The Plant Cell, Vol. 10, 641, May 1998, www.plantcell.
org
IN THIS ISSUE
Factories of the Future? Metabolic Engineering in Plant Cells
Humans have been manipulating the
genetic makeup of agricultural plants
(and animals) for millennia. Recently,
the pace of change has increased
markedly as researchers apply molecu-
lar tools to modify a wide range of ag-
ronomic characteristics and to develop
and market transgenic crops that gen-
erate specific novel products (ap Rees,
1995). Among many others, these in-
clude vaccines and other pharmaceuti-
cals, plastics, and proteins that may
render certain plants effective tools for
environmental decontamination (John
and Keller, 1996; Raskin, 1996; Arntzen,
1997; Hezari and Croteau, 1997).
Related techniques can be adapted
to manipulate endogenous biochemical
pathways. Here, the objective is to gen-
erate transgenic crops in which the
range, scope, or nature of a plant’s ex-
isting natural products is modified to
provide beneficial commercial, agro-
nomic, and/or post-harvest process-
ing characteristics (Kinney, 1998). Such
plants may, for example, accumulate
unusual edible or industrial oils (Broun
and Somerville, 1997; Zou et al., 1997),
lignins with different subunit compo-
sitions that interfere less extensively
with paper production (Campbell and
Sederoff, 1996), starches with different
milling qualities, and so on.
In theory, any biochemical pathway
can be subjected to this “metabolic en-
gineering.” In practice, however, efforts
to modify the biosynthesis of a variety
of plant primary and secondary metab-
olites have been confounded by the
very metabolic flexibility that research-
ers are trying to exploit. Although these
studies may uncover novel information
about a pathway and its regulation that
will inform subsequent attempts to ma-
nipulate it, such difficulties must be
overcome if a genetically engineered
crop is to become commercially viable.
There are four major variables that
appear to affect the outcome of meta-
bolic engineering experiments. First, it
is not always possible to predict the ef-
fects that directed perturbations (i.e.,
changing the level or activity of a sin-
gle biosynthetic enzyme) may have on
the entire pathway—unidentified endoge-
nous feedback and/or feedforward con-
trols may constrain metabolic pathways
in manners that are poorly defined. Sec-
ond, manipulating primary metabolic
pathways can have pleiotropic (and of-
ten detrimental) effects on plant growth
and development. Third, no matter how
well a transgenic plant is characterized
in an experimental setting, crop plants
in the field will be subjected to environ-
mental perturbations that may adversely
affect accumulation of the engineered
product. Fourth, because many appli-
cations of genetic engineering demand
that the engineered products get to the
right place in the plant at the right time,
it is critical to ensure that these prod-
ucts are appropriately targeted within
and/or between cells in the transgenic
plant.
These variables can be addressed by
adapting metabolic engineering strate-
gies in a number of ways. Using cul-
tured plant cells as metabolic factories
instead of whole plants can circumvent
problems associated with inappropriate
targeting and with the vagaries in me-
tabolite accumulation that can occur in
field-grown material. Growing cell cul-
tures in defined media may also bypass
a number of other potential problems,
such as tissue-specific or developmen-
tal regulation of some (or all) of the
genes in a pathway.
Clearly, the more that is known about
the metabolic pathway in question, the
better the chances that metabolic en-
gineering will succeed. However, de-
tailed understanding of the biochemical
changes that take place at each step
and of the enzymes that catalyze the
reactions may not always be sufficient.
It is also critical to determine how the
activities of these enzymes are regu-
lated, either at the level of gene expres-
sion or by their substrates and products
through feedback and/or feedforward
control.
In fact, regulatory genes are becom-
ing valuable tools for metabolic engineer-
ing. Combined genetic and biochemical
studies over the past 20 years suggest
that different transcription factors oper-
ating in a single pathway may affect the
expression of distinct (but sometimes
overlapping) suites of “downstream”
genes. These distinct suites of genes
may encode enzymes that participate in
individual branches of a complex path-
way (see e.g., Grotewold et al., 1994;
Sablowski et al., 1994; Tamagnone et
al., 1998). Thus, by manipulating the
expression of a single transcription fac-
tor, it is theoretically possible to affect
the expression of several coordinately
regulated biosynthetic enzymes. Fur-
thermore, work in a number of systems
suggests that many transcriptional reg-
ulators are capable of faithfully recog-
nizing their homologous target genes in
heterologous species. For example,
Tamagnone et al. (1998) have recently
shown that two transcriptional regula-
tors that repress lignin biosynthesis in
Antirrhinum also affect lignin synthesis
in transgenic tobacco.
