Identification of enterically transmitted hepatitis virus particles by solid phase immune electron microscopy

Journal of Virological Methods, 29 (1990) 177-188

177

Elsevier VIRMET 01049

Identification of enterically transmitted hepatitis virus particles by solid phase immune electron microscopy Charles D. Humphrey’,

‘E.H. Cook Jr.* and Daniel W. Bradley’

’ Ultrastructure Activity and ‘Hepatitis Branch (World Health Organization Collaborating Center for Research and Reference in Viral Hepatitis), Division of Viral and Rickettsial Diseases, Center for Infectious Diseases, Centers For Disease Control, Public Health Service, U.S. Department of Health and Human Services, Atlanta, Georgia, U.S.A.

(Accepted

23 April 1990)

Summary Small ‘featureless’ viruses (c 50 nm) are difficult to identify by routine immune electron microscopy techniques, particularly when they are mixed with debris from stool or cell culture extracts. A combination of conventional immune electron microscopy (IEM) and solid phase IEM (SPIEM) methodologies was ‘used to identify hepatitis A virus (HAV) in stool and cell culture extracts and non-A non-B hepatitis (hepatitis E) in stool extracts. Compared with conventional IEM, the modified SPIEM method resulted in a significant increase in the number of particles observed. Several small aggregates, each containing 2-20 particles, were observed scattered randomly within most grid squares. Similar results were seen with stool extracts from hepatitis E (HEV) infections. The SPIEM method is a simple, highly sensitive specific assay that facilitates rapid identification of enteric hepatitis viruses. Several experiments were done to characterize the effects of altered physical environment within the assay and to evaluate potential modifications. Solid-phase immune electron microscopy; electron microscopy

Hepatitis; Virus identification;

Immune

Correspondence to: C.D. Humphrey, Ultrastructure Activity, Division of Viral and Rickettsial Diseases, Centers for Disease Control, 1600 Clifton Road, Atlanta, GA 30333, U.S.A.

178

Introduction Hepatitis A virus (HAV) and epidemic non-A, non-B hepatitis virus, tentatively named hepatitis E (HEV), are transmitted through fecal passage and oral ingestion by a susceptible host. Consequently, stool suspensions from infected hosts are a major source for the detection of these infectious agents. Although many techniques are available for detection of HAV, including immunofluorescence assay (IFA), enzyme immunoassay (EIA), radioimmunoassay (RIA), immune electron microscopy (IEM), and molecular hybridization (Gust and Feinstone, 1988), the only assay for identifying HEV in stool specimens has been IEM. Gupta et al. (1988) have developed an EIA specific for HEV, and Krawczynski and Bradley, (1989) have identified HEV or HEV antigen (HEV-Ag) in tissue by an IFA. Problems associated with the identification of HEV in feces by routine IEM procedures include low particle counts, deterioration of virus structure, and extraneous debris. In our past IEM studies of HEV (Bradley et al., 1987, 1988), samples that should have been positive for virus particles were often found to contain only debris resembling particle remnants attached to antibody. In addition, each experiment requires large quantities of specimen, limiting the number of studies that can be done with a given sample. Further, the considerable manipulation involved with IEM of labile viruses may reduce the number of intact particles observed. Solid phase IEM (SPIEM) decreases the amount of specimen and reagent required for particle identification and minimizes the need for sample manipulation. We describe a modified SPIEM method for identifying small, labile virus particles. HAV was used for development of the SPIEM assay because of its physical similarity to HEV, its stability and availability, and the ability to quantitate HAV by means other than electron microscopy.

Materials and Methods Viruses Crude lysates of HAV virus strains HM-175 (Cromeans et al., 1987) and partially purified preparations of HAS-15 (Bradley et al., 1984) harvested from FRhK-4 tissue culture cells were used as test viruses. The stock HM-175 antigen contained 10’ RFU/ml of HAV as determined by radioimmunofocus assay (Lemon et al., 1983). Experimental animal and human virus-positive 10% stool suspensions in phosphate-buffered saline, pH 6.6-7.3 or bile specimens were prepared from animals and patients with either hepatitis A or E; diagnoses were determined by observation and history, elevated alanine amino transferase (ALT) activity, elevated isocitrate dehydrogenase (SICD) activity, and positive serologic and pathologic indications for HAV or HEV infection (Bradley et al., 1987, 1988). HEV was isolated from human stools obtained through outbreak case studies or from stool and bile specimens of cynomolgus monkeys (Macaque fuscicularis).

