Efficient introduction of plasmid DNA into human hemopoietic cells by encapsidation in simian virus 40 pseudovirions

Proc. Natl. Acad. Sci. USA

Vol. 83, pp. 6925-6929, September 1986 Genetics

Efficient introduction of plasmid DNA into human hemopoietic cells by encapsidation in simian virus 40 pseudovirions (bone marrow cell/gene transfer/transient expression/simian virus 40 vector/gene therapy)

ARIELLA OPPENHEIM, AVIVA PELEG, EITAN FIBACH, AND ELIEZER A. RACHMILEWITZ Department of Hematology, Hadassah University Hospital, Jerusalem, Israel 91120

Communicated by Helen M. Ranney, May 15, 1986

Introduction of DNA into human hemopoietic ABSTRACT cells is required for the study of regulatory mechanisms operating in these cells, as well as for possible procedures of gene therapy. However, with hemopoietic cells the conventional technique of calcium phosphate precipitation is inefficient. The pathway of encapsidation of plasmid DNA as simian virus 40 (SV40) pseudovirions for the introduction of new genetic material was therefore investigated. Encapsidation was achieved in COS (monkey kidney) cells, which express SV40 large tumor (T) antigen constitutively. The vector, pSO, was introduced to the COS cells by DNA transfection. It carried the SV40 origin of replication (on), to facilitate replication of the plasmid in the COS cells. The SV40 capsid proteins were supplied in trans by a helper SV40 virus. The bacterial chloramphenicol acetyltransferase gene cat was used as a model for gene transmission. After encapsidation, the pseudovirions were used in infection of the human erythroleukemic cell line K562 and of normal human bone marrow cells. The results demonstrate that the cat gene can be transmitted with high efficiency. Over 40% of the infected K562 cells and 30% of the infected bone marrow cells were observed to contain plasmid DNA 48 hr after infection. Moreover, the results suggest that the efficiency of gene transmission by this vector can be improved and so may approach the theoretical 100%.

promoter, into the hemopoietic cells was observed, in comparison with the calcium phosphate precipitation technique. However, that SV40 vector system was found to generate wild-type SV40 virions with high frequency and was not further investigated. SV40 can efficiently infect human lymphoid and erythroid cells; 10-25% of the cells, infected at a multiplicity of 40, expressed high levels of virus-encoded large tumor (T) antigen (A.O., unpublished data). In the present communication, we describe an efficient method that utilizes SV40 pseudovirions to transmit DNA.

METHODS Cell Culture and Viral Stocks. COS (monkey kidney) cells, which are constitutive for SV40 T antigen (12), were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (FBS). Logarithmically growing cultures were used in transfection and infection experiments. Transfections were performed by the DEAE-dextran method (13). During transfection, one culture was treated with DEAE-dextran in parallel to the others, but without DNA (mock transfection). Infections were performed by incubating the cells with viral stocks for 90 min at 37°C, followed by the addition of fresh medium. For CAT assays the cells were harvested by trypsinization after 48 hr. The inducible human erythroleukemic cell line K562 was cultured in alpha medium with 10% FBS. For infection experiments, logarithmically growing cells (5 x 105 per ml) were collected by centrifugation. Virus stock (or a lysate prepared from mock-transfected cells, for mock infection) was added and allowed to adsorb for 2 hr at 37°C. The cells were then centrifuged and fresh medium (with 10% FBS) was added. Cells were counted and harvested after 48 hr. Human bone marrow cells were obtained from four healthy donors for marrow transplantation, designated A-D. The cells were separated on a Ficoll-Hypaque gradient and monocytes were removed by adherence to plastic. The cells were infected with viral stock as described for the K562 cells and harvested 48 hr later. Viral stocks were prepared by transfection of COS cells as described in Results. For high-titer stocks, part of the medium was removed after 3 days and the cultures were incubated in 2.0 ml of medium per 25-cm2 culture flask for 2 more days. Virus stocks were harvested by freeze-thaw (25). Mock-transfected cultures were processed in parallel. CAT Assays. CAT assays were performed essentially as described by Gorman et al. (14). The cells were sonicated in 0.25 M Tris/HCl buffer at pH 7.8. The extract (of 0.5 x 106 cells) was incubated in the presence of [14C]chloramphenicol (0.25 ,uCi; 1 Ci = 37 GBq) and 0.5 mM acetyl-CoA at 37°C for 30 or 60 min, in 0.25 M Tris/HCl buffer (pH 7.5) in a total volume of 180 ,ul. To separate the products from the sub-

