Journal of Reproductive Immunology, 21 (1992) 71-85
71
Elsevier Scientific Publishers Ireland Ltd.
JRI 00746
Isolation and characterisation of a subpopulation of human chorionic cytotrophoblast using a monoclonal anti-trophoblast antibody (NDOG2) in flow cytometry J.J. C a u l f i e l d , I.L. S a r g e n t , B.L. F e r r y , P . M . S t a r k e y a n d C.W.G. Redman Harris Birthright CentreJbr Pre-eclampsia Research, Nuff~eld Department ~[ Ohstetrics and Gynaecology, John Radcli['[~" Itospital, O.\'[brd OX3 9DU ( U.K. ) (Accepted for publication 20 August 1991)
Summary Human cytotrophoblast cells, isolated from term amniochorion by enzymic digestion and Percoll gradient centrifugation, were characterised by flow cytometry. A panel of 12 anti-trophoblast monoclonal antibodies was screened for labelling of these cells in flow cytometry and the results compared with immunoperoxidase labelling ofcytospin preparations and tissue sections. All 12 antibodies were positive for trophoblast on tissue sections, 11/12 were positive on cytospins but only two (NDOG2 and GB25) gave consistent results in flow cytometry. Two-colour labelling with NDOG2 and W6/32, an antibody to HLA-A, -B, -C, demonstrated that 88% of the NDOG2-positive cells also express Class 1 major histocompatibility complex (MHC) antigens. The NDOG2-positive cytotrophoblast subpopulation was isolated by flow cytometry in sufficient purity (>95'¼,) and yield (3.1 x 10 6) for use in functional studies in vitro. Key words: human cytotrophoblast," monoclonal antibody," flow cytometry
Correspondence to." Mr J.J. Caulfield, Harris Birthright Centre for Pre-eclampsia Research, Nuffield Department of Obstetrics and Gynaecology, John Radcliffe Hospital, Oxford OX3 9DU, U.K. 0165-0378/91/$03.50 © 1991 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
72
Introduction
Investigation of the function of cytotrophoblast and the way in which it interacts with maternal tissues requires the isolation of pure populations of cells. The chorion laeve is a readily accessible source of extravillous cytotrophoblast. We have previously developed a method of enzymic digestion to obtain these cells (Ellis et al., 1986) and used flow cytometry to prepare highly purified cytotrophoblast by a technique based on their lack of expression of classical HLA-A, -B and -C antigens (Shorter et al., 1990). The cytotrophoblast isolated by this method contains many cells which express HLA-G (Ellis et al., 1986; Ellis et al., 1990), but they are still a heterogeneous population in terms of their reactivity with W6/32 (an antibody which recognises HLA-G as well as HLA-A, -B, -C) and various trophoblast markers (Shorter et al., 1990). Immunohistological studies of amniochorion have shown the chorionic cytotrophoblast to be comprised of at least three antigenically distinct populations (Bulmer and Johnson, 1985). Population 1 is predominantly located on the fetal aspect and reacts with monoclonal antibodies to cytokeratin and placental alkaline phosphatase (PLAP), but is negative for Class I MHC antigens. Population 2 contains relatively few cells and is positive for cytokeratin and PLAP, and has variable expression of Class I MHC antigens. The third population is located predominantly on the maternal aspect and is negative for PLAP, but positive for cytokeratin and Class I MHC. The difference in antigen expression between the fetal and maternal aspects may be important in relation to the function of these trophoblast subpopulations. These immunohistological studies were carried out using serial cryostat sections to determine the relationship between the trophoblast subpopulations, but this is a subjective technique. A quantitative assessment of the various subpopulations and their relative antigenic expression can only be carried out using double antibody labelling in flow cytometry. The purpose of the present study was to develop methods for analysing and sorting subpopulations of trophoblast from the amniochorion so that their functions can be investigated. An essential first step was to determine which of the monoclonal anti-trophoblast antibodies available to us could be used in flow cytometry. We have therefore screened a panel of antitrophoblast monoclonal antibodies which were reactive with term chorionic cytotrophoblast on tissue sections, against cell dispersions of these tissues, comparing immunoperoxidase labelling of cytospin preparations with immunofluorescence labelling in flow cytometry. Only two antibodies consistently labelled trophoblast in flow cytometry. One antibody (NDOG2), specific for placental alkaline phosphatase (Sunderland et al., 1981) was investigated further. The proportion of NDOG2-positive cells which expressed
73
Class I MHC antigens was measured, and a method was developed for isolating pure populations of NDOG2-positive cells for functional studies. Materials and Methods
Preparation of cytotrophoblast cells Term placentae from normal pregnancies were obtained within 30 min of a Caesarean section or uncomplicated vaginal delivery. The cells were prepared by enzymic digestion of the chorionic membrane after the maternal decidua and amniotic epithelium had been removed, followed by centrifugation over Percoll to remove dead cells and erythrocytes, as previously described (Ellis et al., 1986; Shorter et al., 1990). Briefly, the decidua parietalis was separated from the chorion by gentle scraping with a razor blade, the remaining chorionic membrane was cut into 10 cm squares, weighed and incubated in 20 ml of RPMI 1640 (GIBCO-Europe), containing 25 U protease type XIV (Sigma), per gram of tissue for 1 h at 37°C in 5% CO2 in air. The membrane was washed twice in phosphate buffered saline (PBS) and then incubated in 20 ml RPMI 1640 with 5% fetal calf serum (FCS, GIBCO) containing 1450 U collagenase type IV (Sigma) and 2950 U hyaluronidase type I-S (Sigma) per gram of tissue for a further 90 min. The resultant cell digest was filtered through a 100/~m gauze and washed twice in PBS. The cells were allowed to recover overnight at 4°C and then layered onto a 60%/25% Percoll (Pharmacia), stepwise gradient to remove dead cells and erythrocytes. The cells were counted and their viability determined by phase contrast microscopy.
