0 1991 Wiley-Liss, Inc.

Cytometry 12:373-377 (1991)

BRIEF REPORT

Immunof luorescent Labeling Using Covalently Linked Anti-Phycoerythrin Antibodies and Phycoerythrin Polymers M.R. Wilson,' S. Crowley, G.A. Odgers, and L. Shaw Silenus Laboratories Pty. Ltd., Artarmon, Sydney, New South Wales, Australia 2064 Received for publication January 17, 1990; accepted January 6, 1991

We have developed methods for covalently attaching anti-phycoerythrin (PE) MAbs to other antibodies, and for using PE or polymers of PE in conjunction with these conjugates to rapidly produce specific, high intensity fluorescent labeling of antigens. The performance of these systems was examined on the surface of microspheres and on the cell surface. The noncovalent means by which PE is incorporated into the label complex in this method makes it possible to use crude algal homogenate successfully as a source of PE in immunofluorescence assays. The intensity of labeling achieved using this method is comparable to, or in some cases better than, that obtained using a direct PE conjugate. With the aim of amplifying the intensity of fluorescence

The use of anti-phycoerythrin (PE) antibodies, together with PE, to produce specific immunofluorescent labeling of antigens is a recent development. To the best of our knowledge, the idea was first suggested by Paulus (4)as a possible application for bispecific antibodies. Since then, there have been two reports of the use of anti-PE MAbs in immunofluorescent labeling of cell surface antigens. In both cases, the anti-PE MAbs were noncovalently incorporated into complexes containing MAbs specific for the target antigens (2,101. Both applications suffered from some practical drawbacks. In one case, generation of the fluorescent label required several steps (2). In the other case, subclassspecific linker MAbs were necessary, together with careful optimization of the ratios of the components used to construct the complex (10). The work reported here was undertaken in a n effort to circumvent these

obtained, we produced two types of covalently linked complexes. These were an anti-PE MAb, designated PE6, linked to itself (i.e., PE6-PE6), and PE linked to itself (i.e., PE-PE). When used in a twostage procedure in place of monomeric PE alone, these complexes increased the intensity of fluorescence obtained on the surface of microspheres by more than fourfold. On the cell surface, the performance of this system varied from one antigen to the next but in most cases was restricted to, at best, a 60% increase in the intensity of fluorescence and in many cases only about a 30% increase. Key terms: Flow cytometry, microspheres, monoclonal antibodies, fluorcomplexes

problems and also as a n investigation of the performance of PE in fluorescence amplification systems.

MATERIALS AND METHODS Production of Anti-R-PE MAbs Purified Gastroclonium sp. R-PE (i) was a kind gift from Dr. J. Wotherspoon (Becton Dickinson, Australia). R-PE was purified from Corallina sp. (ii) and from Dasya sp. (iii) by hydroxyapatite and/or anion exchange chromatography. BALB/c mice were injected

'Address reprint requests to Dr. M.R. Wilson, Department of Biochemistry (G08), University of Sydney, Sydney, New South Wales, Australia 2006.

374

WILSON ET AL.

intraperitoneally with a mixture of (i) at time 0, (ii) at time 2, and (iii) a t time 4 weeks. Each injection comprised about 50 pg of PE; initial injections were given in a 1:l mixture of phosphate-buffered saline (PBS) and Freund’s complete adjuvant, secondary injections were given in a similar mixture with Freund’s incomplete adjuvant, and subsequent injections were given in PBS alone. At 6 weeks, one mouse was given a “booster” immunization with 50 pg Dasya PE and a cell fusion performed with spleen cells harvested 3 days after the final injection. The cell fusion was performed according to standard methods ( l ) , using the cell line NS-1 as the fusion partner. Fibroblast conditioned medium was used in the culture medium to enhance hybridoma production (7). Hybridoma supernatants were screened by enzymelinked immunosorbent assay (ELISA), and one anti-PE Mab was selected for further study. This MAb, designated PE6 (IgGlk),was purified from ascites by ammonium sulfate fractionation followed by protein-A or anion exchange chromatography.

