Journal of Yiroiogical Methods, 8 (1984) 243-254

Elsevier

JVM 00306

243

EVALUATION OF IMMUNOADSORBENT ELECTRON MICROSCOPIC TECHNIQUES FOR DETECTION OF SINDBIS VIRUS

DAVID KATZ and YOCHEVED STRAUSSMAN

Department of Virology, Israel Insrirute for Biological Research. Ness-Ziona 70450, Israel

(Accepted 6 February 1984)

Two immunosorbent electron microscopic techniques (ISEM), the protein A coated grid technique

(PA-CGT) and the antibody coated grid technique (AB-CGT) were applied and evaluated for the detection

of Sindbis virus from infected tissue culture fluids. At optimal conditions, the efficiency of trapping the

virions was only about 1.5 higher with the PA-CGT as compared to the AB-CGT, but the PA-CGT was less

dependent on the antiserum dilution used in the test. Both methods were suitable for quantitation

experiments, since the number of virions trapped was proportional to the virus concentration.

The influence of virus incubation time and temperatures, staining solutions, buffers and washing

procedures on the trapping efficiency and specificity was further studied with the PA-CGT. Maximal

trapping on coated grids was obtained after 3 h incubation of the virus. At room temperature, less debris

was found on the grids, as compared to 37”C, and the numbers of virions counted were only slightly lower. The optimal staining sohrtion was alcohol uranyl acetate. The specificity of the PA-CGT was dependent on

washing steps with phosphate buffered saline containing bovine serum albumin. With the standard

procedure, at room temperature around 3 X IO’ virions/ml (1 X IO6 PFU/ml) were specifically detected in

about 1.5 h.

Sindbis virus immunosorbent electron microscopy protein A

INTRODUCTION

‘Classical’ immunoelectron microscopy (IEM) for viral diagnosis is based on the

observation in the electron microscope (EM) of clumps of virions specifically formed

with homologous antibodies (Almeida and Waterson, 1969; Doane et al., 1974;

Flewett and Boxall, 1976; Doane and Anderson, 1977; Milne and Luisoni, 1977;

Almeida, 1980; Van Regenmortel, 1982).

Derrick (1973) described a new IEM method in which grids were coated with

antibodies and used for specific ‘trapping’ of plant viruses. In this method, one of the

reagents (the antibody) is first adsorbed to a ‘solid phase’ (the grid) similar to the

principles of other solid phase immunoassays such as solid phase radioimmunoassay

(SPRIA) (Catt and Tregear, 1967) and enzyme linked immunosorbent assay

(ELISA) (Engvall and Perlman, 1971). Since the introduction of Derrick’s method,

Ol66-0934/84/$03.~ Q 1984 Elsevier Science Publishers B.V

244

other IEM methods that are also based on the ‘solid phase’ principle were published

(Milne and Luisoni, 1975; Shukla and Gough, 1979; Katz et al., 1980).

Roberts et al. (1982) suggested naming Derrick’s method ‘immunosorbent

electron microscopy’ (ISEM). In our opinion, this definition is too narrow and should

include not only methods with antibody coated grids but also all other methods where

a solid adsorbent is introduced.

For the sake of simplicity and uniformity, we propose to replace the many acro-

nyms of the ISEM techniques, with new names. Thus the Derrick method (Derrick,

1973) will be called the antibody coated grid technique (AB-CGT), the method of

Shukla and Gough (1979) the protein A coated grid technique (PA-CGT), the

method of Katz et al. (1980), the protein coated bacteria technique (PA-CBT) and the

‘decoration’ method of Milne and Luisoni (1975) the antigen coated grid technique

(AG-CGT).

ISEM techniques were extensively studied and evaluated for plant viruses (reviewed

by: Milne and Luisoni, 1975; Van Regenmortel, 1982) and much less for animal

viruses (Nicolaieff et al., 1980; Giraldo et al., 1982; Kjeldsberg and Mortensson-

Egnund, 1982; Svensson and Von Bonsdorff, 1982; Rubinstein and Miller, 1983;

Svensson et al., 1983).