Monoclonal antibodies Details of the antibodies used are given in Table 1. Six of the 12 antitrophoblast antibodies detected 'pan-trophoblast' markers (i.e. labelled both cytotrophoblast and syncytiotrophoblast); four labelled cytotrophoblast only and two syncytiotrophoblast only. The antibodies were used as neat culture supernatants, except for the following which were diluted in PBS containing 20 mM glucose and 5% normal human serum (PGN) as shown: W6/32 (1:100), CR3/43 (1:8), NDOG2 (1:4), JMB2 (1:2) and V9 (1:2). For double antibody immunofluorescence labelling studies, antibodies and fluorochromes were diluted in PGN as follows: W6/32-FITC (Serotec), at 1:50, NDOG2-biotin (purified by affinity chromatography on rabbit antimouse Ig) Sepharose and conjugated in this laboratory by the method of Wagener et al. (1983) at 1:30, mouse Ig-FITC (Coulter Clone, Luton, U.K.) at 1:200, mouse Ig-biotin (conjugated in this laboratory as above) at 1:100 and streptavidin-phycoerythrin (Serotec) at 1:50.
74
TABLE 1 Monoclonal antibodies used in study. Antibody
Specificity
Reference
N DOG2
Placental-type alkaline phosphatase Syncytiotrophoblast and cytotrophoblast Syncytiotrophoblast Chorionic cytotrophoblast Syncytiotrophoblast Chorionic cytotrophoblast Syncytiotrophoblast Chorionic cytotrophoblast TLX (CD46), cytotrophoblast and syncytiotrophoblast Syncytiotrophoblast Villous cytotrophoblast Chorionic cytotrophoblast Cytotrophoblast cell islands Villous cytotrophoblast Villous cytotrophoblast
Sunderland et al., 1981
GB25 GB36 H315 H316 ED341 NDOG5 18/BA/5 ED235 HMFG2
ED822 CA2
H u m a n milk fat globule protein epithelial cells (Chorionic cytotrophoblast) Syncytiotrophoblast
W6/32 CR3/43 JMB2 V9
Cancer antigen (Syncytiotrophoblast) H LA-A, - B, -C H L A - D R , -DP, -DQ Cytokeratin Vimentin
NKH1
CD56
Hsi and Yeh, 1986 Hsi et al., 1987 Johnson et al., 1981 Johnson et al., 1981 Contractor and Sooranna, 1986 Shorter et al. (unpublished) Loke and Day, 1984 Contractor and Sooranna, 1986 Burchell et al., 1983
Contractor and Sooranna, 1986 Ashall et al., 1982
Barnstable el al., t979 Naiem et al., 1981 McGee et al., 1982 Mason el al. (unpublished) Griffin et al., 1983
Immunoperoxidase staining of tissue sections and cytospin preparations Cryostat sections (7 ~m thick) were cut from rolls of term amniochorion. Suspensions of cytotrophoblast cells, at a concentration of I x 10S/ml in PBS/0.1% BSA, were centrifuged onto single well PTFE coated slides (Hendley Ltd., Essex, U.K.) at 19 g for 10 min (Shandon Inc, Pittsburgh,
U.S.A.). The slides were air dried overnight, fixed in acetone for 10 min and air dried for a further 30 min. They were then incubated with either 30 ~l of monoclonal antibody or PBS (used as a negative control) in a humid
75
chamber for 60 min at room temperature. After washing in PBS, the slides were incubated for 60 min with 30 ~1 peroxidase-conjugated rabbit anti(mouse Ig) IgG (Dakopatts) diluted at 1:50 in PBS with 10% normal human serum, washed again in PBS and the bound peroxidase visualised by incubating with 50/A 0.06 mg/ml diaminobenzidine tetrahydrochloride (Sigma) plus 0.06% H202 for 5 min. The preparations were counterstained with haematoxylin, dehydrated and mounted in DPX. One hundred cells were examined by light microscopy and the percentage positive for each antibody was recorded.
Immunofiuorescence labelling for flow cytometry For single antibody labelling, 1 x l 0 6 chorionic cytotrophoblast cells were incubated with either 50 #1 of monoclonal antibody or PBS control for 30 min on ice and washed twice by centrifugation at 9000 x g for 5 s in PBS containing 20 mM glucose and 0.1% BSA (PGB). The cells were then incubated with 50 ~1 of fluorescein isothiocyanate-conjugated rabbit antimouse Ig (Ro~M-FITC) (Sigma) at a concentration of 1:50 in PGN for 30 min on ice. For the last 5 min, 2/A of 0.5 mg/ml propidium iodide (Sigma) was added to each sample. For double antibody labelling, 1 x l06 cells were incubated with 50 ~1 W6/32-FITC for 30 min on ice. After washing in PGB, the cells were incubated with 50/~1 NDOG2-biotin for a further 30 min, then washed and incubated with 50 ~1 streptavidin-phycoerythrin (SAPE) for 30 min. The controls used were mouse Ig-FITC, Ig-biotin and for the double labelled samples Ig-FITC and Ig-biotin together. The cells were finally washed twice in PGB and analysed immediately on the flow cytometer.
Flow cytometry Live cells were analysed using a Coulter EPICS 541 flow cytometer fitted with a 100 ~m flow cell tip. The excitation wavelength used was the 488 nm line of the argon-ion laser. In the single labelling experiments, debris was excluded by gating on forward angle light scatter (FALS), a measure of cell size, and log peak 90 ° light scatter (LP90), a measure of cell granularity. Dead cells, which may bind antibody non-specifically, were excluded by gating out on log peak red fluorescence (LPR) those cells which had taken up propidium iodide. Single parameter histograms of log peak green fluorescence (LPG) were then collected on 1 x 10 4 cells, and the percentage positive for each antibody compared to the Ro~M-FITC control was calculated using the Coulter 'Immuno' programme. In the double labelling experiments it was not possible to measure the green fluorescence of FITC and the red fluorescence of both phycoerythrin and propidium iodide simultaneously, as the EPICS 541 has only two
76
fluorescence detectors. To overcome this problem an aliquot of the cells was labelled with propidium iodide and the red fluorescence gate set as before to exclude the dead cells (Fig. lb). This gated sample was then analysed on a FALS and LPg0 histogram (Fig. lc) which when compared to the same histogram from an ungated sample (Fig. la), was found to exclude the population of cells indicated by the arrow (Fig. la). Thus most of the dead cells could be clearly distinguished on the basis of their size and granularity and a multi-sided gate (bit-map) was drawn to exclude them and the remaining debris (Fig. lc). All subsequent analysis in the double labelling exa)
b) LP90
Cell No.
dead cells
17.2% dead cells
I
L
LPR
FALS
c)
d)
LP90
Map 1
8
.
