Quantitative Packaging of Active Enzymes into a Protein Cage

Article abstract of DOI:10.1002/anie.201508414

Genetic fusion of cargo proteins to a positively supercharged variant of green fluorescent protein enables their quantitative encapsulation by engineered lumazine synthase capsids possessing a negatively charged lumenal surface. This simple tagging system provides a robust and versatile means of creating hierarchically ordered protein assemblies for use as nanoreactors. The generality of the encapsulation strategy and its effect on enzyme function were investigated with eight structurally and mechanistically distinct catalysts.

Full text of DOI:10.1002/anie.201508414

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201508414
German Edition: DOI: 10.1002/ange.201508414
Nanoreactors Hot Paper
Quantitative Packaging of Active Enzymes into a Protein Cage
Yusuke Azuma, Reinhard Zschoche, Matthias Tinzl, and Donald Hilvert*
[
8a]
Abstract: Genetic fusion of cargo proteins to a positively
supercharged variant of green fluorescent protein enables their
quantitative encapsulation by engineered lumazine synthase
capsids possessing a negatively charged lumenal surface. This
simple tagging system provides a robust and versatile means of
creating hierarchically ordered protein assemblies for use as
nanoreactors. The generality of the encapsulation strategy and
its effect on enzyme function were investigated with eight
structurally and mechanistically distinct catalysts.
a deca-arginine (R10) tag in vivo. AaLS-neg was further
optimized by directed evolution, resulting in an improved
variant, AaLS-13, with higher loading capacity under phys-
[
8b]
iological conditions.
In addition to R10-tagged proteins,
AaLS-13 efficiently encapsulates the positively supercharged
green fluorescent protein GFP(+ 36) in vitro.
show that GFP(+ 36) can also be used as a genetically
encodable tag to target cargo enzymes to the interior of
AaLS-13 capsids (Figure 1A,B).
[
8c]
Here we
As a model enzyme, we selected a computationally
designed and experimentally optimized retro-aldolase (RA)
that catalyzes the cleavage of (Æ)-4-hydroxy-4-(6-methoxy-2-
naphthyl)-2-butanone (methodol, 1) to 6-methoxy-2-naph-
C
ompartmentalization of enzymes within restricted spaces
is a common strategy for controlling otherwise incompatible
biosynthetic processes in complex intracellular environments.
In bacteria, for example, protein-bounded microcompart-
ments coordinate the activity of catalysts involved in carbon
[
9]
thaldehyde (2) and acetone (Figure 1C). The K210M
[
1]
fixation and small-molecule metabolism. Like the lipid
membrane of eukaryotic organelles, the protein shells of such
assemblies restrict diffusion into and out of the enclosed
chamber, thereby providing a means of regulating the
encapsulated enzymes, concentrating volatile reactants,
sequestering toxic intermediates, and avoiding undesired
[
1]
side reactions.
[
2]
[3]
[4]
Liposomes, synthetic polymers, and protein cages
have been extensively used as artificial microcompartments
to investigate how confinement and spatial organization
affect enzyme activity. Virus-like particles are particularly
attractive in this regard because of their biocompatibility,
spontaneous self-assembly into highly organized structures of
defined size and shape, and ready modification by chemical
Figure 1. A strategy for encapsulating active enzymes in a protein
cage. A) GFP(+36) (green) serves as a directing tag to load appended
cargo proteins (blue) into empty AaLS-13 capsids. B) Designed gene
for the GFP(+36)-fused retro-aldolase (RA). C) The retro-aldol reaction
catalyzed by RA.
[
5]
and genetic methods. Diverse strategies have been exploited
to load virus capsids with cargo molecules, including diffusion
[
4a]
[4b]
[4c]
methods, coiled-coil and electrostatic tagging systems,
[4d]
RNA-mediated loading, and genetic fusion to scaffolding
[
4e–g]
proteins.
Because enzyme stoichiometry can influence
[
6]
pathway output, control over guest loading is highly
desirable in such systems. Experiments with cowpea chlorotic
mottle virus (CCMV) have shown that both covalent and
noncovalent approaches can afford considerable control over
[
9]
À1
À1 À1
variant of RA95.5–8 (kcat = 0.18 s , k /K = 1800m s )
cat
m
was N-terminally fused to GFP(+ 36) through a (Gly-Gly-
Ser) linker (Figure 1B). Despite an approximately 10-fold
5
[
7]
À1
packing density.
decrease in catalytic activity (kcat = 0.018 s , k /K =
cat
m
À1 À1
We recently developed an alternative encapsulation
system based on the non-viral capsid-forming enzyme luma-
170m s ), the unoptimized GFP(+ 36)–RA fusion construct
still efficiently cleaves the chromogenic substrate. The
resulting absorbance change provides a facile means of
monitoring enzymatic activity upon encapsulation.
