Characterization, performance, and applications of a yeast surface display-based biocatalyst

Article abstract of DOI:10.1039/c4ra16304d

This work demonstrates the efficacy and cost effectiveness of yeast surface display (YSD) as a method for producing and purifying enzyme catalysts. Lipase B from Candida antarctica (CalB) and lipase from Photobacterium lipolyticum sp. M37 (M37L) were individually displayed on the surface of yeasts via fusion with alpha-agglutinin. The enzyme is produced, purified, and immobilized in a single step. The population expressing the enzyme was quantified by flow cytometry. After lyophilization, the hydrolytic activity of the biocatalyst was assayed with p-nitrophenyl butyrate and p-nitrophenyl palmitate substrates. Esterification reactions involving octanoic acid and either butanol or octanol were used to evaluate esterification activity. The lyophilized YSD biocatalyst hydrolytic activity matched or exceeded commercial lipase (Novozym 435) immobilized on acrylic resin at equal catalyst loading, and achieved esterification levels 10-50% that of Novozyme 435. Factoring in the cost of production, the YSD biocatalyst represents a considerable savings over traditionally prepared and purchased enzyme catalysts. This promises to significantly expand the catalytic applications of immobilized lipases, and immobilized enzymes more generally, in commercial processes. This journal is

Full text of DOI:10.1039/c4ra16304d

RSC Advances
View Article Online
View Journal | View Issue
PAPER
Characterization, performance, and applications of
a yeast surface display-based biocatalyst†
Cite this: RSC Adv., 2015, 5, 19166
J. M. Eby and S. W. Peretti*
This work demonstrates the efficacy and cost effectiveness of yeast surface display (YSD) as a method for
producing and purifying enzyme catalysts. Lipase B from Candida antarctica (CalB) and lipase from
Photobacterium lipolyticum sp. M37 (M37L) were individually displayed on the surface of yeasts via fusion
with alpha-agglutinin. The enzyme is produced, purified, and immobilized in a single step. The
population expressing the enzyme was quantified by flow cytometry. After lyophilization, the hydrolytic
activity of the biocatalyst was assayed with p-nitrophenyl butyrate and p-nitrophenyl palmitate
substrates. Esterification reactions involving octanoic acid and either butanol or octanol were used to
evaluate esterification activity. The lyophilized YSD biocatalyst hydrolytic activity matched or exceeded
commercial lipase (Novozym 435) immobilized on acrylic resin at equal catalyst loading, and achieved
esterification levels 10–50% that of Novozyme 435. Factoring in the cost of production, the YSD
biocatalyst represents a considerable savings over traditionally prepared and purchased enzyme
catalysts. This promises to significantly expand the catalytic applications of immobilized lipases, and
immobilized enzymes more generally, in commercial processes.
Received 12th December 2014
Accepted 6th February 2015
DOI: 10.1039/c4ra16304d
www.rsc.org/advances
been displayed on the yeast surface and retained their func-
tion, demonstrating the capacity and exibility of the system.
Introduction
The cost of puried enzyme is a primary obstacle to its
Lipases are of particular utility in surface display systems
commercial implementation. Despite the variety of ways that for several reasons: they catalyze many synthetic reactions;
1
enzymes can be prepared for industrial use, up to 80% of the they are relatively thermostable and promiscuous; they require
2,3

nal cost is associated with purication and immobilization.
no cofactors or regeneration; they are stable and active in
Display of the enzyme on the surface of the production some organic solvents; and they are active extracellularly
1
0,11
organism offers the potential for dramatic decreases in cost without modication.
Lipase B from Candida antarctica
and catalyst preparation time. While yeast surface display (CalB) is a widely-studied industrial enzyme whose crystal
1
2
(
YSD) was developed to succeed phage and bacterial display as structure has been solved, but whose application to esteri-
4
a tool for protein engineering, parallel applications in bio- cation is somewhat limited by its susceptibility to methanol
5
13
catalysis were quickly identied. Enzymes displayed on the deactivation. Recently, a lipase, M37L, from a psychrophilic
yeast cell surface take advantage of the yeasts' production bacterium (P. lipolyticum sp. M37) with better tolerance of high
1
4
efficiency and selective protein display. Using the yeast as the methanol concentrations, has been applied to a biodiesel
1
5
production host and as the nal catalytic support circumvents reaction. Most recently, directed evolution was used to
the most costly steps in enzyme preparation. An additional dramatically enhance methanol tolerance and thermal
1
6
benet comes from the fungal cell wall, a rigid structure stability for a lipase from Proteus mirabilis.
6
composed of glucans, mannoproteins, and chitin, which
Investigators have had success immobilizing lipases using
offers physically and chemically robust support. The two most surface display in bacteria, lamentous fungi, and yeast.
prevalent and exible methods for displaying proteins on the Bielen et al. published a systematic review of surface display as it
yeast surface employ either an a-agglutinin fusion (covalent applies to lipolytic enzymes. The use of yeast, especially
cell wall attachment ) or fusion to a truncated yeast occulin Saccharomyces sp., is appealing because of the genus' rapid
17
18
19–21
22
4
7
protein (non-covalent adhesion ). Proteins ranging in size from growth, mapped genome and proteome, and facile genetic
a DNA binding motif (7 kDa) to b-glucosidase (136 kDa) have manipulation. Many fungal lipases have been displayed on yeast,
8
9
23
20,24
including lipases from Rhizomucor miehei, Rhizopus oryzae,
25,26
and Candida antarctica.
North Carolina State University, Department of Chemical and Biomolecular
Engineering, 911 Partners Way, Raleigh, NC 27695-7905, USA. E-mail: peretti@
ncsu.edu
The activity of surface-displayed M37L is reported here for
the rst time. The previous attempt at immobilization used
enzyme cross-linking, and the aggregate produced methyl
†
Electronic supplementary information (ESI) available. See DOI:
0.1039/c4ra16304d
1
19166 | RSC Adv., 2015, 5, 19166–19175
This journal is © The Royal Society of Chemistry 2015

View Article Online
Paper
esters from olive oil at 70% conversion. Since immobiliza-
RSC Advances
2
7
1
9
tion can oen result in an increase in stability and activity,
M37L and wild-type CalB have been incorporated into a YSD
fusion for comparison.
