| Current Opinion in Biotechnology Vol. 7, No. 5, October 1996 Recombinant protein production in transgenic animals [Review article] Yann Echelard Current Opinion in Biotechnology 1996, 7:536-540. |
|




The commercial potential of transgenic animal bioreactors for the production of high-value proteins is appreciable. As a result, a growing body of work has focused on outlining the parameters of mammary gland expression of heterologous proteins. This review will discuss the following three subjects: milk-specific regulatory elements, post-translational modification capabilities of the mammary gland, and physiological effects of transgene expression. Recent developments in the related field of blood-borne production of recombinant proteins and peptides will also be summarized.
Pronuclear microinjection, which is still the only method routinely used to introduce foreign DNA in the germline of farm animals, is, at best, a costly, tedious, and inefficient endeavor. The last part of this review will describe some of the efforts aimed at bypassing pronuclear microinjection, either by developing pluripotent embryonic cell lines, or by developing somatic approaches similar to those currently used in gene therapy trials. These technologies could eventually improve our ability to fine tune the production characteristics of animal bioreactors.

The combination of large daily protein output, excellent post-translational
processing abilities, ease of access to the recombinant protein (milking), and
low capital cost of production plants (farms) as compared to industrial
high-volume fermenters, has made the transgenic mammary gland an excellent
candidate for the production of recombinant protein. Regulatory sequences of the
major milk protein genes have been used, with variable success, to direct the
expression of an array of heterologous proteins in the milk of transgenic mice
(mostly), rats, rabbits, pigs, sheep, goats, and cows (for a recent review, see
[1•]).
Five promoters have been shown to yield high levels (>1mg ml-1) of
foreign protein in milk: sheep
-lactoglobulin
[2] [3] [4], goat
-casein
([5] [6] [7•]; H
Meade et al., abstract 110, Meeting on Mouse Molecular Genetics, Cold
Spring Harbor, New York, September 1994), bovine
s1-casein
[8] [9],
rabbit whey acidic protein (WAP) [10] [11] [12], and
mouse WAP [13] [14••].
Maximum expression levels reported with the mouse WAP promoter have been lower
than with other promoters [5] [13] [14••] [15] [16] [17] [18] [19].
However, this could be a function of the heterologous gene expressed. When the
4.2 kilobase (kb) mouse WAP promoter [18] was
used to express a human
1-antitrypsin
genomic construct in milk of transgenic mice, levels comparable to that reported
with other promoters [2] [10] [19] were
observed (B Wilburn, J Williams, Y Echelard, unpublished data).
In most accounts, it has been difficult to achieve protein yields superior to
1mg ml-1 with cDNA derived constructs [3] [15] [16] [17] [19] [20] [21] [22] and
sometimes the transgenes are transcriptionally silent. However, the use of a
genomic construct does not always ensure high expression levels [23]. The
weak performance of cDNA-containing transgenes could be caused by sensitivity to
the silencing influences of chromosomal sequences surrounding the integration
sites. Two groups [24] [25] have
attempted to correct this situation by co-microinjecting highly expressed
genomic constructs with poorly expressed cDNA constructs in order to obtain
integration sites containing both silent and highly expressed transgenes. In
both cases, although there was increased expression of the cDNA transgenes from
co-integrated sites as compared to transgenic integration sites containing only
cDNA sequences, the cDNA derived expression levels were still 2–3 orders of
magnitude lower than those obtained from genomic transgenes. In another study,
poorly expressed sheep
-lactoglobulin
driven human factor IX (hfIX) cDNA transgenes were shown to contain a cryptic 3'
splice site [26••],
leading to the accumulation of aberrantly spliced hfIX transcripts. With a
construct eliminating the cryptic splice site, transgenic mice expressed hfIX to
levels up to 60µg ml-1. This is much higher than with unmodified hfIX
transgenes and close to commercially acceptable levels if the modified transgene
is expressed similarly in large animals.
Proper post-translational modification of milk-produced recombinant proteins
is also an issue of concern. A recent study [27••]
compares the two N-glycosylation sites of human interferon-
(IFN-
;
Asn25 and Asn97) in recombinant protein samples obtained with three expression
systems: Chinese hamster ovary (CHO) cells, baculovirus-infected SF9 cells, and
the mammary gland of transgenic mice. The transgenic mouse-derived IFN-
had predominantly complex sialylated biantennary N-glycans at Asn25, similar to
the CHO cell-derived IFN-
.
An increased proportion of oligomannose at Asn97 was found in transgenic
mouse-derived material as compared to the CHO cell IFN-
.
Increased incidence of oligomannose glycans could affect the clearance by the
mannose receptor of the recombinant protein.
Transgenically produced human protein C (hPC) from both mouse and swine has
also been well characterized [14••] [17] [28] [29].
Mature hPC is an extensively glycosylated heterodimer that requires the removal
of prepropeptide and propeptide,
-hydroxylation,
as well as
-carboxylation
of nine glutamic acid residues, for full activity. Analysis of transgenic mice
and pig mammary gland-derived hPC, as compared to plasma-derived hPC, showed
reduced anticoagulant activity as well as high proportions of single chain
unprocessed precursors [17] [28] [29],
indicating that the mammary gland carried post-translational modifications
rather inefficiently. It is possible that the high-level expression of hPC was
saturating the post-translational modification machinery of the mammary gland,
because mammary gland expression of lower levels of another extensively modified
protein, hfIX, does not seem to give rise to a high proportion of incompletely
processed protein [26••].
In an effort to ameliorate the processing of recombinant hPC, bi-transgenic mice
coexpressing hPC and the human serine protease furin in their mammary gland were
generated [13].
These mice secreted low amounts (<5%) of unprocessed single chain hPC in
contrast to the 40–60% encountered in animals solely expressing hPC [13] [17].
Coexpression of a modification enzyme with the protein of interest could also be
applied to other post-translational modifications (N-
and O-glycosylation,
-carboxylation).
Moreover, it seems that the high efficiency of processing was obtained by using
what should be a relatively inefficiently expressed human furin cDNA transgene,
making this strategy even more attractive.
In another example [30••],
the modification of the mouse mammary gland post-translational modification
capability by the transgenic expression of a human
1,2-fucosyltransferase
was exploited. Here, the objective was not to better process a recombinant
protein, but to remodel the structure of milk glycoconjugates. This experiment
showed that transgenic dairy animals could become an inexpensive source of
otherwise hard to synthesize oligosaccharide of industrial or therapeutic value,
as well as a source of recombinant pharmaceutical proteins.
The effect of the expression of the transgene on the production animal is
another concern. In problem cases, the expression of the heterologous protein
can either influence the physiology of the mammary gland and disrupt lactation
[6] [31] [32] [33] [34] [35], or
have a systemic effect on the transgenic animal, affecting its viability and
reproductive performance [11] [35] [36] [37]. A
number of milk proteins have been shown to be usually present in blood during
lactation [38•] [39] [40];
hence the detection, in the serum, of a heterologous protein expressed in the
mammary gland was not surprising when it was encountered [10] [11] [12] [19] [41] [42]. RNA
analysis also showed the incidence of low-level ectopic expression of
WAP-derived [10] [11] [12] [18] [23] [38•] [41] [42] [43] and
-lactoglobulin-derived
[44] [45] [46] [47]
transgenes, whereas the expression of
s1-casein
and
-casein
transgenes (when examined) was restricted to the mammary gland [7•] [8] [9] [22] [48] [49] [50].
With the tighter regulation of
s1-casein
and
-casein
promoters, it may be argued that these sequences could be more appropriate than
the current versions of the WAP and
-lactoglobulin
cassettes for driving the milk production of very active proteins such as human
growth hormone (hGH). However, in terms of the health of the transgenic
production animal, low-level ectopic expression of proteins, such as
1-antitrypsin,
human serum albumin, or most monoclonal antibodies, is probably
inconsequential.