It is generally agreed that ISEM methods have several practical advantages over the

‘classical’ IEM methods, yet each virus-antibody system has to be carefully worked

out for optimal results (Milne and Lesemann, 1978; Lesemann et al., 1980; Milne,

1980; Nicolaieff and Van Regenmortel, 1980; Cohen et al., 1982; Nicolaieff et al.,

1982). In the present work, we describe optimal conditions for the application of the

PA-CGT for rapid detection of alpha togaviruses using Sindbis virus as a model. The

performance of the PA-CGT was quantitatively evaluated and compared to the

AB-CGT.

MATERIALS AND METHODS

Virus

Sindbis virus was obtained as a gift from Dr. P. Fuchs from our department. The

virus was grown in a baby hamster kidney tissue culture line (BHK-21) and harvested

in maintenance medium (Eagles’ medium supplemented with 5% fetal calf serum).

This crude preparation of the virus stock was kept in small quantities and frozen at

-20°C until used. The virus had a titer of around 5000 in a solid phase protein A

sandwich indirect radioimmunoassay (SPA-RIA(S)) (Katz and Kohn, 1980) and

around 2 X lo9 plaque forming units per ml (Fuchs, pers. comm.).

Antiserum

Anti Sindbis antibody was prepared in rabbits by repeated intravenous injections

(at 10,29,52 days after first inoculation) with purified viruses, lo9 PFU per inoculum.

A pool that was prepared from the hyperimmune antisera, had a titer of 105.’ in a solid

24s

phase protein A radioimmunoassay (SPA-RIA) (Katz et al., 1981) and a 104.4 in a

plaque neutralization test (Shapira and Lustig, pers. comm.).

Protein A Protein A, a lyophilized derivate from Staphylococcus aureus was bought from

‘Pharmacia’ (Sweden). The powder was reconstituted with 1 ml of sterile distilled

water to a concentration of 5 mg/ml. This preparation was kept frozen (OOC) in the ice

box compartment of a 4*C refrigerator and could be repeatedly thawed and frozen

without detectable loss in activity over a period of at least 1 yr.

Electron microscope (EM) specimen grids Two types of EM grids were used:

(1) 400 mesh carbon fronted formvar coated grids.

(2) Commercial, 400 mesh, carbon coated grids (Polar-on Equipment Ltd., Watford,

U.K.).

Electron microscope staining solutions (1) 2% Phosphotungstic acid.

(2) 2% Uranyl acetate in water.

(3) 1% Uranyl acetate in 47.5% alcohol.

Buffer solutions (1) PB = Phosphate buffer, 0.1 M i- 0.05% sodium azide pH 7.0.

(2) PBS = Phosphate buffer (0.05 M) + saline (0.15 M) + 0.05% sodium azide, pH

7.2.

(3) BB = Bicarbonate buffer, 0.01 M, pH 9.6, i- 0.05% sodium azide.

(4) PBS-BSA = PBS f 0.1% Bovine serum albumin (Sigma).

The PA-CGT Various procedures were used during the process of optimization of the PA-CGT

for the detection of Sindbis virus. Here we describe the procedure we found to be

optimal. Other procedures will be mentioned in the text.

For protein A adsorption, grids were floated for 5 min on 50 1.11 drops containing 1

pg/ml protein A diluted in PBS. The grids were then washed on 2 drops of PBS and

transferred for 15 min incubation onto drops of antiserum diluted I:500 in PBS. The

grids were then washed on 6 drops of PBS containing bovine serum albumin (PBS-

BSA) and transferred to drops of virus suspension (in PBS-BSA). After 60 min

incubation, the grids were again washed on 6 drops of PBS-BSA, drained on filter

paper and immediately stained for 3 min on drops of 1% uranyl acetate in 47.5%

alcohol. All incubations were done at room temperature (unless stated otherwise).

Virus particles were then counted in an JEOL-JEM 100 S electron microscope. For

quantitation, 2 fields in each of 3 squares of the grid were screened at a magnification

246

of 30,O~. Counts were then averaged and multiplied by ten. The results are presented

as the number of viruses trapped per 10 fields which are hereafter defined as a unit area

of the grid (160 u2).

RESULTS

preliminary experiments In preliminary experiments, we compared the trapping efficiency of the commercial

carbon coated grids to grids we coated with carbon-fronted formvar. We found that

the commercial grids trapped more virus particles without regard whether the grids

were untreated or treated with protein A and antibody. We found no significant

difference in results when protein A was diluted in BB, PB or PBS. Also, we did not

observe any difference in results when PB, Tris-NaCl and PBS were used for virus

dilution. We noticed, however, that drying of the viruses on the grids before staining

destroyed them and caused a significant drop in particle counts. A11 the folIowing

experiments were therefore done with commercial grids only, with PBS as the only

diluent and viruses were stained immediately after the desired contact period with the

grids.

