You are on page 1of 5

news and views feature

could considerably benefit patients with
Gene therapy – neurological disease. And finally, we can
consider the transfer of genes to a handful of
stem (or progenitor) cells, which grow and

promises, problems divide to generate millions of progeny. The
range in the number of cells that this technol-
ogy has to cover is vast.
The Achilles heel of gene therapy is gene
and prospects delivery, and this is the aspect that we will
concentrate on here. Thus far, the problem
has been an inability to deliver genes effi-
ciently and to obtain sustained expression.
Inder M. Verma and Nikunj Somia
There are two categories of delivery vehicle
In principle, gene therapy is simple: putting corrective genetic material (‘vector’). The first comprises the non-viral
into cells alleviates the symptoms of disease. In practice, considerable vectors, ranging from direct injection of
obstacles have emerged. But, thanks to better delivery systems, there DNA to mixing the DNA with polylysine or
is hope that the technique will succeed. cationic lipids that allow the gene to cross
the cell membrane. Most of these approaches
n 1990, the first clinical trials for gene- status, accessibility and economics. suffer from poor efficiency of delivery and

I therapy approaches to combat disease
were carried out. Conceptually, the tech-
nique involves identifying appropriate DNA
We also need to consider how much of the
therapeutic protein should be delivered. In
haemophilia B, which is caused by a deficien-
transient expression of the gene6. Although
there are reagents that increase the efficiency
of delivery, transient expression of the
sequences and cell types, then developing cy of a blood-clotting protein called factor transgene is a conceptual hurdle that needs
suitable ways in which to get enough of IX, giving patients just 5% of the normal cir- to be addressed.
the DNA into these cells. With efficient deliv- culating levels of this protein can substan- Most of the current gene-therapy
ery, the therapeutic prospects range from tially improve their quality of life2. Most peo- approaches make use of the second category
tackling genetic diseases and slowing the ple have about 5 mg of factor IX per millilitre — viral vectors. Importantly, the viruses
progression of tumours, to fighting viral of plasma, produced by the 1013 cells that used have all been disabled of any pathogenic
infections and stopping neurodegenerative make up the liver. So we need to deliver a effects. The use of viruses is a powerful tech-
diseases. But the problems — such as the lack correcting gene to 5 2 1011 cells — that is, nique, because many of them have evolved a
of efficient delivery systems, lack of sustained 5% of liver cells. Alternatively, fewer liver specific machinery to deliver DNA to cells.
expression, and host immune reactions — cells would need to be modified if more fac- However, humans have an immune system
remain formidable challenges. tor IX could be produced per cell, without to fight off the virus, and our attempts to
Although more than 200 clinical trials being deleterious. In the brain, however, deliver genes in viral vectors have been

are currently underway worldwide, with gene transfer to just a few hundred cells confronted by these host responses.
hundreds of patients enrolled, there is still no Enhancer–promoter
single outcome that we can point to as a suc- Y
LTR Factor IX LTR Transgene
cess story. To explore why this is the case, we
will use our own experience and other exam- a Transfection
ples to look at the many technical, logistical
and, in some cases, conceptual hurdles that Infection
need to be overcome before gene therapy c
becomes routine practice in medicine.