.
.
.
, ~.-*~7 ¢-
,,:,:..:::.::.:.:,..:..
Cell No. 7.34% dead cells
' ..~J.:~:~.:~i~' ~ L~"r' . .r:":i~'::i'i::i:::~i ' "~i-':::~:%~' - "i':"'i:i:i:?i'!: .
..... •i, ' . ' ' ".I::??:Z~I:i:~:~",~IlU~:":".' .~: ,:.:~::, . ~ ' : . ~ : : : % . - " •:':::::: • , .:,L ::::..'?:;:.:,.%:~.":'-...~:~-"
:~i:i'.,,:. ':: :~'~:f:-"~'-':-" :,i.':i,
%::,::. :
LPR
Fig. 1. Exclusion of dead cells for analysis of double antibody labelling of chorionic cytotrophoblast cell by flow cytometry. (a) Size (FALS) a n d g r a n u l a r i t y (LPg0) histogram of chorionic cytotrophoblast cell dispersions. (b) Red fluorescence (LPR) of cells labelled with propidium iodide, G a t e s e t to exclude dead cells identified as PI positive. (c) FALS and LPg0 histogram of cytotrophoblast g a t e d on LPR as indicated in Fig. lb to exclude dead cells. Bit-map drawn to include live cells only (map 1). (d) LPR of cells within bit-map 1 of Fig. lc labelled with propidium iodide to s h o w t h e e f f e c t i v e n e s s of the procedure in excluding most dead cells.
dispersions
77
periments was performed using this bit-map (Fig. 1, map 1). Figure ld shows the effectiveness of this procedure in excluding dead cells from the analysis as the number of propidium iodide positive cells is substantially reduced. In 8 experiments the median viability of the population gated on map I was 89% (range 80-96). The method of analysis of the double labelled samples is described in the results section.
Cell sorting To sort cytotrophoblast, 50 x l 0 6 chorionic cells were incubated with 2.5 ml NDOG2 antibody (1:5) or mouse-Ig control (1:100) for 30 min at 4°C, washed twice in PGB and then incubated with 2.5 ml of FITC-conjugated goat anti-mouse IgG (1:50) for 30 min. After 25 min, 2 #1 of 0.5 mg/ml propidium iodide was added to each sample and incubated on ice for the remaining 5 min, before being washed twice in PGB and re-filtered through a 100 /~m gauze. The cells were resuspended at 4.0 × 106/ml in PGB and sorted on the flow cytometer at 1500 cells/s using a 100/zm flow-cell tip and one droplet sorting with coincidence on. Results
Screening of anti-trophoblast antibody panel All the anti-trophoblast monoclonal antibodies labelled at least some populations of cytotrophoblast on cryostat sections of amniochorion (Table 2). The pan-trophoblast markers labelled cytotrophoblast cells across the full width of the chorion apart from NDOG2 which was found only on the central and fetal aspects. Of the cytotrophoblast markers, 18/BA/5 and ED235 labelled all of the cytotrophoblast whereas NDOG5 was predominantly on the maternal aspect and H M F G 2 predominantly on the fetal aspect. The syncytiotrophoblast markers (CA2 and ED822) both gave weak staining, with CA2 being localised to the central and fetal aspects and ED822 weakly labelling the maternal aspect. On isolated term chorionic cytotrophoblast the monoclonal antibodies W6/32 and CR3/43 were used both on cytospins and in flow cytometry to demonstrate the purity of the populations obtained. W6/32 recognises not only HLA-A, -B, -C but also the HLA-G on cytotrophoblast (Ellis et al., 1986; Ellis et al., 1990) and therefore labels the majority of cells in these preparations. CR3/43 labels Class II MHC positive cells and reflects the level of contamination of these preparations with maternal and fetal macrophages, which was low. The antibodies to pan-trophoblast and cytotrophoblast specific markers labeled the cytotrophoblast to a similar extent using immunoperoxidase on cytospins (Table 3) but only NDOG2 and GB25 consistently labelled in flow
78
TABLE 2 I m m u n o p e r o x i d a s e labelling o f s e c t i o n s o f a m n i o c h o r i o n : tissue d i s t r i b u t i o n m o n o c l o n a l a n t i b o d i e s . A n t i b o d i e s g r o u p e d a c c o r d i n g to t h e i r specific±ties. Antibody
of antitrophoblast
Aspect of chorionic cytotrophoblast
Pan-trophoblast N DOG2
Maternal
Central
Fetal
-
+
+
GB25 GB36 H315 H316
+ + +
+ + + +
+ + + +
ED341
+
+
+
+
+
-
+ + -
+ + ±
+ + +
± -
+
+
Cytotrophoblast NDOG5 18/BA/5 ED235 HMFG2 Syncytiotrophoblast ED822 CA2
cytometry (Table 4). The proportions of labelled cells were similar between both methods (55"/,, and 58%, respectively for NDOG2 and 59% and 61% for GB25). GB36 also labelled in both systems, but less consistently, with fewer positive cells in flow cytometry. As would be expected, neither of the two antibodies specific for syncytiotrophoblast gave significant staining in flow cytometry, although 50% of the cells on cytospins stained weakly with CA2. TABLE 3 l m m u n o p e r o x i d a s e labelling o f c h o r i o n i c c y t o t r o p h o b l a s t cell d i s p e r s i o n s o n c y t o s p i n s . A n t i b o d i e s g r o u p e d a c c o r d i n g to t h e i r specific±ties. NS = n o s t a i n i n g . Pan-trophoblast
Cytotrophoblast
Syncytiotrophoblast
NDOG2 GB25
55 ( 4 5 - 6 5 ) a 59 ( 4 0 - 6 5 )
NDOG5 18/BA/5
55 ( 5 2 - 7 0 ) 42 ( 0 - 4 4 )
ED822 CA2
NS 50 ( 3 1 - 7 5 )
GB36 H315 H316 ED341
60 50 47 68
ED235 HMFG2
63 ( 4 4 - 7 5 ) 50 ( 2 5 - 6 0 )
MHC W6/32 CR3/43
91 ( 6 5 - 9 5 ) 10 ( 5 - 2 8 )
(45-80) (33-57) (41-60) (50-90)
a p e r c e n t a g e positive cells, m e d i a n ( r a n g e ) f r o m 8 e x p e r i m e n t s .