The production of recombinant human hemoglobin A (HbA) has been achieved in
transgenic mice and pigs [51] to
levels up to 24% completely human HbA and 30% human
–pig
hybrid using a construct containing human
-,
-,
-globin
genes linked to a human
locus control region (LCR). The key to achieving a high expression level was the
construction of a fusion
-globin
gene containing the pig
-globin
promoter linked to the human
-globin
coding region. Previous HbA constructs containing both the human promoter and
-globin
coding sequences permitted the production of, at most, 10% of human HbA in
transgenic pigs [52].
Animals appeared to tolerate well the high levels of human HbA in their
erythrocytes. Structural analysis of the swine-derived human recombinant HbA has
shown that it was equivalent to protein purified from human blood [53].
Another procedure exploiting the recombinant hemoglobin expression system,
and the fact that terminally extended
-globin
can be functional, could become an alternative to the chemical synthesis of
therapeutic peptides. Carboxy-terminal fusions of human
-globin
with either
-endorphin
(16-mer) or magainin (26-mer) were expressed in transgenic mice [54].
Fusion proteins were expressed to levels corresponding to 25% of the total
hemoglobin, seemingly without ill effects on the transgenic mice. Magaining
peptides were cleanly released by enterokinase digestion of purified fusion
protein, using a cleavage site engineered between the two parts of the chimeric
polypeptide. On the other hand, a cryptic enterokinase site caused partial
degradation of
-endorphin
during the digestion step, showing that the release of the recombinant peptides
from the chimeric protein will not always be straightforward.

