Genome shuffling of Lactobacillus for
improved acid tolerance
Ranjan Patnaik1, Susan Louie1, Vesna Gavrilovic1, Kim Perry1, Willem P.C.Stemmer1,
Chris M.Ryan2, and Stephen del Cardayré1*
Fermentation-based bioprocesses rely extensively on strain improvement for commercialization. Whole-cell
biocatalysts are commonly limited by low tolerance of extreme process conditions such as temperature, pH,
and solute concentration.Rational approaches to improving such complex phenotypes lack good models and
are especially difficult to implement without genetic tools. Here we describe the use of genome shuffling to
improve the acid tolerance of a poorly characterized industrial strain of Lactobacillus.We used classical strain-
improvement methods to generate populations with subtle improvements in pH tolerance, and then shuffled
these populations by recursive pool-wise protoplast fusion.We identified new shuffled lactobacilli that grow at
substantially lower pH than does the wild-type strain on both liquid and solid media. In addition, we identified
shuffled strains that produced threefold more lactic acid than the wild type at pH 4.0.Genome shuffling seems
broadly useful for the rapid evolution of tolerance and other complex phenotypes in industrial microorganisms.
Fermentation, the oldest and most sophisticated application of
biocatalysis, represents a $30–50 billion world market. Simple feed-
stocks can be converted to high-value products by multiple enzyme-
catalyzed chemical transformations, all within single cells1–6.
Fermentation products range from high-value natural products and
protein pharmaceuticals to industrial enzymes and commodity
chemicals2,6–15. Academic, government, and industrial researchers
continue to seek biocatalytic routes to produce commercial chemi-
cals from renewable biomass7,8,16,17. To this end, it is often feasible to
identify or construct strains that effect the desired chemical transfor-
mations; however, these first-generation biocatalysts seldom meet
the performance criteria for commercialization. Common limita-
tions of a biocatalyst range from low yield,productivity,and titer,to
poor tolerance to process conditions such as extremes in tempera-
ture, pH, or solvent or solute concentration. A great deal of effort
and resources are therefore committed to the improvement of candi-
date biocatalysts to meet commercial requirements.
Lactobacilli are of interest in the commercial production of lactic
acid, a compound that has long been used as a food additive and has
recently emerged as an important feedstock for the production of
other chemicals such as polylactic acid (PLA), acetaldehyde,
Fermentation pH contributes to the economics of the lactic acid
purification process19,20. Although typical commercial Lactobacillus
fermentations run at a minimum pH of 5.0–5.5, fermentation at or
below the pKaof lactic acid (∼3.8) is desirable.At this low pH,a sub-
stantial proportion of product is in the free-acid form and can be
purified by direct organic extraction of the fermentation broth. At
higher pH values, lactate is the predominant form, requiring a more
expensive and wasteful purification.The commercial goal of improv-
ing the growth and lactic acid production of microorganisms at low
pH could decrease waste and the cost of product purification19.
Growth at low pH involves multiple widely distributed loci.
Characterization of acid-resistant mutants of Lactococcus lactis iso-
lated by insertional mutagenesis suggests the involvement of 18 dif-
ferent loci and multiple mechanisms for protection against low
pH21. Proteomic analysis of acid adaptation in Lactobacillus identi-
fied 63 different proteins that were coordinately induced22.
Similarly, the acid response in Escherichia coli results from the
induction of several global regulators that induce various loci23,24.
Without an understanding of the mechanism of pH tolerance in
bacteria, a rational approach to engineering pH tolerance is
impractical. In addition, the Lactobacillus strain used in this study
is of commercial interest but has not been genetically character-
ized. Thus there exist few genetic tools for manipulating the strain.
Hence, it was an ideal system in which to study the application of
genome shuffling to improve a poorly understood and complex
phenotype of an uncharacterized organism.
Current methods for the improvement of industrial microor-
ganisms range from the random approach of classical strain
improvement (CSI) to the highly rational methods of metabolic
engineering. Although CSI is robust, it is time and resource inten-
sive. Rational approaches are information and tool intensive, and
rely on models that simplify biological systems. All aspects of cell
physiology are inherently complex; however,some are more readily
addressable rationally than others when the appropriate genetic
tools are available. Environmental tolerance, a particularly com-
plex and poorly understood phenotype,is not an obvious target for
rational engineering.We are interested in the application of recom-
bination technologies to the improvement of complex biological
systems. Molecular breeding continues to be a robust method for
the manipulation of gene, enzyme, and pathway function25–31. We
recently described the expansion of molecular breeding to micro-
bial breeding and demonstrated that genome shuffling of
1Codexis (a subsidiary of Maxygen),515 Galveston Drive,Redwood City,CA 94063.2Cargill Dow,LLC,15305 Minnetonka Boulevard,Minnetonka,MN 55345.