Activity is reported here in terms of conversion, normalized
by catalyst dry mass. Activity for YSD-based biocatalysts is
typically reported as a bulk property, in arbitrary units that
neglect catalyst loading, making replication of the results
difficult. These characterizations ignore the dynamics of the
production organism and rely on commonly cited assumptions
1
9,28
about the amount of catalytic protein being produced.
Here,
analysis of the population by ow cytometry has been applied
to optimize the production of the biocatalyst towards
uniformly high surface display of the lipase.
Results
Flow cytometry
Fig. 1 SS vs. FS plots for the ycM37L yeast biocatalyst. Intact cells are
gated based on their characteristic size and granularity (P2), based on
the unlabeled sample (A), and the gate is applied to the histograms in
Fig. 2–4. Small changes in the population's physical characteristics
upon lyophilization (B) are discussed in the text. Immuno-labeling of
the surface construct had no significant effect on either the live (C) or
Cells are selected for analysis according to their forward scatter
and side scatter distributions (roughly correlated to size and
granularity, respectively). In a linear side scatter vs. forward scatter
(SS vs. FS) plot, $80% of the total population is typically gated for
analysis (P2 in Fig. 1). The gated population in P2 decreases the lyophilized cells (D).
approximately 10% aer lyophilization, indicating a partial loss of
cell integrity during the process. Aer lyophilization, a new
Fig. 2 Representative histograms for the yeast biocatalysts, with the percentage of the population above the vertical gate at bottom right. The
population gated in Fig. 1 is labeled with antibodies specific to the C-terminal epitope in the surface construct. From histograms of fluorescence
intensity, the background is determined by setting a gate such that less than 1% of an unlabeled control lies to the right. The population with
higher fluorescence (to the right of the gate) than the control is expressing the SD construct, including the lipase. The population with higher
fluorescence than the control (to the right of the gate) is expressing the SD construct, including the lipase. Background from labeling with only
the secondary antibody is low (A–D). The majority of the population is expressing the surface construct before and after lyophilization (E–H). The
histograms shown are from four independent cultures and cytometry experiments. A reduction in the fraction expressing the construct is
observed if culture conditions are sub-optimal (E, shown for demonstration).
This journal is © The Royal Society of Chemistry 2015
RSC Adv., 2015, 5, 19166–19175 | 19167

View Article Online
RSC Advances
Paper
Scheme 1
c-myc) aer two rounds of cell sorting (Fig. 3). The plasmid
and insert were present and the DNA sequence was conrmed.
The likely cause of failure to detect the construct is over-
saturation of the cellular translation and secretion machinery.
Because little-to-no fusion protein was detected on the cell
surface, the high-copy cells were not tested for lipase activity.
Fig. 3 Histograms for the YSD CalB biocatalyst expressed in a high-
copy vectors. The fraction of the population expressing the construct
after one round (C) or two rounds of sorting (D) is not appreciably
Copy number
higher than untransformed MT8-1 cells (A) or cells labeled without the The measured copy numbers on lyophilized CalB- and ycM37L-
primary antibody (B). The same trend is observed in cells displaying displaying yeast cells are given in Table 1. Approximately 48 000
ycM37L.
and 23 000 copies of the SD construct were observed on the CalB
and ycM37L biocatalysts, respectively.
population appears in the lower le of the SS vs. FS plots, indi-
cating an increase in cell debris (highlighted by a green polygon,
Fig. 1D). The lyophilized cells are permeable to propidium
iodide, which further supports the theory that losses are due to
physical degradation. For single-copy plasmids (pCT-CalB and
pCT-ycM37L), the construct is detected in >69% of the gated
population under the culture and lyophilization conditions
described below (Fig. 2F and H). Aer lyophilization, the
expressing population is typically $90% of that before lyophili-
ꢀ
zation. Storage of the cells in buffer or in culture medium at 4 C
for more than 24 hours also results in a loss of the surface
construct, which suggests that lyophilization is preferable for
long-term storage.
Codon optimization of M37L increased the population
expressing detectable protein on the surface from <1% (ESI†) to the
same range as that of CalB (Fig. 2). However, the average uores-
cent intensity in the ycM37L-expressing population is routinely 2-
to 4-fold lower than that of the CalB-expressing population.
For the high-copy plasmid, no greater than 1% of the pop-
ulation was observed to express the SD fusion (Flo1S-lipase-
Table 1 Observed copy number of lipase on YSD biocatalyst. Values
were determined by interpolating a standard curve, and are the Fig. 4 Assay data for hydrolysis of p-nitrophenyl butyrate at 30 C (A)
average of five samples prepared in parallel
ꢀ
ꢀ
and p-nitrophenyl palmitate at 45 C (B). The activity of either YSD
biocatalyst is comparable to that of the standard on a mass basis. When
Average copy
Standard deviation assay data are normalized by estimated protein loading (C and D), the
Enzyme displayed
number
(N ¼ 5)
specific activities of the YSD biocatalysts are much higher than that of
the resin. Values are average of three samples; error bars are standard
deviations for the same samples. The values for ycM37L and CalB
represent the respective lipases immobilized via YSD. EBY100 is
untransformed, induced yeast.
CalB
ycM37L
48 423
22 739
5983
1314
19168 | RSC Adv., 2015, 5, 19166–19175
This journal is © The Royal Society of Chemistry 2015

View Article Online
Paper
RSC Advances
Hydrolysis activity assays
The yeast biocatalyst successfully produces the indicator (p-nitro-
phenol) from both small (butyrate, pNPB) and large (palmitate,
pNPP) nitrophenol esters (Scheme 1), as demonstrated in Fig. 4.
When used at the same mass loading as a commercial lipase
immobilized on a macroporous resin, the performance of the YSD
biocatalyst is comparable to or better than the commercial catalyst
under the conditions studied. The conversion with lipase-
displaying yeast cells is signicantly higher than untransformed
cells and is higher than that of cells expressing a non-catalytic
protein in the construct (EGFP, ESI†).