79 TABLE 4 Immunofluorescence labelling ofchorionic cytotrophoblast cell dispersions in flow cytometry. Antibodies grouped according to their specificities. Pan-trophoblast NDOG2 GB25 GB36 H315 H316 ED341
58 61 40 7 0 9
(28-76) a (34-70) (5-64) (0-53) (0-0) (0-26)
Cytotrophoblast
Syncytiotrophoblast
NDOG5 18/BA/5 ED235 HMFG2
ED822 CA2
2 (0-66) 0 (0-0) 0 (0-20) 0(0-8)
5 (0-26) 4 (0-281
MttC
W6/32 CR3/43
89 (67-95) 5 (I-6)
apercentage positive cells, median (range) from 8 experiments.
Double labelling with W6/32 and NDOG2 The purpose of these experiments was to investigate the MHC Class I expression of the NDOG2-positive subpopulation of the chorionic cytotrophoblast. W6/32-FITC (green) and NDOG2-biotin detected with streptavidin-phycoerythrin (SAPE) (red) were used as, from immunohistology, it was expected that there would be three cell populations: two positive for only one antibody, and one for both (Bulmer and Johnson, 1985). Analysis of double labelled preparations of chorionic cytotrophoblast directly on histograms of green versus red fluorescence gave no discrete populations (Fig. 2a). This was due to both the variation in the density of antigen expression on the cells and their strong background autofluorescence. This problem was overcome in the following way. Cells labelled with NDOG2-biotin-SAPE, gated on FALS and LP~0 to eliminate dead cells as described above (Fig. 1, map 1), were analysed on a histogram of LPR and LPg0(Fig. 2b). LP90 was used rather than FALS as the cells are more uniform in granularity than size (see Fig. la). A second bit-map (Fig. 2, map 2) was drawn to include only cells with a red fluorescence greater than that of the Ig-biotin-SAPE control (not shown). The NDOG2-positive cells within map 2 were then analysed on a LPG vs. LP~0 histogram to determine their green autofluorescence. A third bit-map (map 3) was drawn to exclude the autofluorescence from the analysis (Fig. 2c). Cells labelled with both W6/32-FITC and NDOG2-biotin-SAPE were then analysed on a LPG vs. LPg0 histogram and the number of cells included within bit-map 3 (those which are both NDOG2 and W6/32-positive), was determined (Fig. 2d). When the Ig-FITC and Ig-biotin-SAPE controls were similarly analysed no cells were found within bit map 3 (data not shown).
LP90
WC
LP(;
Fig. 2. Analysis ofchorionic cytotrophoblast cell dispersions double labelled with NDOG2-biotin-SAPS and W6/32-FITC. The sample was gated on PALS and LP,,, as illustrated in Ftg. Ic. (a) Conventional green versus red fluorescence (LPG vs. LPR) analysis of double antibody labelled sample showing absence of distinct positive and negative populations. (b) Cells labelled with NDOG2-biotin-SAPE analyscd on LPR and LP,,,. Bit-map 2 drawn to include the majority of NDOGI-positive cells. (c) NDOG2-positive cells from bit-map 2 analysed for green tluorescence (LPG) and LP,,,. Bit-map 3 dra\vn to exclude autofluorescent cells and non-specific binding (d) Cells double labelled wtth NDOG?-btottnSAPE and W6/32-FITC analysed using bit-map 3 after gating on bit-map 2. Cells included within hit-map 3 are positive
for both
NDOGZ
and W6!32.
The percentage of cells positive for either W6/32 or NDOGZ was also determined on single labelled samples using the Immuno programme. The results of 8 experiments are given in Table 5. It is clear from both the single and double labelled samples that the majority (88%) of the NDOG2 positive cells analysed are also W6/32 positive.
81
TABLE 5 Immunofluorescence double labelling of chorionic cytotrophoblast with N D O G 2 and W6/32 in flow cytometry. % W6/32 +ve cells a % N D O G 2 +ve cells a % N D O G 2 +ve cells which were also W6/32 +ve b
93 (78-98) c 53 (30-73) 88 (78-91)
aSingle antibody labelled samples. bDouble antibody labelled samples. CMedian percentage positive cells (range) from 8 experiments.
Isolation of NDOG2-positive cells by flow cytometry It was not possible to sort on the basis of granularity (LPg0) and green fluorescence (LPG), the parameters that were used for the double labelling analysis, because it would require a quartz 100/zm flow cell tip which is not available on the EPICS 541 flow cytometer. NDOG2-positive cells were therefore gated for sorting using FALS vs. LPG histograms as follows. The mouse IgG Ro~M-FITC control was first analysed on a two-parameter histogram of FALS and LPG, gated on FALS, LP90 and LPR to exclude dead cells and debris as described above, and a bit-map drawn to define the negative cell population (Fig. 3a). This bit-map was then superimposed on a similar histogram of cells labelled with NDOG2 and Ro~M-FITC, and used to construct a second bit-map around the NDOG2-positive population (Fig. 3b). The NDOG2-positive and negative populations were then sorted using these bit-maps. The purity of the sorted populations was assessed by reanalysis on histograms of LPG and LP90, gated on FALS and LPR (Fig. 3c and 3d) and they were further characterised using a panel of antibodies on cytospins. To re-label the NDOG2-positive cells after sorting, the slides were preincubated in neat NHS for 30 min to prevent non-specific binding, then washed in PBS and labelled with the appropriate monoclonal antibodies. Two additional antibodies were used to characterise the sorted populations: JMB2, which reacts with cytokeratin and is a marker for trophoblast in these tissues (Redman et al., 1984) and V9, which reacts with vimentin, a cytoskeletal component which is absent from trophoblast (Contractor et al., 1984). NKH1 (CD56) was used as a negative control to determine whether there was any binding of the monoclonal antibodies to the Ro~M-FITC previously bound to the NDOG2-positive cells. The results from three preparations are summarised in Table 6. Before sorting, the average viability was 80%, with a purity of 50% NDOG2-positive cells and 79% JMB2-positive. After sorting, the NDOG2-positive cells were
82
a)
b) LPG
LPG
•
.
:
,.::
."
.
Z:
.
" :" " " '
.
.
.+~
"
:.41+
~
.