[
8]
zine synthase from Aquifex aeolicus (AaLS). Introduction
of four glutamate residues on the lumenal surface of the
protein yielded AaLS-neg, which forms inclusion complexes
with complementarily charged guest molecules bearing
Like GFP(+ 36), GFP(+ 36)–RA was readily taken up by
[
10]
purified AaLS-13 capsids.
Following the simple method
[
8c]
described for the encapsulation of GFP(+ 36), GFP(+ 36)–
RA was mixed with empty AaLS-13 capsids in aqueous buffer
[
*] Dr. Y. Azuma, R. Zschoche, M. Tinzl, Prof. Dr. D. Hilvert
Laboratory of Organic Chemistry, ETH Zurich
(pH 8.0, ionic strength = 0.34m) at various ratios. Size-exclu-
sion chromatography of the mixtures, followed by SDS-
PAGE analysis of the capsid fraction, showed that the tagged
aldolase associated with AaLS-13 (Figure 2A,B and Fig-
ure S1 in the Supporting Information), whereas only negli-
8
093 Zurich (Switzerland)
E-mail: hilvert@org.chem.ethz.ch
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201508414.
Angew. Chem. Int. Ed. 2016, 55, 1531 –1534
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1531

Angewandte
Communications
Chemie
the observed ratio of around 45 guests per 180-mer capsid
(1.25 guests per pentamer).
Although an R10 tag has been successfully employed to
direct GFP and HIV protease to the lumen of AaLS-neg and
[8]
AaLS-13 in vivo, it is less effective than GFP(+ 36) for
in vitro encapsulation. When 18 equivalents of an R10-tagged
retro-aldolase (RA–R10) were mixed with AaLS-13, only
2
.7 Æ 1.1 enzymes co-purified with each capsid (Figure S5);
higher cargo/capsid ratios caused precipitation. The low net
charge of RA–R10 (+ 4.3 at pH 8.0 versus + 29.9 for GFP
(+ 36)–RA and + 20.0 for the originally studied HIV
protease–R10 dimer) may account for its comparatively
poor in vitro encapsulation efficiency.
The activity of the encapsulated aldolase was found to be
sensitive to packing density. At low loading (e.g., ca. 8 guests
per capsid), bound GFP(+ 36)–RA exhibited nearly the same
À1
kcat value (0.018 Æ 0.001 s ) as free enzyme in solution but
a 4.5-fold higher K value (470 Æ 120 mm; Figure 3A).
m
Increasing the concentration of enzyme within the capsid
had little effect on k /K but led to a gradual decrease in k
cat
cat
m
(Figure 3B), likely owing to inactivation of the enzyme by the
aldehyde product 2, which can form a covalent Schiff base
[13]
adduct with the catalytic lysine.
The generation of high
concentrations of this reactive metabolite in the vicinity of the
Figure 2. Quantitative loading of GFP(+36)–RA into empty AaLS-13
capsids. Size-exclusion chromatogram of A) empty capsids and B) cap-
sids mixed with 45 equivalents of GFP(+36)–RA. Traces and bars
correspond to l=280 nm absorbance and fluorescence readings from
individual fractions, respectively. Insets show TEM images of A) empty
and B) filled AaLS-13 capsids (scale bar: 100 nm); a representative
particle from each sample is enlarged. C) Average number of encapsu-
lated GFP(+36)–RA fusion proteins per AaLS-13 capsid as a function
of input mixing ratio. Errors indicate the standard deviation for three
independent experiments. The linear fit (excluding the point in
parentheses) has a slope of 0.995 (dashed line).
gible association was observed for untagged RA (Figure S2).
Negative-staining transmission electron microscopy (TEM)
of the capsid particles confirmed that the fusion protein
localized to the capsid interior (Figure 2B, inset). Host–guest
stoichiometry was estimated from the l = 280 nm/488 nm
absorbance ratio. A linear correlation between input and
output host/guest ratios was observed up to 45 GFP(+ 36)–
RAs per capsid, assuming that AaLS-13 adopts a 180-subunit
[8b]
Figure 3. Retro-aldolase activity in AaLS-13. A) Michaelis–Menten plots
for GFP(+36)–RA in solution (circles) and in AaLS-13 (diamonds; 8.6
GFP(+36)–RA fusion proteins per capsid). B) kcat (circles) and kcat/Km
T= 3 structure
(Figure 2C). Increasing the input ratio
further did not afford higher loading densities owing to co-
precipitation of both proteins. These results show that cargo
encapsulation is essentially quantitative up to around 45
guests per T= 3 capsid. The observed encapsulation efficiency
(bars) parameters for encapsulated GFP(+36)–RA as a function of
catalyst loading density. Errors indicate the standard deviation for
three independent experiments. ****: P<0.001.