Comparison of staining solutions for electron microscopy on control uncoated and coated (for PA-CGT) grids

Three staining solutions were compared: 2% uranyl acetate in water, 1% uranyl

acetate in 47.5% ethanol and 2% of phosphotungstic acid in water, pH 6.0.

Three grids coated with 25 ug/ml of protein A in PB and a l:SOO dilution of rabbit

antiserum in PBS, were incubated for 60 min on drops of a 1: 10 dilution in PBS of the

stock Sindbis virus. As controls, three uncoated grids were similarly incubated. One

grid from each group was stained immediately (without drying) by a 2 min incubation

on drops of one of the staining solutions. From the results shown in Fig. 1, we

concluded that the uranyl acetate stains are signi~cantly superior to the phospho-

tungstate stain. We have chosen the alcoholic uranyl acetate stain for the following

experiments since the contrast obtained with this stain was superior to the aqueous

stain. The PA-CGT permitted the detection of approximately 10 times more particles

on the coated grids as compared to the amount trapped directly on non-treated grids.

Comparison of the effect of dQ$erent concentrations ofprotein A and antibodies used for coating grids on the virus trapping capacity

Twelve grids were arranged in three groups. The four grids in each group were first

coated with 0, 1, S and 25 pg/ml of PA in PB. Each group was then coated with

different antiserum dilutions (1: 100, 1:500 and 1:2SOO).

The trapping efficiency of the grid was examined after a 60 min incubation on drops

of a l:SO Sindbis virus dilution. After draining the excess virus with filter paper, the

grids were stained with 1% alcoholic uranyl acetate for 3 min.

247

The results summarized in Fig. 2 show that without protein A on the grid, the

highest number of virions was seen with an antiserum dilution of 1:2500; however, on

grids that were first coated with protein A at any concentrations, similar numbers of

viruses were trapped independently of the dilution of antiserum. From these results,

we concluded that the AB-CGT (grids without protein A) under best conditions

trapped about 1.5 times less viruses than that of the PA-CGT. The major advantage of

the PA-CGT was that this technique is less dependent on optimal antiserum dilutions.

Since there was no significant effect of the concentration of protein A used, we have

chosen, for further experiments with the PA-CGT, the combination of 1 ug/ml and

1:500 antiserum dilution.

Specificity of fhe PA-CGT Experiment A In this experiment, 4 grids were coated with 1 ug/ml protein A. Two of

them were then coated with 1:2500 dilution of antiserum and the other two with

normal rabbit serum (NRS). All grids were thereafter incubated for 60 min on drops of

a 1:50 dilution of Sindbis virus. One of the grids coated with antiserum and one of the

grids coated with NRS were first washed on 6 drops of PBS and then stained; the

“W

r’ N 5 25

-

Fig. 1. The effect of coating grids with protein A and antiserum (PA-CGT) and the use of different staining

solutions on the amount of virions trapped. A 1:lO dilution of the Sindbis stock solution was incubated on

control, not treated (NT) grids, and on grids coated with 25 ug/mI protein A and 1500 antiserum dilution

(T). For comparison, the grids (NT and T) were stained with three staining solutions: 2% uranyl acetate in

water (UW), 1% uranyl acetate in 47.5% ethanol (UE) and 2% phosphotungstic acid (PT).

Fig. 2. The effect of different concentrations of protein A (0, I, 5,25 @g/ml) and different antibody dilutions

(l:lOO, 1:500. 1:2500) used for coating grids on the virus trapping efficiency. The Sindbis virus dilution in

this experiment was 1:50 in PBS.

248

remaining two grids were stained without washing. Staining was performed for 2 min

with 1% uranyl acetate in 47.5% ethanol. Table 1 shows that the viruses adsorbed

equally well to antibody coated grids as to the NRS coated grids. The number ofvirus

particles was slightly less on washed grids as compared to the unwashed ones, though

the difference was not significant. Viruses on washed grids seemed to be stained

weaker than those on the unwashed.

