputed to stabilize cells almost indefinitely, the scatter profiles were not different from those before fixation.
Smaller size of colloidal gold particles Cells were also labeled with GAM G20 and GAM G5. The 20 nm gold produced about 20% weaker signal enhancement than the 40 nm gold, and the 5 nm gold showed almost no signal enhancement. Higher centrifugation speeds and longer incubation times during the labeling procedure did not help. Electron microscopic inspection showed that the cells were labeled much more densely with the 5 nm gold than with the 40 nm gold. Therefore, it would appear that it is colloidal particle size which is critical for the enhancement of the right-angle scatter signal at longer wavelengths.
Other light scatter angles A preliminary study was performed (data not shown) in order to find whether the 90° light scatter analysis as provided by the commercial flow cytometer represents the optimum for detection of gold label. For this purpose an optical fibre (light conductor) was placed close to the flow chamber with a solid angle of light acceptance similar to the one of the standard detection system, and the mean detection angle then varied between approximately 60 ° and 120 °. The optimum discrimination between labeled and non-labeled cells was found in the 90 o position. In a further study, the light acceptance angle of the standard system was narrowed by placing either an iris or narrow vertical slots in front of the 90 o light collection lens. No improvement in positive/negative signal discrimination could be found as compared to the standard optical setting.
Discussion
It is shown in this paper that immuno-gold labeling of cells resulted in a drastic enhancement of the 90° red light scatter (factor 10-12 for spleen B-lymphocytes and factor 16 for peritoneal exudate B-lymphocytes) which was sufficient for the quantitative distinction between positive and negative cells. Furthermore, gold and FITC label could be applied together in a two-parametric label analysis. In this work we did not attempt to compare quantitatively the sensitivities of gold and FITC labels, which needs to be done by optimizing both labels for the same antigenic determinants, because our flow cytometer was not yet equipped for fully exploiting the potentials of the gold technique (see below). Thus we have not as yet been able to determine whether the new immuno-gold technique will be superior to FITC staining for detecting low densities of antigenic determinants, or whether it will simply provide a useful additional label for multi-parameter flow cytometry.
56 Preliminary results obtained with non-fixed cells from a permanent cell culture line indicated that gold-labeled cells retained their ability to proliferate in culture. In the case of vital cell labeling the red low-angle scatter signal offered the additional feature of a much better discrimination between vital and dead cells than is provided by the blue or green low-angle light scatter. The usual ethidium bromide counter-stain for dead cells, which would interfere with the red 90 ° scatter signal, is, therefore, not required. Further studies upon the correlation between dye exclusion and low-angle red light scatter are in progress. It seems to us that this non-fluorescent cell surface label offers a variety of possible combinations of simultaneous cell labels for multi-parametric flow cytometry. Colloidal gold particles of defined size may be prepared easily in the laboratory and coupled to a broad range of antibodies, growth factors, lectins, etc. (Geoghegan and Ackerman, 1977; Horrisberger, 1981). Our studies on the basic technical factors influencing the results of this technique were limited by the lack of a tunable red laser. Fig. 4 seems to indicate that an even better than 12-fold signal enhancement might have been achieved with wavelengths beyond 632.8 nm. Furthermore, since the red light angle signals are weak as compared to scatter signals at other angles, and due to the weak output of our laser (0.5 mW), we were analysing extremely weak signals. At the standard setting of the machine, most fluorescent dyes specific for D N A or other cell contents would significantly add to the weak scatter signals at this wavelength. There are three ways of reducing the relative contribution of fluorescence light to the measured red scatter signal: (1) Use of very narrow band-pass filter for the light scatter signal. (2) Separate the two incident laser beams, so that the scatter signal and the corresponding fluorescence signal become separated in time, and correct electronically for this delay. (3) Increase the power of the red laser. Higher power would be provided by krypton or dye lasers, which would also provide the possibility of screening different red wavelengths for optimizing signal discrimination.
Acknowledgement We wish to thank Dr. A.W. Burgess, Dr. J.D. Gardner and Dr. R.G. Ashcroft for their support during the experiments and preparation of the manuscript, and Mr. J. Papaioannou for skilfully running the flow cytometer.
References Bergroth, V., Y.T. Konttinen and S. Reitamo, 1983, J. Histochem. Cytochem. 31, 837. B6hmer, R.-M., 1984, Cell Tissue Kinet. in press. De Mey, J., M. Moeremans, G, Geuens, R. Nuydens and M. De Brabander, 1981, Cell. Biol. Int. Rep. 5, 889.
57 Geoghegan, W.D. and G.A. Ackerman, 1977, J. Histochem. Cytochem. 25, 1187. Goding, J., 1976, J. Immunol. Methods 13, 215. Horrisberger, M., 1981, Scanning Electron Microscopy, (Sem Inc., AMF O'Hare, Chicago) pp. 9-31. King, N.J.C. and E.L. Parr, 1982, Aust. J. Exp. Biol. Med. Sci. 60, 655. Loken, M.R., 1980, Cytometry 1,136. Loken, M.R., R.D. Stout and L.A. Herzenberg, 1979, Flow Cytometry and Sorting, eds. Melamed, Mullaney and Mendelsohn (Wiley, New York). McLean, I.W. and P.K. Nakane, 1974, J. Histochem. Cytochem. 22, 1077. Shapiro, H.M., 1983, Cytometry 3, 227. Steinkamp, J.A., C.C. Stewart and H.A. Crissman, 1982, Cytometry 2, 226.