*Corresponding author (firstname.lastname@example.org).
Streptomyces fradiae was an effective means to achieve rapid
improvement in tylosin production32. Genome shuffling
takes advantage of the diversity that exists within a popula-
tion and provides a means to eliminate deleterious muta-
tions that accumulate in strains derived from asexual
approaches such as classic and mutator-mediated strain
improvement. Here we describe its application to improv-
ing acid tolerance in a strain of Lactobacillus with commer-
Isolation of initial population diversity. Genome shuffling
accelerates directed evolution by facilitating recombina-
tion between members of a diverse selected population32.
Because a single Lactobacillus strain was the starting point
of the evolution program, an improved population for
breeding was required. We used two classical methods to
generate populations of acid-tolerant variants of our wild-
type Lactobacillus (LB-WT). The first population (pop-adap) was
obtained by chemostat-mediated adaptation of LB-WT strain to
low pH in a 1 liter fermenter. The second population was enriched
for acid-tolerant mutants of the wild-type strain using a
nitrosoguanidine (NTG) mutant library (pop-NTG) selected on
pH gradient plates.The chemostat adaptation of the LB-WT strain
was achieved by slowly decreasing the fermentation pH from 6.0
to 4.1 over a period of 1,200 h. A stable population of Lactobacilli
had taken over the fermenter and was growing at pH 4.1,a pH that
severely inhibits growth of LB-WT. Pop-adap and pop-NTG were
grown anaerobically side by side along with LB-WT in indepen-
dent experiments on a rectangular agar plate whose pH ranged
from 6 to 3. All populations resulted in a confluent lawn of cells
that spanned from the neutral edge to the middle of the plate.
Pop-adap and pop-NTG populations, however, colonized a more
acidic region of the plate than that survived by LB-WT. Cells that
grew at more acidic pH than LB-WT were scraped from the plate,
subcultured in yeast extract (YE) medium and replated on pH gra-
dient plates for further enrichment. Subpopulations pop1 and
pop2 were isolated from the pop-adap and pop-NTG cultures by
two rounds of enrichment on pH gradient plates (Table 1). These
acid-tolerant populations were then used as breeding stock for
Genome shuffling. Genome shuffling amplifies the genetic
diversity within a selected population through extensive recombi-
nation between the individual members32. The resulting new pop-
ulation represents a larger combinatorial library of the original
genetic diversity. We shuffled the two pH-tolerant populations
(pop1 and pop2) by means of five rounds of pool-wise recursive
protoplast fusion32. Samples of the regenerated protoplasts from
each round were saved for analysis and then used as a source of
protoplasts for subsequent rounds of fusion. The resulting popu-
lations were screened for individuals with improved acid tolerance
using the acid gradient plates. Figure 1A shows a comparison of
LB-WT, pop1, pop2, and samples from regenerated protoplasts
from the first (F1), third (F3), and fifth (F5) round of fusions. Of
these six populations, cells from the F5 population colonized
medium of the lowest pH, followed, in order of increasing pH, by
the F3,F1,pop2,pop1,and LB-WT.Thus,the shuffled populations
contained new strains that were improved in growth at low pH as
compared with LB-WT, pop1, and pop2, and those that had
undergone more recombination (F5) seemed to contain better
performers than those that had undergone less recombination
(F1 and F3).
To insure that improvements in strains resulted from shuffling
as opposed to recursive protoplast-induced mutagenesis, two sets
of control experiments were carried out. The first measured the
efficiency of homologous recombination effected by protoplast
fusion, and the second measured the contribution of mutagene-
sis resulting from the formation and regeneration of protoplasts.
Recombination efficiency was estimated as 0.1–1% per round of
fusion, using two NTG-induced auxotrophs as marker strains
(data not shown). It should be noted that the measured efficiency
is only an estimate,as one of the auxotrophs was leaky and formed
small colonies on minimal media, thus impairing exact measure-
ment of true prototrophs resulting from fusion. To estimate the
contribution of protoplast-mediated mutagenesis, protoplasts
were formed and regenerated either from LB-WT once or from a
population of lactic acid–tolerant NTG-induced mutants (LAT-1)
five times (without exposure to polyethylene glycol (PEG), which
promotes fusion). The populations resulting from these, LB-WT-
P1 and LAT-P5, were plated on lactic acid gradient plates (YE
containing 0–60 g/l, pH 4.5) alongside LB-WT, LAT-1, and LAT-
F5 (LAT-1 recursively fused five times). The formation and
regeneration of protoplasts of LB-WT was mutagenic, as expect-
ed, and the LB-WT-P1 population grew to approximately the
same region of the plate as LAT-1. Similarly,
LAT-P5 showed a slight improvement in lactic
acid tolerance relative to LAT-1 (likely the
result of additional mutagenesis). However,
LAT-F5 grew to regions of dramatically higher
lactic acid concentration in which all other
populations were completely inhibited. Thus
substantial improvements in the protoplast
populations are associated with shuffling
(recursive PEG-mediated fusion and recombi-
nation) as opposed to recursive protoplast-
induced mutagenesis. This is consistent with
our previous studies32.