Approaches combining the potential of clonal expansion offered by nuclear transfer technology, with the possibilities of gene addition and replacement achievable in embryonic stem (ES) cell lines (or ES-like cells), could be particularly beneficial to the generation of transgenic bioreactors (potential advantages are reviewed in [55]).
Pluripotent ES cells able to contribute to the germline have only been derived in mice. Recently, there have been descriptions of chimeric animals generated with rat [56] and pig [57] ES cells, but, in both cases, no evidence of germline transmission from these cells has been reported. Cow fetuses obtained following the nuclear transfer of ES cell-derived nuclei in recipient oocytes died in utero, exhibiting major defects in placental development [58•]. In a related breakthrough, two healthy phenotypically female lambs were born from embryos generated by transferring nuclei isolated from embryo-derived epithelial cells into enucleated oocytes [59••]. Microsatellites analysis demonstrated that all the nuclear transfer lambs and fetuses were derived from the cell line. It is not yet clear whether foreign DNA can be easily introduced in this type of cell line, and there are questions about the health and reproductive fitness of the recovered lambs. However, this experiment is the first demonstration that it is possible to obtain large farm animals from embryo derived-cell lines, and it is certainly a step towards the possibility of performing gene replacement in large animals.
Another group reported a strategy that bypasses entirely the transgenic route to obtain recombinant protein production in animals [60]. Replication-defective retroviral vectors carrying the hGH cDNA were used to infect the mammary gland of goats hormonally induced to lactate. Expression levels of up to 118ng of hGH per milliliter of milk were obtained on the first lactation day. By day 3, expression had stabilized to 3–12 ng ml-1. The advantage of this method is that expression of the recombinant protein is obtained quickly, without the delay caused by the generation interval required to generate a producing transgenic animal (18 months for goats). However, production levels reported in this paper [60] are very low. In the conditions described, less than 1 mg of crude recombinant hGH would be obtained per lactation per goat. If this technology could also be applied to other potential production species (rabbit, sheep, pigs, and cows), it could permit the quick determination, for specific proteins, of which system is most appropriate for the high-level production of correctly processed polypeptides.

The expression of recombinant proteins in transgenic animals has come of age.
At least two products (sheep-derived human
1-antitrypsin
and goat-derived antithrombin III) will be submitted to phase I trials within
two years. Transgenic animals have the ability to process highly modified
proteins, although not always with total efficiency. There are species-specific
and tissue-specific characteristics. However, it seems that transgenic systems
are flexible and that, when needed, processing enzymes can be coexpressed with
foreign products to obtain more humanized recombinant proteins. In the future,
the further characterization of each bioreactor's capability, and an improved
ability to introduce specific genetic modifications, should lead to more
sophisticated transgenic production herds.

I wish to thank Barbara Davis for her help with manuscript preparation, and Harry Meade, Carol Ziomek, and Li-How Chen for comments and many stimulating discussions.

1
antitrypsin in the milk of transgenic sheep.
and mammary-tissue specific expression of a caprine
-casein-encoding
minigene driven by a
-casein
promoter in transgenic mice.
-casein
content.
S1-casein
gene sequences direct expression of active human urokinase in mouse
milk.
the upstream region of the rabbit whey acidic protein (Wap) gene
targets transgene expression to the mammary gland.
produced in different animal expression systems.
and tissue-specific nature of glycosylation and the need to thoroughly analyze
the carbohydrate composition of recombinant proteins.
1.
Leu
mutant p53 in the mammary gland of transgenic mice results in altered
lobulalveolar development.
1
expression in the secretory mammary epithelium induces early senescence of the
epithelial stem cell population.
-casein
under the control of the bovine
-lactalbumin
5' flanking region.
-lactoglobulin
in transgenic mice.
-lactoglobulin
gene in transgenic mice.
-lactoglobulin/human
serum albumin fusion genes in transgenic mice: hormonal regulation and in
situ localization.
-lactoglobulin
transgenes.
s1-casein-derived
transgenes in mice.
-,
s2-,
and
-casein
genes in transgenic mice.
-endorphin
and magainin in the erythrocytes of transgenic mice: a production system for
therapeutic peptides.