,
"~+rk-
.,z,8',.'.,~:;~::~.~.~:
71
Elsevier Scientific Publishers Ireland Ltd.
JRI 00746
Isolation and characterisation of a subpopulation of human chorionic cytotrophoblast using a monoclonal anti-trophoblast antibody (NDOG2) in flow cytometry J.J. C a u l f i e l d , I.L. S a r g e n t , B.L. F e r r y , P . M . S t a r k e y a n d C.W.G. Redman Harris Birthright CentreJbr Pre-eclampsia Research, Nuff~eld Department ~[ Ohstetrics and Gynaecology, John Radcli['[~" Itospital, O.\'[brd OX3 9DU ( U.K. ) (Accepted for publication 20 August 1991)
Summary Human cytotrophoblast cells, isolated from term amniochorion by enzymic digestion and Percoll gradient centrifugation, were characterised by flow cytometry. A panel of 12 anti-trophoblast monoclonal antibodies was screened for labelling of these cells in flow cytometry and the results compared with immunoperoxidase labelling ofcytospin preparations and tissue sections. All 12 antibodies were positive for trophoblast on tissue sections, 11/12 were positive on cytospins but only two (NDOG2 and GB25) gave consistent results in flow cytometry. Two-colour labelling with NDOG2 and W6/32, an antibody to HLA-A, -B, -C, demonstrated that 88% of the NDOG2-positive cells also express Class 1 major histocompatibility complex (MHC) antigens. The NDOG2-positive cytotrophoblast subpopulation was isolated by flow cytometry in sufficient purity (>95'¼,) and yield (3.1 x 10 6) for use in functional studies in vitro. Key words: human cytotrophoblast," monoclonal antibody," flow cytometry
Correspondence to." Mr J.J. Caulfield, Harris Birthright Centre for Pre-eclampsia Research, Nuffield Department of Obstetrics and Gynaecology, John Radcliffe Hospital, Oxford OX3 9DU, U.K. 0165-0378/91/$03.50 © 1991 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
72
Introduction
Investigation of the function of cytotrophoblast and the way in which it interacts with maternal tissues requires the isolation of pure populations of cells. The chorion laeve is a readily accessible source of extravillous cytotrophoblast. We have previously developed a method of enzymic digestion to obtain these cells (Ellis et al., 1986) and used flow cytometry to prepare highly purified cytotrophoblast by a technique based on their lack of expression of classical HLA-A, -B and -C antigens (Shorter et al., 1990). The cytotrophoblast isolated by this method contains many cells which express HLA-G (Ellis et al., 1986; Ellis et al., 1990), but they are still a heterogeneous population in terms of their reactivity with W6/32 (an antibody which recognises HLA-G as well as HLA-A, -B, -C) and various trophoblast markers (Shorter et al., 1990). Immunohistological studies of amniochorion have shown the chorionic cytotrophoblast to be comprised of at least three antigenically distinct populations (Bulmer and Johnson, 1985). Population 1 is predominantly located on the fetal aspect and reacts with monoclonal antibodies to cytokeratin and placental alkaline phosphatase (PLAP), but is negative for Class I MHC antigens. Population 2 contains relatively few cells and is positive for cytokeratin and PLAP, and has variable expression of Class I MHC antigens. The third population is located predominantly on the maternal aspect and is negative for PLAP, but positive for cytokeratin and Class I MHC. The difference in antigen expression between the fetal and maternal aspects may be important in relation to the function of these trophoblast subpopulations. These immunohistological studies were carried out using serial cryostat sections to determine the relationship between the trophoblast subpopulations, but this is a subjective technique. A quantitative assessment of the various subpopulations and their relative antigenic expression can only be carried out using double antibody labelling in flow cytometry. The purpose of the present study was to develop methods for analysing and sorting subpopulations of trophoblast from the amniochorion so that their functions can be investigated. An essential first step was to determine which of the monoclonal anti-trophoblast antibodies available to us could be used in flow cytometry. We have therefore screened a panel of antitrophoblast monoclonal antibodies which were reactive with term chorionic cytotrophoblast on tissue sections, against cell dispersions of these tissues, comparing immunoperoxidase labelling of cytospin preparations with immunofluorescence labelling in flow cytometry. Only two antibodies consistently labelled trophoblast in flow cytometry. One antibody (NDOG2), specific for placental alkaline phosphatase (Sunderland et al., 1981) was investigated further. The proportion of NDOG2-positive cells which expressed
73
Class I MHC antigens was measured, and a method was developed for isolating pure populations of NDOG2-positive cells for functional studies. Materials and Methods
Preparation of cytotrophoblast cells Term placentae from normal pregnancies were obtained within 30 min of a Caesarean section or uncomplicated vaginal delivery. The cells were prepared by enzymic digestion of the chorionic membrane after the maternal decidua and amniotic epithelium had been removed, followed by centrifugation over Percoll to remove dead cells and erythrocytes, as previously described (Ellis et al., 1986; Shorter et al., 1990). Briefly, the decidua parietalis was separated from the chorion by gentle scraping with a razor blade, the remaining chorionic membrane was cut into 10 cm squares, weighed and incubated in 20 ml of RPMI 1640 (GIBCO-Europe), containing 25 U protease type XIV (Sigma), per gram of tissue for 1 h at 37°C in 5% CO2 in air. The membrane was washed twice in phosphate buffered saline (PBS) and then incubated in 20 ml RPMI 1640 with 5% fetal calf serum (FCS, GIBCO) containing 1450 U collagenase type IV (Sigma) and 2950 U hyaluronidase type I-S (Sigma) per gram of tissue for a further 90 min. The resultant cell digest was filtered through a 100/~m gauze and washed twice in PBS. The cells were allowed to recover overnight at 4°C and then layered onto a 60%/25% Percoll (Pharmacia), stepwise gradient to remove dead cells and erythrocytes. The cells were counted and their viability determined by phase contrast microscopy.