179

Antibodies Monoclonal antibody to HAV (14-Hlc18, Abbott Laboratories, North Chicago, IL)* was diluted from a stock concentration of 3 mg/ml. Polyclonal anti-HAV was obtained either from HAV-infected patients or from chimpanzees (Pan troglodytes) (Bradley et al., 1984). Whole antisera to HEV were obtained from infected marmosets during the acute phase of disease and several weeks following the peak of liver enzyme activity (convalescent phase) (Bradley et al., 1987). Electron microscopy Direct electron microscopy (DEM), immune electron microscopy, and traditional solid phase immune electron microscopy procedures were performed as described previously (Almeida and Waterson, 1969; Derrick, 1973; Shukla and Gough, 1979). Freshly prepared grids, grids that had been stored for several weeks, commercially prepared grids from several sources, and glow-discharge prepared grids were compared for their relative capabilities to trap either HAV or HEV. The modified SPIEM technique was begun by inverting carbon-Formvar coated 300 or 400 mesh copper grids (The Lanum Company, Homewood, IL) onto 10 ~1 drops of protein A (Staphylococcus aureus, Polysciences Inc., Warrington PA) in 0.1 M phosphate buffer, pH 7.4, for 20 min. Coating, rinsing, and incubations were contained in hydrophobic wells formed by indenting Parafilm (American Can Co:, Greenwich, CT) stretched across empty micropipette tip boxes (Ranin-yellow tips l-200 ~1 type, Ranin Inst. Co., Wobum MA). The paper was left on the Parafilm during indentation and removed after the wells were formed. After incubation on protein A, each grid was transferred across ten 20 ,~l drops of phosphate buffer, 0.1 M, pH 7.4, blotted on edge, and immediately incubated on the antibody-antigen mixture. Forcep tines were cleaned and blotted with filter paper before transferring the grids. The antibody-antigen mixture was prepared during the protein A coating step by mixing 3-10 ~1 of antibody with 3-10 ,~l of 10% stool suspensions, cell culture virus preparations, or bile dilutions of 1:lO or greater. Short incubations were done at room temperature, while incubations longer than 3 h were done at 4°C under humidity controlled conditions. Grids were blotted on edge with filter paper after incubation, or rinsed gently with a stream of distilled water from a wash bottle, and then blotted. Each grid was stained with 2% phosphotungstic acid (PTA) in water, pH adjusted to 6.5 with 1 M potassium hydroxide, and viewed. Contamination of the PTA was minimized by blotting the forceps tines before reintroducing them to the PTA drop. In some experiments, 2% PTA was prepared in water containing 25 ,@ml bacitracin to maintain grid hydrophilicity during drying (Gregory and Pirie, 1973).

* Use of tradenames is for identification only and does not imply endorsement by the Public Health Service or the U.S. Department of Health and Human services.

180

Particle counting and data analysis Particles were counted at an approximate screen magnification of x43 000 in a Philips 201C transmission electron microscope at 60 or 80 KV. Three to five grid squares were examined per grid. In some experiments, both the number of particles per aggregate and per grid square were counted. Grid squares were selected randomly for viewing, provided the spreading quality of the virus-antibody mixture was adequate; grid squares were excluded when the virus-antibody mixture had pooled or was too dense to permit viewing. The mean, standard deviation, and standard error of the mean for particles per grid square were determined for each sample. The coefficients of variation were determined for the different methods. Problems related to large complexes with resultant overlapping particles were unusual with the low concentrations of HAV or HEV used in this study.