At present, gene therapy is being RNA Reverse
contemplated only on somatic (essentially, Nucleus
non-reproductive) cells. Although many b

somatic tissues can receive therapeutic Viral RNA
Env d
DNA, the choice of cell usually depends on
Gag RNA–DNA hybrid
the nature of the disease. Sometimes a clear Pol
definition of the target cell is needed. For Packaging e
Double-stranded DNA
example, the gene that is defective in cystic cell Integration
fibrosis has been identified, and clinical Factor IX protein
trials to deliver DNA as an aerosol into the
lung have already begun1. Although cystic cell Translation
fibrosis is manifest in this organ, it is still not Cytoplasm Nucleus DNA
clear that delivery of a correcting gene by
this method will reach the right type of cell. Figure 1 To create the retroviral vectors that are used in gene therapy, the life-cycles of their
On the other hand, to correct blood-clot- naturally occurring counterparts are exploited. a, The transgene (in this case, the gene for factor IX)
ting disorders such as haemophilia, all that in a vector backbone is put into a packaging cell, which expresses the genes that are required for viral
is needed is a therapeutic level of clotting integration (gag, pol and env). b, The transgene is incorporated into the nucleus, where it is
protein in the plasma2. This protein may be transcribed to make vector RNA. This is then packaged into the retroviral vector, which is shed from
supplied by muscle or liver cells, fibroblasts, the packaging cell. c, The vector is delivered to the target cell by infection. The membrane of the viral
or even blood cells3–5. The choice of tissue in vector fuses with the target cell, allowing the vector RNA to enter. d, The virally encoded enzyme
which to express the therapeutic protein reverse transcriptase converts the vector RNA into an RNA–DNA hybrid, and then into double-
will also ultimately depend on considera- stranded DNA. e, The vector DNA is integrated into the host genome, then the host-cell machinery
tions such as the efficiency of gene deliv- will transcribe and translate it to make RNA and, in this case, factor IX protein.
ery, protein modifications, immunological LTR, long terminal repeat; c, packaging sequence.

NATURE | VOL 389 | 18 SEPTEMBER 1997 Nature © Macmillan Publishers Ltd 1997 239
news and views feature
Retroviral vectors grown in vitro, and infected with the recom- of trial and error for a given type of cell.
Retroviruses are a group of viruses whose binant retroviral vector. The target cells pro- Another formidable challenge to the ex
RNA genome is converted to DNA in the ducing the foreign protein are then trans- vivo approach is the efficiency of transplan-
infected cell. The genome comprises three planted back into the animal. This procedure tation of the infected cells. Attempts to
genes termed gag, pol and env, which are has been termed ‘ex vivo gene therapy’ and repeat the long-term myoblast transplanta-
flanked by elements called long terminal our group has used it to infect mouse pri- tion in haemophiliac dogs led to only short-
repeats (LTRs). These are required for inte- mary fibroblasts or myoblasts (connective- term expression, because the infected dog
gration into the host genome, and they define tissue and muscle precursors, respectively) myoblasts could not fuse with the muscle
the beginning and end of the viral genome. with retroviral vectors producing the factor fibres. So perhaps successful animal models
The LTRs also serve as enhancer–promoter IX protein. But within five to seven days of will prove inadequate when the same proto-
sequences — that is, they control expression transplanting the infected cells back into cols are extended to humans. Moreover,
of the viral genes. The final element of the mice, expression of factor IX is shut off 3,5,9. these models are based on inbred animals —
genome, called the packaging sequence (c), This transcriptional shut-off has even been the outbred human population, with indi-
allows the viral RNA to be distinguished from observed in mice lacking a functional vidual variation, will add yet another degree
other RNAs in the cell (Fig. 1)7. immune system (nude mice), and it cannot of complexity. The haematopoietic (blood-
By manipulating the viral genome, viral be due to cell loss or gene deletion5 because producing) system may offer an advantage
genes can be replaced with transgenes — such the transplanted cells can be recovered. for ex vivo gene therapy because resting stem
as the gene for factor IX (Table 1). Tran- What is the mechanism of this unexpect- cells can be stimulated to divide in vitro
scription of the transgene may be under ed but intriguing problem? We do not yet using growth factors and the transplanta-
the control of viral LTRs or, alternatively, know, but the exceptions may provide some tion technology is well established. But there
enhancer–promoter elements can be engi- clues. To obtain sustained expression in is still a lack of good enhancer–promoter
neered in with the transgene. The chimaeric mouse muscle following the transplantation combinations that allow sustained produc-
genome is then introduced into a packaging of infected myoblasts, we used the muscle tion of high levels of protein in these cells.