The temperature dependence of the yeast biocatalyst's
performance is similar to lipase immobilized on other
supports. Cells displaying CalB have an optimum temperature
ꢀ
Fig. 6 Use of yeast biocatalyst in multiple assays at 30 C. Cells and
resin were washed twice with 1 mL potassium phosphate buffer
ꢀ
around 40 C, which aligns with literature reports of a Topt for
ꢀ
between rounds. Pelleted cells and resin were stored at 4 C overnight
ꢀ
19
the enzyme between 40 and 50 C. Cells displaying ycM37L
after rounds 2 and 4. Conversion (bars) is the average of three
have a similar temperature prole, conicting with reports in experiments. The dry masses of the resin and YSD catalysts (squares
ꢀ
29
and triangles, respectively) were measured before and after the
experiment.
the literature of a Topt near 25 C for free enzyme, which
1
suggests that immobilization has a stabilizing effect. The
lower activity at 30 is due to a twofold increase in micelle
ꢀ
diameter (from 20 to 30 degrees), which results in a much lower
accessible surface area for the reaction. This was conrmed by
combination of heat and detergent begin to disrupt the yeast
cell and denature the enzyme. This effect leads to the steady
rise in activity from the untransformed cells, as well as the
decrease in activity from both the experimental and control
samples (Fig. 5A). For this reason, pNPB assays were run at
lower temperatures, while the solubility of p-NPP dictated a
higher temperature.
dynamic light scattering experiments with the assay emulsion
ꢀ
(
ESI†). At 40 C and higher, the emulsion is unordered,
rendering micelle size irrelevant. Adding iso-propanol to the
assay at 5% by volume stabilizes the micelles, and a steady
increase in activity is observed between 20 and 30 degrees
ꢀ
(
Fig. 5B). Above 40 C, with and without iso-propanol, a
The yeast biocatalyst can be re-used in the pNPB assay. Cells
were washed with potassium phosphate buffer between assays
and recovered by centrifugation. While loss of the biocatalyst
mass was observed in the wash and recovery steps, little
decrease in conversion over multiple runs was observed
(Fig. 6). The conversion is shown normalized by catalyst
loading in Fig. 7.
Esterication activity assays
Esterication reactions are of greater commercial importance;
once the hydrolysis performance of the two YSD systems was
Fig. 5 Hydrolysis of p-nitrophenyl butyrate at different temperatures.
A: without alcohol added. B: assay with 5% v/v iso-propanol added.
CalB: diamonds, ycM37L: triangles, untransformed EBY100: squares, Fig. 7 Use of yeast biocatalyst in multiple assays at 30 C. Conversions
ꢀ
macroporous resin-immobilized lipase: circles.
(Fig. 7) were normalized with estimated catalyst dry mass.
This journal is © The Royal Society of Chemistry 2015
RSC Adv., 2015, 5, 19166–19175 | 19169

View Article Online
RSC Advances
Paper
benchmarked relative to Novozym 435 (N435), an esterication
reaction was investigated to establish the relevance of the YSD
approach. Under the conditions studied, both YSD CalB and
ycM37L produced the butyl or octyl esters. Conversion of n-
butanol and fatty acids of varying length to their respective esters
is presented in Fig. 8A. The YSD biocatalysts displaying CalB and
ycM37L exhibited similar preference for acyl donor chain length,
with the highest conversion to the butyl ester from the shortest
fatty acid (octanoic acid, C8). Further tests of the catalysts' effi-
cacy with a water-insoluble alcohol (1-octanol) also yielded an
octanoate ester (Fig. 8B). In contrast to the hydrolysis reactions,
the untransformed control did not consume any fatty acid. The
commercial resin apparently reached equilibrium within six
hours regardless of substrate; that is, no further fatty acid was
consumed. Based on the extent of the octyl octanoate reaction
before equilibrium, the turnover rate for N435 is 15–20 times
that of the YSD biocatalyst.
Discussion
Flow cytometry
Under the culture conditions routinely used to produce the
biocatalyst, the majority (>60%) of cells carrying the single-
copy vectors are carrying the SD construct. Sub-optimal
lyophilization conditions (inadequate vacuum, solids loading
too high or too low, or temperature too high), reduce the SD-
carrying population fraction by approximately 10%.
However, reductions of the same magnitude are observed aer
ꢀ
storage of the live cells at 4 C aer only three days, suggesting
that lyophilization leads to better long-term stability. A similar
decrease is observed in the number of cells that fall into the
“
normal, healthy” gate (P2 in Fig. 1). It is assumed that the
reduction is due to cell lysis or physical deformation, and that
the impact is the same for lipase-displaying and non-
displaying cells. Optimized lyophilization reduces the SD-
carrying population fraction less than 5% (ESI†). However,
all lyophilized cells are permeable to propidium iodide, which
suggests partial disruption of the cell membrane and cell wall.
Lyophilization's impact on the activity of the biocatalyst as a
bulk material is not clear, but it is assumed that less damage to
the cells corresponds to less damage to the surface-displayed
protein. Lyophilization with a cryoprotectant may improve
cell integrity, but additives would likely complicate down-
stream applications.
The observed copy number agrees with previously repor-
5
28
ted values for the Aga2-based SD system (5 ꢁ 10 ). The
amount of enzyme in the bulk YSD biocatalyst was calcu-
lated, based on the factors given in Table 2. Molecular
weights for the enzymes were calculated from their putative
amino acid sequences. The number of cells per gram of dry
weight (g dw) was taken from vendor estimates for brewers'
Fig. 8 The YSD biocatalyst synthesized esters of saturated fatty acids
and butanol (A) or octanol (B and C). Dry catalyst was added to reaction
vials. A solution of alcohol and fatty acid (both 100 mM in heptane) was yeast. The expressing population fractions were taken from
added. Water (0.5% v/v [A] or 0.4% v/v [C]) was slowly added. Samples
were taken after six hours (A) or as indicated (B and C). Data with error
bars are averages of at least three trials, and standard deviations are
shown as error bars.
Fig. 2, and those fractions are assumed to display the average
number of enzymes. The copy number and protein loading of
the YSD ycM37L biocatalyst was approximately half that of
YSD CalB.