Table 1. Lactobacillus strains used in this study
Strain designationHistory of isolation
F5-37, 42, 67, 79, 95
Lactic acid production host from Cargill Dow
Population derived by adaptation of LB-WT to low pH
NTG mutant library of LB-WT
pop-adap enriched on pH gradient plates
pop-NTG enriched on pH gradient plates
Population generated after one round of fusion between pop1 and pop2
Three rounds of fusion between pop1 and pop2
Five rounds of fusion between pop1 and pop2
Colonies from the F5 library selected for shake-flask characterization
Figure 1. Comparison of genome-shuffled libraries.(A) Protoplast fusion libraries
generated after one (F1), three (F3), and five rounds (F5) of recursive protoplast
fusion are compared against initial starting populations pop1 and pop2 on pH
gradient plates.The pH of the gradient plate ranges from pH 6.0 on the left to pH 3.0
on the right. (B) Serial dilutions of LB-WT, pop1, pop2, and F5 were plated onto YEC
medium of defined pH, and the total cfu/ml for each population was determined at
each pH.At pH 3.8, only members of the F5 population survived.
Screening and characterization of individual strains. To
achieve better quantification of the improved pH tolerance of the
shuffled strains, serial dilutions of LB-WT, pop1, pop2, and F5
each were anaerobically incubated on YEC solid medium at pH
4.4, 4.2, 4.0, or 3.8. The total colony-forming units (cfu)/ml of
each population at each pH is shown in Figure 1B.All populations
survived equally well at pH 4.4 and 4.2,although at pH 4.2,LB-WT
colonies were smaller compared with the other populations.At pH
4.0, members from only pop1, pop2, and F5 survived, although
members of pop1 grew as tiny colonies and took
longer to grow as compared with pop2 and F5.At pH
3.8, only members of the F5 population survived.
Thus, consistent with the results of the gradient plate
(Fig. 1A), the shuffled population contained new
strains that grew at pH 3.8, below the pH level
inhibitory to LB-WT, pop1, and pop2.
Characterization of growth and lactic acid produc-
tion. To characterize strain performance in liquid cul-
ture, small-scale fermentations were carried out in YE
medium containing 20 mM citrate buffer (YEC medi-
um) of defined pH in 96-well plates. Cultures were
monitored for cell density and lactic acid production.
Individual isolates from the F5 population growing on
solid medium at pH 4.0 were compared with individual
isolates of the LB-WT strain in liquid medium at pH
4.4 and pH 4.0. Data from a typical 96-well plate com-
paring F5 isolates to LB-WT after 40 h of growth at the
two pH values is shown in Figure 2. Although growth
and lactic acid production by the LB-WT isolates were
similar, these parameters were more diverse for the
members of the F5 population. However, all of the F5
acid-tolerant strains grew to higher optical densities
and produced more lactic acid than did the LB-WT
strain at both pH 4.4 and 4.0. Of the best-performing
shuffled strains, 71 hits were then compared with
LB-WT in 10 ml fermentations at pH 4.0.After 96 h of
growth at pH 4.0, the LB-WT strain had produced just
∼1.6 g/l of lactic acid, whereas the majority of the
F5 hits produced in the range of 4–5 g/l of lactic acid
(Fig.3).The best hit produced >5 g/l lactic acid,a three-
fold improvement over the LB-WT strain.
Shake-flask characterization of isolates from acidic
pH. Five isolates from the 10 ml fermentations were
retested in shake-flask assays against the LB-WT strain
at a pH of 3.9.Consistent with culture tube results,the
isolates from the F5 library produced about two- to
threefold more lactate as compared with LB-WT.
Lactate production also correlated with growth as
shown in Figure 4. Similar observations were also
made for the other pH values tested (data not shown).
After 96 h, the culture was spun down and the end-
point pH of the supernatant was measured. All of the
mutants had reduced the medium pH to 3.5–3.7,
whereas LB-WT effected no detectable change in
medium pH (data not shown).