[11]
is comparable to that of GFP(+ 36) alone. Recent studies
on the loading mechanism suggest that GFP(+ 36) associates
[12]
tightly with the interior surface of the AaLS-13 capsid shell,
densely packed enzymes would be expected to exacerbate this
problem. In analogy to the cascade reactions promoted in
thus leaving ample space to accommodate a fusion partner in
the central cavity (Figure 2B inset). Approximately 1.5 GFP
molecules would fit on the lumenal surface of one pentameric
subunit of AaLS (Figure S4), which is in good agreement with
[
1]
natural bacterial micro-compartments, however, further
processing of the aldehyde by additional co-encapsulated
enzymes should be possible.
1
532
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1531 –1534

Angewandte
Communications
Chemie
The generality of this simple targeting system was
assessed by fusing GFP(+ 36) to seven other enzymes:
tag. Further fine-tuning of the electrostatic interaction
between guests and host capsids will be necessary for efficient
[14]
[12, 21]
a
computationally designed Kemp eliminase (KE),
encapsulation of such proteins.
[15]
a TEM b- lactamase (bLac), a cyclohexylamine oxidase
The model enzymes in this study act on a range of
substrates that vary in both size and charge. Remarkably, and
in contrast to encapsulated GFP(+ 36)–RA, which was
around 5-fold less efficient than the corresponding free
enzyme, GFP(+ 36)–KE, GFP(+ 36)–CHAO, GFP(+ 36)–
KatG, and GFP(+ 36)–NOX retained nearly full activity
with their native substrates upon complexation with AaLS-13
(Figure 4B and Table S2). Although the full scope and
mechanism of molecular transport still needs to be elucidated,
[
16]
[17]
(
CHAO),
a catalase-peroxidase (KatG),
an NADH
[18]
[19]
oxidase (NOX),
and a monoamine oxidase (MAO). These proteins cover
a broad range of properties in terms of molecular mass
an aldehyde dehydrogenase (AldH),
[20]
(
29 kDa to 78 kDa), quaternary state (monomer, dimer, and
tetramer), and net charge (À23.9 to + 2.5 at pH 8.0; Table S1).
Monomeric cargo enzymes with intermediate charge, like
GFP(+ 36)–KE and GFP(+ 36)–bLac, were encapsulated by
AaLS-13 as efficiently as GFP(+ 36)–RA (Figure 4A and
Figure S6A,B, Table S1). Encapsulation of very negatively
charged proteins, like CHAO and KatG (Table S1), was less
straightforward, however. Although GFP(+ 36)–CHAO and
GFP(+ 36)–KatG quantitatively associated with AaLS-13
the high turnover of encapsulated GFP(+ 36)–KE (k
=
cat
À1
380 s ) is especially notable, since it implies fast diffusion
of the neutral 5-nitrobenzisoxazole substrate across the capsid
wall. Retention of NOX activity additionally shows that the
negatively charged capsid shell does not prevent the uptake of
negatively charged substrates like NADH. Molecules capable
of reacting directly with the host or guest proteins, like
aldehyde 2, are potentially problematic of course. Activity
could also be adversely affected by unfavorable interactions
between the encapsulated enzyme and the negatively charged
capsid lumen that block the active site and/or populate an
inactive form of the catalyst. Such effects may be responsible
for the more than 10-fold drop in catalytic efficiency observed
for encapsulated GFP(+ 36)–bLac, for example (Figure 4B).
These limitations notwithstanding, the facility with which
assembled AaLS-13 cages rapidly take up a wide range of
cargo molecules irrespective of size and charge promises to be
broadly useful for efforts to engineer artificial microcompart-
ments for novel applications.
(
Figure 4A), TEM images of isolated particles suggest that
the guests partially attached to the exterior of the capsid
Figure S6C, D). Conversely, cargo molecules with too much
(
In summary, this study establishes GFP(+ 36) as a useful,
genetically encodable tag for efficiently packaging active
enzymes in AaLS-13 protein cages. The encapsulation
procedure is easy and robust. Simply mixing host and cargo
under mild aqueous conditions leads to internalization. No
pH or temperature changes are required. Loading is nearly
quantitative up to around 45 guest enzymes per T= 3 capsid,
thereby affording precise control over the density of guest
enzymes in the lumenal space. Although encapsulation is
straightforward for most monomeric enzymes, modulating the
surface charge of the enzyme or GFP may be needed to avoid
difficulties encountered with very negatively charged or
oligomeric enzymes. Nevertheless, the properties of this
simple encapsulation system set the stage for creation of
more complex nanoreactors by co-encapsulation of sequen-
tially acting enzymes. The competitive advantage such
cascades provide to organisms is currently under investigation
in our laboratory.