Monoclonal antibodies Details of the antibodies used are given in Table 1. Six of the 12 antitrophoblast antibodies detected 'pan-trophoblast' markers (i.e. labelled both cytotrophoblast and syncytiotrophoblast); four labelled cytotrophoblast only and two syncytiotrophoblast only. The antibodies were used as neat culture supernatants, except for the following which were diluted in PBS containing 20 mM glucose and 5% normal human serum (PGN) as shown: W6/32 (1:100), CR3/43 (1:8), NDOG2 (1:4), JMB2 (1:2) and V9 (1:2). For double antibody immunofluorescence labelling studies, antibodies and fluorochromes were diluted in PGN as follows: W6/32-FITC (Serotec), at 1:50, NDOG2-biotin (purified by affinity chromatography on rabbit antimouse Ig) Sepharose and conjugated in this laboratory by the method of Wagener et al. (1983) at 1:30, mouse Ig-FITC (Coulter Clone, Luton, U.K.) at 1:200, mouse Ig-biotin (conjugated in this laboratory as above) at 1:100 and streptavidin-phycoerythrin (Serotec) at 1:50.
74
TABLE 1 Monoclonal antibodies used in study. Antibody
Specificity
Reference
N DOG2
Placental-type alkaline phosphatase Syncytiotrophoblast and cytotrophoblast Syncytiotrophoblast Chorionic cytotrophoblast Syncytiotrophoblast Chorionic cytotrophoblast Syncytiotrophoblast Chorionic cytotrophoblast TLX (CD46), cytotrophoblast and syncytiotrophoblast Syncytiotrophoblast Villous cytotrophoblast Chorionic cytotrophoblast Cytotrophoblast cell islands Villous cytotrophoblast Villous cytotrophoblast
Sunderland et al., 1981
GB25 GB36 H315 H316 ED341 NDOG5 18/BA/5 ED235 HMFG2
ED822 CA2
H u m a n milk fat globule protein epithelial cells (Chorionic cytotrophoblast) Syncytiotrophoblast
W6/32 CR3/43 JMB2 V9
Cancer antigen (Syncytiotrophoblast) H LA-A, - B, -C H L A - D R , -DP, -DQ Cytokeratin Vimentin
NKH1
CD56
Hsi and Yeh, 1986 Hsi et al., 1987 Johnson et al., 1981 Johnson et al., 1981 Contractor and Sooranna, 1986 Shorter et al. (unpublished) Loke and Day, 1984 Contractor and Sooranna, 1986 Burchell et al., 1983
Contractor and Sooranna, 1986 Ashall et al., 1982
Barnstable el al., t979 Naiem et al., 1981 McGee et al., 1982 Mason el al. (unpublished) Griffin et al., 1983
Immunoperoxidase staining of tissue sections and cytospin preparations Cryostat sections (7 ~m thick) were cut from rolls of term amniochorion. Suspensions of cytotrophoblast cells, at a concentration of I x 10S/ml in PBS/0.1% BSA, were centrifuged onto single well PTFE coated slides (Hendley Ltd., Essex, U.K.) at 19 g for 10 min (Shandon Inc, Pittsburgh,
U.S.A.). The slides were air dried overnight, fixed in acetone for 10 min and air dried for a further 30 min. They were then incubated with either 30 ~l of monoclonal antibody or PBS (used as a negative control) in a humid
75
chamber for 60 min at room temperature. After washing in PBS, the slides were incubated for 60 min with 30 ~1 peroxidase-conjugated rabbit anti(mouse Ig) IgG (Dakopatts) diluted at 1:50 in PBS with 10% normal human serum, washed again in PBS and the bound peroxidase visualised by incubating with 50/A 0.06 mg/ml diaminobenzidine tetrahydrochloride (Sigma) plus 0.06% H202 for 5 min. The preparations were counterstained with haematoxylin, dehydrated and mounted in DPX. One hundred cells were examined by light microscopy and the percentage positive for each antibody was recorded.
Immunofiuorescence labelling for flow cytometry For single antibody labelling, 1 x l 0 6 chorionic cytotrophoblast cells were incubated with either 50 #1 of monoclonal antibody or PBS control for 30 min on ice and washed twice by centrifugation at 9000 x g for 5 s in PBS containing 20 mM glucose and 0.1% BSA (PGB). The cells were then incubated with 50 ~1 of fluorescein isothiocyanate-conjugated rabbit antimouse Ig (Ro~M-FITC) (Sigma) at a concentration of 1:50 in PGN for 30 min on ice. For the last 5 min, 2/A of 0.5 mg/ml propidium iodide (Sigma) was added to each sample. For double antibody labelling, 1 x l06 cells were incubated with 50 ~1 W6/32-FITC for 30 min on ice. After washing in PGB, the cells were incubated with 50/~1 NDOG2-biotin for a further 30 min, then washed and incubated with 50 ~1 streptavidin-phycoerythrin (SAPE) for 30 min. The controls used were mouse Ig-FITC, Ig-biotin and for the double labelled samples Ig-FITC and Ig-biotin together. The cells were finally washed twice in PGB and analysed immediately on the flow cytometer.
Flow cytometry Live cells were analysed using a Coulter EPICS 541 flow cytometer fitted with a 100 ~m flow cell tip. The excitation wavelength used was the 488 nm line of the argon-ion laser. In the single labelling experiments, debris was excluded by gating on forward angle light scatter (FALS), a measure of cell size, and log peak 90 ° light scatter (LP90), a measure of cell granularity. Dead cells, which may bind antibody non-specifically, were excluded by gating out on log peak red fluorescence (LPR) those cells which had taken up propidium iodide. Single parameter histograms of log peak green fluorescence (LPG) were then collected on 1 x 10 4 cells, and the percentage positive for each antibody compared to the Ro~M-FITC control was calculated using the Coulter 'Immuno' programme. In the double labelling experiments it was not possible to measure the green fluorescence of FITC and the red fluorescence of both phycoerythrin and propidium iodide simultaneously, as the EPICS 541 has only two
76
fluorescence detectors. To overcome this problem an aliquot of the cells was labelled with propidium iodide and the red fluorescence gate set as before to exclude the dead cells (Fig. lb). This gated sample was then analysed on a FALS and LPg0 histogram (Fig. lc) which when compared to the same histogram from an ungated sample (Fig. la), was found to exclude the population of cells indicated by the arrow (Fig. la). Thus most of the dead cells could be clearly distinguished on the basis of their size and granularity and a multi-sided gate (bit-map) was drawn to exclude them and the remaining debris (Fig. lc). All subsequent analysis in the double labelling exa)
b) LP90
Cell No.
dead cells
17.2% dead cells
I
L
LPR
FALS
c)
d)
LP90
Map 1
8
.