It was concluded that the test as performed in this experiment was nonspecific,

in spite of the washing procedure.

Experiment B To obtain specific results, we decided to wash the grids with PBS-BSA

after incubation with the antiserum (or NRS as controls) and after the virus incuba-

tion step. Moreover, the virus was diluted in PBS-BSA instead of PBS. Staining time

in the alcoholic uranyl acetate solution was increased from 2 to 3 min to obtain heavier

staining of the viral particles.

The experiment was carried out with two concentrations of protein A (1 and 5

ug/ml in PBS) and two dilutions of antiserum and NRS (I:500 and I:2500 in PBS). The

virus dilution was 1:50.

The results depicted in Fig. 3 show that in this experiment the trapping was specific

since NRS coated grids did not trap any significant number of viruses while the

antiserum control grids did. It is also seen that optimal conditions for specific trapping

were obtained on grids coated with 1 ug/ml and 1:500 dilution of antiserum.

The effect of virus incubation time and temperature on the sensitivity of the PA-CGT Virus dilutions as indicated in Fig. 4 were tested with the PA-CGT on grids coated

with 1 ug/ml protein A and a I:500 dilution antiserum, using the procedure described

in the previous experiment and in ‘Materials and methods’. Two sets of virus dilutions

were incubated at room temperature, one for 1 h and the other for 3 h. Another two

TABLE 1

The effect of PBS washing steps on the specificity of the PA-CGT

Coating sera” Number of trapped virionsd per unit area f SD

Washede Not washed

NRSb 178 It 48 226 f 34

AS” 140 f 39 206 f 37

Sera were diluted I:2000 in PBS and incubated on grids that were previously coated with I &ml

protein A.

b Normal rabbit serum.

’ Anti Sindbis virus antiserum.

d Sindbis virus dilution used here was 150 in PBS.

’ PBS washing on grids after virus incubation.

249

Fig. 3. Specificity of the PA-CGT performed on grids coated with 1 and 5 ug/ml protein A (PA) and with

I:500 and I:2500 antiserum (AS) or normal rabbit serum (NRS). The Sindbis virus dilution in this

experiment was I:50 in PBS-BSA. Washings of the grids after the serum coating step were done with

PBS-BSA.

Fig. 4. The effect of virus incubation time and temperature on the sensitivity of the PA-CGT. The

experiment was performed according to the procedure described in ‘Materials and methods’, on grids

coated with 1 ug/ml protein A and a 1:500 dilution of antiserum.

a

Fig. 5. Comparison of the effect of two different incubation temperatures (a = room temperature. b =

37”C), on the morphology of the viruses and background debris. The incubation timewas 3 h, and the virus

dilution 1: IQ: otherwise the procedure was as described in ‘Materials and methods’, for the PA-CGT. Bars

represent 200 rim.

251

sets of virus dilutions were incubated in a 37°C humidified incubator, the first set for 1

h and the second for 3 h. The minimum amount of virions that was detected with the

PA-CGT standard procedure (1 h at 24°C) in this experiment, was 3 X 10’ virions/ml

or 1 X lo6 PFU/ml. About 2-fold higher counts were obtained when the virus was

incubated with the coated grids for 3 h. The effect of temperature was less marked,

though at 37°C the counts were slightly higher than those obtained at room tempera-

ture. At any temperature and incubation time combination, a linear log-to-log rela-

tionship was obtained between the number of viral particles trapped on the grid and

the virus dilution.

In Fig. 5, trapped viruses by the PA-CGT are shown. Results in 5a were obtained

after 3 h of floating of the grids on drops of a 1: 10 virus dilution at room temperature,

while results in 5b were obtained after 3 h at 37°C. At 37°C (Fig. 5b), the viruses were

damaged and the surface of the grid was covered with debris. On the grids prepared at

room temperature (Fig. 5a), the viruses were intact and the background clearer.

Counting of the viruses on these grids was easier and possible even at a magnification

of 10,000.

In a similar experiment, we examined the influence of virus incubation time at room

temperature and 37°C on the performance of the PA-CGT. One group of protein A

and antiserum coated grids (1 ug/ml protein A and 1:500 antiserum dilution) were

floated on a 1: 150 viral dilution for 1,2,3 and 5.5 h at room temperature and the other

for the same periods of time at 37°C. Results are shown in Fig. 6. At 2 and 3 h

incubation at 37°C significantly more viruses are trapped on the grid as compared to

room temperature. The optimal time of incubation for best trapping was 3 h; longer or

shorter incubation times were less efficient.