Lignin is a secondary metabolite, and
metabolic engineering strategies aimed
at manipulating such compounds have
generally been less complicated than
those in which primary metabolic path-
ways are targeted for modification
(Kinney, 1998). Many secondary me-
tabolites are not essential for plant via-
bility, and so it is possible that their
biosynthesis is less tightly constrained
642 The Plant Cell
IN THIS ISSUE
than is that of critical primary metabo-
lites such as starch and lipids.
Despite the fact that they are dis-
pensable, many secondary metabolites
play important roles in plant biology.
For example, specific secondary me-
tabolites have been reported to function
in defending plants from pathogens, as
insecticidal compounds, and as UV
protectants (Dixon and Paiva, 1995;
Byrne et al., 1996; Jansen et al. 1998).
Secondary metabolites are also the
source of many components in the
pharmaceutical arsenal against human
pathogens and diseases (Kutchan, 1995).
Others are food colorings, and still others
contribute to the scent and color of cut
flowers (Holton and Cornish, 1995; Wang
et al., 1997).
For these reasons, secondary meta-
bolic pathways have been subjected
to intense genetic and biochemical in-
vestigation over the years. In fact, the
biosynthetic pathways of many plant
secondary metabolites are generally well
characterized and, in some cases, a
number of genes that could be used as
targets for metabolic engineering strat-
egies have been cloned.
One well-characterized pathway di-
rects the synthesis of anthocyanins.
Experiments in maize, Antirrhinum, pe-
tunia, and Arabidopsis have uncovered
not only structural genes that encode
enzymes in this pathway but also a
number of regulatory genes. The antho-
cyanins are flavonoids, one of the two
major classes of plant secondary me-
tabolites that are derived from the shiki-
mate pathway, the primary products of
which include phenylalanine and ty-
rosine (Herrmann, 1995; Shirley, 1996).
There are several chemically distinct fam-
ilies of flavonoids in cereals, ranging from
the anthocyanins and other 3-hydroxy
flavonoids to the related 3-deoxy fla-
vonoids (Shirley, 1996). Many flavonoids
and phenylpropanoids (the other major
class of shikimate pathway–derived
secondary metabolites) are of commer-
cial importance, either in their own right
or by virtue of the agronomic benefits
they confer on plants in which they ac-
cumulate (Shirley, 1996; Jansen et al.,
1998).
In maize plants, enzymes carrying
out steps along each of these major
branches appear to be under the sepa-
rate control of distinct transcriptional reg-
ulators. Upregulation of C1 and R leads
to anthocyanin accumulation, whereas P
is required for 3-deoxy flavonoid biosyn-
thesis. Research with mutants has shown
that P and C1/R operate independently
in maize floral organs, and it has been
shown that P actually binds to the pro-
moter of a1, one of the flavonoid biosyn-
thetic genes that it regulates (Grotewold
et al., 1994; Sainz et al., 1997).
On pages 721–740 of this issue,
Grotewold et al. bring together fla-
vonoid biosynthesis, transgenic cell
lines, and metabolic engineering to
demonstrate that the ectopic expres-
sion of transcription factors is a viable
method for engineering secondary me-
tabolism in cultured plant cells. Their
work also significantly extends our un-
derstanding of and appreciation for the
flexibility (and complexity) of flavonoid
biosynthesis in maize.
Grotewold et al. begin by investigating
changes in flavonoid and phenylpro-
panoid biosynthesis that are provoked
by overexpressing P or C1 and R in cul-
tured Black Mexican Sweet (BMS) maize
cells. The corresponding genes ( c1, r,
and p1) are not ordinarily expressed in
these cells, which gives the authors a
clean metabolic slate upon which to
identify some of their downstream tar-
get genes.
They do this by analyzing the fla-
vonoids and related compounds that
accumulate in the transgenic lines and
by investigating the effects of ectopic
C1/R or P expression on the accumula-
tion of transcripts that encode flavonoid
biosynthetic enzymes. By marrying their
two data sets with what is known of fla-
vonoid biosynthesis, Grotewold et al.
come up with some solid conclusions
regarding the downstream targets of P
and C1/R as well as some surprising in-
formation on the potential role of P in
phenylpropanoid metabolism and sec-
ondary metabolite targeting.