.
.
.
, ~.-*~7 ¢-
,,:,:..:::.::.:.:,..:..
Cell No. 7.34% dead cells
' ..~J.:~:~.:~i~' ~ L~"r' . .r:":i~'::i'i::i:::~i ' "~i-':::~:%~' - "i':"'i:i:i:?i'!: .
..... •i, ' . ' ' ".I::??:Z~I:i:~:~",~IlU~:":".' .~: ,:.:~::, . ~ ' : . ~ : : : % . - " •:':::::: • , .:,L ::::..'?:;:.:,.%:~.":'-...~:~-"
:~i:i'.,,:. ':: :~'~:f:-"~'-':-" :,i.':i,
%::,::. :
LPR
Fig. 1. Exclusion of dead cells for analysis of double antibody labelling of chorionic cytotrophoblast cell by flow cytometry. (a) Size (FALS) a n d g r a n u l a r i t y (LPg0) histogram of chorionic cytotrophoblast cell dispersions. (b) Red fluorescence (LPR) of cells labelled with propidium iodide, G a t e s e t to exclude dead cells identified as PI positive. (c) FALS and LPg0 histogram of cytotrophoblast g a t e d on LPR as indicated in Fig. lb to exclude dead cells. Bit-map drawn to include live cells only (map 1). (d) LPR of cells within bit-map 1 of Fig. lc labelled with propidium iodide to s h o w t h e e f f e c t i v e n e s s of the procedure in excluding most dead cells.
dispersions
77
periments was performed using this bit-map (Fig. 1, map 1). Figure ld shows the effectiveness of this procedure in excluding dead cells from the analysis as the number of propidium iodide positive cells is substantially reduced. In 8 experiments the median viability of the population gated on map I was 89% (range 80-96). The method of analysis of the double labelled samples is described in the results section.
Cell sorting To sort cytotrophoblast, 50 x l 0 6 chorionic cells were incubated with 2.5 ml NDOG2 antibody (1:5) or mouse-Ig control (1:100) for 30 min at 4°C, washed twice in PGB and then incubated with 2.5 ml of FITC-conjugated goat anti-mouse IgG (1:50) for 30 min. After 25 min, 2 #1 of 0.5 mg/ml propidium iodide was added to each sample and incubated on ice for the remaining 5 min, before being washed twice in PGB and re-filtered through a 100 /~m gauze. The cells were resuspended at 4.0 × 106/ml in PGB and sorted on the flow cytometer at 1500 cells/s using a 100/zm flow-cell tip and one droplet sorting with coincidence on. Results
Screening of anti-trophoblast antibody panel All the anti-trophoblast monoclonal antibodies labelled at least some populations of cytotrophoblast on cryostat sections of amniochorion (Table 2). The pan-trophoblast markers labelled cytotrophoblast cells across the full width of the chorion apart from NDOG2 which was found only on the central and fetal aspects. Of the cytotrophoblast markers, 18/BA/5 and ED235 labelled all of the cytotrophoblast whereas NDOG5 was predominantly on the maternal aspect and H M F G 2 predominantly on the fetal aspect. The syncytiotrophoblast markers (CA2 and ED822) both gave weak staining, with CA2 being localised to the central and fetal aspects and ED822 weakly labelling the maternal aspect. On isolated term chorionic cytotrophoblast the monoclonal antibodies W6/32 and CR3/43 were used both on cytospins and in flow cytometry to demonstrate the purity of the populations obtained. W6/32 recognises not only HLA-A, -B, -C but also the HLA-G on cytotrophoblast (Ellis et al., 1986; Ellis et al., 1990) and therefore labels the majority of cells in these preparations. CR3/43 labels Class II MHC positive cells and reflects the level of contamination of these preparations with maternal and fetal macrophages, which was low. The antibodies to pan-trophoblast and cytotrophoblast specific markers labeled the cytotrophoblast to a similar extent using immunoperoxidase on cytospins (Table 3) but only NDOG2 and GB25 consistently labelled in flow
78
TABLE 2 I m m u n o p e r o x i d a s e labelling o f s e c t i o n s o f a m n i o c h o r i o n : tissue d i s t r i b u t i o n m o n o c l o n a l a n t i b o d i e s . A n t i b o d i e s g r o u p e d a c c o r d i n g to t h e i r specific±ties. Antibody
of antitrophoblast
Aspect of chorionic cytotrophoblast
Pan-trophoblast N DOG2
Maternal
Central
Fetal
-
+
+
GB25 GB36 H315 H316
+ + +
+ + + +
+ + + +
ED341
+
+
+
+
+
-
+ + -
+ + ±
+ + +
± -
+
+
Cytotrophoblast NDOG5 18/BA/5 ED235 HMFG2 Syncytiotrophoblast ED822 CA2
cytometry (Table 4). The proportions of labelled cells were similar between both methods (55"/,, and 58%, respectively for NDOG2 and 59% and 61% for GB25). GB36 also labelled in both systems, but less consistently, with fewer positive cells in flow cytometry. As would be expected, neither of the two antibodies specific for syncytiotrophoblast gave significant staining in flow cytometry, although 50% of the cells on cytospins stained weakly with CA2. TABLE 3 l m m u n o p e r o x i d a s e labelling o f c h o r i o n i c c y t o t r o p h o b l a s t cell d i s p e r s i o n s o n c y t o s p i n s . A n t i b o d i e s g r o u p e d a c c o r d i n g to t h e i r specific±ties. NS = n o s t a i n i n g . Pan-trophoblast
Cytotrophoblast
Syncytiotrophoblast
NDOG2 GB25
55 ( 4 5 - 6 5 ) a 59 ( 4 0 - 6 5 )
NDOG5 18/BA/5
55 ( 5 2 - 7 0 ) 42 ( 0 - 4 4 )
ED822 CA2
NS 50 ( 3 1 - 7 5 )
GB36 H315 H316 ED341
60 50 47 68
ED235 HMFG2
63 ( 4 4 - 7 5 ) 50 ( 2 5 - 6 0 )
MHC W6/32 CR3/43
91 ( 6 5 - 9 5 ) 10 ( 5 - 2 8 )
(45-80) (33-57) (41-60) (50-90)
a p e r c e n t a g e positive cells, m e d i a n ( r a n g e ) f r o m 8 e x p e r i m e n t s .