Results Several preliminary experiments were done to evaluate established solid phase IEM methods (Derrick, 1973; Shukla and Gough, 1979). We determined that the sensitivity of assays designed for trapping and viewing HAV for electron microscopy could be significantly increased by coating the grids with protein A before adsorbing them with antigen and antibody (Shukla and Gough, 1979). Initial experiments involving our modified protein A technique were done to refine the protein A grid coat procedure for trapping HAV. To determine an optimized concentration of monoclonal antibody (mAb) for the assay, we tested ten-fold dilutions ranging from no dilution to 1: 100 of stock HAV HM- 175 samples ( lo7 RFU/ml) in a checkerboard manner with stock HAV monoclonal antibody (3 m&ml) in two-fold dilutions ranging from 1: 10 to 1: 160 and from 1: 125 to 1:4000. The pH optimum for virus aggregation was determined on stock HAV HM-175 by altering the pH of both the antigen and the antibody preparations in increments of 0.5 pH units from pH 5.0-8.0. Phosphate buffer (0.1 M) and Tris buffer (0.1 M) were each tested. SPIEM of HAV resulted in a random dispersion of particles in small clumps (Fig. 1) and permitted statistically valid counting and measurement. A 1:500 dilution (6 &ml) of stock mAb was the best dilution for the antibody preparation (Fig. 2), and a dilution of 2.5-10 pg/ml protein A was satisfactory. Monoclonal antibody as the aggregating antibody resulted in capturing more particles than the human antisera (Fig. 2). Bacitracin in the negative stain improved reproducibility of spreading of virus-antibody mixtures on the grid surface (Gregory and Pirie, 1973). Typical sensitivity of the assay for crude tissue culture preparations of HAV was lo5 RFU. Virus could be detected in purified preparations of HAV at a concentration of 5-10 ng/ml. Consistent results could be obtained with as little as 2 1~1of antigen and 2 ~1 of antibody. HAV trapping varied considerably at different pH levels. HAV aggregates formed (Fig. 3) and were trapped most efficiently when the pH of both antigen and antibody were nearly identical; the best trapping occurred when both were pH

Fig. 1. Solid-phase IEM of highly purified HAV HAS-U, x 24,550.

700 @ml protein. Original magnification

7-8. Maintaining of antibody pH at 7-8 improved trapping more than maintaining antigen pH at 7-8. The use of phosphate buffer resulted in more consistent virus trapping than the use of tris buffer. Tris buffer also reacted with copper grids, forming copper-salt contaminants that interfered with viewing.

HUl

ANTISERA

ANTIBODY

40

100

5001ooo2ow4ooo8ooa ANTIBODY

DILUTION

(1 X)

Fig. 2. Effects of antibody dilution on SPIEM trapping of HAV HM-175, 10’ RFU. Error bars represent standard error of the means.

182

0’

5

7 6 pH OF HAV MONOCLONAL ANTIBODY

8

Fig. 3. Effects of monoclonal antibody pH on aggregation of HAV HM-175 at pH 5-8. +, HAV in phosphate buffer at pH 8; & HAV at pH 7; 0, HAV at pH 6, and + HAV at pH 5.

Direct comparisons were made among several procedures for trapping HAV onto formvar-carbon coated grids (Fig. 4). Trapping of stock HAV-HM-175 was tested either by direct attachment, by IEM in which mAb, as aggregant, was diluted to

0

loo

200

300

400

500

600

700

800

Boo

looo

PARTICLESPERQS

Fig. 4. Comparison of HAV trapping among DEM, IEM, and SPIEM methods. Error bars represent standard error of the means.