Continuous monitoring of fermentation pH. One
of the improved acid-tolerant strains,F5-95,was com-
pared with the LB-WT strain in a 1 liter fermenter.The
fermentation was initiated at pH 4.0; medium pH and
total lactic acid concentrations were monitored over
time. The fermentation profiles are shown in Figure 5.
Consistent with the solid and liquid medium results,
the genome-shuffled strain grew faster,produced more
lactic acid, and brought the medium to a lower pH.
Within 10 h, the shuffled strain had produced more lactic acid and
brought the medium pH below that reached by LB-WT at 24 h
(3.75),and by 24 h it had brought the pH to 3.5 and produced more
than twofold more lactic acid.
Low tolerance of extreme process conditions is a common limita-
tion for many industrial biocatalysts. For example, commercial
ethanol- and lactic acid–producing organisms could all benefit
Figure 2. Characterization of libraries in 96-well megatiter plates.LB-WT and individual
isolates from the acid-tolerant F5 population were grown in 96-deep-well fermentations,
and growth and lactate production at pH 4.4 (A) and 4.0 (B) were determined after 40 h.
Growth and lactate production by LB-WT was determined as the average of 96 individual
isolates and is represented as the left bar on each graph.Dotted line represents the LB-WT
level.Coefficient of variance (COV) for OD600and lactate measurements are 12% and 10%,
from improved tolerance towards acids, higher fermentation
product titers, and other toxic compounds present in the fermen-
tation33–37. We used genome shuffling to address this common
challenge and here describe its application to improving the acid
tolerance of a strain of Lactobacillus with commercial potential.
As a single Lactobacillus strain was the starting point of the evolu-
tion program, an improved population was required for breeding.
Classical methods such as chemostat adaptation, NTG mutagene-
sis, and selection on acid gradient plates were sufficient to gener-
ate improved populations or “breeding stocks” of genetically
diverse strains with slight improvements in acid tolerance
(Fig. 1A). Genome shuffling of these populations by five rounds
of recursive pooled protoplast fusion generated a new population
of strains with additional improvements in acid tolerance; the
shuffled population (F5) contained members that grew on solid
medium of pH 3.8 (Fig. 1B). The successive improvement of pop-
ulations that had undergone successively more recombination
(F5 > F3 > F1; Fig. 1A) illuminates the importance of recombina-
tion in the improvement process.
Characterization in small-scale liquid cultures of strains from
F5 growing well on plates at pH 3.8 showed that many grew at up
to 70% higher densities and produced 40% more lactic acid than
did LB-WT at pH 4.4, a pH at which LB-WT grew well. At pH 4.0,
the relative performance of the F5 strains was enhanced, with
many strains growing to fourfold higher densities and producing
twofold more lactic acid. In 10 ml cultures most strains grew to
twice the optical density and many produced over threefold more
Shake-flask comparisons further highlighted the improvements
and the diversity within the selected library.One mutant was partic-
ularly interesting.F5-42 consistently grew to a lower cell density than
F5-79 and F5-95, yet produced more lactic acid. Lactobacillus is
known to have a “two-stage” production mode, one during cell
growth and one after growth halts. Although growth and lactic acid
production correlate well for the LB-WT and the other shuffled
strains, F5-42 maintains its lactic acid productivity after growth
slows. Second-stage lactic acid production can be useful as theoreti-
cally more carbon and energy can be dedicated to lactic acid produc-
tion and export than to cell growth and maintenance.A comparison
of F5-95 to LB-WT in a 1 liter fermentation starting at pH 4.0 with
pH continuously monitored for 24 h clearly demonstrates the
improved acid tolerance of the shuffled strain (Fig. 5). The shuffled
strain grows faster and produces more lactic acid at the initial low
pH,bringing the pH down more quickly and to a pH lower than that
achieved by LB-WT.
A pH of 3.8 is a critical threshold for lactic acid–fermenting
microorganisms as it is the pKaof the acid38. Free lactic acid is
believed to be the ultimate cause of Lactobacillus growth inhibi-
tion at low pH19,38,39, presumably because it can diffuse back into
the cell, short-circuiting proton export and inhibiting enzymes in
the cell. However, more general mechanisms could be contribut-
ing to the inhibitory effect,such as stress responses induced by low
pH of the growth medium40. On the basis of a pKaof 3.8 for lactic
acid, the genome-shuffled mutants are tolerant to approximately
fivefold more of the protonated form of the lactic acid as com-
pared with LB-WT.
The commercial utility of the evolved strains will require fur-
ther characterization under true process conditions. However,
because we maintain a diverse population of improved strains,
further shuffling and evolution of this population can be pursued
to meet a particular commercial goal. It may be possible to isolate
Figure 3. Screening in 10 ml culture tubes.The best-performing members
of the F5 library as determined by the 96-deep-well analysis were further
tested in 10 ml culture tubes at 35°C with mild shaking.Controls consisted
of 24 LB-WT isolates as shown in the figure.Lactate (g/l) and OD600are
reported for cultures growing in YEC at pH 4.0, 96 h after inoculation.Hits
F5-37, -42, -67, -79, and -95 (?) were selected for shake-flask
Figure 4. Fermentations in shake flasks at pH 3.9.Genome-shuffled (F5)
isolates and LB-WT were compared in 250 ml shake-flask cultures
containing 50 ml of YEC medium.Lactate and OD600data were plotted as
averages from four independent shake-flask experiments.The error bars
represent 1 s.d.Final pH for LB-WT was 3.9, whereas that for the mutants
ranged from 3.5 to 3.7.
Figure 5. Characterization of an isolate in a 1 liter bioreactor. LB-WT and
F5-95 were compared in 1 liter fermenters in YEC medium at pH 4.0.
Profiles of pH (continuously monitored) and lactate production (monitored
at 0, 18, and 24 h) were plotted.
and characterize the mutations responsible for improved tolerance
and begin a more rational approach; however, such an approach
would likely be more time-consuming than the shuffling approach
we have taken.
We have recently described the application of genome shuffling
for the improvement of antibiotic production from a commercial
streptomycete32. Here we demonstrate its utility in the improve-
ment of acid tolerance in a poorly characterized strain of
Lactobacillus. Our methodology uses homologous recombination
based on recursive protoplast fusion to facilitate the evolution of
the Lactobacillus population.Although protoplast fusion is broad-
ly applicable for parasexual mating within prokaryotes and
eukaryotes, other recombination formats may be preferred,
depending on the target organism. We have described other
formats based on both homologous and nonhomologous recom-
bination to address a variety of organisms41, and we believe this
technology will prove catalytic in the development of future
Strains.LB-WT is a high-performing Lactobacillus,and was the gift of Cargill
Dow, LLC (Minnetonka, MN). A mutant library of LB-WT was generated
using NTG as the mutagen42.
Media components and composition. All frozen stocks of strains and cul-
tures were propagated on Morrison,Rogosa,and Sharp medium (MRS) agar
plates at 35°C under anaerobic conditions before inoculation into liquid fer-
mentation medium. Yeast extract fermentation medium (YE) contained
100 g/l of glucose and 15 g/l of yeast extract. YEC medium was YE medium
containing 20 mM citrate buffer. pH was controlled using 30% (vol/vol)
ammonium hydroxide (NH4OH) and 4 M hydrochloric acid (HCl). MRS
Broth was obtained from Becton Dickinson (Sparks, MD); BSA, lysozyme,
mutanolysin, and gelatin were purchased from Sigma (St. Louis, MO).
Polyethylene glycol (PEG) 6000 was purchased from Fluka Chemicals
(Milwaukee,WI). Regeneration medium (RM) consisted of MRS agar medi-
um supplemented with 20 mM magnesium chloride, 2.5% gelatin, 0.5 M
sucrose,and 10.5% BSA.
Assay for lactic acid and glucose. Lactic acid in culture supernatants was
spectrophotometrically measured by coupling lactate oxidase and peroxidase
and following the peroxidase-catalyzed oxidative condensation of a chro-
mogen (Sigma Diagnostics, St. Louis, MO). L-Lactate in the fermentation
broths was also measured using the YSI 2700 Select Biochemistry Analyzer
(YSI, Yellow Springs, OH). All dilutions and liquid handling for high-
throughput (HTP) screens were robotically automated.
Chemostat adaptation of LB-WT to low pH. Draw–fill semicontinuous fer-
mentations were carried out in a BioFlow-2000 fermentation system from
New Brunswick Scientific (NBS, Edison, NJ). Culture adaptation was
achieved through a continuous and controlled decrease in medium pH. The
working volume of the reactor was set at 900 ml, and the feed contained YE
medium. pH was controlled by adding 30% (vol/vol) NH4OH and 4 M HCl
For scenarios in which the imposed pH was not tolerated, and the culture
steady state could not be maintained, the fermentation was switched to a
fill–draw mode of operation. Here, 90% of the culture was removed and
replaced with fresh YE medium of lower pH. The culture was allowed to
adapt in batch mode.Samples were removed periodically to follow OD600and
stored as frozen stocks to maintain genetic diversity for genome shuffling.