79 TABLE 4 Immunofluorescence labelling ofchorionic cytotrophoblast cell dispersions in flow cytometry. Antibodies grouped according to their specificities. Pan-trophoblast NDOG2 GB25 GB36 H315 H316 ED341
58 61 40 7 0 9
(28-76) a (34-70) (5-64) (0-53) (0-0) (0-26)
Cytotrophoblast
Syncytiotrophoblast
NDOG5 18/BA/5 ED235 HMFG2
ED822 CA2
2 (0-66) 0 (0-0) 0 (0-20) 0(0-8)
5 (0-26) 4 (0-281
MttC
W6/32 CR3/43
89 (67-95) 5 (I-6)
apercentage positive cells, median (range) from 8 experiments.
Double labelling with W6/32 and NDOG2 The purpose of these experiments was to investigate the MHC Class I expression of the NDOG2-positive subpopulation of the chorionic cytotrophoblast. W6/32-FITC (green) and NDOG2-biotin detected with streptavidin-phycoerythrin (SAPE) (red) were used as, from immunohistology, it was expected that there would be three cell populations: two positive for only one antibody, and one for both (Bulmer and Johnson, 1985). Analysis of double labelled preparations of chorionic cytotrophoblast directly on histograms of green versus red fluorescence gave no discrete populations (Fig. 2a). This was due to both the variation in the density of antigen expression on the cells and their strong background autofluorescence. This problem was overcome in the following way. Cells labelled with NDOG2-biotin-SAPE, gated on FALS and LP~0 to eliminate dead cells as described above (Fig. 1, map 1), were analysed on a histogram of LPR and LPg0(Fig. 2b). LP90 was used rather than FALS as the cells are more uniform in granularity than size (see Fig. la). A second bit-map (Fig. 2, map 2) was drawn to include only cells with a red fluorescence greater than that of the Ig-biotin-SAPE control (not shown). The NDOG2-positive cells within map 2 were then analysed on a LPG vs. LP~0 histogram to determine their green autofluorescence. A third bit-map (map 3) was drawn to exclude the autofluorescence from the analysis (Fig. 2c). Cells labelled with both W6/32-FITC and NDOG2-biotin-SAPE were then analysed on a LPG vs. LPg0 histogram and the number of cells included within bit-map 3 (those which are both NDOG2 and W6/32-positive), was determined (Fig. 2d). When the Ig-FITC and Ig-biotin-SAPE controls were similarly analysed no cells were found within bit map 3 (data not shown).
LP90
WC
LP(;
Fig. 2. Analysis ofchorionic cytotrophoblast cell dispersions double labelled with NDOG2-biotin-SAPS and W6/32-FITC. The sample was gated on PALS and LP,,, as illustrated in Ftg. Ic. (a) Conventional green versus red fluorescence (LPG vs. LPR) analysis of double antibody labelled sample showing absence of distinct positive and negative populations. (b) Cells labelled with NDOG2-biotin-SAPE analyscd on LPR and LP,,,. Bit-map 2 drawn to include the majority of NDOGI-positive cells. (c) NDOG2-positive cells from bit-map 2 analysed for green tluorescence (LPG) and LP,,,. Bit-map 3 dra\vn to exclude autofluorescent cells and non-specific binding (d) Cells double labelled wtth NDOG?-btottnSAPE and W6/32-FITC analysed using bit-map 3 after gating on bit-map 2. Cells included within hit-map 3 are positive
for both
NDOGZ
and W6!32.
The percentage of cells positive for either W6/32 or NDOGZ was also determined on single labelled samples using the Immuno programme. The results of 8 experiments are given in Table 5. It is clear from both the single and double labelled samples that the majority (88%) of the NDOG2 positive cells analysed are also W6/32 positive.
81
TABLE 5 Immunofluorescence double labelling of chorionic cytotrophoblast with N D O G 2 and W6/32 in flow cytometry. % W6/32 +ve cells a % N D O G 2 +ve cells a % N D O G 2 +ve cells which were also W6/32 +ve b
93 (78-98) c 53 (30-73) 88 (78-91)
aSingle antibody labelled samples. bDouble antibody labelled samples. CMedian percentage positive cells (range) from 8 experiments.
Isolation of NDOG2-positive cells by flow cytometry It was not possible to sort on the basis of granularity (LPg0) and green fluorescence (LPG), the parameters that were used for the double labelling analysis, because it would require a quartz 100/zm flow cell tip which is not available on the EPICS 541 flow cytometer. NDOG2-positive cells were therefore gated for sorting using FALS vs. LPG histograms as follows. The mouse IgG Ro~M-FITC control was first analysed on a two-parameter histogram of FALS and LPG, gated on FALS, LP90 and LPR to exclude dead cells and debris as described above, and a bit-map drawn to define the negative cell population (Fig. 3a). This bit-map was then superimposed on a similar histogram of cells labelled with NDOG2 and Ro~M-FITC, and used to construct a second bit-map around the NDOG2-positive population (Fig. 3b). The NDOG2-positive and negative populations were then sorted using these bit-maps. The purity of the sorted populations was assessed by reanalysis on histograms of LPG and LP90, gated on FALS and LPR (Fig. 3c and 3d) and they were further characterised using a panel of antibodies on cytospins. To re-label the NDOG2-positive cells after sorting, the slides were preincubated in neat NHS for 30 min to prevent non-specific binding, then washed in PBS and labelled with the appropriate monoclonal antibodies. Two additional antibodies were used to characterise the sorted populations: JMB2, which reacts with cytokeratin and is a marker for trophoblast in these tissues (Redman et al., 1984) and V9, which reacts with vimentin, a cytoskeletal component which is absent from trophoblast (Contractor et al., 1984). NKH1 (CD56) was used as a negative control to determine whether there was any binding of the monoclonal antibodies to the Ro~M-FITC previously bound to the NDOG2-positive cells. The results from three preparations are summarised in Table 6. Before sorting, the average viability was 80%, with a purity of 50% NDOG2-positive cells and 79% JMB2-positive. After sorting, the NDOG2-positive cells were
82
a)
b) LPG
LPG
•
.
:
,.::
."
.
Z:
.
" :" " " '
.
.
.+~
"
:.41+
~
.
,
"~+rk-
.,z,8',.'.,~:;~::~.~.~:
