Prospects for Human Gene Therapy

W. French Anderson

Summary. Procedures have now been developed for inserting functional genes into the bone marrow of mice. The most
effective delivery system at present uses retroviral-based vectors to transfer a gene into murine bone marrow cells in
culture. The genetically altered bone marrow is then implanted into recipient animals. These somatic cell gene therapy
techniques are becoming increasingly efficient. Their future application in humans should result in at least partial
correction of a number of genetic disorders. However, the safety of the procedures must still be established by further
animal studies before human clinical trials would be ethical.

Gene therapy the insertion into an organism of a normal gene which then corrects a genetic defect
has been carried out in fruit flies (1) Drosophila melanogaster) and mice (2). How soon gene therapy
might be available for the treatment of human genetic diseases and what criteria should be used in
determining when clinical trials should begin are issues examined in this article. Several investigators
are now preparing protocols for clinical trials of gene therapy in seriously ill patients (3). Since most
of these protocols will be based on the use of retroviral vectors as a delivery system, these structures
will be emphasized. It may well be, however, that one of the other delivery systems described below,
or a new one not yet developed, will be the procedure of choice in the future.

Gene Therapy in Lower Species

The most elegant system thus far demonstrating successful gene therapy is the work in Drosophila
(/). The transposable genetic element, the P factor, has been used to transfer a normal gene coding
for the enzyme that produces the wild-type red eve color in Drosophila embryos which have a
genetically defective gene. The result is that the treated flies acquire normal eye color. Similar
transfer experiments under way use other genes. Despite considerable searching, transposon-like
elements have not been identified in vertebrates. Retroviruses, however, are structurally and
functionally similar in many ways to the mobile genetic elements found in lower organisms, and
retroviral vectors have now been used to transfer functioning genes into mouse bone marrow cells.

The first genetic "cure" reported in a mammal (2) was in a strain of mice, called little. These have a
mutation that results in reduced serum levels of growth hormone, and the mice are therefore dwarfs.
The equivalent human disease is pituitary dwarfism. Hammer et al. (2) succeeded in inserting a rat
growth hormone gene into the cells of these mice in such a way that the gene is expressed at a high
level. The deficiency in growth hormone was corrected and the animals grew rapidly but the gene
was not controlled appropriately and gigantism resulted namely a mouse one-and-a-half times as
large as a normal animal. A major research effort is focusing on how to correctly regulate
transferred genes.

Gene Therapy in Humans

Human disease candidates for gene therapy. Pituitary dwarfism in humans is not a reasonable initial
candidate. Genes making hormones that circulate in the bloodstream are probably not appropriate for
early attempts at gene therapy in humans. First the normal feedback controls in DNA that regulate
the expression of hormone genes in the body are not now known. Therefore, physiologically correct
levels of hormone production would probably not be possible. Second, it would be easier and safer to
use recombinant DNA manufacturing techniques to produce sufficiently large quantities of hormone
so that the active polypeptide itself could be given to the patient. Hormone levels could then be
titrated precisely.
At first, clinical investigators thought that the human genetic diseases most likely to be the initial
ones successfully treated by gene therapy would be the hemoglobin abnormalities (specifically, B-
thalassemia) because these disorders are the most obvious ones carried by blood cells, and bone
marrow is the easiest tissue to manipulate in vitro (4). Regulation of globin synthesis, however, is
unusually complicated. Not only are the embryonic, fetal and adult globin chains carefully regulated
during development, but also the a- and B-globin-like chains are always maintained in a 1 to 1 ratio
despite the fact that the a- and B- globin loci are on different chromosomes. To understand the
regulatory signals that control such a complicated system and to develop means for obtaining
controlled expression of an exogenous B-globin gene will take considerably more research effort.
The recent development of a mouse model for B-thalassemia should aid these investigations (5).

Gene therapy should be beneficial primarily for the replacement of a defective or missing enzyme or
protein that must function inside the cell that makes it, or of a deficient circulating protein whose
level does not need to be exactly regulated (for example factor VIII). Early at- tempts at gene
therapy will almost certainly be done with genes for enzymes that have a simple "always-on" type of
regulation. Three genes are the initial prime candidates: hypoxanthine-guanine phosphoribosyl
transferase (HPRT), the absence of which results in Lesch-Nyhan disease: purine nucleoside
phosphorylase (PNP), the absence of which results in a severe immunodeficiency disease: and
adenosine deaminase (ADA), the absence of which results in severe combined immunodeficiency
disease. For all three the clinical syndrome is profoundly debilitating. The defect in each is found in
the patient's bone marrow (although the severe central nervous system manifestations of Lesch-
Nyhan disease are due to absence of HPRT in brain cells and probably cannot be corrected with
current techniques). In all three there is no or minimal detectable enzyme in marrow cells from
patients homozygous (or hemizygous) for the defect and the production of a small fraction of the
normal enzyme level should be beneficial. Furthermore a mild over production of enzyme should
not be harmful to the cell. In addition in all three the gene has been cloned and a complementary
DNA is available.

Since combined immunodeficiency due to a defect in the ADA gene in Tlymphocytes can be
corrected by infusion of normal bone marrow cells from a histocompatible donor selective replication
of the normal T cells appears to take place (6). This observation offers hope that defective bone
marrow can be removed from a patient, the normal ADA gene inserted into a number of cells through
gene therapy and the treated marrow reimplanted into the patient where it may have a selective
growth advantage. There is also evidence that marrow cells containing HPRT (HPRT-) may have a
selective advantage (in both mice and humans) over cells that do not (HPRT-) (7). If selective
growth occurs no ablation of the patient's own marrow would be necessary. If however, corrected
stem cells have no growth advantage over endogenous ones, then partial or complete marrow
destruction (either by irradiation or by other means) may be required in order to allow the corrected
marrow cells an environment favorable for expansion.

Ethics. The ethics of gene therapy in humans has been discussed for many years (8) and is being
widely debated at present (9). Essentially all observers have stated that they believe that it would be
ethical to insert genetic material into a human being for the sole purpose of medically correcting a
severe genetic defect in that patient that is somatic cell gene therapy. Attempts to correct germ cells
(that is to permit the new gene to be passed on to the patient's children) or to enhance or improve a “
normal" person by gene manipulation do not have societal acceptance at this time (9). However,
somatic cell gene therapy for a patient suffering a serious genetic disorder would be ethically
acceptable if carried out under the same strict criteria that cover other new and experimental medical
procedures (10). The techniques that are now being developed for human application are for somatic
cell, not germ line gene therapy.

The question examined here is: What criteria should be used in evaluating gene therapy protocols?
Three general requirements first presented in 1980 (10) are that it should be shown in animal studies
that (I) the new gene can be put into the correct target cells and will remain there long enough to be
effective (II) the new gene will be expressed in the cell at an appropriate level and (III) the new gene
will not harm the cell or by extension, the animal. These three requisites summarized as delivery
expression and safety will each be examined in turn.

Delivery

At present the only human tissues that can be used for gene transfer are bone marrow and skin cells.
No other cells can be extracted from the body grown in culture to allow manipulation and then
successfully reimplanted into the patient from whom the tissue was taken. In the future as more is
learned on how to package the injected DNA and to make it tissue or even cell type specific the
intravenous route would be the simplest and most desirable. Attempting to give a foreign gene by
injection directly into the bloodstream is not advisable with our present state of knowledge, since the
procedure would be enormously inefficient and there would be little control over the DNA's fate
(11).

Studies are considerably more advanced with bone marrow than skin cells as a recipient tissue for
gene transfer. Bone marrow consists of a heterogeneous population of cells, most of which are
committed to differentiation into erythrocytes, lymphocytes, megakaryocytes and so on. Only a small
proportion (0. 1 to 0.5 percent) of nucleated bone marrow cells are stem cells (that is cells that have
not yet differentiated into specific cell types and which divide as needed to maintain the marrow
population. In gene therapy, stem cells would be the primary target. Because they are low in number
and are not recognizable a delivery system for transferring a gene into stem cells must be efficient.

Techniques for transferring cloned genes into cells can be grouped in four categories: (i) viral both
RNA viruses (or retroviruses) and DNA viruses (for example SV40 adenovirus and bovine
papilloma); (ii) chemical such as calcium phosphate-mediated DNA uptake. (iii) fusion, that is
fusion of DNA-loaded membranous vesicles, such as liposomes, red blood cell ghosts or proto- plasts
to cells: and (iv) physical, that is, microinjection or electroporation. Each technique is valuable for
certain types of experiments but none can yet be used to insert a gene into a specific chromosomal
site in a target cell. Fusion techniques are the least well characterized and will not be discussed. As
noted retroviral-based vectors appear to be the most promising approach at present for use in
humans.

Viral Techniques

RNA viruses (retroviruses). There are a number of advantages of vectors derived from retroviruses as
a gene delivery system. First up to 100 Percent of cells can be infected and can express the integrated
viral (and exogenous) genes: this is in contrast to chemical methods where although most cells take
in the DNA as shown by positive assays after 48 hours only one cell in 10^3 to 10^7 stably expresses
the exogenous gene. Second as many cells as desired can be infected simultaneously: 10^6 to 10^7
is a convenient number for a Simple protocol. Third under appropriate conditions the DNA can
integrate as a single copy at a single albeit random site, whereas the chemical and physical
techniques often result in the insertion of multiple copies of the transferred gene all linked head-to-
tail in tandem repeats. Fourth although integration is random with respect to the host genome it is
precise with respect to the viral genome-that is the structure of the integrated DNA is known. Fifth
the infection and long-term harboring of the retroviral vector usually does not harm cells. Finally a
wide and controllable host range is available. A number of retroviral vector systems have been
developed. Here we concentrate on vectors based on Maloney murine leukemia virus (MoMLV).

1) Life cycle and structure. The details of the life cycle of retroviruses have been reviewed recently
(12). In brief the retrovirus composed of an RNA-protein core and a glycoprotein envelope enters a
cell where the RNA acts as a template for the reverse transcription of the genetic information into a
double strand of DNA.. This DNA can precisely integrate as a single copy called a provirus at a
random location in the genome of the host.

Although much has been learned about the regulatory features of retroviruses uncertainties remain.
Those features of the proviral structure that are thought to be necessary for transcription and
transmission of the viral genome are (see Fig. 1): a long terminal repeat (LTR) sequence on each end
containing regulatory signals for initiating and terminating transcription sequences required for
reverse transcription and others for PF(viral integration: short sequences (called here for short. r - and
r ' ) immediately adjacent to each LTR and necessary for reverse transcription: the packaging
sequence called / in MoMLV necessary for the viral RNA to be packaged into an infectious viral
particle and the donor (D) and acceptor (A) splice sites.

Retroviral RNA is synthesized from the proviral DNA by the host cell's own RNA polymerase. A
portion of this RNA is used in the cell's translational machinery to synthesize the viral proteins that
go into the final viral particles along with the genomic RNA. These viral particles bud off from the
cell and can then infect other cells.

From experimental studies as well as the existence of a number of naturally occurring defective
viruses, it is known that almost all of the regions coding for viral proteins (gag, pol and env in Fig.
1) can be deleted and some or all of these sequences replaced with other DNA. Once the viral genes
are deleted the retroviral vector becomes defective. In order to obtain infectious viral particles a cell
harboring a defective provirus must be infected with a "helper" virus which carries all the viral
functions needed that is, the genes for gag, pol. and env.

Fig. 1. Simplified structure of MoMLV retroviral provirus DNA. Abbreviations: E. enhancer: P. promoter: 1. initiation
(Cap) site for viral RNA synthesis : ( r): replication initiation site for minus DNA strand (transfer RNA binding site): D.
donor splice site: ! packaging sequence: A,. the major acceptor splice site: r-. replication initiation site for plus DNA
strand (purine-rich site): T. terminal [poly(A) addition] site for viral RNA synthesis: LTR. long terminal repeat: U3. R.
and U5 are portions of the LTR: gag. group-specific (that is viral core) antigens: pl5. pl2. p30. and p10: pol RNA-
dependent DNA polymerase (reverse transcriptase): and env. envelope proteins: gp70. pl5E and R. (Not drawn to scale)
2) Use as gene delivery system. The proviral DNA for the desired retrovirus [commonly either
MoMLV or murine sarcoma virus (MSV)L is isolated and inserted into a convenient plasmid. The
viral genes can then be replaced with the exogenous genes of choice by standard recombinant DNA
techniques. This construct is used to transfect tissue culture cells (for example NIH 3T3 cells) by a
convenient gene transfer procedure (for example. calcium phosphate). After infecting the cells with a
helper virus (such as intact MoMLV) infectious viral particles, possessing both the retroviral vector
and the helper virus bud off from the cells into the surrounding medium. This particle-containing
supernatant is collected and used to infect bone marrow cells in culture or more simply freshly
extracted bone marrow is incubated directly with the cells budding the viral particles. The marrow
cells are removed and injected intravenously into a mouse whose bone marrow has been killed by x-
rays (lethally irradiated). The animal is then studied to determine if the transferred marrow cells
express the desired gene from the vector.

3) Successful gene transfer into adult mice. Joyner et al. (13) have successfully used this procedure
to transfer a functional gene for neomycin resistance (neo') into mouse hematopoietic progenitor cells
by use of a MoMLV retroviral vector. The presence and expression of this gene in granulocytic
progenitor cells rendered these cells resistant to the neomycin-like antibiotic
G418 as determined by in vitro colony assays. Treated cells were injected into lethally irradiated
mice: Southern blot analysis, and colony assays showed that the neo' gene is present and functional
in the spleens of the recipient animals (13).

An improvement on this procedure would be to treat bone marrow cells with a retroviral particle that
could deliver the vector but which would not itself produce a spreading infection. Mann et al. (14)
have developed a technique for accomplishing this goal. The regulatory signal ! (Fig. 1) contains a
sequence, the exact size and structure of which are not yet known (15). That must be present in the
viral RNA for it to be packaged into a viral particle. A helper virus was
constructed with this sequence deleted (!-) by making use of convenient restriction endonuclease
sites (Bal I and Pst I) flanking the ! sequence in MoMLV. The !- helper is able to produce all the
viral proteins required to make a particle but the particle does not package its own RNA. Since the
retroviral vector has a ! sequence it is packaged. Consequently, the particle can just infect once: it is
only a delivery system for the vector, not an infectious agent.

In order to use the !- helper virus conveniently, a line of NIH 3T3 cells was established with the
helper provirat DNA permanently integrated (14): !- helper viral RNA is produced constitutively.
The transfection of this cell line (called !-2) with the retroviral vector DNA results 48 hours later in a
supernatant that contains viral particles with only the vector.

Williams et al. (16) have used the !-2 cell line to place a functioning neo' gene into the hematopoietic
cells of adult mice. Freshly extracted murine bone marrow was layered onto !-2 cells producing a
retroviral vector called MSV DHFR-NEO, which contains genes for dihydrofolate reductase (DHFR)
and neo’ in an MSV backbone. The marrow and !-2 cells were incubated for 48 hours under standard
incubation conditions: similar results were obtained when bone marrow cells with the !-2 MSV
DHFR- NEO cell layer were incubated for 6days under Dexter-type conditions (17). The viral
particles that budded off into the supernatant contained the MSV DHFR. N EO vector but to the
extent that could be determined. no !- helper viral RNA. Ten days after lethally irradiated mice were
injected with the treated bone marrow cells analysis of the regenerating spleens showed that the mice
carried the MSV DHFR-NEO proviral DNA in their hematopoietic cells. Individual spleen colonies
each arising from a single stem cell were generated by injecting an estimated one to ten stem cells
into another group of lethally irradiated mice. Cells from individual colonies were able to produce
spleen colonies in a secondary group of lethally irradiated animals. These mice also were shown to
carry MSV DHFR-NEO DNA in their total spleen DNA and in each case to have the same
integration site restriction pattern as the colony from the primary mouse. These data show that the
delivery system is effective at least for mouse bone marrow cells. Preliminary evidence indicates that
the neo' gene is expressed (16).

Southern blot analysis of total spleen DNA with a number of restriction enzymes revealed in some
cases a small number of proviral integration sites. This result suggests that only a few infected stem
cells were proliferating to repopulate the irradiated spleen. Secondary transfers of individual colonies
showed that only 7 of 48 colonies (15 percent) contained the neo’ gene. This is a lower limit since an
occasional colony might have been formed from endogenous stem cells that survived the irradiation.
These data suggest that the present bone marrow procedure might still be made more efficient as a
delivery system.

A similar retroviral vector system based primarily on MoMLV has been developed by Verma and his
co-workers (18). In their ! - helper virus they substituted an amphotrophic ( that is wide host range)
env gene for the MoMLV env gene which produces a particle coat with a narrow host range. This
helper viral construct is called pSAM. Miller et al. (19) built a retroviral vector containing a full-
length complementary DNA for the human enzyme HPRT. This vector called PLPL was
cotransfected along with the !- helper. pSAM. into HPRT- BALB/3T3 cells. One clone (c7cl) was
obtained, that produced high levels of viral particles containing the HPRT
vector. Injection of these cells into lethally irradiated mice resulted in animals that continued to
produce HPRT-vector particles for at least 6 months (19). The infectious particles resulted from the
presence of low levels (<0.l percent of the HPRT-vector virus) of packageable helper virus along
with the injection of MoMLV as additional helper which led to multiple rounds of replication in the
host. In addition human HPRT enzyme was detected in spleen cells.

4) Shortcomings of retroviral delivery systems. The evidence indicates that retroviruses can be used
as a reasonably efficient delivery system. A gene therapy procedure however, also requires a reliable
system. In most of the work reported to date a number of cells are found to contain altered proviral
DNA. The biggest problem appears to be that retroviruses have a strong propensity for deleting
sequences during virus replication (19a). Many vectors have been ineffective because the foreign
DNA is partially or totally removed from the construct or is rearranged. For example Joyner and
Bernstein (20) have used the Friend spleen focus-forming virus as a potential vector system for
hematopoietic cells. Constructs containing a thymidine ki- nase (TK) gene in the gag region and an
intact env gene (gp55) were used along with MoMLV as helper to obtain viral particles. The particles
were injected into lethally irradiated mice and also layered onto rat TK- (LTA) cells. Southern blot
analysis of the integrated proviral DNA in erythroleukemic spleens demonstrated vector
constructs with intact gp55 genes but deleted TK sequences, whereas TK - LTA clones possessed
intact TK genes but deleted or rearranged gp55 sequences . In other words in no case could a
provirus be found that still contained both the TK and gp55 genes. Even the successful MSV DHFR-
NEO vector which produces neo’ expression in mice has lost a portion of its DHFR gene during
production of the viral particles (16) Several approaches are being tried to circumvent this problem of
instability (19a).

5) Properties still needed for an optimal delivery system. An ideal delivery system not only would be
stable but also would be tissue-specific. When a genetic disorder is in the hematopoietic system, then
the isolated bone marrow can be treated. But no other tissue, except skin cells can be removed,
treated and replaced at present. Since many viruses are known to infect only specific tissues (that is
to bind to receptors that are present only on certain cell types) a retroviral particle containing a coat
glycoprotein that recognizes only human hematopoietic stem cells would permit the retroviral vector
to be given intravenously with little danger that cells other than those in the marrow would be
infected. Such specificity could permit the liver and brain for example to be treated individually. In
addition, the danger of inadvertently infecting germ cells could be eliminated. One problem however,
is that cell replication appears to be necessary for integration. It would not be possible to infect
nondividing brain cells for example as far as we now know.