Production of PE6-PEG and PE-PE Conjugates Most conjugates of PE6 or PE were made using SPDP (Pharmacia LKB, Uppsala, Sweden) using a protocol based on the manufacturer’s instructions. Purified Dasya PE was used to produce all PE conjugates described here. MAb or PE, in coupling buffer (0.1 M Na,HPO,, 0.1 M NaC1, pH 7 . 5 ) ,was reacted with SPDP (29 pg/mg MAb, 33 pgimg PE) for 30 min at room temperature. The reaction was terminated by passing the protein over a PDlO column (Pharmacia LKB, Uppsala, Sweden) equilibrated in coupling buffer. Both partners in any conjugation scheme were treated similarly until this stage. At this stage, in order to introduce free -SH groups, one partner was reduced with 50 mM dithiothreitol (DTT) for 10 min a t room temperature. The reduction of SPDP-PE was done in coupling buffer; however, SPDP-PEG was reduced in 0.1 M Na acetate, 0.1 M NaCl, pH 4.5 to protect the immunoglobulin disulfide bridges from cleavage. The SH- protein was removed from DTT by passage over a PDlO column equilibrated in coupling buffer. The two reaction partners were then mixed together in a molar ratio of 1 : l and incubated overnight at room temperature. The reaction was terminated with the addition of N-ethylmaleimide a t a 100-fold molar excess over -SH groups. On a few occasions the conjugates PE6-PE6 and PEPE were made using disuccinimidyl suberate (DSS). In these reactions, the proteins, a t a concentration of about 5 mgiml in 0.1 M phosphate buffer, pH 7.5, were reacted with a 50-fold molar excess of DSS a t room temperature for 2 h. Unreacted DSS was removed by dialysis. MALMAb or PE-PE conjugates were purified by gel filtration over Ultrogel ACA-22. The size of the conjugates was estimated by analytical high-performance liquid chromatography (HPLC). Small samples of each of the conjugates and molecular weight standards were

run over a TSK-3000 and a TSK-4000 column connected in tandem and driven by a single-pump LKB HPLC. In all cases, the conjugates eluted at the void volume; they were therefore greater than lo6 daltons in size.

Other Antibodies and Antibody Conjugates Bi2 is a n IgG,, isotype MAb reactive with bovine immunoglobulin. The MAbs HuLy.ml (IgG,,; CD2), AMD-T3 (IgG,,; CD3), AMD-RPA-T4 (IgG,; CD4), AMD-T8 (IgM; CD8) and AMD-B21 (IgG,,; CD21) all react with the surface of human peripheral blood lymphocytes. In some experiments, a polyclonal goat-antimouse immunoglobulin antibody, reactive with both IgG and IgM isotype antibodies [GaMIg(G+ M); Tago], was used to crosslink murine MAbs noncovalently. Sheepantimouse Ig was obtained from Silenus Laboratories (Melbourne, Australia) and conjugated with PE using the SPDP method described above (to give SaMIg-PE). This same method was also used to produce the PE conjugates Bi2-PE, HuLy.ml-PE, and T4-PE. We made T4-biotin using NHS-X-biotin from Calbiochem (Australia). A streptavidin-PE conjugate (SA-PE) was also obtained from Calbiochem (Australia). Preparation of Lymphocytes and Antigen-CoatedMicrospheres Peripheral blood lymphocytes (PBL) from heparinized blood of normal donors were isolated by centrifugation over Ficoll-Hypaque and prepared fresh before each experiment. Bovine immunoglobulin (BIg) was covalently attached to polyacrylamide Immunobeads (IB; BioRad) by adding 1 mg (for “low antigen density beads”) or 10 mg (for “high antigen density beads”) of protein to 5 mg of beads in 10 ml of 3 mM NaH,PO, buffer, pH 6.3, containing 20 mg of EDAC [l-ethyl3-(3-dimethyl-aminopropyl)carbodiimide hydrochloride] and shaking the suspension overnight a t 4°C. The beads were then washed repeatedly by centrifugation (5 min a t 1,OOOg) and finally suspended in 2-4 ml of storage buffer and kept a t -20°C. The composition of the storage buffer was 50% (viv) glycerol in water, pH 7.2, containing 50 mM HEPES, 0.02% (wiv) NaN,, 5% (wiv) bovine serum albumin (BSA), radioimmunoassay (RIA) grade, 2.5 mM EDTA, 2 mM PMSF (phenylmethylsulfonyl fluoride), and 2 mM iodoacetic acid. All staining reactions, whether with beads (-20 ~1 of stock suspensionkest) or cells (1-2 x 106itest), were done in 100-pl reaction volumes a t .Q”Cfor 30 min. The reaction mixture consisted of PBS containing 1% (wiv) BSA and 0.1% (wiv) NaN,, together with the reagent(s1. Washings were done by centrifugation (beads: 5 min a t 1,OOOg; cells: 5 min at 300g). At the completion of the reaction sequence, the beads or cells were washed and then suspended in 1% paraformaldehyde in PBS and stored at 4°C in the dark until analyzed by flow cytometry.

LABELING WITH MAb COMPLEXES AND PE POLYMERS

Flow Cytometry Flow cytometric analysis was performed using a FACS 440 (Becton Dickinson, Mountain View, CA) equipped with a 5-W argon ion laser tuned to 488 nm a t 200 mW. Microspheres were selected for analysis by electronic gating on forward angle laser scatter. The orange fluorescence, collected using a 570 ? 15-nm bandpass filter, was analyzed a s histograms of the log of the intensity of fluorescence. [For more details regarding the flow cytometry of Immunobeads, refer to Wilson and Wotherspoon (9).] For comparisons of the intensity of fluorescence we transformed data collected by the FACS 440 in logarithmic form to a linear scale. This was done using the equation of Schmid et al. ( 5 ) and assumed that the three-decade log amplifier of the FACS 440 was behaving ideally.

RESULTS Staining With Primary Ab-PEG Conjugates Specific immunofluorescent staining of BIg-IB and the CD2 and CD4 cell surface markers on PBL was demonstrated with the Bi2-PE6, HuLy.ml-PEG and T4-PE6 conjugates, respectively (Fig. 1;CD2 and CD4 results not shown). A crude Dasya homogenate was able to provide PE for successful incorporation into the label complexes (Fig. 1A; PBL data not shown). In subsequent experiments reported here, we used purified PE a t 50 kg/ml. Amplification of Immunofluorescence Using P E 6 P E 6 and PE-PE Complexes Microsphere surface. Substantial amplification of fluorescence on the surface of microspheres was obtained when either PE-PE or PEG-PEGIPE-PE complexes were substituted for PE in a two-stage staining procedure (Fig. 1B). With a primary Ab-PEG conjugate as the first stage, substituting PE-PE for PE as the second stage approximately doubled the intensity of fluorescence obtained (Fig. 1B). Substituting PE6PEG/PE-PE complexes for PE, the intensity of fluorescence was increased by about 4.2-fold (Fig. 1B). Other permutations of the system (e.g., using PEGIPE-PE complexes or adding all components separately) did not achieve a s great a n increase in fluorescence. Adding all components simultaneously resulted in a drastic loss in fluorescence, the final signal being only about 5% of that obtained using the optimum two-stage method (data not shown). Cell surface. We tested the efficacy of PE6-PE6 and PE-PE in amplifying specific immunofluorescence associated with five cell surface antigens on human PBL (see Materials and Methods). For the four cases other than the CD3 antigen, any application of PE6PE6 and PE-PE produced at best about a 60% increase in specific fluorescence. For many attempts, this increase was as little as 30%. The result was similar using either a primary Ab-PEG conjugate or a GaMIg-

A v)

t-

375

LOG OF FLUORESCENCE

1

3

V p O pk2

z w

2;

&GoPk3

8

4;

d

z

B

LOG OF FLUORESCENCE

FIG.1. Flow cytometry profiles. (A) “High antigen density” BIg-IB (see Materials and Methods) reacted with (1)20 pgiml T4-PE6 followed by 50 Fgiml purified PE, (2) 20 pgiml Biz-PE6 followed by 50 pg/ml crude PE (estimated by A,,,; crude algal homogenate), or ( 3 )20 pgiml Bi2-PE6 followed by 50 pgiml purified PE. (B) “Low antigen density” BIg-IB (see Materials and Methods) reacted with (1)50 pg/ ml PE alone, (2)20 pgiml Bi2-PE6 followed by 50 pgiml PE, ( 3 )as for (2)except substituting 50 pgiml PE-PE for PE, or (4)as for (2) except substituting preformed PEG-PE6IPE-PE complex (20 pgiml PE6PE6 + 50 pgiml PE-PE, preincubated for 1min at room temperature) for PE. The illustration to the right of the profiles in B is a schematic representation of the antibody-PE complex generated on the microsphere surface in each case. *, BIg-IB; 0 , PE.

linked noncovalent complex. Similar modest increases in fluorescence were obtained when PE6-PE6 and PEPE conjugates were used to amplify immunofluorescence obtained using T4-PE, HuLy.ml-PE, and B21PE conjugates (data not shown). For the CD3 antigen, using a GaMIg “linker” system, substituting preformed PEG-PEGIPE-PE complex for PE resulted in a 4.4-fold increase in the intensity of fluorescence, comparable to the best results obtained on the microsphere surface. An additional “cycle” of PEG-PEG/PE-PE complex yielded a further 30% increase in fluorescence (Table 1). The relative intensity of fluorescent labelling of the CD4 antigen on the surface of human PBL by a variety of techniques using PE as the fluor is given in Table 2.

DISCUSSION Theoretically, there are four alternative ways in which PE might be used to produce specific immunofluorescence. These options are represented schemati-

376

WILSON ET AL.

Table 1 Flow Cytometry Results for H u m a n PBL Reacted i n Series With 10 pgIml A M D - T 3 , 5 0 Fgiml GaMIg, and Then as IndicatedaSb Final stages PE6, PE PE6, PE-PE PE6-PE6, PE PE6-PE6, PE-PE PEG-PE6IPE complex PE6/PE complex PEG-PE6IPE-PE complex PEG-PEGPE-PE, more PEG-PE6PE-PE

Fold increase 1.00 1.58 1.72 2.19 3.75 1.58 4.41 5.78

“In all cases, PE6 or PE6-PE6 were used at 20 pgiml and PE or PE-PE were used at 50 pg/ml. The complexes were generated by mixing the components and incubating for 1min at room temperature before adding to the PBL. bThe result in each case is expressed as a -fold increase in fluorescence relative to that obtained with the final stages being PE6 followed by PE. All samples were done in duplicate. The results shown were calculated from the mean intensity of fluorescence for each pair. Table 2 Flow Cytometry Results of H u m a n PBL Stained for the CD4 Antigen Using a Variety of Immunofluorescent Techniques“,b Techniaue T4-PE T4, SaMIg-PE T4-biotin. SA-PE T4-PE6, PE T4-PE6. PE6 -PEGIPE-PE

Fold increase 1.00 2.14 2.95 1.17 1.89

aThe result in each case is expressed as a -fold increase in fluorescence relative to that obtained with the direct T4-PE conjugate. All samples were done in duplicate. bThe results shown were calculated from the mean intensity of fluorescence for each pair. The performance of these techniques will vary with the concentration and batch of each individual reagent used, so these results should only be interpreted as a guide to the relative performance of each system. T4-PE, SA-PE, SaMIg-PE, T4-PE6, and PE6-PE6 were all used at 20 pgiml; PE-PE and PE were used at 50 FgIml; T4 and T4-biotin were used at 10 FgIml.

