Fabrication of Pascal-triangle Lattice of Proteins by Inducing Ligand Strategy

Article abstract of DOI:10.1002/anie.202000771

A protein Pascal triangle has been constructed as new type of supramolecular architecture by using the inducing ligand strategy that we previously developed for protein assemblies. Although mathematical studies on this famous geometry have a long history, no work on such Pascal triangles fabricated from native proteins has been reported so far due to their structural complexity. In this work, by carefully tuning the specific interactions between the native protein building block WGA and the inducing ligand R-SL, a 2D Pascal-triangle lattice with three types of triangular voids has been assembled. Moreover, a 3D crystal structure was obtained based on the 2D Pascal triangles. The distinctive carbohydrate binding sites of WGA and the intralayer as well as interlayer dimerization of RhB was the key to facilitate nanofabrication in solution. This strategy may be applied to prepare and explore various sophisticated assemblies based on native proteins.

Full text of DOI:10.1002/anie.202000771

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Fabrication of Pascal-triangle Lattice of Proteins by Inducing
Ligand Strategy
Authors: Guosong Chen
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202000771
Angew. Chem. 10.1002/ange.202000771
Link to VoR: http://dx.doi.org/10.1002/anie.202000771
http://dx.doi.org/10.1002/ange.202000771

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
Fabrication of Pascal-triangle Lattice of Proteins by Inducing
Ligand Strategy
Rongying Liu[a], Zdravko Kochovski[b], Long Li[a], Yue-wen Yin[c], Jing Yang[a], Guang Yang[a], Guoqing
Tao[a], Anqiu Xu[a], Ensong Zhang[a], Hong-ming Ding*[c], Yan Lu*[b,d], Guosong Chen*[a,e], Ming Jiang[a]
complexity of itself; i.e. it still remains a great challenge to
fabricate Pascal-triangle entities via self-assembly strategy.
Recently, through protein self-assembly, lots of nanostructures
with high complexity have been obtained,[7] including hollow
tubes,[8] ring-like structures,[9] multiscale layers and crystals,[10]
which fully exhibited the high capability of protein self-assembly
to construct sophisticated nano-sized objects. We proposed and
developed a strategy to construct protein assemblies via the dual
non-covalent interactions participated by the small molecular
“inducing ligand”, with which different lectins have been piled up
to form architectures at nanoscale via carbohydrate-protein
interactions and dimerization of rhodamine B (RhB).[11] However,
as far as we know, biomacromolecular Pascal triangles have not
been achieved by using native proteins as building blocks. In our
previous work,[11] the identical carbohydrate-binding sites tended
to result in isotropic protein self-assemblies. While the key to
Pascal triangle construction seems to be distinctive multiple
interactions among building blocks. Hence, the proteins with
distinctive carbohydrate binding sites rather than the identical
carbohydrate-binding sites we previously employed, are probably
capable of constructing Pascal triangle. In this paper, we would
like to probe the feasibility of utilizing self-assembly of native
proteins to fabricate Pascal triangle. Herein, we sought to
leverage wheat germ agglutinin (WGA) to assemble with
sialyllactoside-linked RhB as the small molecular ligand; the
unique nature of eight carbohydrate-binding sites of WGA made
the specific binding among WGA accessible. Through controlling
the ratio of inducing ligand to WGA, Pascal-triangle 2D lattices
with three types of triangular voids, 3D crystals of WGA were
constructed. To the best of our knowledge, such fabrication of
Pascal-triangle 2D lattices from native proteins has not been
reported in the literature. The results reported here demonstrate
great potential of the inducing ligand strategy for construction of
protein Pascal triangle pattern with high complexity and provide
in-depth understanding on the design principles of sophisticated
geometries in living organisms.
Abstract: In this paper, a new type of supramolecular architecture,
protein Pascal triangle has been constructed by using the inducing
ligand strategy that we developed for protein assembly previously.
Although mathematical studies on this famous geometry has a long
history, no work on such Pascal triangle fabricated from native
proteins appeared so far due to their structural complexity. In this work,
by carefully tuning the specific interactions between the native protein
building block WGA and the inducing ligand R-SL, Pascal-triangle 2D
lattice with three types of triangular voids has been assembled.
Moreover, 3D crystal structure was obtained based on the 2D Pascal
triangles. The distinctive carbohydrate binding sites of WGA and
intralayer & interlayer dimerization of RhB was the key to make such