19170 | RSC Adv., 2015, 5, 19166–19175
This journal is © The Royal Society of Chemistry 2015

View Article Online
Paper
RSC Advances
Table 2 Estimated enzyme loading in YSD biocatalyst. Calculation of protein in the biocatalyst (in mg of protein per gram dry weight of bio-
catalyst) is based on flow cytometry data for copy number and fraction of the population expressing the construct
Molecular weight
Da)
Enzymes per
cell
Number of cells
expressing
mg protein per
g dw biocatalyst
(
Cells per g dw
1
0
0
CalB
ycM37L
33 030
38 026
48 422
22 739
70%
65%
2.00 ꢁ 10
2.00 ꢁ 10
0.037
0.019
1
SD expression
where the activity of the control cells approaches that of the
lipase-displaying cells. Presumably, the cells are lysing and
releasing endogenous hydrolases and other lytic proteins into
the local environment. Judging by the performance of the
untransformed cells here, that threshold is above the upper
limit of the proteins' thermal stability, and will not signicantly
affect the performance of the yeast biocatalyst in most
applications.
The multi-copy vectors, under the control of a strong promoter,
appear to have over-burdened the cells' synthesis and secretion
pathways (Fig. 3). Weaker promoters or smaller fusion
constructs may improve expression; however, others reported
success with a YSD biocatalyst using Flo1S in a high-copy vector
7
under the control of an isocitrate lyase promoter, a promoter of
30
comparable strength to Gal7. With no quantitative surface
coverage or population data, comparisons with these results are
without merit.
The resin (N435) produces esters of butanol and octanol
more quickly, irrespective of the fatty acid or alcohol chain
length. No acyl donor size preference is discernible for
N435 because the reactions reach equilibrium before the six-
hour sampling time. The addition of water to the reaction
hydrates the enzyme, but also causes aggregation of the cells.
The additional water is necessary for ycM37L to function in an
Activity assays
The YSD biocatalyst performs as well as commercial lipase
immobilized on a macroporous resin in the assays under the
conditions tested. The activity of CalB and ycM37L are compa-
31
organic environment, but not for CalB. The necessity of water
15
rable, as previously reported for free enzyme. The activity has
been normalized with the mass of the catalyst so that compar-
isons with other lipase-based catalysts may be made on an equal
basis (catalyst mass loading). Since the protein is immobilized
on an insoluble particle, recovery of the catalyst is a matter of
simple centrifugation or ltration, both of which are routinely
scaled to industrial applications. The catalyst's reusability was
demonstrated here. The robustness of the yeast cell as a support
is underpinned by the durability of the polysaccharide and
mannoprotein cell wall, which is tolerant to solvents as well as
relevant temperature and pH ranges. However, there will be
some threshold where the cell support begins to break down,
for ycM37L activity is apparent in Fig. 8C when compared to
Fig. 8B. Water also benets CalB to a lesser extent, as a slight
improvement is observed between both the YSD CalB- and
N435-catalyzed reactions (Fig. 8B and C). The octyl octanoate
reaction reaches equilibrium within 2 hours if catalysed by
N435, while the YSD biocatalysts need 24–48 hours (data for
longer experiments not shown). The reaction is signicantly
impeded if cells aggregate to an extent that their uniform
suspension in the reaction suffers (see CalB results in Fig. 8B,
which suffered from catalyst clumping). The disparity between
the esterication activity of the YSD biocatalysts and N435 is
Table 3 Cost estimates for production of the YSD biocatalyst
Current (lab-scale) production
Theoretical production
a
e
Nutrient requirement
Current price
Bulk cost
Bulk cost
%8.33
%1.61
%0.19
%108
Nutrient source
Galactose
40 g/batch
10 g/batch
3.4 g/batch
4 g/batch
4 L/batch
%15 per kg
%50 per kg
%1 per g
%310 per kg
<%2 per MT
%50
%42
Beet sugar, %2.50 per kg
Fish meal, %0.8 per kg
Ammonium sulfate
Nitrogen/vitamin base
Amino acids
%283
%103
%0.19
%478
%108
%4.60
z0
Water (sterile + disposal)
Media
b
Lyophilization
ꢀ
c
Coolant, ꢂ50 C
%13.11 per GJ
%0.35 per std m
%0.06 per kWh
3
Plant air
140 L chamber
d70% efficiency
Electricity (3/4 hp vacuum)
Maintenance
Total
%3.45
%100
%10 000 per year
%586
%118
a
b
Assumes yield of 30 g dry cells per 100 g of galactose, scaled up for 1 kg dry cells. Assumes single 3 day cycle for 1 kg of dry cells, dried to 5%
c
d
e
moisture. Assumed cell slurry is 5% solids, removed 95% of moisture. Assumed 300 working days per year. Bulk prices quoted from vendors for
00 kg, current as of April 2014.
1
This journal is © The Royal Society of Chemistry 2015
RSC Adv., 2015, 5, 19166–19175 | 19171

View Article Online
RSC Advances
Paper
intriguing, particularly in light of the comparable performance equipment used in this study. Additional costs for waste treat-
in hydrolysis assays. This is a particularly interesting result ment and peripheral utilities (plant air and DI water) are also
because the YSD CalB and N435 are both CalB-centric systems. included. Assuming a carbon efficiency of 30% (kg dry cells per
Comparable esterication data for ycM37L could not be found kg galactose), the cost of one kilogram of the YSD biocatalyst is
in the literature. It is unclear whether the difference is one of %590 to %640 for the galactose prices mentioned previously.
specic protein modications introduced to N435, the relative The theoretical production reects potential cost savings from
protein loadings, or the surface chemistry of the acrylic bead complex nutrient sources. Compared to %11 250 per kg, the
versus the yeast cell.
bulk price for the lipase resin used in this study, the YSD bio-
There are signicant differences between the resin and the catalyst represents signicant cost savings without changing
yeast cell in terms of surface environment and protein the catalyst loading or batch productivity—irrespective of the
concentration. The difference in surface polarity may play a application. Comparisons with other commercial lipase sources
greater role in the hydrophobic ester synthesis reactions than it (Table 4) are less favorable, but the conservative theoretical
does in a surfactant-mediated hydrolysis assay. Previous production cost of the YSD biocatalyst is still less than that of
attempts to demonstrate a YSD biocatalyst in ester synthesis other readily available commercial lipases.