183

1:500 (Almeida and Wat~~on et al., 1969), by p~incubation of the coated grid for 20 min with mAb diluted to 1:500, followed by overnight incub~ion with stock HM-175 HAV (Derrick, 1973), or by our SPIEM procedure in which protein A was used to coat the grid for 20 min before exposing it to a 10 ~1 mixture of stock HM-175 HAV antigen and 10 ~1 of monoclonal antibody diluted to 1500. Incubation times for the protein A method were 20 min and overnight. SPIEM was decidedly the most sensitive and reproducible method used in these experiments. The coefficient of variation for SPIEM with aggregation and trapping was 26%, but SPIEM with trapping only was 44%. IEM had a variation of 67%, and direct adsorption to grids without antibody trapping had a variation coefficient of 100%. Virus appearance and virus aggregates were mo~hologica~y altered when different methods were used to capture virus (Fig. 5). DEM resulted in less obscuration of the virus surface while SPIEM with mAb on the grid surface caused only slight obscuration (Derrick, 1973). However, considerable obs~u~tion and coating of virus surface structure were seen by SPIEM done by aggregation and trapping. Aggregation by the mAb and trapping of the aggregates by protein A increased the number of virus particles attached to the grid surface in SPIEM. Unfortunately, some loss of detail occurred. Surface detail was enhanced by washing with a stream of water from a wash bottle for 15-30 s. Occasionally there was some loss of antigen-antibody complex attachment to the grid and increased deterioration of the grid-coat in the electron microscope beam when this washing method was used. The kinetics of HAV trapping onto protein A coated grids (Fig. 6) were studied by incubating 14 formvar-carbon coated grids with 10 ~1 of 2.5 &ml protein A for 20 min, rinsing with phosphate buffer, and then incubating with a mixture of 10 ~1 of stock HAV HM-175 antigen and 10 ~1 of mAb diluted to 1:SOO. Two grids were removed from the drops at intervals of 1 min, 5 min, 20 min, 30 mitt, 1 h, 2 h, 3 h, and 20 h. The number of particles per grid square and particles per complex were observed and recorded. Virus trapping was apparent within 20 min after the incubation of antigen and antisera wtis begun; thereafter, aggregation and trapping of HAV progressed logarithmically. Aggregation equilibrium was reached within 2 h, while the attachment rate of particles to the grid, though decreased, had not reached equilibrium within 20 h.

Fig! 5. Comparison of TEM PTA negative stain images of HAV trapped on carbon-formvar coated grids by DEM (a), SPIEM, Ig capture only (b), and SPIEM, Ig aggregation and protein A trapping (c). Original magnifications: a, ~355,600, b, ~320,400, c, x331,750.

184

18 16

14 12

t, I 9

lo

fi! VI

8

2

6

I

4 2

r

P....,.................................,

0

4

a

NclBAnoN

0 12

16

a

(laJRs)

Fig. 6. Kinetics of HAV aggregation and trapping by SPIEM. Error bars represent standard error of the means. - - - -, particles per aggregate; -, particles per square.

The IgG fraction of convalescent HAV animal sera, two convalescent HAV antisera, and mAbs were compared for their relative ability to aggregate and trap HAV. HAS-15 and HM-175 tissue culture strains of HAV and HAV in 10% stool samples from an HAV infected chimpanzee were tested with several dilutions of the antisera and mAb preparations. More complex assays were attempted in which either rabbit or goat anti-human IgG in dilutions ranging from 1: 10 to 1: lo6 was used as grid coat. Anti-human Ig was used instead of and following protein A precoat before incubation with the HAV antigen-antisera complex. There were no significant differences observed in our mAb-SPIEM method when either formvarcarbon, collodion-carbon, commercially prepared formvar-carbon grids, or grids stored for several weeks were used with tissue-culture-derived virus, However, differences resulting from the method used to prepare grids were seen when 10% stools were assayed for HAV (Fig. 7). Freshly prepared collodion-carbon grids were better, under these conditions, than formvar-carbon grids for trapping virus from 10% stool suspensions. Neither protein A nor anti-human IgG, when used as the grid coat, changed this finding. Protein A was more effective than either rabbit or goat anti-human IgG alone as a grid coat for trapping HAV-antibody complexes. The rabbit anti-human IgG used in these experiments was more effective than goat anti-human IgG. However, incubation of the collodion-carbon grids with 2.5 pg/ml protein A, followed by rinsing with phosphate buffer, reincubation with anti-human IgG (rabbit), and washing, resulted in trapping of virus that was similar to trapping when protein A was used alone (Fig. 8). The IEM and SPIEM methods were compared to determine the capability of each for identifying HEV in stool extracts from infected patients in outbreak cases, in stool extracts obtained from experimentally infected non-human primates, and in sucrose gradients. IEM was done as described previously (Almeida and Waterson,

185

FCCC 1000 900 800 g

700

&- 600 cl. t:

500

6 F

400

% n

300

FCCC

FCCC

FCCC

FCCC

FCCC

200 100 0 1x10

1x10’ lxlol 5x103 ANTI HUMAN IgG (RABBIT) DILUTION (1 :X)

10’

;5/Jg/ml ;BSA ~PAOTEl~5mg/ml : A I

Fig, 7. Comparison of relative capabilities among grid preparations and grid coats for trapping HAV from 10% stools of infected human and non-human primates. Direct comparisons were made between formvar-carbon (F) and collodion-carbon (C) grids with human convalescent HAV antisera (HUl) (final dilution = 1:30). Direct comparisons were made between whole antisera and its IgG fraction with convalescent antisera obtained from non-human primate, Faylene (final dilution = 150). Standard error bars represent standard error of the means.