A plethora of studies on the regulation of inserted genes in a variety of cultured cells has contributed enormously to the understanding of gene regulation in eukaryotic cells in recent years. However, the lack of an adequate method for introduction of cloned genes into hemopoietic cells in general, and human bone marrow cells in particular, has hindered the investigation of regulatory mechanisms in these cells. With hemopoietic cells, the conventional technique of calcium phosphate precipitation (1) is inefficient for introduction of DNA, as only 1 in 105 cells becomes transfected (2). Therefore, methods that are more efficient with hemopoietic cells have been developed. Protoplast fusion has increased the frequency of transfection in some cell lines to 10-4 (3). With electroporation (electric field-mediated transfer), DNA is introduced at an efficiency of 10-3 to 10-4 (4). Retroviral vectors (5, 6) were reported to be very efficient in introducing genetic material into murine (7, 8) and human hemopoietic stem cells (9, 10). The use of simian virus 40 (SV40) virions carrying the bacterial chloramphenicol acetyltransferase (CAT; acetylCoA:chloramphenicol 03-acetyltransferase, EC 2.3.1.28) gene cat to transmit genetic material into hemopoietic cells was explored by Karlsson et al. (11), who also investigated the use of adenovirus vectors. A great increase in the efficiency of introduction of the bacterial cat gene, fused to the Rous sarcoma virus (RSV) long terminal repeat (LTR) The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: CAT, chloramphenicol acetyltransferase; cat, gene encoding CAT; SV40, simian virus 40; bp, base pair(s).

6925

6926

Genetics:

Oppenheirn et al.

Proc. Natl. Acad. Sci. USA 83 (1986)

strate, the reaction mixture was fractionated by TLC. Radioactivity in the spots was quantitated in a liquid scintillation counter. The assays were performed at the linear range for enzyme activity and were later normalized to percent acetylation catalyzed by an extract of 106 cells during a 60-min incubation. In Situ Hybridization. To assay for the number of cells that incorporated plasmid DNA, individual cells were immobilized on nitrocellulose membrane filter for in situ hybridization as follows. The infected cells were harvested 48 hr after infection, washed in Dulbecco's phosphate-buffered saline without Ca2l and Mg2+ (PBS), and counted. The cells were serially diluted in serological plates, and known aliquots were spotted in triplicate onto nitrocellulose membrane filter. The nitrocellulose filter was washed in PBS and processed for in situ hybridization of DNA. This was accomplished by gentle cell lysis and DNA denaturation in alkali followed by neutralization and baking at 80'C under reduced pressure (15). The filters were hybridized to nick-translated pML2 (16) and not to pSO3cat, since the latter plasmid contains SV40 sequences (see Fig. 1) and would have cross-hybridized to the SV40 helper. pML2 did not cross-hybridize to the SV40 helper (data not shown). Autoradiography was performed by exposing the filter to x-ray film. Radioactive signals visualized on the film represent single cells (or clumps of cells) that contain pSO3cat DNA. The percentage of cells that incorporated plasmid DNA was computed by counting the signals. For example, in Fig. 5 Lower there are a total of 8 signals (2 + 3 + 3) in the 3 replicate spots that contain 8 cells each (altogether, 24 cells). Therefore, 8 of 24 cells (or 33%) incorporated pSO3cat DNA. The number of radioactive signals was not reduced by treating the cells with trypsin or by repeated washings with EDTA before they were spotted onto nitrocellulose, indicating that the plasmid DNA was contained within the cells and not in pseudovirions adsorbed to the cell surface.

RESULTS Encapsidation of Plasmid DNA as SV40 Pseudovirions. The

experimental design included encapsidation of the plasmid in SV40 coat proteins and infection of the hemopoietic cells with the plasmid-containing pseudovirion preparation. In the initial experiments, we used as a model for encapsidation and gene transmission the plasmid pSV2cat (ref. 14 and Fig. 1), constructed for expression of the bacterial cat gene in eukaryotic cells by fusion to SV40 expression signals (18). The experiments were performed as follows. COS cells were

cotransfected with pSV2cat and SV40 DNA by the DEAEdextran procedure (13). It was expected that pSV2cat DNA would replicate in these cells and would become encapsidated in the SV40 capsid proteins supplied by the helper viral DNA. Virion mixture, presumably containing pSV2cat pseudovirions and wild-type SV40, was harvested 5 days later. The virion mixture was analyzed as detailed below and used to transmit the cat gene into the hemopoietic cells by infection. To determine whether pSV2cat becomes encapsidated, logarithmically growing COS cells (25-cm2 cultures) were transfected in duplicate as follows: (a) pSV2cat plus SV40 DNA (0.5 ,ug each), (b) pSV2cat (1 ,ug), (c) SV40 DNA (1 ,g), and (d) mock transfection (i.e., no DNA). After 48 hr, one of each pair of duplicate cultures was harvested and assayed for CAT enzymatic activity. As expected, CAT activity was observed only in cultures a and b. The other duplicate culture was incubated for 5 days to allow for DNA replication and encapsidation. These cultures were harvested by repeated freeze-thaw to break up the cells and liberate viral particles. When the cell lysates were viewed under the electron microscope, SV40-like particles were observed only in the cultures transfected with SV40 and with pSV2cat plus SV40. Curiously, more empty capsids were observed in the SV40 transfection as compared with the cotransfection by pSV2cat and SV40. Empty SV40 capsids were reported to constitute a large part of the SV40 viral stock prepared in monkey kidney cells (19). For estimation of the titer, virion particles were counted at several dilutions together with a standard of known titer (bacteriophage T4) that was mixed with the cell lysates. The titer of the pSV2cat/SV40 virion mixture was computed to be 2 x 108 "full" particles per ml. Encapsidation of pSV2cat was inferred from transmission of the cat gene into fresh cultures of COS cells (Fig. 2A). Duplicate cultures were each infected by incubation with 0.5 ml of the cell lysate. One set was incubated for 5 days to allow new virion particles to form, and the other was harvested after 48 hr and assayed for CAT activity (Fig. 2A, infection cycle 1). The virus and pseudovirus were thus repassaged in COS cells several times. The results, summarized in Fig. 2A, show that the cat gene was not transmitted in the absence of the helper SV40. Therefore, we concluded that encapsidation was taking place and was obligatory for the process. It can be observed that CAT activity was rapidly declining with every subsequent infection. Analysis of Hirt supernatants (20) obtained from the infected cultures revealed that SV40 was overtaking the infection. Hirt supernatants of the cells 100 _

EcoRl

A

100 _

B

(Cla I/Hpa II)

Hilndlll U>~

E pSO3cat z

H>50 tz

O Hiindlll

(Hae II) BarnHI

FIG. 1. Structure of plasmids carrying the cat gene. pSV2cat (14) kindly provided by C. Gorman. pSO3cat was constructed from pBRM (ref. 17; obtained from J. Hartman), a derivative of pBR322 (BamHI-EcoRI); pBR322 (EcoRI-Cla I); SV40 (Hpa II-HindIII); and pSV2cat (HindIII-BamHI). pBR322 sequence is represented by and SV40 sequence, by - *. Tn9 is represented by a thick line. Direction of transcription of cat and the ampicillin-resistance gene amp are shown. Parentheses indicate restriction sites that were destroyed in constructing the plasmid. bp, Base pairs. was

-

4072 bp

(a Hfae 11

50

-

L

a o -0 T

1

2

T

1

2

Infection cycle FIG. 2. Encapsidation of the plasmids with SV40 as a helper. The experiments are described in detail in the text. Logarithmically growing COS cells (25-cm2 cultures) were transfected as follows. (A) *, pSV2cat and SV40 DNA (0.5 ,g each); o, pSV2cat (1 ug); A, SV4O DNA (1 ,ug). (B) *, pSO3cat and SV40 DNA (0.5 jig each); o, pS03cat (1 ,g); A, SV40 DNA (1 ,ug). CAT activity is expressed as the percentage of chloramphenicol acetylated during 1 hr by an extract of 106 cells. T, transfection.

Genetics:

Proc. Natl. Acad. Sci. USA 83 (1986)

Oppenheirn et al.

infected in the second cycle contained more than 80% SV40 DNA and little pSV2cat DNA (results not shown). Infection of hemopoietic cells-MEL (murine erythroleukemia), K562, and human bone marrow cells-showed that the cat gene was introduced into those cells by the pSV2cat pseudovirus. However, transmission was not very efficient, as deduced from CAT activity in transient-expression experiments. Only 2-5% of the chloramphenicol substrate became acetylated during a 1-hr incubation at 370C with an extract of 106 cells. To improve the efficiency of gene transmission, a new plasmid carrying the cat gene was constructed, pSO3cat (Fig. 1). pBR322 sequences that may interfere with replication initiated at the SV40 origin of replication (ori) (16) were removed by transferring the cat gene and SV40 regulatory signals into pBRM (17). In addition, some SV40 late sequences were removed and the orientation of cat with respect to the vector sequences was reversed. As will be described below, pSO3cat was significantly more efficient than pSV2cat in transmission of cat. This could be due to any one of the differences between the two plasmids. To find out whether replication of pSO3cat was more efficient, we transfected COS cells with each of the two plasmids, at two different DNA levels, and assayed for CAT after 48 hr. The rationale was that a higher starting level of pSV2cat DNA might compensate for its poor replication. However, the results (not shown) indicated that this was not the case. This question was not further investigated. Encapsidation of pSO3cat was studied by gene transmission in COS cells as described for pSV2cat (Fig. 2B). The new plasmid appeared satisfactory and was used in subsequent experiments. Since CAT activity was declining after the first infection cycle, the first viral mixture obtained from the transfected COS cells was used in infection of hemopoietic cells. Infection of K562 Cells. Logarithmic cultures of K562 cells were infected with viral stocks as described in Methods. The number of cells that incorporated plasmid DNA was assayed by in situ hybridization of individual cells immobilized on nitrocellulose membrane filter. The number of cells that incorporated plasmid DNA, as well as total CAT activity in the transient expression assay, increased with the multiplicity of infection (Fig. 3). At the highest multiplicity used in these experiments, more than 40% of the cells that were infected incorporated the plasmid DNA. The efficiency of the infection probably could be. improved by increasing the multiplicity of infection and by optimizing the conditions for adsorption. Similar results (not shown) were obtained with MEL cells.

6927

Infection of Human Bone Marrow Cells. Several human bone marrow samples were infected with the same virion mixture and analyzed for CAT activity (Fig. 4) and the number of cells that incorporated plasmid DNA (Fig. 5). The results are summarized in Table 1. Although differences between different bone marrow samples were observed, it can be seen that about 30% of the cells incorporated pSO3cat DNA. In addition, CAT activity in these transient-expression assays was reasonably high: 6-40% of the chloramphenicol substrate was acetylated during 1 hr at 37°C. The results suggest that this vector can be used for transient-expression studies in human bone marrow cells.

DISCUSSION The studies reported here describe encapsidation of plasmid DNA as SV40 pseudovirions and the use of this approach to introduce DNA into human hemopoietic cells. Our results demonstrate that the plasmid pSO3cat is significantly more efficient than pSV2cat in supporting the expression of cat and in gene transmission. Additional experiments are required to establish which DNA signals are responsible for this difference. The pSO construct reported here may be superior to pSV2 (18) in the expression of other genes as well. In pSO3cat, the cat gene is transcribed from the SV40 early promoter, which has been found to be several times less efficient than the Rous sarcoma virus (RSV) long terminal repeat (LTR) promoter in a variety of cells (14). It will be interesting to find whether combining the RSV promoter with the pSO construct will increase gene transcription even further. The results demonstrate that pSO is an efficient vector for introducing genetic material into human hemopoietic cells. In the experiments described here, 30-40% of the hemopoietic cells became infected. The results in Fig. 3 and Table 1 suggest that the efficiency of the vector can be improved by increasing the multiplicity of infection and may approach the theoretical 100%. With this method, one could study the _ CM 1,3-diacetate

-CM 3-acetate 60

60~

>0'=40

40~

U

20 0

A~~~~~~

-CM 1-acetate

*

*

*

1

2

3

_-CM ..

20

0

1.0 0.2 0.4 Multiplicity of infection FIG. 3. Effect of multiplicity of infection (ml of virus stock per 106 cells) on infection of K562 cells. CAT activity (e) is expressed as % chloramphenicol acetylated during 1 hr by an extract of 106 cells. The

percentage of cells with pSO3cat DNA of the total cells that were infected (A, % infected cells) was assayed by in situ hybridization as in Fig. 6.

4

FIG. 4. CAT activity in human bone marrow cells. Bone marrow samples from donors C and D (Table 1) were harvested 48 hr after infection. Extracts of 106 cells were incubated with [14C]chloramphenicol (CM) and acetyl-CoA for 1 hr at 37°C. The reaction mixture was analyzed by TLC followed by autoradiography. Lanes 1 and 2: mock-infected and infected bone marrow cells, respectively, from donor D. Lane 3: infected bone marrow cells from donor C. Lane 4: control CAT extract from Escherichia coli, prepared by sonicating E. coli cells harboring cat in Tn9.

Genetics:

6928

Proc. Natl. Acad. Sci. USA 83 (1986)

Oppenheirn et al. Cells per spot .dailk,

AWOL-

-1000

"",6, .i

W:Wv'' O

-

500

-

250

-

125

00*

0 0 w -:90LAk. .a.

M.

40 *

.014,..

-62

i 9

;4.

-

31

-

16

-8 C

D

mockinfected

mockinfected

+

c pSO3cat (0.5 ml)

*^s

D + pSO3cat

(2.0 ml)

*411-31

FIG. 5. Analysis of the number of infected cells by in situ hybridization. Human bone marrow cells (donors C and D) were harvested 48 hr after infection, counted, diluted in triplicate, and spotted on nitrocellulose. (Upper) The filter was hybridized to nick-translated pML2 DNA and autoradiographed on x-ray film for 2 days. (Lower) The bottom-right portion of a 7-day exposure of the same filter, magnified. The numerical results appear in Table 1.

regulation of foreign genes introduced into hemopoietic cells by relying on transient-expression studies. Tissue-specific expression of genes in different hemopoietic lineages could be explored in conjunction with appropriate tissue culture techniques. By avoiding the need for selection in tissue culture, the complications that usually arise in interpretation of the results due to random integration of the incoming DNA into the host chromosome would be avoided. In long-term experiments, we anticipate that the plasmid will integrate at random chromosomal locations, as has been observed before for other SV40 vectors. Table 1. Infection of human bone marrow cells with

pSO3cat pseudovirus Pseudoviral preparation, ml per 106 cells

%

% cells with chloramphenicol pSO3cat DNA acetylated Donor 0 0 0 A 0 0 0.4 (mock) 6 3 0.4 (pSO3cat) 0 0 2.0 (mock) B 8 23 2.0 (pSO3cat) 0 0 0.5 (mock) C 9 4 0.5 (pSO3cat) 0 0 2.0 (mock) D 41 33 2.0 (pSO3cat) Experiments are described in the text. Chloramphenicol acetylation was assayed for 1 hr at 370C with an extract of 106 cells. The percentage of cells that incorporated pSO3cat DNA was calculated from in situ hybridization experiments, as shown in Fig. 5.