26,32–34
have used more polar reactants.
unsaturated fatty acids (log P from 3.1 for octanoic acid to different applications, the predicted cost of manufacture was
.2 for hexadecanoic acid), and 1-octanol are highly hydro- normalized with the measured activity in order to provide a rational
Heptane (log P ¼ 4.3),
Because the apparent activity of the YSD biocatalysts changes in
7
phobic. While butanol is more polar (log P ¼ 0.84), it is not basis for comparison. In an aqueous environment (geared towards
miscible with water. The local concentration of the hydrophobic hydrolysis), the calculus based on mass loading does not differ
substrates near the cell surface—which likely collects any meaningfully. However, in non-polar environments geared toward
available water—could be lower than that near the surface of the organic synthesis, the order-of-magnitude difference in conversion
hydrophobic resin. A locally water-rich environment on the cell rate puts the YSD biocatalyst behind N435 by a factor of nearly two.
ꢂ1 35
surface could create a transport barrier between the enzyme and
The estimated protein loading in N435 is 30 mg per g dw .
the bulk solvent. Despite the difference in apparent esterica- This gure puts the amount of protein in the commercial system
tion activity, the performance on a bulk mass basis reects a at roughly three orders of magnitude more than the YSD system.
signicant improvement in terms of cost.
Given this inequality, an order of magnitude difference in rate is
unsurprising. Cost comparison on a per-enzyme basis will not
favor the YSD system, nor will comparisons in synthetic appli-
cations. However, the relative activity on a per-enzyme basis
Cost comparison
Since the YSD biocatalyst is as active (hydrolytically) as
commercialized formulations of immobilized lipase, the
production cost can be compared to the commercial purchase
price to inform the sourcing decision. A plasmid-based
production strain is unlikely since chromosomal integration
is typical, so the strains used in this work provide a useful base-
case. Excluding the costs associated with cloning and screening,
the estimated cost of production from seed culture to lyophi-
lized nal product are indicated in Table 3. This analysis
(
Fig. 4C and D) suggests that the YSD system is a far more effi-
cient scaffold for lipase production. This holds true in non-polar
environments, where N435 is only an order of magnitude faster,
despite presenting three orders of magnitude more enzyme.
Experimental
Strains and media
assumes that galactose is the limiting nutrient; the bulk price Plasmid DNA was screened and stored in Escherichia coli DH5a.
ꢂ1
for the grade used in these studies is %15–%30 per kg, which The untransformed strain was cultivated in LB media (10 g L
ꢂ1
ꢂ1
was used in the cost analysis. The costs of the other media tryptone, 5 g L yeast extract, 5 g L sodium chloride, 1 mL
ꢀ
components are assumed equal to or less than the largest lab- 1 N sodium hydroxide) at 37 C and made electrocompetent in-
36
scale quantities available. The energy and projected mainte- house and transformed as described in Molecular Cloning.
nance costs for lyophilization are included, based on the Aer electroporation, cells were recovered in SOC media
Table 4 Cost comparison for the YSD biocatalyst and commercial lipase sources
e
e
e
Price per kg-U
Price per kg-U
Price per kg-U
(synthesis)
Price per kg
(hydrolysis, pNPB)
(hydrolysis, pNPP)
a
YSD biocatalyst (lab-scale)
YSD biocatalyst (theoretical)
%586
%118
%11 250
%1200
%150–300
%16
%3
%330
%35
Not tested
%28
%6
%516
%55
Not tested
%151
%30
%153
%16
a
b
Novozym 435 (bulk, from Sigma Aldrich)
Novozym 435 (bulk, commercial)
c
d
TransBiodiesel
Not tested
a
b
c
d
Current work. Quote from Sigma-Aldrich, Feb. 2014.e Glycerin Carbonate Market Research and Analysis, Piedmont Biofuels, 2010. http://
www.transbiodiesel.com/faq1, accessed March 9, 2014. Normalized price (per kg) by average of YSD biocatalyst activities in Fig. 4 and 20 min
time point in Fig. 8B.
19172 | RSC Adv., 2015, 5, 19166–19175
This journal is © The Royal Society of Chemistry 2015

View Article Online
Paper
RSC Advances
ꢂ1
ꢂ1
ꢀ
(
20 g L tryptone, 5 g L yeast extract, 10 mM magnesium T4 ligase overnight at 16 C. Ligation products were transformed
ꢂ1
sulfate, 0.5 g L sodium chloride, 250 mM potassium chloride, into E. coli by electroporation. Putative transformants were
10 mM magnesium chloride, 20 mM glucose, 1 mL 1 N sodium screened for the insert by PCR (OneTaq Quick-Load Master Mix,
hydroxide) and plated with appropriate antibiotics (LB with New England Biolabs, Ipswich, MA). Aer the insert was detec-
ꢂ1
ꢂ1
1
5 g L agar and 50 mg L ampicillin). Cultures for plasmid ted using PCR, the identied colonies were picked for plasmid
preparation were grown on or in LB, with antibiotics. Saccharo- purication, restriction digest screening, and sequencing. The
ꢂ1
myces strain EBY100 (ref. 4) was propagated in 2ꢁ YAPD (20 g L
c-myc epitope was added to o1S-lipase fusions and the
ꢂ1
ꢂ1
yeast extract, 100 mg L adenine hemisulfate, 40 g L peptone, p426 GALL backbone via PCR. The insert and backbone were
0 g L glucose). Two strains of Saccharomyces cerevisiae, recombined with one-step ISO assembly, transformed into
ꢂ1
37
42
4
S288C (MATa sta1 sta2 sta3 STA10) and MT8-1 (MATa ade his3 E. coli, and screened as before.
leu2 trp1 ura3), were purchased from the American Type Culture
ꢀ
Collection (ATCC) and propagated in/on YAPD at 30 C. Yeasts Production of the yeast surface-display biocatalyst
38
were made electrocompetent in-house, with modied wash and
recovery buffers. Aer electroporation, cells were recovered in
Yeasts carrying the putative construct for the lipase-alpha-
agglutinin fusion protein under GAL promoter control (Fig. 9)
were propagated in SC/Dex (SC with 2% glucose) without tryp-
tophan. Yeasts carrying the putative construct for the Flo1S-
lipase fusion under GALL control were propagated in SC/Dex
without uracil. Batches were passaged by inoculating 3 mL
cultures 1 : 500 with cell slurry from previous generations and
39
1
: 1 2ꢁ YAPD : 1 M sorbitol and plated on minimal media or
used to inoculate liquid media. Selection is maintained through
the strain's tryptophan (EBY100) or uracil (MT8-1) auxotrophy.