cell, which produces all of the viral proteins creatine kinase enhancer–promoter to con- Another problem is the possibility of
(such as the products of the gag, pol and env trol transcription of the transgene. Unfortu- random integration of vector DNA into the
genes), but these have been separated from nately, this is a weak promoter, and only low host chromosome. This could lead to activa-
the LTRs and the packaging sequence. So, levels of protein were produced. So, we tion of oncogenes or inactivation of tumour-
only the chimaeric viral genomes are assem- generated a chimaeric vector in which the suppressor genes. Although the theoretical
bled to generate a retroviral vector. The cul- muscle creatine kinase enhancer was linked probability of such an event is quite low, it is of
ture medium in which these packaging cells to a strong promoter from cytomegalovirus. some concern (see section on clinical trials).
have been grown is then applied to the target Using this vector, sustained and high levels of
cells, resulting in transfer of the transgene. factor IX were expressed when the infected Lentiviral vectors
Typically, a million target cells on a culture myoblasts were transplanted back into Lentiviruses also belong to the retrovirus
dish can be infected with one millilitre of the mice. Remarkably, these expression levels family, but they can infect both dividing
viral soup. remained unchanged for more than two and non-dividing cells10. The best-known
A critical limitation of retroviral vectors years (the life of the animal). So we can lentivirus is the human immunodeficiency
is their inability to infect non-dividing cells8, override the ‘off switch’ and achieve higher virus (HIV), which has been disabled and
such as those that make up muscle, brain, levels of expression by using an appropriate developed as a vector for in vivo gene
lung and liver tissue. So, when possible, the enhancer–promoter combination. But the delivery. Like the simple retroviruses, HIV
cells from the target tissue are removed, search for such combinations is a case has the three gag, pol and env genes, but it
Table 1 Candidate diseases for gene therapy also carries genes for six accessory proteins
Disease Defect Incidence Target cells termed tat, rev, vpr, vpu, nef and vif 11.
Genetic Using the retrovirus vectors as a model,
Severe combined Adenosine deaminase Rare Bone-marrow cells or lentivirus vectors have been made, with the
immunodeficiency (ADA) in ~25% of SCID T lymphocytes transgene enclosed between the LTRs and a
(SCID/ADA) patients
packaging sequence12. Some of the accessory
A Factor VIII deficiency 1:10,000 males Liver, muscle, fibroblasts
Haemophilia { B Factor IX deficiency 1:30,000 males
or bone-marrow cells proteins can be eliminated without affecting
production of the vector or efficiency of
Familial Deficiency of low-density 1:1 million Liver infection. The env gene product would
hypercholesterolaemia lipoprotein (LDL) receptor
restrict HIV-based vectors to infecting only
Cystic fibrosis Faulty transport of salt in 1:3,000 Caucasians Airways in the lungs
lung epithelium.
cells that express a protein called CD4+ so, in
Loss of CFTR gene the vectors, this gene is substituted with env
Haemoglobinopathies: Structural defects in 1:600 in certain Bone-marrow cells, giving sequences from other RNA viruses that have
thalassaemias/ a- or b-globin gene ethnic groups rise to red blood cells a broader infection spectrum (such as glyco-
sickle-cell anaemia
protein from the vesicular stomatitis virus).
Gaucher’s disease Defect in the enzyme 1:450 in Bone-marrow cells,
glucocerebrosidase Ashkenazi Jews macrophages
These vectors can be produced — albeit on a
a1-antitrypsin deficiency: Lack of a1-antitrypsin 1:3,500 Lung or liver cells
small scale at the moment — at concentra-
inherited emphysema tions of >109 virus particles per ml (ref.12).
Acquired When lentivirus vectors are injected into
Cancer Many causes, 1 million/year in USA Variety of cancer-cell rodent brain, liver, muscle, eye or pancrea-
including genetic types; liver, brain, tic-islet cells, they give sustained expression
and environmental pancreas, breast, kidney
for over six months — the longest time test-
Neurological diseases Parkinson’s, Alzheimer’s, 1 million Parkinson’s Direct injection in the
spinal-cord injury and 4 million Alzheimer’s brain, neurons, glial cells, ed so far13,14. Unlike the prototypical retrovi-
patients in USA Schwann cells ral vectors, the expression is not subject to
Cardiovascular Restinosis, 13 million in USA Arteries, vascular ‘shut off ’. Little is known about the possible
arteriosclerosis endothelial cells immune problems associated with lentiviral
Infectious diseases AIDS, hepatitis B Increasing numbers T cells, liver, macrophages vectors, but injection of 107 infectious units
240 Nature © Macmillan Publishers Ltd 1997 NATURE | VOL 389 | 18 SEPTEMBER 1997
news and views feature
does not elicit the cellular immune response
at the site of injection. Furthermore, there What makes an ideal vector?
seems to be no potent antibody response.
So, at present, lentiviral vectors seem to offer All of the current without attendant to be successfully
an excellent opportunity for in vivo gene methods of gene delivery immunological problems maintained as a stable
delivery with sustained expression. — whether viral or non- is more desirable. An episome;
viral — have some ideal vector may have to ● A transcriptional unit
Adenoviral vectors limitation. So, the choice borrow properties from that can respond to
The adenoviruses are a family of DNA virus- of vector will often be both viral and synthetic manipulation of its
es that can infect both dividing and non- dictated by the need. If systems, and it should regulatory elements;
dividing cells, causing benign respiratory- expression of the gene is have: ● Ability to target the
tract infections in humans11. Their genomes required for only a short ● High concentration desired type of cell;
contain over a dozen genes, and they do not time (for example, (>10 8 viral particles per ● No components that
usually integrate into the host DNA. Instead, expression of a toxic ml), allowing many cells elicit an immune
they are replicated as episomal (extrachro- gene-product in cancer to be infected; response.
mosomal) elements in the nucleus of the cells), then the adenoviral ● Convenience and Although no such
host cell. Replication-deficient adenovirus vectors are ideal. But if reproducibility of vector is currently
vectors can be generated by replacing the E1 sustained expression is production; available, all of these
gene — which is essential for viral replica- needed (such as for ● Ability to integrate in a properties exist,
tion — with the gene of interest (for exam- most genetic diseases), site-specific location in individually, in disparate
ple, that for factor IX) and an enhancer–pro- then an integrating vector the host chromosome, or delivery systems.
moter element. The recombinant vectors are
then replicated in cells that express the prod- ther with the generation of ‘gut-less’ vectors tially20 into human chromosome 19. To pro-
ucts of the E1 gene, and they can be generat- — all of the viral genes are deleted, leaving duce an AAV vector, the rep and cap genes are
ed in very high concentrations (>1011–1012 only the elements that define the beginning replaced with a transgene. Up to 1011–1012
adenovirus particles per ml)15. and the end of the genome, and the viral viral particles can be produced per ml, but
Cells infected with recombinant adeno- packaging sequence. The transgenes carried only one in 100–1,000 particles is infectious.
virus can express the therapeutic gene but, by these viruses were expressed for 84 days19. Moreover, preparation of the vector is labori-
because essential genes for viral replication There are considerable immunological ous and, due to the toxic nature of the rep gene
are deleted, the vector should not replicate. problems to be overcome before adenoviral product and some of the adenoviral helper
These vectors can infect cells in vivo, causing vectors can be used to deliver genes and pro- proteins, we currently have no packaging
them to express very high levels of the trans- duce sustained expression. The incoming cells in which all of the proteins can be stably
gene. Unfortunately, this expression lasts for adenoviral proteins that package DNA can provided. Vector preparations must also be
only a short time (5–10 days post-infection). be transported to the cytoplasm where they carefully separated from any contaminating
In contrast to the retroviral vectors, long- are processed and presented on the cell sur- adenovirus, and AAV vectors can accommo-
term expression can be achieved if the face, tagging the cell as infected for destruc- date only 3.5–4.0 kilobases of foreign DNA —
recombinant adenoviral vectors are intro- tion by cytotoxic T cells. So adenoviral vec- this will exclude larger genes. Finally, we need
duced into nude mice, or into mice that tors are extremely useful if expression of the more information about the immunogeni-
are given both the adenoviral vector and transgene is required for short periods of city of the viral proteins, especially given that
immunosuppressing agents16. This indicates time. One promising approach is to deliver 80% of the adult population have circulating
that the immune system is behind the short- large numbers of adenoviral vectors contain- antibodies to AAV. These considerations
term expression that is usually obtained ing genes for enzymes that can activate cell notwithstanding, AAV vectors containing
from adenoviral vectors. killing, or immunomodulatory genes, to human factor IX complementary DNA have
The immune reaction is potent, eliciting cancer cells. In this case, the cellular immune been used to infect liver and muscle cells
both the cell-killing ‘cellular’ response and response against the viral proteins will in immunocompetent mice. The mice
the antibody-producing ‘humoral’ response. augment tumour killing. Finally, although produced therapeutic amounts of factor IX
In the cellular response, virally infected cells immunosuppressive drugs can extend the protein in their blood for over six months21,22,
are killed by cytotoxic T lymphocytes16,17. The duration of expression, our goal should be to confirming the promise of AAV as an in vivo
humoral response results in the generation manipulate the vector and not the patient. gene-therapy vector.
of antibodies to adenoviral proteins, and it
will prevent any subsequent infection if the Adeno-associated viral vectors Other vectors
animal is given a second injection of the A relative newcomer to the field, adeno-asso- Among the other viruses being considered
recombinant adenovirus. Unfortunately for ciated virus (AAV) is a simple, non-patho- and developed, is herpes simplex virus, which
gene therapy, most of the human population genic, single-stranded DNA virus. Its two infects cells of the nervous system23. The virus
will probably have antibodies to adenovirus genes (cap and rep) are sandwiched between contains more than 80 genes, one of which
from previous infection with the naturally inverted terminal repeats that define the (IE3) can be replaced to create the vector.
occurring virus. beginning and the end of the virus, and con- Around 108–109 viral particles are produced
It is possible that the target cell contains tain the packaging sequence20. The cap gene per ml, but the residual proteins are toxic to
factors that might trigger the synthesis of encodes viral capsid (coat) proteins, and the the target cell. Additional genes can be delet-
adenoviral proteins, leading to an immune rep gene product is involved in viral replica- ed, allowing more than one transgene to be
response. To try to get around this problem, tion and integration. AAV needs additional included. But if essentially all of the viral
second-generation adenoviral vectors were genes to replicate, and these are provided by proteins are deleted (gut-less vectors), only
developed, in which additional genes that are a helper virus (usually adenovirus or herpes around 106 viral particles are produced per
implicated in viral replication were deleted. simplex virus). ml. And, again, many people have an immu-
These vectors showed longer-term expres- The virus can infect a variety of cell types, nity to components of herpes simplex virus,
sion, but a decline after 20–40 days was still and — in the presence of the rep gene product having already been infected at some time.
apparent18. This idea has now been taken fur- — the viral DNA can integrate preferen- Vaccinia-virus-based vectors have also
NATURE | VOL 389 | 18 SEPTEMBER 1997 Nature © Macmillan Publishers Ltd 1997 241
news and views feature
been explored, largely for generating vac- procedures for infection and transplantation, vectors. Some of these are being character-
cines24. The Sindbis and Semliki Forest virus and the protocols for monitoring patients and ized; for example, the adenoviral E3 protein,
is being exploited as a possible cytoplasmic analysing data. The disappointing outcome the herpes simplex ICP47 protein and the
vector25 which does not integrate into the probably lies in the inefficient gene-delivery cytomegalovirus US11 protein30. Systems
nucleus. Although most of these systems system. We need a system in which the vector from other pathogens may also be borrowed
produce the foreign protein only transiently, — containing the ADA gene — is efficiently and incorporated into vectors. In some cases,
they add diversity to the available systems of delivered to progenitor cells, leading to sus- the correcting protein will be sensed as
gene transfer (Table 2). tained expression of high levels of the ADA foreign, eliciting an immune response.