The optimal system not only would deliver the vector specifically into the cell type of choice but
would also direct the vector to a predetermined chromosomal site. Specific insertion into a selected
site of a chromosome by means of homologous recombination can be readily achieved in lower
organisms but appears to be a formidable task in mammals, whether retroviral vectors or plasmid-
based vectors are used. Present evidence suggests that homologous site- specific integration occurs at
a very low level. when it occurs at all in mammals (21).

DNA viruses. Viruses such as SV40 with DNA as the nucleic acid in their core have been employed
for several years as gene transfer vectors (22). A conditionally nonreplicating adenoviral vector has
recently been developed that will efficiently infect animal and human cells (including hematopoietic
cells) with the result that one or a few copies of the recombinant virus are integrated into the host
cell's genome (23). Whether adenoviral vectors will be as efficient as retroviral vectors or will offer
other advantages as a gene transfer delivery system remains to be determined. One subcategory of
DNA viruses should be mentioned: bovine papilloma virus (BPV) (24). This viral DNA replicates
extra-chromosomally so that BPV-based vectors may prove to be useful for maintaining genes in
cells in a nonintegrated manner. Transfection of hematopoietic cells with BPV-vectors has not yet
been reported.

Chemical Techniques

The Other Procedure under active consideration for insertion of genes into human bone marrow cells
is calcium phosphate-mediated DNA uptake. The original procedure of Graham and van der Eb (25)
was modified by Wigler et al. (26) in order to insert into the genome of mammalian cells growing in
culture a fragment of DNA carrying one or more genes. A number of genes have been used including
the herpes simplex TK gene complementing TK- cells the DHFR gene protecting against the drug
methotrexate and the neo' gene protecting against the antibiotic G418.

Procedure. Transfection is carried out by pipetting a suspension of DNA complexed into small
precipitates with calcium phosphate onto a monolayer of cells growing in a tissue culture dish (26). A
number of techniques are used to increase the efficiency of transfection in different cell types: for
example diethylaminoethyl dextran can be employed instead of calcium phosphate or the cells can be
shocked with glycerol after 2 hours of incubation (27). The efficiency of the process varies with the
cell line. Under optimal conditions and very receptive cells (for example mouse L cells) one cell in
10^2 to I0^3 can be obtained that has integrated and expressed the exogenous DNA. Because the
usual efficiency is 10^-5 to 10^-7 a procedure is required to detect the occasional transfected cell. In
other words a gene must be present that can protect the cell from a lethal selective agent that is added
to the incubation medium or that complements a genetic defect (HPRT or TK. for example). The
transfected cell will survive while all others are killed. Attempts to obtain transfected cells without
selective pressure have generally been unsuccessful.
Transfection appears to work poorly in suspension cells namely bone marrow cells. Efficiencies can
only be estimated but the value is probably one cell in 10^6 or 10^7. Using the powerful selection
system offered by the mutant DHFR gene (isolated from 3T6-R400 cells) that provides exceptional
resistance to methotrexate. Carr et al. (28) reported that the calcium phosphate transfer technique can
be successfully employed to obtain mouse bone marrow cells that contain a functional exogenous
DHFR gene. The permanently transfected cells can partially repopulate a lethally irradiated mouse.
These results support the studies of Cline et al (29) who reported successful transfer of a functional
DHFR gene into the bone marrow of mice. However the presence of the DHFR gene has not been
confirmed with DNA hybridization studies and. until such experiments are reported. the efficiency of
the calcium phosphate procedure is uncertain.

Shortcomings of chemical techniques. If a chemical technique for gene transfer were used in a
protocol designed for humans the predicted results appear discouraging. Recovery from bone marrow
of approximately 10^10 nucleated cells (of which 10^7 to 10^8 are stem cells) can routinely be
obtained from patients for marrow transplantation. Efficiency of 1 to 10^6 would mean that only 10
to 100 stem cells would be transfected. Reinsertion of these cells into the total stem cell pool of 10^8
to 10^9 cells would be very unlikely to have any noticeable effect on a patient's course unless there
was an extraordinary selective advantage for the treated cells. Any human gene therapy protocol that
uses chemical means for transfection would have to establish therefore that either a few transfected
stem cells might have a detectable beneficial effect on the patient’s course or that the investigator has
improved substantially the efficiency of the procedure for human bone marrow cells.
.

Physical Techniques

Microinjection (30) and electroporation (31) are the two principal classes of physical techniques.
Electroporation a relatively new technique is the transport of DNA directly across a cell membrane
by means of an electric current. It has been used to transfer a variety of genes into a number of
different cells including the immunoglobulin K gene into B cells (31). Its potential for human gene
therapy is uncertain.

Microinjection has been used for a number of years and has the advantage of high efficiency (up to
one cell in five injected can be permanently transfected). However, the distinct disadvantage is that
only one cell at a time can be injected. Transfection of a large number of hematopoietic stem cells is
not feasible. Even if a stem cell could be recognized it would have to be fixed to a slide for injecting.
The effect of attaching, injecting and subsequent detaching is unknown. Microinjection of mouse
erythroleukemia (MEL) cells is difficult, although possible (32) and these cells are much easier to
manipulate in culture than are bone marrow cells.

Transfer of genes into mouse eggs. An area where microinjection has had spectacular success is in
transferring genes into fertilized mouse eggs (33) Gordon et al. (34) first demonstrated that if
plasmid DNA is microinjected into one of the two pronucler of a recently fertilized mouse egg and
the ovum is then placed into the oviduct of a pseudopregnant female, the egg could develop into a
normal mouse carrying the plasmid DNA in every cell of its body. Furthermore the injected DNA
can be transmitted to offspring in a normal Mendelian manner. Mice carrying an erogenous gene in
their genome are called "transgenic."
Hammer et al. (2) used this technique to partially correct a mouse with a defect in its growth
hormone production. By attaching a rat growth hormone gene to an active regulatory sequence
(specifically the promoter that normally directs the synthesis of metallothionein messenger RNA in
mice) they obtained a recombinant DNA construct that actively produces growth hormone in the
genetically defective mouse. Although the level of growth hormone production is inappropriately
controlled-that is influenced by signals that normally regulate metallothionein synthesis-these
experiments do show that microinjection can be used as a delivery system that can put a gene into
every cell of an animal's body.

Nonapplicability for humans. Should the technique of microinjecting a fertilized egg be employed
for human gene therapy at the present time? The answer is no on three grounds: the procedure has a
high failure rate, can produce a deleterious result and would have limited usefulness. Microinjection
has a high failure rate because the majority of eggs are damaged by the microinjection and transfer
procedures so that they do not develop into live offspring. In one recent experiment involving
microinjection of an immunoglobulin gene (35). 300 eggs were injected. 192 (64 percent) were
judged sufficiently healthy to be transferred to surrogate mothers only 11 (3.7 percent) proceeded to
live birth and 6 (2 percent) carried the gene. These results are from a highly experienced laboratory
in which thousands of identical eggs from the same hybrid cross of inbred mice have been injected
over a number of years. The mice were chosen precisely because they gave the best results for gene
transfer by microinjection. Experience with attempts to microinject growth hormone genes into
livestock eggs have met with a number of major biological and technical problems (36). Successful
gene transfer by microinjection of human eggs without a long period of trial and error
experimentation is extremely unlikely.

Second . Microinjection of eggs can produce deleterious results because there is no control over
where the injected DNA will integrate in the genome. Lacy et al. (37) showed that the integration of
an exogenous rabbit B-globin gene in transgenic mice could sometimes occur into a chromosomal
location that results in expression of the B-globin gene in inappropriate tissue, namely muscle or
testes. There have been a number of cases reported where integration of microinjected DNA has
resulted in a pathological condition (38). Although there is no control over where exogenous DNA
will integrate in any gene transfer procedure the damaging effect caused by a harmful insertion site
could be great when it occurs in the egg but may be negligible when it occurs in one or a few of a
large number of bone marrow cells.

Third is limited usefulness. Not only is it of questionable ethics to experiment on human eggs
because of the expected losses but even if “success" were obtained it would be applicable primarily
when both parents are homozygous for the defect. When the parents are both carriers only one
fertilized egg out of four would result in an affected child (39). Since a homozygous defect cannot
yet be recognized in an ovum and since the procedure itself carries such a high risk it would be
improper to attempt any manipulation in this situation. Further more, most of the very serious genetic
disorders result in infertility (or death before reproductive age) in homozygous patients.
Consequently there would be little use for the procedure even if it were available. A different
approach for human gene therapy is required.

Expression

The second criterion for evaluating a human gene therapy protocol is that there is appropriate
expression of the new gene in the target cells. Even when a delivery system can transport an
exogenous gene into the DNA of the correct cells of an organism it has been a major problem to get
the integrated DNA to function. A vast array of cloned genes have been introduced into a wide range
of cells by the several gene transfer techniques discussed above. "Normal'' expression of exogenous
genes is the exception rather than the rule.

Active exogenous promoters in tansgenic mice. Microinjection of fertilized eggs with exogenous
DNA to obtain transgenic mice carrying an expressing gene has resulted in several spectacular
successes but also in a considerable number of unpublished failures. Thus far only four genomic
promoters have been reported to show significant activity: metallothionein (2). transferrin (33).
immunoglobulin (35) and elastase (40). However essentially any complementary DNA can be
attached to an active promoter such as metallothionein and the coding sequence will usually be
expressed in a transgenic mouse under the control of that promoter.

Why are most promoters inactive after microinjection into mouse oocytcs? Al least one promoter has
been examined in this regard: mouse B^ma) globin. The sequences are found to be heavily
methylated in mouse tissues where they are inactive but relatively unmethylated in tissue culture
cells where they are active (41). Therefore, the mouse zygote appears to respond to this foreign
DNA by covering it with methyl groups, which remain on the DNA throughout the life- time of the
animal. Attempts to decrease the methylation of the genomic DNA by treating adult mice carrying an
exogenous B-globin promoter with the hypomethylating drug 5-azacytidine have been essentially
unsuccessful (41). The metallothionein promoter, however, even if methylated can remain active
(42). Why some promoters are inactivated by methylation or other mechanisms while others are not
is not known.

Expression from retroviral vectors. If a retroviral vector is used for gene transfer the transcriptional
signals in the retrovirus's own LTR's can be used (Fig. I). Expression of exogenous genes carried by
retroviral vectors into bone marrow cells has been reported by three laboratories. The two studies in
which a neo' gene was expressed in mouse bone marrow were described above (13. 16). The most
extensive data, however, are from Willis et al. (43). A homozygous Lesch-Nyhan (LN) lymphoblast
cell line was used to determine whether an HPRT- human hematopoietic cell could be corrected by a
retroviral vector containing a functional HPRT gene. The LN cells have all the characteristics of a
cell line totally defective in HPRT, specifically a disruption in their inosinate cycle that leads to a
high purine production and a number of other metabolic abnormalities (44). LN cells infected with
viral particles containing the HPRT vector could be rescued in selective medium. Seventeen HPRT-
clones were isolated and studied. These cell lines had HPRT levels ranging from 4 to 23 percent of
the normal level and the abnormalities associated with a deranged inosinate cycle were partially to
nearly completely corrected (43). In a corollary study viral particles containing the HPRT -vector ere
used to infect mouse bone marrow cells that were then injected into lethally irradiated mice (19) Both
human HPRT proteins and chronic production of HPRT-vector particles were detected in the
hematopoietic tissue of the mice.

A problem must still be overcome, however. Even though expression of HPRT and neo' genes has
been obtained in the hematopoietic tissue of irradiated mice, the efficiency of the combined
delivery-expression system is poor. If 15 percent of stem cells can be infected and if 4 to 23 percent
of normal expression can be obtained in them, can sufficient enzyme be synthesized to be of benefit
to a patient? The issue, once again is whether or not the treated cells will have a selective growth
advantage in the patient's body. If they do not, then, either the patient’s own bone marrow must be
partially or totally eliminated before re-implantation of the treated cells or the gene therapy protocol
must demonstrate at least some expression in nonirradiated animals. It must be recognized, however,
that in the absence of a true animal model for a given genetic disease, it might be difficult or
impossible to demonstrate selective growth advantage except in human patients.

Use of enhancers to increase expression. How can the level of expression be increased and properly
regulated? One key element may be the enhancers. These are DNA sequences usually 50 to 150 base
pairs in length that increase the expression of the adjacent gene 10 to 1000 times (45). A retrovirus
has its own enhancer immediately upstream from its promoter in the LTR (Fig. 1). Enhancers are
known to be species-specific (46). A primate enhancer (for example, the 72 base pair repeat from
SV40) is several times more active in primate tissue culture cells than in rodent cells. Likewise, a
mouse enhancer (for example, the 73 base pair repeat from MSV) is more active in rodent cells than
in primate cells. The promoter acted upon does not influence the species specificity (a mouse B-
globin promoter and a primate SV40 promoter are both activated more by a primate enhancer in
primate cells than in rodent cells), although different promoters can be enhanced to different extents
(47). Retroviral vectors designed for therapeutic application in humans may need primate, or even
human, enhancing sequences rather than the mouse ones that are now used.

Some enhancers may even be tissue-specific (48). With a tissue-specific enhancer it may not be
necessary to develop a cell-specific delivery system. The DNA could be integrated into all cells
but only be expressed significantly in that tissue in which the enhancer is active. Even more precision
may be achieved if one could place a tissue- specific coat on a retroviral particle that would direct the
virus into the target cell, along with a tissue-specific (and possibly even a
developmental-time-period-specific) enhancer in the construct itself.

Systems like globin undoubtedly have other regulatory regions in addition to enhancers which
recognize cellular factors that are involved in control. Much information still needs to be learned
about the regulatory signals in these multigene families.

Expression from plasmid-based expression vectors. If a chemical gene transfer technique is used as a
delivery system, then the gene must be inserted into an appropriate expression vector. An expression
vector is a plasmid (usually pBR322) in which the complementary DNA (or genomic gene) of
interest is inserted together with regulatory signals. A typical expression vector would be composed
of a promoter (for example, from the mouse metallothionein gene), the complementary DNA of
choice, a splice site and polyadenylation site (necessary for correct processing of the transcribed
RNA), and an enhancer.

Plasmid-based expression vectors containing an enhancer have not yet been used to transfect bone
marrow cells. Therefore, how effective expression might be is unknown. The inefficiency of the
presently available delivery systems for these vectors was discussed above.

One additional complication is that calcium phosphate-directed transfection, as well as

microinjection does not usually result in the integration of a single copy of the expression vector. The
plasmid DNA vector appears to be ligated or replicated, or both, inside the cell to form a long
head-to-tail structure called a tandem repeat (49). This tandem repeat, which can be a few or up to
hundreds of copies in length, is randomly inserted usually in one site in the genome. The tandem
repeats may produce problems for genes requiring intricate regulation because of the uncertainty as
to how many of the copies are active.

Regulation by genomic control signals. Can either plasmid-based expression vectors or retroviral
vectors be used to transfer genes that are controlled by the gene's own genomic regulatory
sequences? Plasmid-based expression vectors in transgenic mice do respond to normal physiological
control signals in some cases. Metallothionein-promoted genes express primarily in the liver, the
normal tissue for metallothionein synthesis, and can be induced by cadmium, as occurs in vivo for
the endogenous gene: however, they do not respond to steroids, which are another physiologic
inducer in vivo (50). An immunoglobulin gene is expressed in the spleen, the correct in vivo tissue,
and not in liver (35). A mouse-human B-globin fusion gene expresses in hematropoietic tissue (51).

In tissue culture cells, a number of plasmid-based expression vectors have demonstrated at least a
degree of normal regulation. For example, the human B-globin gene with approximately 1 kilobase
of genomic 5’ flanking sequence can be induced (along with endogenous mouse globin) in a
transfected MEL cell (52). The level of expression is not as high as that of the normal endogenous
B-globin gene, suggesting that other regulatory signals are needed. However, transfection of MEL
cells with cosmids carrying 30 to 40 kilobases of human genomic DNA containing the human
B-globin gene does not result in higher expression of human B-globin messenger RNA (53).

Miller et al. (54) obtained encouraging results when they placed a rat growth hormone
complementary DNA together with 237 bases of genomic 5' flanking sequence into the env, region of
the HPRT-vector already described above. This growth hormone gene was regulated in rescued
HPRT- fibroblast cells by its own genomic promoter and regulatory sequences as shown by (i)
stimulation by glucocorticoid and thyroid hormones, which are normal in vivo regulators. and (ii)
equal activity whether the fragment was placed in the same direction or opposite to the vector's
LTR’s (54). Expression of the vector in an animal has not yet been studied.

These data provide hope that vectors can be built with all the genomic regulatory signals necessary to
produce correctly controlled expression in target cells. In the future, one might use only selected
portions of a retrovirus in order to construct a delivery and integration system that would place one
copy of the vector DNA into the target cell's genome. Expression would be controlled by the
exogenous gene’s own genomic regulatory signals. One possible problem is size: it appears that
MoMLV constructs must not be over 9 to 12 kilobases in order to be packaged. Since 2 or 3
kilobases are necessary for essential function, only 6 to 9 kilobases are available for insert. This
amount may be adequate, but further studies are needed to determine the answer (55).

Importance of chromosomal location. A major question that remains is. How important is
chromosome location? Integration of a proviral structure can in some cases activate a downstream
gene, as can occur with oncogenes. This problem could be eliminated by deleting the enhancer and
promoter regions from the 3' (right-hand) LTR in the retroviral vector. One round of reverse
transcription could then occur which would result in double-stranded retroviral DNA with both
LTR's defective. The retroviral vector DNA would then integrate with no transcription initiation
signals. Therefore, expression would have to be controlled by exogenous signals in the inserted gene,
and no downstream activation of other genes could take place.

Certainly an integration site that disrupts an important gene or regulatory sequence would normally
be detrimental. How often this would occur must be determined by experiment. It is probable though
that in most such cases the insertional event would diminish the fitness of the recipient cell so that it
would be outgrown by normal cells.

Are there only certain active chromatin regions that can allow expression of a gene? Or could an
expression vector take its own -active domain- with it so that essentially any location would be
acceptable? The answers to these questions are still not known.
Safety

The third and final criterion for evaluating a human gene therapy protocol is that the
delivery-expression system be safe.

Retroviral vectors. Although retroviruses have many advantages for gene transfer, they also have
disadvantages. One problem is that they can rearrange their own structure as well as exchange
sequences with other retroviruses. In the future it might be possible to modify retroviral vectors in
such a way that they become less unstable. At present, however, there is the possibility that a
retroviral vector might recombine with an endogenous viral sequence (56) to produce packageable,
infectious recombinant virus. Properties that such a recombinant would have are unknown, but the
potential homology between retroviral vectors and as-yet unknown primate cancer retroviruses or
human T-cell leukemia viruses might be sufficiently close so that possible recombinants should be
sought. There is, however, a built-in safety feature with the mouse retroviral vectors now in use.
These mouse structures have a very different sequence from known primate retroviruses, and there
appears to be little or no homology between the two (57). Therefore, a potentially "safe” proviral
vector construct might be one composed of mouse LTR's with their enhancer and promoter regions
deleted and a human gene controlled by the appropriate human genomic regulatory signals.

With the present constructs, three types of experiments ought to be carried out before any
retrovirus-treated bone marrow is injected into a patient. These protocols, designed to test the safety
of the delivery-expression system are necessary since once treated bone marrow is reinserted into a
patient, it and all retroviruses that it contains are irretrievable.

First studies in vitro with human bone marrow are needed. Marrow cultures infected with the
therapeutic vector should be tested for a period of time for the production of recombinant viruses.
Any infectious virus isolated should be studied for possible pathogenicity.

Second studies in vivo with mice are needed. Since many retroviral vectors are constructed from
mouse retroviruses, and expression studied in mouse bone marrow transplanted into lethally
irradiated (or nonirradiated) mice, these animals should be followed to determine if genomic
rearrangement or the site of chromosomal integration has resulted in any pathologic manifestations or
the production of any infectious viruses.

Third studies in vivo with primates are needed. A protocol similar to the one planned for human
application should be carried out in primates, not just mice, because the endogenous proviral
sequences in primate, including human. DNA are different from those in mouse DNA. Therefore, the
nature of any viral recombinants would be different. Treated bone marrow should be reimplanted into
primates, the successful transfer of intact vector DNA into hematopoictic cells demonstrated, the
expression of at least small amounts of gene product verified, and the existence of infectious
recombinant viruses sought and, if found, analyzed.

Plasmid-based expression vectors. The calcium-phosphate procedure for transferring a

plasmid-based expression vector into human bone marrow has not yet been demonstrated to be an
effective delivery system. However, the procedure itself does not appear to represent a significant
risk of harm. In theory, of course, a stem cell could be altered to make it carcinogenic so that it would
still be necessary to follow treated mice over time to determine the likelihood of pathology. Primate
studies, however, would appear not to be necessary.
The initial clinical protocols designed to carry out gene therapy in patients will probably be evaluated
in the following way. Under current Department of Health and Human Services regulations for the
protection of human research subjects, a human gene therapy protocol must be reviewed by the
Institutional Review Board at the investigator’s home institution. In addition, because of the
widespread public interest and concern in this area, the National Institutes of Health has announced
(58) that any federally funded gene therapy experiment involving recombinant DNA must first be
approved by the NIH after review by the Recombinant DNA Advisory Committee (RAC). Prior to
review by RAC, proposals will be examined by a special RAC working group on human gene ther-
apy (59). In addition, the Food and Drug Administration could regulate the DNA used in a human
trial as a biological drug, analogous to polynucleotide interferon inducers, interferons, and vaccines
(60). The Food and Drug Administration is currently exploring its regulatory responsibilities in this
area (61).

Representative Albert Gore’s proposal for a President's Commission on the Human Applications of
Genetic Engineering (62) has just passed both houses of Congress in a modified form. This
commission, if signed into law, would probably concern itself primarily with matters of policy and
procedure rather than the review of individual recombinant DNA research proposals (63): the initial
protocols, however, might be of particular interest to the commission in helping it to define the scope
of its efforts.

Conclusion

It now appears that effective delivery expression systems are becoming available that will allow
reasonable attempts at human gene therapy. These systems are based on treatment of bone marrow
cells with retroviral-vectors carrying a normal gene. The safety of the procedures is the remaining
major issue. Patients severely debilitated by being homozygous for a defect in the gene for one of the
enzymes HPRT, PNP, or ADA are the most likely first candidates for gene therapy.

It is unrealistic to expect a complete cure from the initial attempts at gene therapy. Many patients,
who suffer from severe genetic diseases, as well as their families, are eager to participate in early
clinical trials even if the likelihood is low that the original experiments will alleviate symptoms.
However, for the protection of the patients, particularly since those with the most severe diseases and
therefore, the most ethically justifiable first candidates are children, gene therapy trials should not be
attempted until there are good animal data to suggest that some amelioration of the biochemical
defect is likely. Then it would be necessary to weigh the potential risks to the patient, including the
possibility of producing a pathologic virus or a malignancy against the anticipated benefits to be
gained from the functional gene. This risk to benefit determination, a standard procedure for all
clinical research protocols, would need to be carried out for each patient.

In summary, institutional review boards should carefully evaluate therapeutic protocols to ensure that
the delivery system is effective. That sufficient expression can be obtained in bone marrow cultures,
and in laboratory animals to predict probable benefit, even if small, for the patient, and that safety
protocols have demonstrated that the probability is low for the production of either a malignant cell,
or a harmful infectious retrovirus. Once these criteria are met, I believe that it would be unethical to
delay human trials. Patients with serious genetic diseases have little other hope at present for
alleviation of their medical problems. The issues of germ line therapy and enhancement engineering
need to be debated widely in society; but arguments that genetic engineering might someday be
misused do not justify the needless perpetuation of human suffering that would result from an
unnecessary delay in the clinical application of this potentially powerful therapeutic procedure (64).

References and Notes

1. A. C. Spradling and G. M. Rubin, Science 218, 341 (1982); G. M. Rubin and A. C. Spradling,
ibid., p. 348.
2. R. E. Hammer, R. D. Palmiter, R. L. Brinster, Nature (London) 311, 65 (1984).
3. G. Kolata, Science 223, 1376 (1984); M. Chase, Wall Street J. (26 January 1984), p. 1; H. M.
Schmeck, Jr., The New York Times (10 April 1984), p. C1.
4. K. E. Mercola and M. J. Cline, N. Engl. J. Med. 303, 1297 (1980).
5. L. C. Skow et al., Cell 34 1043 (1983)
6. R. Parkman, E. W. Gelfand, F. S. Rosen, A. Sanderson, R Hirschhorn, N. Engl. J. Med. 292, 714
(1975); R. Hirschhorn, V. Roegner-Maniscalco, L. Kuritsky, F. S. Rosen, J. Clin. Invest. 68 1387
(1981).
7. W. L. Nyhan, B. Bakay, J. D. Connor, J. F. Marks, D. K. Keele, Proc. Natl. Acad. Sci. U.S.A. 65,
214 (1970); J. A. McDonald and W. N. Kelley, Biochem. Genet. 6, 21 (1972); R. O. McKeran et
al., J. Neurol. Sci. 22, 183 (1974); M. J. Dewey, D. W. Martin, G. R. Martin, B. Mintz, Proc.
Natl. Acad. Sci. U.S.A. 74, 5564 (1977).
8. M. W. Nirenberg, Science 157, 633 (1967); W. F. Anderson, in The New Genetics and the Future
of Man, M. Hamilton, Ed. (Eerdmans, Grand Rapids, Mich., 1972), p. 109; T. Friedmann and R.
Roblin, Science 175, 949 (1972).
9. President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and
Behavioral Research, Splicing Life (Government Printing Office, Washington, D.C., 1982); U.S.
Congress, House, Hearings on Human Genetic Engineering Before the Subcommittee on
Investigations and Oversight of the Committee on Science and Technology, 97th Cong. 2d sess.,
No. 170 (Government Printing Office, Washington, D.C., 1982); W. F. Anderson, J. Am. Med.
Assoc. 246, 2737 (1981); A. G. Motulsky, Science 219, 135 (1983); J. C. Fletcher, Va. Law Rev.
69, 515 (1983); T. Friedmann, Gene Therapy: Fact and Fiction (Cold Spring Harbor Laboratory,
 Cold Spring Harbor, N.Y., 1983); J. Rifkin, Algeny (Viking, New York, 1983); Y. Baskin, The
Gene Doctors (Morrow, New York, 1984).
10. W. F. Anderson and J. C. Fletcher, N. Engl. J. Med. 303, 1293 (1980).
11. The most obvious problems would be degradation by nucleases in the blood and elicitation of an
immune response to the foreign material. Any DNA that could get into cells would do so
randomly, perhaps even into germ cells.
12. H. Varmus and R. Swanstrom, in RNA Tumor Viruses, R. Weiss, N. Teich, H. Varmus, J. Coffin,
Eds. (Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., 1982), p. 369; H. E. Varmus,
Science 216, 812 (1982); R. Risser, J. M. Horowitz, J. McCubrey, Annu. Rev. Genet. 17, 85
(1983).
13. A. Joyner, G. Keller, R. A. Phillips, A. Bernstein, Nature (London) 305, 556 (1983); A.
Bernstein, personal communication.
14. R. Mann, R. C. Mulligan, D. Baltimore, Cell 33, 153 (1983).
15. The packaging signal has been mapped by Bal 31 digestion to a region less than 160 bases and
lies completely outside the LTR (D. Armentemo and E. Gilboa, personal communication).
16. D. A. Williams, I. R. Lemischka, D. G. Nathan, R. C. Mulligan, Nature (London) 310, 476
(1984); D. Nathan, personal communication.
17. T. M. Dexter, T. D. Allen, L. G. Lajtha, R. Schofeld, B. I. Lord, J. Cell. Physiol. 82, 461 (1973);
P. Mauch, J. S. Greenberger, L. Botnick, E. Hannon, S. Hellman, Proc. Natl. Acad. Sci. U.S.A. 77
2927 (1980).
18. I. M. Verma, F. van Straaten, M. Jones, R. A. Bosselman, C. Van Beveren, in Tumor Viruses and
Cell Differentiation, E. Scolnick and A. Levine, Eds. (Liss, New York, 1983), p. 251; I. M.
Verma, personal communication.
19. A. D. Miller, D. J. Jolly, T. Friedmann., I. M. Verma, Proc. Natl. Acad. Sci. U.S.A. 80, 4709
(1983); A. D. Miller, R. J. Eckner, D. J. Jolly, T. Friedmann, I. M. Verma, Science 225, 630
(1984).
(A) P. K. Bandyopadhyay and H. M. Temin, Mol. Cell. Biol. 4 749 (1984); M. Emerman and
H. M. Temin, J. Virol. 50, 42 (1984).
20. A. L. Joyner and A. Bernstein, Mol. Cell. Biol. 3, 2191 (1983).
21. P. E. Berg, A. Henderson, S. Ripley, J.-K. Yu, W. F. Anderson, Biochem. Biophys. Res.
Commun. 116, 959 (1983); R. E. Steele, A. H. Bakken, R. H. Reeder, Mol. Cell. Biol. 4 576
(1984); A. J. H. Smith and P. Berg, Recombination at the DNA Level (Cold Spring Harbor
Laboratory, Cold Spring Harbor, N.Y., in press).
22. R. C. Mulligan, B. H. Howard, P. Berg, Nature (London) 277, 108 (1979); D. H. Hamer, K. D.
Smith, S. H. Boyer, P. Leder, Cell 17, 725 (1979); Y. Gluzman, Ed., Eukaryotic Viral Vectors
(Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., 1982).
23. K. Van Doren, D. Hanahan, Y. Gluzman, J. Virol. 50, 606 (1984); K. Van Doren and Y.
Gluzman, Mol. Cell. Biol. 4, 1653 (1984); S. Karlsson, R. K. Humphries, Y. Gluzman, A. W.
Nienhuis, Proc. Natl. Acad. Sci. U.S.A., in press.
24. N. Sarver, P. Gruss, M.-F. Law, G. Khoury, P. M. Howley, Mol. Cell. Biol. 1 486 (1981).
25. F. L. Graham and A. J. van der Eb, Virology 52, 456 (1973).
26. M. Wigler et al.,Cell 11, 223 (1977).
27. C. M. Corsaro and M. L. Pearson, Somatic Cell Genet. 7, 603 (1981); Y.-M. Shen et al., Mol.
Cell. Biol., 2, 1145 (1982); M.A. Lopata, D. W. Cleveland, B. Sollner-Webb, Nucleic Acids Res.
12, 5707 (1984).
28. F. Carr, W. D. Medina, S. Dube, J. R. Bertino, Blood 62, 180 (1983).
29. M. J. Cline, H. Stang, K. Mercola, L. Morse, R. Ruprecht, J. Brown, W. Salser, Nature (London)
284, 422 (1980); K. E. Mercola, H. D. Stang, J. Browne, W. Salser, M. J. Cline, Science 208,
1033 (1980); W. Salser et al., in Organization and Expression of Globin Genes, G.
Stamatoyannopoulos and A. W. Nienhuis, Eds. (Liss. New York, 1981), pp. 313-334; M. Bar-Eli,
H. D. Stang, K. E. Mercola, M. J. Cline, Somatic Cell Genet. 9, 55 (1983).