Yeasts are grown in selective (i.e., lacking tryptophan/uracil)
synthetic complete (SC ) media with 2% glucose for a carbon
source and penicillin-streptomycin (100
35
ꢂ1
U
mL
ꢀ
and
ꢀ
grown at 30 C, 250 rpm in a 15 mL Falcon tube to an OD600 of
ꢂ1
100 mg mL , respectively) through two passes at 30 C before
2–3. A 200 mL culture was inoculated 1 : 200 from this culture
being transferred to selective SC media with 2% galactose as a
and grown under the same conditions in baffled shaker asks to
ꢀ
carbon source for induction at 20 C.
7
a nal OD600 of 6–8 (5 ꢁ 10 cells per mL). Media was removed
ꢀ
by centrifugation (3000 g for 15 min at 4 C) and the cell pellet
was re-suspended in induction media. The suspension was used
to inoculate a 2 L bioreactor (Sartorius AG, Goettingen,
Cloning and transformation
All T4 ligase, phosphatase, and polymerase were obtained from
New England Biolabs, Ipswich, MA. The bacterial protein
sequence for M37L (NCBI accession # AAS78630) was optimized
for expression in yeast by GENEWIZ (Research Triangle Park,
NC) and provided in a bacterial stock containing the plasmid
pUC57-ycM37L. DNA for the CalB mature peptide was provided
in pET22-mCalB in bacterial stocks. Genomic DNA from S. c.
S288C was puried using a modied protocol for a blood and
40
tissue kit (QIAGEN, Valencia, CA). The gene for FLO1S was
amplied from genomic DNA using high-delity Phusion
7
polymerase and previously reported primers. The PCR (poly-
merase chain reaction) product was digested with BamHI/XhoI
and ligated into BamHI/XhoI-digested pYES with T4 ligase. The
construct was transformed into E. coli for storage. The insert
was conrmed by restriction digest and sequencing. The o1S
gene was puried from an agarose gel aer digesting the
plasmid with BamHI/XhoI, using a spin kit (Thermo Fisher
Scientic, Pittsburgh, PA). Plasmid DNA was puried from
bacterial cultures with a spin kit. The lipase sequences were
amplied from their plasmid templates with high-delity PCR.
41
Plasmid backbones (pCT-EGFP and p426 GALL, ATCC Fig. 9 Vector maps for the plasmids used in this study. A and B: YSD
vectors containing ycM37L and CalB, respectively; includes bacterial
origin and ampicillin resistance marker for propagation in E. coli,
single-copy yeast replication origin (CEN6/ARSH4) and selection
marker (TRP10). Transcription is induced by the GAL promoter.
Secretion and surface attachment are accomplished via the fusion to
8
7341) were puried from bacterial stocks, digested with
BamHI/NheI (pCT plasmids) or BamHI/EcoRI (p426 plasmids),
and treated with phosphatase. For pCT-based plasmids, the PCR
products of the lipases were digested with BamHI/NheI,
combined individually with the plasmid backbones in Aga2. Immuno-labeling detects the C-terminal c-myc epitope tag. C
and D: YSD vectors containing ycM37L and CalB, respectively; includes
3
: 1 ratios, and ligated with T4 ligase. For p426-based plasmids,
bacterial origin and selection marker, multi-copy yeast replication
origin (2m) and selection marker (URA3). Transcription is induced by the
GAL promoter. Secretion and surface attachment are accomplished
via the fusion to Flo1S. Immuno-labeling detects the C-terminal c-myc
the PCR products of the lipases were digested with XhoI/EcoRI
and combined with gel-puried o1S (BamHI/XhoI) and the
digested plasmid backbone in 3 : 3 : 1 molar ratios
(
lipase : o1S : p426). The components were ligated with epitope tag.
This journal is © The Royal Society of Chemistry 2015
RSC Adv., 2015, 5, 19166–19175 | 19173

View Article Online
RSC Advances
Germany) and operated at 20 C, 250 rpm with 4 L min
Paper
ꢀ
ꢂ1
ꢀ
ꢂ1
ꢀ
ꢀ
ꢂ1
5 min , and to 300 C at 25 min . The concentration of the
sparged air during induction. Samples were removed at regular fatty acid was determined by comparing the FID peak area to
intervals (4–6 hours), and the sample volume (typically 10 mL) standard curves prepared in heptane.
was replaced with galactose (20% weight/volume in DI water,
sterilized by ltration). Induction ended aer 18–20 hours, with
Flow cytometry
7
a nal cell count of 1–3 ꢁ 10 cells per mL.
6
7
ꢀ
Yeast cells (8 ꢁ 10 to 5 ꢁ 10 ) were suspended in 1 mL phos-
phate buffered saline supplemented with bovine serum albumin
(PBS–BSA, 137 mM sodium chloride, 2.7 mM potassium chlo-
ride, 10 mM dibasic sodium phosphate, 2 mM monobasic
Following induction, the bioreactor was chilled to 6 C and
the culture siphoned into chilled centrifuge bottles. Cells were
ꢀ
harvested by centrifugation (4000 g for 30 min at 4 C) and
washed once with 1 L sterile, cold water. Cells were re-suspended
ꢂ1
9
potassium phosphate, 1 g L BSA in water). The suspension was
in a dense slurry in DI water (3–5 ꢁ 10 cells per mL), distributed
centrifuged at 10 000 g for 30 s (subsequent centrifugations used
the same conditions) and re-suspended in 250 mL PBS–BSA.
Primary antibody (chicken anti-c-myc, Invitrogen) was added at a
into 2 mL microcentrifuge tubes, frozen, lyophilized in a
ꢀ
SpeedVac (Thermo SPD-111V) for >8 hours, and stored at 4 C.
1
: 250 dilution and mixed by vortexing. Aer >15 minutes at
Assays for esterase (hydrolytic) activity
room temperature, the cells were diluted with 1 mL PBS–BSA,
centrifuged, and the supernatant was aspirated. The cells were
washed with 1 mL PBS–BSA, centrifuged, and aspirated again.
Secondary antibody (goat anti-chicken Alexa 633, Invitrogen) was
added at a 1 : 250 dilution in 250 mL PBS–BSA. Aer >15 minutes
at room temperature in the dark, the cells were diluted and
washed with PBS–BSA in the same way. Cells were re-suspended
in 1 mL PBS–BSA and vortexed immediately before data collec-
tion. Data were collected on an Accuri C6 cytometer with at least
Lipase from C. antarctica immobilized on acrylic resin was
purchased from Sigma Aldrich (St. Louis, MO) and used as a
positive control. Lyophilized YSD biocatalyst or resin was
weighed into clean 2 mL microcentrifuge tubes. Two substrates
were used for the assays: p-nitrophenyl butyrate (pNPB) and p-
nitrophenyl palmitate (pNPP) (Sigma Aldrich, St. Louis, MO). An
emulsion of substrate (1–4 mM in 50 mM potassium phosphate
buffer, pH 6.5, with 0.5% Triton-X 100) was added to a nal
ꢂ1
catalyst loading of 20 mg dw mL . Iso-propanol was added to p-
30 000 events per run and slow uidics. The manufacturer's
nitrophenyl palmitate assays at 5% by volume. Samples were
gently agitated in a temperature-controlled water bath during
soware was used for analysis. Normal, healthy cells were gated
in the FSvSS plots based on an unlyophilized, untreated,
untransformed control (Fig. 1). This gate was applied to histo-
grams of Alexa 633 uorescence (Fig. 2 and 3), which were used
to determine the fraction of the population expressing the
surface display (SD) construct. The baseline for expression was
determined by experimental samples incubated without the
primary antibody (Fig. 2).
ꢀ
the assay. The bath was maintained at 30 C for assays using
ꢀ
pNPB and at 45 C for assays using pNPP (see Scheme 1), unless
otherwise indicated. Aliquots of at least 20 mL were removed
periodically and spun down at >10 000 g for >1 minute and the
absorbance of the supernatant at 400 nm was measured on a
NanoDrop 1000 (Thermo Scientic, Wilmington, DE). The
maximum observed absorbance within 10 minutes was recor-
ded, since the acrylic resin was observed to absorb some of the
hydrolysis product at higher conversions, leading to under-
reporting of the activity of the resin-immobilized lipase. The
concentration of p-nitrophenol was calculated based on the
absorbance and standard curves prepared in appropriate buffer.
Copy number was determined by comparing labeled cells to a
standard curve drawn through four groups of antibody coated
beads with known antibody binding capacities (Simply Cellular
anti-Mouse IgG, Bangs Labs, Fishers, IN). Lyophilized cells
6
(
2 ꢁ 10 ) and the anti-mouse IgG microspheres were suspended
in PBS–BSA. Fluorescein-conjugated mouse anti-c-cymc antibody
Sigma) was added, 10 mg to 200 mL of cell suspension or to 100
(
Esterication activity
mL of bead suspension. Aer incubating for 30 minutes at room
temperature and protected from light, the beads and cells were
washed twice with PBS–BSA and re-suspended in 500 mL for
analysis on an Accuri C6 cytometer. According to the manufac-
turer's instructions, half-height, full-width gates were drawn on
the standards and labeled cell populations. The mean uores-
cence intensity of the gated standards was used to draw the
standard curve from which the copy number of the labeled cells
was interpolated.
Chemicals were GC grade or better. Dry catalyst was weighed into
clean 1.5 dram glass vials. Solutions of a fatty acid and either n-
butanol or n-octanol (1 mL of each, both 200 mM in heptane)
were added to the vials. Palmitic acid solutions were warmed to
ꢀ
3
0 C to dissolve the fatty acid. For the fatty acid specicity trials,
water was added to the reaction at 0.5% vol/vol. The vials were
sealed and moved to a magnetically stirred heat block with
digital temperature control. The reactions were stirred at 200
rpm for six hours, beginning when the reaction was rst heated.
Samples of 50 mL were added to 1 mL of heptane and analyzed by
GC-FID. The samples were analyzed on a Shimadzu GC2010
Conclusions
equipped with an autosampler and a DB-5 capillary column (30 In terms of hydrolytic activity, the YSD catalyst is comparable to
m ꢁ 0.25 mm ꢁ 0.25 mm). One mL of each sample was injected at commercial immobilized lipase catalysts at equal mass load-
a 30 : 1 split ratio with helium as the carrier gas. The injector and ings, but not as effective in non-aqueous applications. The
ꢀ
detector were held at 300 C. The oven temperature was held at biocatalyst can be re-used in the same manner as a bead-based
ꢀ ꢀ ꢀ ꢀ
ꢂ1
9
0 C for 2 minutes, ramped to 160 C at 20 min , to 250 C at catalyst, with the additional benet of a greater surface-to-
19174 | RSC Adv., 2015, 5, 19166–19175
This journal is © The Royal Society of Chemistry 2015

View Article Online
Paper
RSC Advances
volume ratio, but at a size that can easily be removed from 17 S. H. Lee, J. Choi, M. Han, J. H. Choi and S. Y. Lee, Biotechnol.
suspension by centrifugation or routine ltration. The yeast cell Bioeng., 2005, 90, 223–230.
appears to confer signicant thermal stability upon ycM37L, 18 S. Tamalampudi, S. Hama, T. Tanino, M. R. Talukder,
and the thermal stability of YSD CalB was similar to that of a
commercial acrylic bead-based system. Lipase produced with
A. Kondo and H. Fukuda, J. Mol. Catal. B: Enzym., 2007, 48,
33–37.
YSD is more efficient on a per-enzyme basis than lipase in N435. 19 M. Kato, J. Fuchimoto, T. Tanino, A. Kondo, H. Fukuda and
M. Ueda, Appl. Microbiol. Biotechnol., 2007, 75, 549–555.
2
2
2
2
2
2
0 S. Shiraga, M. Kawakami, M. Ishiguro and M. Ueda, Appl.
Environ. Microbiol., 2005, 71, 4335–4338.
1 G. Su, X. Zhang and Y. Lin, Biotechnol. Lett., 2010, 32, 1131–
Acknowledgements
The authors gratefully acknowledge Sonia Herrero from
Professor Margo Daub's lab at NCSU for donating a lab strain of
E. coli; Professor Balaji Rao's lab for donating the pCT-EGFP
vector, the stocks for EBY100, and time on the cell sorter; and
Kerri Cushing for donating the pET22-CalB vector. This work
1136.
2 A. Bielen, R. Teparic, D. Vujaklija and V. Mrsa, Food Technol.
Biotechnol., 2014, 2014, 16–34.
3 W. Zhang, S. Han, D. Wei, Y. Lin and X. Wang, J. Chem.
Technol. Biotechnol., 2008, 83, 329–335.
4 A. Yoshida, S. Hama, K. Nakashima and A. Kondo, Enzyme
Microb. Technol., 2011, 48, 334–338.
5 T. Tanino, T. Aoki, W. Chung, Y. Watanabe, C. Ogino,
H. Fukuda and A. Kondo, Appl. Microbiol. Biotechnol., 2009,
(JE) was supported in part by NSF award 0832498 through the
Center for BioEnergy Research and Development, by the GAANN
Interdisciplinary Doctoral Program in Molecular Biotechnology,
award P200A090129-11, and by the Eastman Chemical Center of
Excellence. The authors declare no conicts of interest.
82, 59–66.
2
6 S. Han, Z. Pan, D. Huang, M. Ueda, X. Wang and Y. Lin,
J. Mol. Catal. B: Enzym., 2009, 59, 168–172.
7 J. Han and H. Kim, J. Microbiol. Biotechnol., 2011, 21, 1159–
1165.
Notes and references
2
1
C. Mateo, J. M. Palomo, G. Fernandez-Lorente, J. M. Guisan
and R. Fernandez-Lafuente, Enzyme Microb. Technol., 2007, 28 S. Shibasaki, M. Ueda, T. Iizuka, M. Hirayama, Y. Ikeda,
4
0, 1451–1463.
N. Kamasawa, M. Osumi and A. Tanaka, Appl. Environ.
Microbiol., 2001, 55, 471–475.
29 H. S. Ryu, H. K. Kim, W. C. Choi, M. H. Kim, S. Y. Park,
N. S. Han, T. K. Oh and J. K. Lee, Appl. Microbiol.
Biotechnol., 2005, 70, 321–326.
2
3
4
A. Illanes, Enzyme biocatalysis: principles and applications,
Springer, New York, 2008.
P. Tufvesson, J. Lima-Ramos, M. Nordblad and
J. M. Woodley, Org. Process Res. Dev., 2010, 15, 266–274.
E. T. Boder and K. D. Wittrup, Nat. Biotechnol., 1997, 15, 553– 30 K. Umemura, H. Atomi, T. Kanai, Y. Teranishi, M. Ueda and
57. A. Tanaka, Appl. Microbiol. Biotechnol., 1995, 43, 489–492.
5
5
6
M. Ueda and A. Tanaka, J. Biosci. Bioeng., 2000, 90, 125–136. 31 J. Eby and S. W. Peretti, RSC Adv., submitted
C. E. Ballou, Cold Spring Harbor Monograph Archive, 1982, vol. 32 Z. Pan, Z. Yang, L. Pan, S. Zheng, S. Han and Y. Lin, J. Ind.
1
1, pp. 335–360.
T. Matsumoto, H. Fukuda, M. Ueda, A. Tanaka and A. Kondo, 33 Z. Jin, J. Ntwali, S. Han, S. Zheng and Y. Lin, J. Biotechnol.,
Appl. Environ. Microbiol., 2002, 68, 4517–4522.
2012, 159, 108–114.
N. Gera, M. Hussain, R. C. Wright and B. M. Rao, J. Mol. Biol., 34 Z. Jin, S. Liang, X. Zhang, S. Han, C. Ren, Y. Lin and S. Zheng,
011, 409, 601–616. Biotechnol. Bioprocess Eng., 2013, 18, 365–374.
Y. Fujita, S. Takahashi, M. Ueda, A. Tanaka, H. Okada, 35 Z. Cabrera, G. Fernandez-Lorente, R. Fernandez-Lafuente,
Microbiol. Biotechnol., 2014, 41, 711–720.
7
8
9
2
Y. Morikawa, T. Kawaguchi, M. Arai, H. Fukuda and
A. Kondo, Appl. Environ. Microbiol., 2002, 68, 5136–5141.
J. M. Palomo and J. M. Guisan, J. Mol. Catal. B: Enzym.,
2009, 57, 171–176.
1
0 P. Villeneuve, J. M. Muderhwa, J. Graille and M. J. Haas, 36 J. Sambrook, Molecular cloning: a laboratory manual, Cold
J. Mol. Catal. B: Enzym., 2000, 9, 113–148.
1 A. M. Klibanov, Trends Biochem. Sci., 1989, 14, 141–144.
Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.,
2001.
1
12 J. Uppenberg, N. Oehrner, M. Norin, K. Hult, G. J. Kleywegt, 37 D. R. Gietz and R. A. Woods, in Methods in Enzymology, ed. C.
S. Patkar, V. Waagen, T. Anthonsen and T. A. Jones,
Biochemistry, 1995, 34, 16838–16851.
3 Y. Shimada, Y. Watanabe, T. Samukawa, A. Sugihara,
Guthrie and G. R. Fink, Academic Press, 2002, pp. 87–96.
38 D. M. Becker and V. Lundblad, in Current Protocols in
Molecular Biology, John Wiley & Sons, Inc., 2001.
1
H. Noda, H. Fukuda and Y. Tominaga, J. Am. Oil Chem. 39 N. Gera, M. Hussain and B. M. Rao, Methods, 2013, 60, 15–26.
Soc., 1999, 76, 789–793.
40 C. S. Hoffman, in Current Protocols in Molecular Biology, John
1
1
4 J. H. Yoon, Int. J. Syst. Evol. Microbiol., 2005, 55, 335–339.
Wiley & Sons, Inc., 2001.
5 K. S. Yang, J. Sohn and H. K. Kim, J. Biosci. Bioeng., 2009, 107, 41 M. F. Canbolat, N. Gera, C. Tang, B. Monian, B. M. Rao,
5
99–604.
B. Pourdeyhimi and S. A. Khan, ACS Appl. Mater. Interfaces,
2013, 5, 9349–9354.
1
6 T. P. Korman, B. Sahachartsiri, D. M. Charbonneau,
G. L. Huang, M. Beauregard and J. U. Bowie, Biotechnol. 42 D. G. Gibson, in Methods in Enzymology, Academic Press,
Biofuels, 2013, 6, DOI: 10.1186/1754-6834-6-70. 2011, pp. 349–361.
This journal is © The Royal Society of Chemistry 2015
RSC Adv., 2015, 5, 19166–19175 | 19175