Figure 4. Scope of the targeted encapsulation strategy. A) Fraction of
the GFP(+36) fusion enzyme associated with AaLS-13. B) Activity of
the GFP(+36)–enzyme/AaLS-13 complexes relative to that of free
enzyme. *: includes some guests associated with the capsid exterior
(see Figure S6C,D). n.d., not determined due to protein precipitation.
Experimental details are summarized in Tables S1 and S2.
Acknowledgements
positive charge led to precipitation and hence lower loading
efficiencies. This was observed with the homodimeric enzyme
GFP(+ 36)–NOX, which has a calculated net charge of + 78.2
at pH 8.0 (Figure 4A and Table S1). Precipitation was even
more severe with the tetrameric enzymes AldH and MAO
bearing four copies of the positively supercharged GFP(+ 36)
We thank Peter Tittmann at the Scientific Center for Optical
and Electron Microscopy (ScopeM), ETH Zurich for his help
with electron microscopy experiments. This work was gen-
erously supported by the ETH Zurich and the European
Research Council (Advanced ERC Grant ERC-dG-2012-
321295 to D.H.). Y.A. is grateful for an Uehara Memorial
Angew. Chem. Int. Ed. 2016, 55, 1531 –1534
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1533

Angewandte
Communications
Chemie
Foundation Research Fellowship and an ETH Zurich Post-
doctoral Fellowship (co-funded by the Marie Curie Actions
program). R.Z. was awarded a scholarship from the Stipen-
dienfonds der Schweizerischen Chemischen Industrie (SSCI).
[7] W. F. Rurup, F. Verbij, M. S. T. Koay, C. Blum, V. Subramaniam,
J. J. L. M. Cornelissen, Biomacromolecules 2014, 15, 558 – 563.
[
8] a) F. P. Seebeck, K. J. Woycechowsky, W. Zhuang, J. P. Rabe, D.
Hilvert, J. Am. Chem. Soc. 2006, 128, 4516 – 4517; b) B.
Wçrsdçrfer, K. J. Woycechowsky, D. Hilvert, Science 2011, 331,
5
89 – 592; c) B. Wçrsdçrfer, Z. Pianowski, D. Hilvert, J. Am.
Keywords: enzymes · host-guest systems · nanoreactors ·
Chem. Soc. 2012, 134, 909 – 911.
protein engineering · synthetic biology
[9] L. Giger, S. Caner, R. Obexer, P. Kast, D. Baker, N. Ban, D.
Hilvert, Nat. Chem. Biol. 2013, 9, 494 – 498.
How to cite: Angew. Chem. Int. Ed. 2016, 55, 1531–1534
Angew. Chem. 2016, 128, 1555–1558
[10] Although packaging of GFP(+ 36) could also be achieved
starting from unassembled capsid components, mixing GFP(+
3
6)–RA with AaLS-13 fragments only afforded low yields of
capsid–cargo complexes (Figure S3). The fusion partner may
sterically interfere with capsid assembly.
[
1] a) A. H. Chen, P. A. Silver, Trends Cell Biol. 2012, 22, 662 – 670;
b) T. O. Yeates, C. A. Kerfeld, S. Heinhorst, G. C. Cannon, J. M.
Shively, Nat. Rev. Microbiol. 2008, 6, 681 – 691; c) T. O. Yeates,
M. C. Thompson, T. A. Bobik, Curr. Opin. Struct. Biol. 2011, 21,
[11] We previously reported that AaLS-13 encapsulates up to 100
GFP(+ 36) proteins based on Bradford analysis of total protein
concentration and determination of GFP(+ 36) concentration by
[8c]
fluorescence. Loading yields calculated independently from
l = 280 nm/488 nm absorbance ratios suggests that the earlier
values were overestimated by a factor of around 2, probably
because the staining efficiency of the supercharged GFP(+ 36)
and AaLS-13 proteins differs from that of bovine serum
albumin, which was used as a standard in the Bradford assay.
12] R. Zschoche, D. Hilvert, J. Am. Chem. Soc. DOI:10.1021/
jacs.5b10588.
2
23 – 231.
[
[
2] a) P. Walde, S. Ichikawa, Biomol. Eng. 2001, 18, 143 – 177; b) B.