Fig. 8. SPIEM of HEV. Bar, 50 nm.

186

1969; Gravelle et al., 1975) and SPIEM was done as described above. We were able to trap and identify HEV (Fig. 8) by SPIEM in every assay (50-100 assays) in which IEM was successful. However, when conventional IEM was used for aliquots of the same specimens, we were often unable to isolate and identify virus particles with the morphologic characteristics of HEV. When IEM was used, unidentifiable debris, perhaps disrupted virus and antibody, was frequently observed, and intact virions were rarely seen.

Discussion The SPIEM procedure described here is a simple, highly sensitive assay that should facilitate rapid identification of small, labile virus particles. Monoclonal antibodies were superior to our best convalescent antisera or their IgG fractions for complexing HAV. The SPIEM method differs from those described by others since it combines classic aggregation procedures (Almeida and Waterson, 1969) with solid-phase techniques (Derrick, 1973; Shukla and Gough, 1979). The combination of aggregation and antibody decoration of virus particles by our assay permits rapid identification of small hepatitis viruses. The most important advantages, though, are the sparing of reagents required for the assay and the apparent decrease in virus particle disruption due to specimen manipulation. The only disadvantages are that an additional protein layer of protein A, or anti-Ig, on the carbon-plastic grid coat increases the background and that the aggregating Ig may obscure virus surface detail. Simple adsorption of antigen-antibody complex to grids or washing the grids with water after incubation with antigen-antibody complex decreases these obscuration problems when there is sufficient concentration of complex available. We encountered no difficulties with nonspecific aggregation in the low concentrations of virus used in these studies. However, when highly purified, highly concentrated preparations of HAV were viewed, considerable nonspecific aggregation was seen. Occasional nonspecific aggregation of antigen in stool preparations should be expected because of the variable nature of these specimens. Several considerations for negative-stain SPIEM of viruses were addressed during this study. The value of using a particular grid coating could not be shown with tissue culture virt+s. However, the capability to complex and trap virus from 10% stool preparations was enhanced by using freshly carbon-coated collodion grids. The probable reason for the difference in these preparations was that the tissue culture virus was from a crude’cell lysate and contained considerable serum protein and peptides fr
187

tides (Pegg-Feige and Doane, 1983). Concerns about grid hydrophilicity may also be minimized by using protein A as the first protein layer on the grid for SPIEM. Equally important as the initial wetting of the grid surface is the manner in which it dries after staining. We observed that bacitracin in the negative stain (Gregory and Pirie, 1973) was effective for providing hydrophilicity to the grid surface during staining, blotting, and drying. Buffer type and pH levels affected the capability of SPIEM to trap virus. The most effective trapping occurred when the pH of antigen and antisera each had a pH of 7-8. When using copper grids, Tris buffer should be avoided if possible because it reacts with these grids. The contamination difficulties are minimized if Tris buffer is blotted from both sides of the copper grids before staining, or if incubation time is no longer than 3 h. Nickel or gold grids should be considered as an alternative whenever Tris buffer is used since these metals are resistant to chemical oxidation. The kinetics experiments revealed changes in the numbers of particles per aggregate suggesting a means for observing and comparing the relative capabilities of different antisera to complex virus particles. The method of rating IEM virusantibody preparations by subjective scoring of the antisera coating density when viewed after negative staining has caused some concern for the reproducibility of the assay (Almeida et al., 1974; Lewis et al., 1988) since its development (Kapikian et al., 1972). The SPIEM experiments with standardized antigen and antisera suggest that interpretations based on the rate of particle aggregation in antigen-antibody mixtures are more objective and quantitative, and perhaps more reproducible than the rating systems used previously in IEM studies. In addition, the free interaction of antigen and antibody, as it occurs in our assay, may permit valid comparisons of these interactions in different antigen-antibody mixtures. Finally, more complex methods in which protein A is incubated with the grid surface, followed by anti-human Ig, and then incubated with the aggregating mixture may offer some gain in virus trapping in specimens with low titers of virus. These procedures may be useful for virus identification when simpler methods are unsuccessful.