protein. But, encouragingly, despite being Animal models should help us to under-
Clinical trials repeatedly injected with highly concentrated stand this and, where necessary, to develop
Over half of the 200 clinical trials that have recombinant viruses, the patients have strategies for tolerance.
been launched in the United States involve suffered no untoward effects to date. Cell biology is involved because, in many
therapeutic approaches to cancer. Nearly 30 cases, the goal of gene therapy is to correct
of them involve correction of monogenic Future prospects differentiated cells, such as epithelial cells in
diseases (Table 1) such as cystic fibrosis, a1- We now need a renewed emphasis on the cystic fibrosis and lymphoid cells in ADA
antitrypsin deficiency and severe combined basic science behind gene therapy — deficiency. However, because these cells are
immunodeficiency (SCID). Most of the particularly the three intertwined fields of continuously replaced there has to be either
trials are phase I (safety) studies and, by and vectors, immunology and cell biology. continued therapy or an attempt to target the
large, the existing delivery systems have no All of the available viral vectors arose from stem cells. We first need to develop further
major toxicity problems. Moreover, lessons understanding the basic biology of the struc- the technologies for identifying and isolating
can be learned from previous clinical ture and replication of viruses. Clearly, existing these cells, to understand their regulation,
trials26,27. For example, seven years ago two vectors need to be streamlined further (see box and to target infection to them in vivo.
patients were enrolled in a trial to correct on page 241), and vectors that target specific So how far have we come since clinical
deficiencies in adenosine deaminase (ADA, types of cell are being developed. The use of trials began? The promises are still great, and
which leads to severe immunodeficiency). antibody fragments, ligands to cell-specific the problems have been identified (and they
One of the patients improved, and makes receptors, or chemical modifications to the are surmountable). But what of the prospects?
25% of normal ADA in her T cells, five years vector need to be explored systematically. And Our view is that, in the not too distant future,
after the last infusion of infected T cells advances such as the report — published only gene therapy will become as routine a
(although she is still treated with PEG–ADA; last week29 — of the three-dimensional struc- practice as heart transplants are today.
bovine adenosine deaminase mixed with ture of the Env protein from mouse leukaemia Inder M. Verma and Nikunj Somia are in the
polyethylene glycol). But in the other virus (a commonly used vector), will facilitate Laboratory of Genetics, The Salk Institute for
patient, the infected T cells could not the design of targeted vectors. A better under- Biological Studies, La Jolla, California 92037, USA.
produce enough of the deficient enzyme. standing of gene transcription will allow us to 1. Crystal, R. G. Science 270, 404–410 (1995).
The efficacy of gene therapy cannot be design regulatory elements that permit pro- 2. Scriver, C. R. S., Beaudet, A. L., Sly, W. S. & Valle, D. V. The
Metabolic Basis of Inherited Disease (McGraw-Hill, New York,
evaluated until patients are completely taken moter activity to be modulated, and develop-
off alternative treatments (in the above exam- ment of tissue-specific enhancer–promoter 3. Dai, Y., Roman, M., Naviaux, R. K. & Verma, I. M. Proc. Natl
ple, PEG–ADA). In another trial28, weaning a elements should be vigorously pursued. Final- Acad. Sci. USA 89, 10892–10895 (1992).
patient away from PEG–ADA reduced the ly, we need to begin merging some of the quali- 4. Kay, M. A. et al. Science 262, 117–119 (1993).