Enrichments using pH gradient plates. pH gradient agar plates were used to
enrich for mutants of LB-WT that could grow at an acidic pH inhibitory to
LB-WT. Two layers of solid medium (150 ml of YEC–agar) differing in their
initial pH (3.0 or 6.0) were used to create the gradient.The acidic lower layer
was poured first into a rectangular plate raised on one side by ∼0.5 cm. The
solidified medium formed a wedge.The plate was returned to a level position,
and the neutral upper layer was then poured onto the solidified lower layer
such that it formed a complementary wedge (reminiscent of the yin–yang
pattern).The plate was incubated at room temperature overnight to establish
the pH gradient across the plate. The maintenance of the gradient over the
course of the experiment was monitored on control plates using pH paper.
The solid gradient plates were divided into six lanes using sterilized plastic
dividers, and 100 µl of different culture samples were compared side by side
in each lane.
Genome shuffling. Cells were cultured overnight at 35°C in MRS broth in
a 50 ml screw-cap tube,subcultured to an OD600of 0.05 in MRS broth con-
taining 1.2% glycine, and then incubated for 12–18 h at 35°C to OD600> 5
before harvesting for the preparation of protoplasts. The method for gen-
erating protoplasts was essentially as described by Cocconcelli et al.43,44,
except that enzyme treatment was for 2 h with 10 mg/ml of lysozyme in
Lactobacillus protoplasting buffer (LBP), which consisted of 10 mM Tris-
HCl, pH 6.3, containing 20 mM CaCl2and 0.5 M sucrose. The appearance
of spherical cells as judged by light microscopy was used as an indicator of
To generate fusions between protoplast preparations, an approximately
equal number of protoplasts from different populations were mixed, cen-
trifuged, and resuspended in 50 µl LBP. Nine volumes of 60% PEG 6000 in
LBP were added to the resuspended protoplast mix and incubated for 6 min
at room temperature. LBP was added to 5 ml, and the fused protoplast
preparation was centrifuged, washed with LBP, resuspended in 100 µl LBP,
and serial dilutions were anaerobically regenerated on RM at 35°C. The
plates were scraped to generate a pooled fusion library. The formation of
protoplasts, their fusion, and subsequent regeneration was repeated five
times with the pooled regenerated cells from one fusion being the inoculum
for the subsequent protoplast culture. Samples from each of the pooled
fusion libraries (F1–F5) were saved for later analysis. Nonshuffled controls
were prepared by the recursive formation and regeneration of protoplasts
without exposure to PEG.
Small-scale fermentations. HTP fermentations were initiated by robotically
picking colonies from solid YE agar plates into 96-deep-well plates containing
1.5 ml of rich MRS medium.The plates were anaerobically incubated at 35°C
for 48–60 h, and 40 µl (1:50 dilution) was used to inoculate similar cultures
containing 2 ml of YEC at pH 4.4 and 4.0. Profiles of growth and lactic acid
production for each culture were then followed.Periodically,150 µl of culture
was removed,OD600was read,the sample was centrifuged,and the lactic acid
content of the supernatant was determined.
Shake flask analysis. Colonies from MRS plates were used to inoculate
overnight cultures in MRS medium.YEC medium (50 ml,pH 3.9) in capped
shake flasks was inoculated with 500 µl of the overnight cultures and shaken
at 65 rpm (orbital shaker with 2.5 cm throw) and 35° C. Lactate and growth
were periodically monitored up to 96 h. Each strain was grown in at least
three separate shake flasks; the data for each time point was averaged and is
shown in Figure 4.
Bioreactor fermentation. Characterization of strains at 1 liter scale was
carried out in the BioFlow-2000 fermenter. Fermentation was initiated by
inoculating 20 ml of an overnight culture in MRS into 1 liter of YEC medi-
um, pH 4.0. The fermenters were maintained at 35°C with an agitation
speed of 100 rpm. The pH of the culture was monitored and recorded
automatically throughout the duration of the fermentation. The cultures
were periodically sparged with nitrogen to maintain an anaerobic environ-
ment. Samples were periodically drawn from the reactors and monitored
We thank Keith Powell and Tony Cox for their useful suggestions and discus-
sions and Ken Zahn for proofreading the manuscript.Special thanks go to Amy
Giver,Henry Garcia,and Rob Pak for technical support.Funding for this pro-
ject was provided in part by the US National Institute of Standards and
Technology Advanced Technology Program.
Competing interests statement
The authors declare competing financial interests: see the Nature
Biotechnology website (http://biotech.nature.com) for details.