SV40 sequences have beep used in many different vectors. Extrapolation of previous results suggests that genes that will be carried by pSO will not undergo excessive rearrangements after introduction into eukaryotic cells. In particular, intron and flanking sequences will not be destabilized. This is most important in view of the findings that regulatory signals sometimes reside within introns (21-24). In addition, available information suggests that SV40 pseudovirions will have a wide, perhaps unlimited, host range. The prokaryotic pBRM DNA can be removed from the plasmid before introduction into the COS cells for encapsidation. In several cases, prokaryotic vector sequences have been found to interfere with regulatory processes in eukaryotic cells (16, 26). This procedure will increase the cloning capacity of the vector, which is limited by size requirement for encapsidation in SV40 to about 5.5 kbp. The only part of the vector essential for the pseudoviral preparation is the SV40 ori region, which is contained in a fragment of '200 bp, allowing additional cloning space of =5 kbp. In the present construct, pSO3cat, the SV40 enhancer and expression signals are included to facilitate expression of the bacterial cat gene in eukaryotic cells. However, in constructs that carry eukaryotic genes, these DNA sequences could be removed. Under these conditions, when the vector part will include only the SV40 ori, we anticipate minimal interference of vector signals with the regulation of the cloned gene. Thus this vector will be most suitable for studies on the regulation of normal genes in hemopoietic cells. Efficient introduction of DNA into human bone marrow cells is also a prerequisite for gene therapy of thalassemia, sickle cell anemia, and some enzyme deficiencies. Our method may be suitable for this purpose when a helper-free pseudoviral preparation becomes available. For this, a helper plasmid that will supply the necessary SV40 functions in trans but will not be encapsidated, and therefore will not be contained in the pseudoviral preparation, is needed. It is anticipated that the use of SV40 will facilitate rigorous control and routine screening for absence of SV40 virions in the helper-free preparations, to ensure the level of safety required for gene therapy. We wish to thank Ms. A. Treves and M. Yanuv for assistance in some of the experiments. This research was supported by the Harry and Jenny Klein Fellowship in Medical Research and by the Joel Ostrowicz Foundation. 1. Graham, F. L. & van der Eb, A. J. (1973) Virology 52, 456-467. 2. Miller, G., Wertheim, P., Wilson, G., Robinson, J., Geelen, J. L. M. C., van der Noordaa, J. & van der Eb, A. J. (1979) Proc. Natl. Acad. Sci. USA 76, 949-953. 3. Oi, V. T., Morrison, S. L., Herzenberg, L. A. & Berg, P. (1983) Proc. Natl. Acad. Sci. USA 80, 825-829. 4. Potter, H., Weir, L. & Leder, P. (1984) Proc. Natl. Acad. Sci. USA 81, 7161-7165. 5.' Joyner, A., Keller, G., Phillips, R. A. & Bernstein, A. (1983) Nature (London) 305, 556-558. 6. Cepko, C. L., Roberts, B. E. & Mulligan, R. C. (1984) Cell 37, 1053-1062. 7. Williams, D. A., Lemischka, I. R., Nathan, D. G. & Mulligan, R. C. (1984) Nature (London) 310, 476-480. 8. Dick, J. E., Magli, M. C., Huszar, D., Phillips, R. A. & Bernstein, A. (1985) Cell 42, 71-79. 9. Gruber, H. E., Finley, K. D., Hershberg, R. M., Katzman, S. S., Laikind, P. K., Seegmiller, J. E., Friedmann, T., Yee, J. K. & Jolly, D. J. (1985) Science 230, 1057-1061. 10. Hock, R. A. & Miller, A. D. (1986) Nature (London) 230, 275-277. 11. Karlsson S., Humphries, R. K., Gluzman, Y. & Nienhuis, A. W. (1985) Proc. Natl. Acad. Sci. USA 82, 158-162. 12. Gluzman, Y. (1981) Cell 23, 175-182. 13. McCutchan, J. H. & Pagano, J. S. (1968) J. Natl. Cancer Inst. 41, 351-357.

Genetics: Oppenheim et al. 14. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol. Cell. Biol. 2, 1044-1051. 15. Winocour, E. & Keshet, I. (1980) Proc. Nati. Acad. Sci. USA 77, 4861-4865. 16. Lusky, M. & Botchan, M. (1981) Nature (London) 293, 79-81. 17. Hartman, J. R., Nayak, D. P. & Fareed, G. C. (1982) Proc. Natl. Acad. Sci. USA 79, 233-237. 18. Mulligan, R. & Berg, P. (1980) Science 209, 1422-1427. 19. Black, P. H., Crawford, E. M. & Crawford, L. V. (1964) Virology 24, 381-387. 20. Hirt, B. (1967) J. Mol. Biol. 26, 365-369.

Proc. Nati. Acad. Sci. USA 83 (1986)

6929

21. Gillies, S. D., Morrison, S. L., Oi, V. T. & Tonegawa, S. (1983) Cell 33, 717-728. 22. Banerji, J., Olson, I. & Schaffner, W. (1983) Cell 33, 729-740. 23. Queen, C. & Baltimore, D. (1983) Cell 33, 741-748. 24. Moore, D. D., Marks, A. R., Buckley, D. I., Kapler, G., Payvar, F. & Goodman, H. M. (1985) Proc. Natl. Acad. Sci. USA 82, 699-702. 25. Chou, J. Y., Avila, J. & Martin, R. G. (1974) J. Virol. 14, 116-124. 26. Chada, K., Magram, J., Raphael, K., Radice, G., Lacy, E.'& Constantini, F. (1985) Nature (London) 314, 377-380.