References Almeida, J.D. and Waterson, A.P. (1969) The morphology of virus-antibody interaction. Adv. Virus Res. 15, 307-338. Almeida, J.D., Gay, F.W. and Wreghitt, T.G. (1974) Pitfalls in the study of hepatitis A. Lancet 2, 748-750. Bradley, D.W., Schable, C.A., McCaustland, K.A., Cook, E.H., Murphy, B.L., Fields, H.A., Ebert, J.W., Wheeler, C.W. and Maynard, J.E. (1984) Hepatitis A virus: growth characteristics of in vivo and in vitro propagated wild and attenuated virus strains. J. Med. Virol. 14, 373-386. Bradley, D.W., Krawczynski, K., Cook, E.H., McCaustland, K.A., Humphrey, C.D., Spelbring, J.E., Myint, H. and Maynard, J.E. (1987) Enterically transmitted non-A, non-B hepatitis: serial passage of disease in cynomolgus macaques and tamarins and recovery of disease associated 27-34 nm viruslike particles. Proc. Natl. Acad. Sci. 84, 62776281. Bradley, D.W., Andjaparidze, A., Cook, E.H., McCaustland, K., Balayan, M., Stetler, H., Velazquez, O., Robertson, B., Humphrey, C., Kane, M. and Weisfuse, I. (1988) Aetiological agent of enterically transmitted non-A, non-B hepatitis. J. Gen. Virol. 69, 731-738.

Cromeans, T., Sobsey, M.D. and Fields, H.A. (1987) Development of a plaque assay for a cytopathic, rapidly replicating isolate of hepatitis A virus. J. Med. Virol. 22, 45-56. Derrick, KS. (1973) Quantitative assay for plant viruses using serologically specific electron microscopy. Virology 56, 652-653. Gravelle, C.R., Bradley, D.W., Cook, E.H. Jr. and Maynard, J.E. (1975) Hepatitis A: report of a common source outbreak with recovery of a possible etiologic agent II. Laboratory studies. J. Infect. Dis. 131, 167-171. Gregory, D.W. and Pirie, B.J.S. (1973) Wetting agents for biological electron microscopy. I. General considerations and negative staining. J. Microsci. 99, 251-265. Gupta, H., Joshi, Y.K. and Tandon, B.N. (1988) An enzyme-linked immunoassay for the possible detection of non-A, non-B viral antigen in patients with epidemic viral hepatitis. Liver 8, 11 l-l 15. Gust, I.D. and Feinstone, SM. (1988) In: Hepatitis A, pp. 21-61. CRC Press, Inc., Boca Raton, Florida. Kapikian, A.Z., Wyatt, R.D., Thomhill, T.S., Kalica, A.R. and Chanock, R.M. (1972) Visualization by immune electron microscopy of a 27-nm particle associated with acute infectious nonbacterial gastroenteritis. J. Virol. 10, 1075-1081. Krawczynski, K., and Bradley, D.W. (1989) Enterically transmitted non-A, non-B hepatitis: identification of virus associated antigen in experimentally infected cynomolgous macaques. J. Infect. Dis. 159, 1042-1049. Lemon, S.M., Binn, L.N. and Marchwicki, R.H. (1983) Radioimmunofocus assay for quantitation of hepatitis A virus in cell cultures. J. Clin. Microbial. 17, 834-839. Lewis, D.C., Lightfoot, N.F. and Pether, J.V.S. (1988) Solid phase immune electron microscopy with human immunoglobulin M for serotyping of Norwalk-like viruses. J. Clin. Microbial. 26, 938-942. Pegg-Feige, K. and Doane, F.W. (1983) Effect of specimen support film in solid phase immunoelectron microscopy. J. Virol. Methods 7, 315-319. Shukla, D.R. and Gough, K.H. (1979) The use of protein A from Sruphylococcus auras, in immune electron microscopy for detecting plant virus particles. J. Gen Virol. 45, 533-536.