Time (hours)

Fig. 6. The effect of incubation time and temperature on the trapping capacity of the grids. Grids were

treated according to the procedure described in ‘Materials and methods’ for the PA-CGT. The virus

dilution in this experiment was 1:150 in PBS-BSA.

252

DISCUSSION

In the PA-CGT, the antibody molecules (IgG) bind to the protein A through their

Fc part, but leaving the antibody binding part (Fab) free. It was therefore believed that

this method should be much more sensitive, as compared to the AB-CGT. Shukla and

Gough (1979) in their publication stated that their PA-CGT was more than 300-fold

more sensitive than the AB-CGT for sugar cane mosaic virus. However, it was later

recognized that the conditions for the AB-CGT were not optimal. Under optimal

conditions, for both techniques, the PA-CGT was only 25 fold more sensitive than the

AB-CGT (Gough and Shukla, 1980).

In the present work, we adapted the two ISEM methods for the specific identifica-

tion of Sindbis virus. The effect of different buffers, EM-grids, staining solutions,

concentrations of antiserum and protein-A, virus incubation time and temperature

was quantitatively evaluated.

We found that under our experimental conditions, commercial carbon coated grids

were superior to our own carbon-fronted formvar-coated grids. The alcohol uranyl

acetate stain was used by other investigators who used similar formvar coated grids

(Nicolaieff and van Regenmortel, 1980; Nicolaieff et al., 1980) without noticing any

undesirable effect. It seems therefore unlikely that the alcohol solvent may have

damaged the formvar films. This possibility, however, cannot be ruled out. In our

PA-CGT system uranyl acetate stain (in water or alcohol) was superior to the

phosphotungstic stain. Differential sensitivities of viruses to stains were observed for

certain plant viruses (Roberts and Harrison, 1979).

We also found that the AB-CGT was dependent on the antiserum dilution. Best

results were obtained with an antiserum dilution of 1:2500, while with the PA-CGT,

similar results were obtained with a wide range of dilutions. The sensitivity of the

PA-CGT was only 1.5 higher as compared to the AB-CGT. Thus, the main advantage

of the PA-CGT over the AB-CGT is its being less dependent on antiserum concentra-

tion. These results and conclusions are in agreement with other authors (Lesemann

and Paul, 1980; Milne, 1980; Nicolaieff et al., 1980) who compared the AB-CGT and

the PA-CGT for different viruses.

The optimal incubation time of the virus in the PA-CGT was 3 h. The trapping

efficiency of the coated grids was only around two-fold higher at 3 h as compared to 1

h incubation. Longer incubation times caused a decrease in the trapping efficiency,

presumably because of damages caused to the viruses due to the prolonged incuba-

tion. Incubation of the viruses with the coated grids at 37°C as compared to room

temperature, did not have a marked effect on the trapping efficiency, although the

counts were always slightly higher after 37°C than after room temperature incuba-

tions. In spite of slightly higher counts with grids incubated at 37”C, these grids were

covered with debris, and the virions on them seemed to be damaged. We therefore

prefer, for practical purposes, a routine procedure for the PA-CGT in which the virus

is incubated with the coated grids, for 1 h at room temperature.

253

Quantitative data were easily obtained with the PA-CGT, since there was a linear

relationship between the log of the number of virions trapped on the grids and the log

of the reciprocal of the virus dilution.

The minimal amount of virions that we detected with the PA-CGT was around 3 X

10’ virions/ml (1 X lo6 PFU/ml). This sensitivity is comparable to the sensitivity

obtained with an ELISA used in our laboratory, but is about IO-fold higher than

direct electron microscopy.

In a comparative experiment ‘classical’ IEM was only about 3- to 4-fold less

sensitive than the PA-CGT (D. Katz, Y. Straussman and A. Shahar, unpubl. results).

However, the PA-CGT was much simpler and easier to perform.