5. St Louis, D. & Verma, I. M. Proc. Natl Acad. Sci. USA 85,
ability of the T cells to respond invitroto a chal- ties of viral vectors with non-viral vectors, 3150–3154 (1988).
lenge by pathogens. Clearly, in these cases the to generate a safe and efficient gene-delivery 6. Felgner, P. L. Sci. Am. 276, 102–106 (1997).
retroviral vectors were not optimal, and the system. 7. Verma, I. M. Sci. Am. 263, 68–72 (1990).
number of infected blood-progenitor cells was With respect to immunology, viruses still 8. Miller, D. G., Adam, M. A. & Miller, A. D. Mol. Cell. Biol. 10,
4239–4242 (1990).
extremely low. However, these trials did help to have many secrets to be unravelled. Viral 9. Palmer, T. D., Rosman, G. J., Osborne, W. R. & Miller, A. D.
establish the technology for generating clini- systems that have evolved to escape immune Proc. Natl Acad. Sci. USA 88, 1330–1334 (1991).
cal-grade recombinant retroviral particles, the surveillance can be incorporated into viral 10. Lewis, P., Hensel, M. & Emerman, M. EMBO J. 11, 3053–3058
Table 2 Comparison of properties of various vector systems
11. Field, B. N., Knipe, D. M. & Howley, P. M. (eds) Virology
Features Retroviral Lentiviral Adenoviral AAV Naked/ (Lippincott-Raven, Philadelphia, PA, 1996).
lipid–DNA 12. Naldini, L. et al. Science 272, 263–267 (1996).
Maximum 7–7.5 kb 7–7.5 kb ~30 kb 3.5–4.0 kb Unlimited size 13. Miyoshi, H., Takahashi, M., Gage, F. H. & Verma, I. M. Proc.
insert size Natl Acad. Sci. USA 94, 10319–10323 (1997).
Concentrations >108 >108 >1011 >1012 No limitation 14. Blömer, U. et al. J. Virol. 71, 6641–6649 (1997).
(viral particles 15. Yeh, P. & Perricaudet, M. FASEB J. 11, 615–623 (1997).
per ml) 16. Dai, Y. et al. Proc. Natl Acad. Sci. USA 92, 1401–1405 (1995).
Route of gene Ex vivo Ex/In vivo Ex/In vivo Ex/In vivo Ex/In vivo 17. Yang, Y., Ertl, H. C. & Wilson, J. M. Immunity 1, 433–442 (1994).
delivery 18. Engelhardt, J. F., Ye, X., Doranz, B. & Wilson, J. M. Proc. Natl
Acad. Sci. USA 91, 6196–6200 (1994).
Integration Yes Yes No Yes/No Very poor
19. Chen, H. H. et al. Proc. Natl Acad. Sci. USA 94, 1645–1650
Duration of Short Long Short Long Short (1997).
expression 20. Muzyczka, N. Curr. Top. Microbiol. Immunol. 158, 97–127
in vivo (1992).
Stability Good Not tested Good Good Very good 21. Snyder, R. O. et al. Nature Genet. 16, 270–276 (1997).
Ease of Pilot scale up, Not known Easy to scale up Difficult to purify, Easy to scale up 22. Herzog, R. W. et al. Proc. Natl Acad. Sci. USA 94, 5804–5809
preparation up to 20–50 l difficult to (1997).
(scale up) scale up 23. Fink, D. J., DeLuca, N. A., Goins, W. F. & Glorioso, J. C.
Annu. Rev. Neurosci. 19, 265–287 (1996).
Immunological Few Few Extensive Not known None
problems 24. Moss, B. Proc. Natl Acad. Sci. USA 93, 11341–11348 (1996).
25. Berglund, P. et al. Biotechnology (NY) 11, 916–920 (1993).
Pre-existing Unlikely Unlikely, except Yes Yes No 26. Blaese, R. M. et al. Science 270, 475–480 (1995).
host immunity maybe AIDS
27. Bordignon, C. et al. Science 270, 470–475 (1995).
28. Kohn, D. B. et al. Nature Med. 1, 1017–1023 (1995).
Safety Insertional Insertional Inflammatory Inflammatory None 29. Fass, D. et al. Science 277, 1662–1665 (1997).
problems mutagenesis? mutagenesis? response, toxicity response, toxicity 30. Wiertz, E. J. et al. Cell 84, 769–779 (1996)

242 Nature © Macmillan Publishers Ltd 1997 NATURE | VOL 389 | 18 SEPTEMBER 1997
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.