Received 16 October 2001; accepted 9 May 2002
1. Choi, J. & Lee, S.Y. Economic considerations in the production of poly(3-hydroxy-
butyrate-co-3-hydroxyvalerate) by bacterial fermentation.
Biotechnol. 53, 646–649 (2000).
2. Hashimoto, S. & Ozaki, A. Whole microbial cell processes for manufacturing
amino acids, vitamins or ribonucleotides. Curr. Opin. Biotechnol. 10, 604–608
3. Ho, N.W., Chen, Z., Brainard, A.P. & Sedlak, M. Successful design and develop-
ment of genetically engineered Saccharomyces yeasts for effective cofermenta-
tion of glucose and xylose from cellulosic biomass to fuel ethanol. Adv. Biochem.
Eng.Biotechnol.65, 163–192 (1999).
4. Ikeda, M. & Katsumata, R. Hyperproduction of tryptophan by Corynebacterium
glutamicum with the modified pentose pathway. Appl. Environ. Microbiol. 65,
5. Katsumata, R. & Ikeda, M. Hyperproduction of tryptophan in Corynebacterium
glutamicum by pathway engineering.Biotechnology (NY) 11, 921–925 (1993).
6. Perkins, J.B. et al. Genetic engineering of Bacillus subtilis for the commercial pro-
duction of riboflavin.J.Ind.Microbiol.Biotechnol. 22, 8–18 (1999).
7. Aristidou, A. & Penttila, M. Metabolic engineering applications to renewable
resource utilization.Curr.Opin.Biotechnol.11, 187–198 (2000).
8. Braunegg, G., Lefebvre, G. & Genser, K.F. Polyhydroxyalkanoates, biopolyesters
from renewable resources: physiological and engineering aspects. J. Biotechnol.
65, 127–161 (1998).
9. Chontani, G. et al. The commercial production of chemicals using pathway engi-
neering.Biochim.Biophys.Acta 1543, 434–455 (2000).
10. Donnelly, M.I., Millard, C.S., Clark, D.P., Chen, M.J. & Rathke, J.W. A novel fer-
mentation pathway in an Escherichia coli mutant producing succinic acid, acetic
acid, and ethanol.Appl.Biochem.Biotechnol.70–72, 187–198 (1998).
11. Ho, N.W., Chen, Z.& Brainard, A.P.Genetically engineered Saccharomyces yeast
capable of effective cofermentation of glucose and xylose. Appl. Environ.
Microbiol.64, 1852–1859 (1998).
12. Kautola, H., Vassilev, N. & Linko, Y.Y. Continuous itaconic acid production by
immobilized biocatalysts.J.Biotechnol.13, 315–323 (1990).
13. Lee, S.Y.& Choi, J.I.Production of microbial polyester by fermentation of recombi-
nant microorganisms.Adv.Biochem.Eng.Biotechnol.71, 183–207 (2001).
14. Park, J.H., Kim, G.J. & Kim, H.S. Production of D-amino acid using whole cells of
recombinant Escherichia coli with separately and coexpressed D-hydantoinase
and N-carbamoylase.Biotechnol.Prog.16, 564–570 (2000).
15. Schulze, B. & Wubbolts, M.G. Biocatalysis for industrial production of fine chemi-
cals.Curr.Opin.Biotechnol.10, 609–615 (1999).
16. Gong, C.S., Cao, N.J., Du, J. & Tsao, G.T. Ethanol production from renewable
resources.Adv.Biochem.Eng.Biotechnol.65, 207–241 (1999).
17. McLaren, J. & Faulkner, D.The technology roadmap for plant/crop-based renew-
able resources 2020 (DOE/GO-10099-706) 1–41 (US Department of Energy,
Indianapolis, IN,) 1999.
18. Varadarajan, S. & Miller, D.J. Catalytic upgrading of fermentation-derived organic
acids.Biotechnol.Prog. 15, 845–854 (1999).
19. Porro, D. et al. Replacement of a metabolic pathway for large-scale production of
lactic acid from engineered yeasts. Appl. Environ. Microbiol. 65, 4211–4215
20. Baniel, A.M.et al.Process for isolating lactic acid.US patent 6,087,532 (2000).
21. Rallu, F., Gruss, A., Ehrlich, S.D. & Maguin, E. Acid- and multistress-resistant
mutants of Lactococcus lactis: identification of intracellular stress signals. Mol.
Microbiol. 35, 517–528 (2000).
22. De Angelis, M., Bini, L., Pallini, V., Cocconcelli, P.S.& Gobbetti, M.The acid-stress
response in Lactobacillus sanfrancisscensis CB1. Microbiology 147, 1863–1873
23. Arnold, C.N., McElhanon, J., Lee, A., Leonhart, R.& Siegele, D.A.Global analysis
of Escherichia coli gene expression during the acetate-induced acid tolerance
response.J.Bacteriol.183, 2178–2186 (2001).