Grotewold et al. demonstrate that
many of the effects of overexpressing P
and C1/R on flavonoid biosynthesis in
BMS cells reflect previous analyses of
genetically analogous tissues in maize
plants. For example, the authors show
that ectopic expression of C1/R leads
to the accumulation of anthocyanins and
a number of other related 3-hydroxy
flavonoids. Similarly, P overexpression
correlates with the accumulation of
3-deoxy flavonoids such as luteoforol
and C-glycosyl flavones.
These data are significant for two
reasons. First, a role for P in inducing
the synthesis of 3-deoxy flavonoids had
not been demonstrated previously in
tissues in which P is not normally ex-
pressed. Second, although Grotewold
et al. show that C-glycosyl flavones ac-
cumulate in the BMS cells overexpress-
ing P, it is the closely related rhamnosyl
derivatives that ordinarily accumulate in
maize tissues. This distinction is impor-
tant because rhamnosylated flavones
such as maysin, which accumulates in
maize silks, are thought to possess
insecticidal properties (Byrne et al.,
1996). The authors suggest that the
rhamnosylated derivatives do not accu-
mulate in transgenic BMS cells simply
because they do not express a rham-
nosyl transferase that is required for the
final biosynthetic step. If this is true,
then it is reasonable to assume that P
may induce the accumulation of insec-
ticidal compounds in maize reproduc-
tive tissues, which, in turn, could confer
a selective advantage on plants ex-
pressing p1.
That this advantage may be quantita-
tive is suggested by Grotewold et al.’s
data showing that in P-expressing
lines, the transcript levels of two fla-
vonoid biosynthetic enzymes, a1 and
c2, correlate with the level of P expres-
sion. These data support the authors’
contention and previous suggestions
(e.g., Byrne et al., 1996) that p1 may

operate as a quantitative trait locus by
virtue of the presumed insecticidal ac-
tivity of the rhamnosylated 3-deoxy fla-
vone derivatives whose synthesis it
regulates. Thus, allelic variation of p1
may derive from the level of P expres-
sion and/or from the distinct P tissue
specificities that are known to correlate
with alterations in the C-terminal do-
main of P (Chopra et al., 1996).
Not all of the data reported by
Grotewold et al. reflect parallels be-
tween flavonoid biosynthesis in BMS
cells and maize plants, and the authors
point out some potentially interesting
differences. For example, they show
that the phenylpropanoids ferulic acid
and chlorogenic acid accumulate in cell
lines that overexpress P. This observa-
tion is surprising because P has not
been reported to affect the levels of these
compounds in plants. Moreover, neither
chlorogenic acid nor ferulic acid is a
known precursor of the 3-deoxy flavo-
noids whose synthesis is regulated by P.
In support of their contention that the
accumulation of ferulic acid in BMS cells
has uncovered a novel regulatory role
for P, Grotewold et al. point out that the
chemical structure of ferulic acid is anal-
ogous to that of the C-glycosyl flavone
isocoparin. Given the rather broad sub-
strate preferences of some of the en-
zymes on the flavonoid biosynthetic
pathway, they suggest that P may up-
regulate a catechol O-methyl trans-
ferase that can utilize either a flavonoid
or a phenylpropanoid as substrate.
However, it is also conceivable that fe-
rulic acid accumulates as a nonspecific
consequence of P-induced changes in
flux through the flavonoid and phenyl-
propanoid pathways. Regardless of the
underlying mechanism, the possibility
that P expression may affect phenyl-
propanoid biosynthesis is supported by
recent work showing that chlorogenic
acid, an esterified derivative of ferulic
acid, accumulates in immature maize
silk tissue (Byrne et al., 1996).
In addition to their efforts to explain the
molecular basis for the changes in fla-
vonoid and phenylpropanoid metabolism
that are induced by expression of C1/R
and/or P in BMS cells, Grotewold et al.
also take some important steps toward
understanding how secondary metabo-
lites are trafficked in plant cells. Indeed,
they find that the changes associated
specifically with C1/R or P expression
at the molecular level reflect C1/R– and
P-specific differences in cell biology.
The authors show that in C1/R–
expressing cells, anthocyanins accumu-
late first in vesicles called anthocyano-
plasts, which are derived from the
smooth endoplasmic reticulum. In time,
the anthocyanoplasts coalesce to form
a single large vacuole. By contrast,
Grotewold et al. detect two distinct
types of green fluorescent organelles in
cells overexpressing P, neither of which
appears to be closely related to antho-
cyanoplasts. Each type of organelle tar-
gets P-induced secondary metabolites
to a distinct cellular location—either to
the cell wall (the destination of most of
the fluorescent material in P-expressing
BMS cells) or to the vacuole.