Chaize, J.-P. Colletier, M. Winterhalter, D. Fournier, Artif. Cells
Blood Substitutes Immobilization Biotechnol. 2004, 32, 67 – 75.
3] a) M. Marguet, C. Bonduelle, S. Lecommandoux, Chem. Soc.
Rev. 2013, 42, 512 – 529; b) R. J. R. W. Peters, M. Marguet, S.
Marais, M. W. Fraaije, J. C. M. van Hest, S. Lecommandoux,
Angew. Chem. Int. Ed. 2014, 53, 146 – 150; Angew. Chem. 2014,
[
[
13] J. K. Lassila, D. Baker, D. Herschlag, Proc. Natl. Acad. Sci. USA
2010, 107, 4937 – 4942.
[14] R. Blomberg, H. Kries, D. M. Pinkas, P. R. E. Mittl, M. G.
Grütter, H. K. Privett, S. L. Mayo, D. Hilvert, Nature 2013, 503,
418 – 421.
1
26, 150 – 154.
[
4] a) M. Comellas-Aragon›s, H. Engelkamp, V. I. Claessen,
N. A. J. M. Sommerdijk, A. E. Rowan, P. C. M. Christianen,
J. C. Maan, B. J. M. Verduin, J. J. L. M. Cornelissen, R. J. M.
Nolte, Nat. Nanotechnol. 2007, 2, 635 – 639; b) I. J. Minten,
L. J. A. Hendriks, R. J. M. Nolte, J. J. L. M. Cornelissen, J. Am.
Chem. Soc. 2009, 131, 17771 – 17773; c) J. E. Glasgow, S. L.
Capehart, M. B. Francis, D. Tullman-Ercek, ACS Nano 2012, 6,
[
15] X. Raquet, J. Lamotte-Brasseur, E. FonzØ, S. Goussard, P.
Courvalin, J. M. Fr›re, J. Mol. Biol. 1994, 244, 625 – 639.
[16] H. Iwaki, M. Shimizu, T. Tokuyama, Y. Hasegawa, J. Biosci.
Bioeng. 1999, 88, 264 – 268.
[17] H. Forkl, J. Vandekerckhove, G. Drews, M. H. Tadros, Eur. J.
Biochem. 1993, 214, 251 – 258.
[18] H.-J. Park, C. O. A. Reiser, S. Kondruweit, H. Erdmann, R. D.
Schmid, M. Sprinzl, Eur. J. Biochem. 1992, 205, 881 – 885.
19] J.-E. Jo, S. M. Raj, C. Rathnasingh, E. Selvakumar, W.-C. Jung, S.
Park, Appl. Microbiol. Biotechnol. 2008, 81, 51 – 60.
20] N. J. Turner, Chem. Rev. 2011, 111, 4073 – 4087.
21] T. Beck, S. Tetter, M. Künzle, D. Hilvert, Angew. Chem. Int. Ed.
2015, 54, 937 – 940; Angew. Chem. 2015, 127, 951 – 954.
8
658 – 8664; d) J. D. Fiedler, S. D. Brown, J. L. Lau, M. G. Finn,
Angew. Chem. Int. Ed. 2010, 49, 9648 – 9651; Angew. Chem. 2010,
22, 9842 – 9845; e) A. O’Neil, C. Reichhardt, B. Johnson, P. E.
Prevelige, T. Douglas, Angew. Chem. Int. Ed. 2011, 50, 7425 –
428; Angew. Chem. 2011, 123, 7563 – 7566; f) D. P. Patterson, B.
1
7
[
Schwarz, K. El-Boubbou, J. van der Oost, P. E. Prevelige, T.
Douglas, Soft Matter 2012, 8, 10158 – 10166; g) D. P. Patterson,
P. E. Prevelige, T. Douglas, ACS Nano 2012, 6, 5000 – 5009.
5] a) C. Ludwig, R. Wagner, Curr. Opin. Biotechnol. 2007, 18, 537 –
[
[
[
[
5
45; b) T. Douglas, M. Young, Science 2006, 312, 873 – 875.
6] a) J. E. Dueber, G. C. Wu, G. R. Malmirchegini, T. S. Moon, C. J.
Petzold, A. V. Ullal, K. L. J. Prather, J. D. Keasling, Nat.
Biotechnol. 2009, 27, 753 – 759; b) C. J. Delebecque, A. B.
Lindner, P. A. Silver, F. A. Aldaye, Science 2011, 333, 470 – 474.
Received: September 8, 2015
Revised: November 3, 2015
Published online: December 23, 2015
1
534
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1531 –1534