24. Audia, J.P., Webb, C.C. & Foster, J.W. Breaking through the acid barrier: an
orchestrated response to proton stress by enteric bacteria. Int. J. Med. Microbiol.
291, 97–106 (2001).
25. Arnold, F.H. & Volkov, A.A. Directed evolution of biocatalysts. Curr. Opin. Chem.
Biol.3, 54–59 (1999).
26. Crameri, A., Raillard, S.A., Bermudez, E.& Stemmer, W.P.DNA shuffling of a fam-
ily of genes from diverse species accelerates directed evolution. Nature 391,
27. Jaeger, K.E., Eggert, T., Eipper, A.& Reetz, M.T.Directed evolution and the creation
of enantioselective biocatalysts.Appl.Microbiol.Biotechnol.55, 519–530 (2001).
28. Ness, J.E., Del Cardayré, S.B., Minshull, J. & Stemmer, W.P. Molecular breeding:
the natural approach to protein design.Adv.Protein.Chem.55, 261–292 (2000).
29. Schmidt-Dannert, C., Umeno, D. & Arnold, F.H. Molecular breeding of carotenoid
biosynthetic pathways.Nat.Biotechnol.18, 750–753 (2000).
30. Soong, N.W.et al. Molecular breeding of viruses.Nat.Genet.25, 436–439 (2000).
31. Wackett, L.P. Directed evolution of new enzymes and pathways for environmental
biocatalysis.Ann.NY Acad.Sci.864, 142–152 (1998).
32. Zhang, Y., Perry, K., Powell, K., Stemmer, P.C. & del Cardayré, S.B. Evolution of
Streptomyces fradiae by whole genome shuffling.Nature 415, 644–646 (2002).
33. Larsson, S., Quintana-Sainz, A., Reimann, A., Nilvebrant, N.O. & Jonsson, L.J.
Influence of lignocellulose-derived aromatic compounds on oxygen-limited
growth and ethanolic fermentation by Saccharomyces cerevisiae. Appl. Biochem.
Biotechnol.84–86, 617–632 (2000).
34. Martinez, A., Rodriguez, M.E., York, S.W., Preston, J.F. & Ingram, L.O. Effects of
Ca(OH)2treatments (“overliming”) on the composition and toxicity of bagasse
hemicellulose hydrolysates.Biotechnol.Bioeng.69, 526–536 (2000).
35. Lawford, H.G. & Rousseau, J.D. Improving fermentation performance of recombi-
nant Zymomonas in acetic acid-containing media. Appl. Biochem. Biotechnol.
70–72, 161–172 (1998).
36. Zaldivar, J., Martinez, A. & Ingram, L.O. Effect of alcohol compounds found in
hemicellulose hydrolysate on the growth and fermentation of ethanologenic
Escherichia coli.Biotechnol.Bioeng.68, 524–530 (2000).
37. Gonzalez-Vara, Y.R.A.et al. Enhanced production of L-(+)-lactic acid in chemostat
by Lactobacillus casei DSM 20011 using ion-exchange resins and cross-flow fil-
tration in a fully automated pilot plant controlled via NIR. Biotechnol. Bioeng. 67,
38. Benninga, H.A. A History of Lactic Acid Making (Kluyver Academic Publishers,
Dordrecht, The Netherlands, 1990).
39. Hongo, M., Nomura, Y. & Iwahara, M. Novel methods of lactic production by elec-
trodialysis fermentation.Appl.Environ.Microbiol. 32, 227–234 (1986).
40. O’Sullivan, E.& Condon, S.Intracellular pH is a major factor in the induction of tol-
erance to acid and other stresses in Lactococcus lactis. Appl. Environ. Microbiol.
63, 4210–4215 (1997).
41. Del Cardayré, S.B.et al.Molecular breeding of transposable elements.PCT appli-
cation WO 02/04629 A (2002).
42. Miller, J.H. in Experiments in Molecular Genetics. (ed. Miller, J.H.) 125–129 (Cold
Spring Harbor Laboratory, Cold Spring Harbor, NY, 1972).
43. Cocconcelli, P.S., Morelli, L., Vescovo, M. & Bottazzi, V. Intergeneric protoplast
fusion in lactic acid bacteria.FEMS Microbiol.Lett. 35, 211–214 (1986).
44. Morelli, L. et al. Lactobacillus protoplast transformation. Plasmid 17, 73–75