Grotewold et al. suggest that these
data reflect the inherent complexity that
is likely to be a hallmark of secondary
metabolite targeting in plants. Because
very little is currently known about this
topic, subsequent investigations of the
P-induced fluorescent compounds and
the organelles in which they accumulate
are bound to be particularly informative.
Moreover, continuing cell biological in-
vestigations of flavonoid targeting in
BMS cells may complement some ex-
isting data on anthocyanin targeting.
For example, one of the maize genes
that has been defined as playing a role
in anthocyanin synthesis, bronze-2, was
recently shown to encode a glutathione
S-transferase that both conjugates an-
thocyanin with glutathione and plays a
role in targeting the conjugates to vacu-
oles (Marrs et al., 1995).
The work reported by Grotewold et
al. represents an informative and pro-
ductive evaluation of the feasibility of
metabolic engineering in plant cells.
May 1998 643
IN THIS ISSUE
The majority of their data suggest that
the flavonoid pathway is regulated sim-
ilarly in BMS cells and in maize plants,
with some of the more subtle differ-
ences likely being accounted for by
deficiencies in the complement of bio-
synthetic genes that are expressed in
the cell line. This suggests that conclu-
sions drawn from experiments with
cells lines can generally be extrapo-
lated to whole plants.
In the future, Grotewold et al. can fur-
ther refine their analyses of flavonoid
biosynthesis by focusing their efforts on
a series of constructs that allow C1/R
and/or P expression to be induced in
BMS cells upon estradiol application. In
conjunction with 13C-labeling studies,
this estradiol-inducible system should
provide a more dynamic view of fla-
vonoid biosynthesis in BMS cells by
facilitating the identification of interme-
diates along the distinct branches of
flavonoid biosynthesis that are con-
trolled by C1/R or P.
Grotewold et al. may also be able to
use the transformed BMS cell lines to
further investigate the biological func-
tions of the compounds that accumulate
as a result of the overexpression of P
and/or C1/R. For example, preliminary
indications that some of these cell lines
may exhibit increased UV tolerance
should be pursued vigorously.
Finally, the authors’ apparent identifi-
cation of two distinct trafficking path-
ways, each of which appears to correlate
with the expression of a distinct regula-
tory gene (or gene pair), is most illumi-
nating. Do these transcription factors
directly regulate trafficking, or are the
two trafficking pathways associated
with the intrinsic qualities of the specific
flavonoids whose biosynthesis is trig-
gered by the ectopic expression of
each one? Because it will allow real-
time imaging of flavonoid transport, the
estradiol-inducible system described
by Grotewold et al. should also facili-
tate efforts to address this question.
Clearly, Grotewold et al.’s data repre-
sent a promising start in the engineering
644 The Plant Cell
IN THIS ISSUE
of secondary metabolite biosynthesis in
homogeneous plant cell cultures. How-
ever, whether the benefits of manipu-
lating metabolic pathways in cultured
cell lines will outweigh the expense of
their production and maintenance, and
whether it will become economically
preferable to use cell cultures as bio-
logical factories instead of plants in the
field, remain to be seen. My guess is
that, for certain products, they will.
Crispin B. Taylor
REFERENCES
ap Rees, T. (1995). Prospects of manipulat-
ing plant metabolism. Trends Biotechnol.
13, 375–378.
Arntzen, C.J. (1997). High-tech herbal medi-
cine: Plant-based vaccines. Nat. Biotech-
nol. 15, 221–222.
Broun, P., and Somerville, C.R. (1997).
Accumulation of ricinoleic, lesquerolic,
and densipolic acids in seeds of trans-
genic Arabidopsis plants that express a
fatty acyl hydroxylase cDNA from castor
bean. Plant Physiol. 113, 933–942.
Byrne, P.R., McMullen, M.D., Snook, M.E.,
Musket, T.A., Theuri, J.M., Widstrom,
N.W., Wiseman, B.R., and Coe, E.J., Jr.
(1996). Quantitative trait loci and meta-
bolic pathways: Genetic control of the
concentration of maysin, a corn earworm
resistance factor, in maize silks. Proc.
Natl. Acad. Sci. USA 93, 8820–8825.
Campbell, M.M., and Sederoff, R.R. (1996).
Variation in lignin content and composi-
tion. Plant Physiol. 110, 3–13.
Chopra, S., Athma, P., and Peterson, T.