The specificity of the PA-CGT in our hands, was dependent on the use of BSA for

washing steps and for the dilution of the virus. Other authors obtained specific results

without the introduction of BSA or other additives to their buffers, but more washing

steps were used (Milne, 1980; Lesemann and Paul, 1980; Milne, 1980). On the other

hand, Svensson et al. (1983), used a non-specific PA-CGT for the diagnosis of

rotaviruses but based their identification on morphology.

In this work, we demonstrated the application of the PA-CGT for the diagnosis of

Sindbis virus as a model for other togaviruses. On the basis of this work and from the

accumulated data obtained with plant viruses and other animal viruses, we believe that

the PA-CGT can be easily applied as a simple and rapid diagnosis method for any

virus, provided that optimal conditions are first worked out for each new system. The

combination of the specific immunological trapping and the direct morphological

examination of the viruses is most advantageous for a reliable diagnosis when

compared to other solid phase immunoassays (such as ELISA or SPRIA).

REFERENCES

Almeida, J.D., 1980, Yale J. Biol. Med. 53, 5.

Almeida, J.D. and A.P. Waterson, 1969, Adv. Vir. Res. IS, 307.

Catt, K. and G.W. Tregear, 1967, Science 158, 1570.

Cohen, J.. G. Loebenstein and R.G. Mime, 1982, J. Viral. Methods 4, 155.

Derrick, K.S., 1973, Virology 56, 652.

Doane, F.W. and N. Anderson, 1977. in: Comparative Diagnosis of Virus Diseases, Vol. 2, rds. E.

Kurstack and C. Kurstack (Academic Press, New York) p. 506.

Deane, F.W., N. Anderson, J. Chao and A. Noonan, 1974, Appi. Microbial. 27, 407.

Engvall, E. and P. Perlman, 1971. Immunochemistry 8, 871.

Flewett, T.H. and E, Boxall, 1976, Clinics in Gastroenterology 5, 359.

Giraldo, G.. E. Beth, J. Lee, E. de Harven and M. Chernesky, 1982, J. Clin. Microbial. 15, 517.

Gough, K.H. and D.D. Shukla. 1980, J. Gen. Viral. 51, 415.

Katz, D. and A. Kohn, Second International Conference on the Impact of Viral Diseases on the Develop--

ment of African and Middle East Countries, Nairobi, Kenya, p. 96.

Katz, D., E. Gruber, S. Lehrer and A. Kohn. Fifth International Congress ofvirology. Strasbourg, France,

p. 169.

Katz, D., Y. Straussman, A. Shahar and A. Kohn, 1980, J. Immunol. Methods 38, 171-174.

Kjeldsberg, E. and K. Montesson-Egmund, 1982, J. Viral, Methods, 4, 45.

254

Lesemann, D.E., R.F. Bozarth and R. Koemg, 1980, J. Gen. Viral. 48, 257.

Milne, R.G., 1980, Acta Horticulturae 110, 129.

Milne, R.G. and E. Luisoni, 1975, Virology 68, 270.

Mime, R.G. and E. Luisoni, 1977. in: Methods in Virology, Vol. 6, eds. H. Maramorosch and H. Koprowski

(Academic Press, New York) p. 265.

Mime, R.G. and D.E. Lesemann, 1978, Virology 90, 299.

Nicolaieff, A. and M.H.V. van Regenmortel, 1980, Ann. Virol. (Inst. Pasteur) 131, 95.

Nicolaieff, A., G. Obert and M.H.V. van Regenmortel, 1980. J. Clin. Microbial. 12, 101.

Nicolaieff, A., D. Katz and M.H.V. van Regenmortel, 1982, J. Viral. Methods 4, 155.

Roberts, I.M. and Harrison, B.D., 1979, Ann. Appl. Biol. 93, 289.

Roberts, I.M., R.G. Mime and M.H.V. van Regenmortel. 1982, Intervirology 18, 147.

Rubinstein, A.S. and M.F. Miller, 1982, J. Clin. Microbial. 15, 938.

Shukla, D.D. and K.H. Gough, 1979, J. Gen. Viral. 45. 533.

Svensson, L. and C.H. von Bonsdorff, 1982, J. Med. Viral. 10, 243-253.

Svensson, L., M. Grandien and C.A. Petterson, 1983, J. Clin. Microbial. 18, 1244-1249.

Van Regenmortel, M.H.V.. 1982. Serology and Immunochemistry of Plant Viruses (Academic Press, New

York).