cally in Figure 2. Option A, the direct PE conjugate, is the conventional application of PE in immunofluorescence assays. Two variants of option B, using antimouse Ig antibodies to link the complex, have been described previously (2,101. We have successfully used another variant of this option, using protein A a s the linker (data not shown). Our development of option C, the primary A b P E 6 conjugate, makes i t possible to generate simply and rapidly specific immunof luorescence, using noncovalently incorporated PE, in one or (optimally) two steps. With this technique, even crude algal homogenate can be used successfully as a source of PE for immunofluorescence assays. In a n effort to increase the specific fluorescence obtained with the primary A b P E 6 conjugate we developed the PE6-PE6 and PE-PE conjugates. Our think-

7 A

B

C

D

FIG.2. Schematic representation of four theoretical options available for the incorporation of PE into immunofluorescence assays. Option A represents the conventional direct PE conjugate. Option B represents a variety of schemes in which some type of linker molecule (stippled box) is used to join noncovalently the primary Ab to an anti-PE Ab. Option C represents the covalently linked primary Abanti-PE Ab complex described in this paper. Option D represents a bispecific Ab capable of binding simultaneously to both the antigen and PE.

ing in producing PE6-PE6 and PE-PE was guided in part by the principles of the option B scheme (Fig. 2) and also by the peroxidase-anti-peroxidase (PAP) enzyme amplification method widely used in immunochemistry (6). A similar approach was taken by Leary et al. (3), who incorporated biotinylated alkaline phosphatase polymers into alkaline phosphatase-avidin complexes. They found that the crosslinked polymers gave a four- to fivefold increase in assay “sensitivity” relative to monomeric alkaline phosphatase. Using a two-stage procedure, we were able to show that substituting preformed complexes of PE6PEG/PE-PE for PE gave better than a fourfold increase in specific fluorescence on the surface of microspheres (see Fig. 1B).Using a conventional two-stage approach, it is possible to generate specific fluorescence on the microsphere surface comparable to that obtained with the two-stage “amplified” method described here. This can be achieved, for example, by using primary mouse antibody followed by a second-stage antimouse Ig-PE conjugate (data not shown). A possible advantage of the system presented here lies in its selectivity. While a second-stage PE-conjugated antibody will often recognize any and all bound primary antibodies (subject to its own particular species and subclass specificity), the PEG-PEGIPE-PE complex will only bind to those primary antibodies already “tagged” with a n anti-PE antibody (or PE). Therefore, in this context, this method offers a n alternative to the avidin-biotin system. On the surface of human PBL, the best performance from the primary Ab-PEG, PE6-PEG/PE-PE system was in most cases intermediate between the levels obtained with the direct PE conjugate and a conventional second-stage conjugate (e.g., SaMIg-PE) (see Table 2).

LABELING WITH MAb COMPLEXES AND PE POLYMERS

This relatively modest performance may result from the effects of greater steric “crowding” at the cell surface, compared with the surface of microspheres. Since substituting the PEG-PEG/PE-PE complex for PE-PE doubles the intensity of fluorescence obtained on the microsphere surface (see Fig. lB), it is reasonable to assume that the PEG-PEG/PE-PE complex contains at least one PE6-PE6 associated with a minimum of two PE-PE conjugates. This means that the minimum mass of the primary Ab-PE6/PE6-PEG/PE-PE assembly is about 4 x lo6 daltons. If the cell surface is a sterically “crowded” environment, it is not surprising that attempts at binding an entity of this size met with limited success. The success of the PEG-PEG/PE-PE complex in amplifying the intensity of fluorescence obtained for the CD3 antigen (see Table 1) suggests that this antigen may be unusual in having relatively few immediately adjacent cell surface structures, and thus fewer steric barriers. If this interpretation is correct, then it might be possible to improve amplification of PE fluorescence at the cell surface by “trimming down” the various components to reduce the overall bulk of the labelling assembly. The recent development of “single domain antibodies (dAbs)” by Ward et al. (8) may eventually find application in this area. The very small size of dAbs (