(1996). Alleles of the maize P gene with
distinct tissue specificities encode Myb-
homologous proteins with C-terminal
replacements. Plant Cell 8, 1149–1158.
Dixon, R.A., Lamb, C.J., Masoud, S.,
Sewalt, V.J., and Paiva, N.L. (1996). Met-
abolic engineering: Prospects for crop
improvement through the genetic manipu-
lation of phenylpropanoid biosynthesis
and defense responses—A review. Gene
179, 61–71.
Grotewold, E., Drummond, B.J., Bowen,
B., and Peterson, T. (1994). The myb-
homolgous P gene controls phlobaphene
pigmentation in maize floral organs by
directly activating a flavonoid biosynthetic
gene subset. Cell 76, 543–553.
Grotewold, E., Chamberlin, M., Snook, M.,
Siame, B., Butler, L., Swenson, J.,
Maddock, S., St. Clair, G., and Bowen,
B. (1998). Engineering secondary metab-
olism in maize cells by ectopic expression
of transcription factors. Plant Cell 10,
721–740.
Herrmann, K.M. (1995). The shikimate path-
way: Early steps in the biosynthesis of aro-
matic compounds. Plant Cell 7, 907–919.
Hezari, M., and Croteau, R. (1997). Taxol
biosynthesis: An update. Planta Medica
63, 291–295.
Holton, T.A., and Cornish, E.C. (1995).
Genetics and biochemistry of anthocya-
nin biosynthesis. Plant Cell 7, 1071–1083.
Jansen, M.A.K., Gaba, V., and Greenberg,
B.M. (1998). Higher plants and UV-B radi-
ation: Balancing damage, repair and accli-
mation. Trends Plant Sci. 3, 131–135.
John, M.E., and Keller, G. (1996). Metabolic
pathway engineering in cotton: Biosynthe-
sis of polyhydroxybutyrate in fiber cells.
Proc. Natl. Acad. Sci. USA 93, 1268–1277.
Kinney, A.J. (1998). Manipulating flux
through plant metabolic pathways. Curr.
Opin. Plant Biol. 1, 173–178.
Kutchan, T.M. (1995). Alkaloid biosynthe-
sis—the basis for metabolic engineering of
medicinal plants. Plant Cell 7, 1059–1070.
Marrs, K.A., Alfenito, M.R., Lloyd, A.M.,
and Walbot, V. (1995). A glutathione
S-transferase involved in vacuolar trans-
fer encoded by the maize gene Bronze-2.
Nature 375, 397–400.
Raskin, I. (1996). Plant genetic engineering
may help with environmental cleanup.
Proc. Natl. Acad. Sci. USA 93, 3164–3166.
Sablowski, R.W.M., Moyano, E., Culianez-
Macia, F.A., Schuch, W., Martin, C., and
Bevan, M. (1994). A flower-specific Myb
protein activates transcription of phenyl-
propanoid biosynthetic genes. EMBO J.
13, 128–137.
Sainz, M.B., Grotewold, E., and Chandler,
V.L. (1997). Evidence for direct activation
of an anthocyanin promoter by the maize
C1 protein and comparison of DNA bind-
ing by related Myb domain proteins. Plant
Cell 9, 611–625.
Shirley, B.W. (1996). Flavonoid biosynthesis:
‘New’ functions for an ‘old’ pathway.
Trends Plant Sci. 1, 377–382.
Tamagnone, L., Merida, A., Parr, A.,
Mackay, S., Culianez-Macia, F.A.,
Roberts, K., and Martin, C. (1998). The
AmMYB308 and AmMYB330 transcription
factors from Antirrhinum regulate phenyl-
propanoid and lignin biosynthesis in trans-
genic tobacco. Plant Cell 10, 135–154.
Wang, J., Dudareva, N., Bhakta, S., Raguso,
R.A., and Pichersky, E. (1997). Floral
scent production in Clarkia breweri (Onagra-
ceae). II. Localization and developmental
modulation of the enzyme S-adenosyl-L-
methionine:(iso)eugenol O-methyltransfer-
ase and phenylpropanoid emission. Plant
Physiol. 114, 213–221.
Zou, J., Katavic, V., Giblin, E.M., Barton,
D.L., MacKenzie, S.L., Keller, W.A., Hu,
X., and Taylor, D.C. (1997). Modification
of seed oil content and acyl composition
in the Brassicaceae by expression of a
yeast sn-2 acyltransferase gene. Plant Cell
9, 909–923.