nanofabrication in solution possible. This strategy may be applied to
prepare and explore various sophisticated assemblies based on
native proteins.
Introduction
Fractal geometries such as Sierpinski triangle,[1] tree-like
architectures[2] and other self-similar entities[3] have attracted
increasing attention of chemists due to their importance in
mathematics, engineering and life science.[4] However, Pascal
triangle, as another significant type of pattern which appears
similar to Sierpinski triangle but is not fractal, has been rarely
favored by chemists. Virtually, Pascal triangle, to some extent,
can be regarded as the origin of Sierpinski triangle.[5] In-depth
investigating on Pascal triangle not only provides access to further
ascertain the origin of fractals but also holds great potential for
gaining a coherent description of the design principles underlying
living organisms. [6] Whereas the study of Pascal triangle so far is
still limited to the field of mathemathics as a consequence of high
[a]
R.Y. Liu, L. Li, J. Yang, G.Yang, G.Q. Tao, A.Q. Xu, E.S. Zhang,
Prof. Dr. G.S. Chen, Prof. M. Jiang.
The State Key Laboratory of Molecular Engineering of Polymers and
Department of Macromolecular Science
Fudan University, Shanghai 200433, China
E-mail: guosong@fudan.edu.cn (G. C.)
Results and Discussion
Choices of protein building block and inducing ligand. Given
Pascal triangle shares similar construction rules with Sierpinski
triangle, the two critical rules for construction of the latter are
available to consider:[1] (1) The building blocks with anisotropic
shapes are favored, i.e. V-shape or K-shape. (2) The optimal
selection of driving force is a combination of two or more
interactions with different strength, detailly, the strong interaction
is likely to be responsible for initially assembling the small
molecules into basic building blocks, and then the weak one tends
to enable the basic building blocks to further assemble into
Sierpinski triangle. According to these design rules, both the
[b]
[c]
Dr. Z. Kochovski, Prof. Dr. Y. Lu.
Soft Matter and Functional Materials, Helmholtz-Zentrum Berlin für
Materialien und Energie, 14109 Berlin, Germany
E-mail: yan.lu@helmholtz-berlin.de (Y. L.)
Y. W. Yin, Prof. Dr. H. M. Ding.
Center for Soft Condensed Matter Physics and Interdisciplinary
Research, School of Physical Science and Technology
Soochow University, Suzhou 215006, China
E-mail: dinghm@suda.edu.cn (H. D.)
[d]
[e]
Prof. Dr. Y. Lu.
Institute of Chemistry, University of Potsdam, 14476 Potsdam,
Germany
Prof. Dr. G.S. Chen
Multiscale Research Institute of Complex Systems, Fudan
University, Shanghai 200433, China
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
two.[12] In terms of interactions, the four high-affinity carbohydrate-
binding sites on the top of WGA are roughly identical, and the
same for the four low-affinity ones. Therefore, there are two sets
of carbohydrate-binding sites: four strong binding sites distributed
at two positions present on the top of WGA (labeled as 1 in Fig.
1b) and four weak binding sites distributed at another two
positions locate at the bottom of WGA (labeled as 2 in Fig. 1c).
[12a, 12c]
The isothermal titration calorimetry (ITC) experimently
indicated the presence of two sets of binding sites on a WGA (Fig.
S1).
To fabricate Pascal-triangle 2D lattice, WGA is supposed to be
capable of initially forming a triangular basic building block after
addition of inducing ligand. Based on our previous strategy,[11]
here we synthesized a inducing ligand composed of two parts:
α2,3-sialyllactose, which is responsible for binding with WGA;
RhB, whose dimerization connects the ligand-attached WGA.
Here this inducing ligand was denoted as R-SL (Fig. 1d). The all-
atom molecular dynamics (MD) simulation was firstly employed to
examine whether WGA can form triangular complex after binding
with R-SL. Given that the strong carbohydrate-binding sites
distributed at two positions would be firstly occupied and the steric
effect (detailed discussion will be given in the next part) could
arise from pre-binding R-SL at one position, the WGA attached
by two R-SL was firstly used in simulation. After a 20-ns
equilibrium, the final orientation of two R-SL was not parallel and
their relative orientation to the protein was also different (Fig. 2a).
Such property may not favor the formation of closed dimer but
make triangular trimer construction possible (Fig. 2b, 2c).
Aiming at further estimating the probability of WGA forming
triangular trimers, the coarse-grained (CG) Brownian dynamics
(BD) simulation was applied, where each amino acid was treated
as one CG bead in the simulation (Fig. 2d). Initially, 25 WGA
proteins containing two R-SL (on the strong sites) were placed in
the simulation box (Fig. S2). As time proceeded, the triangular
trimer firstly appeared at 43800  (Fig. S3c), then more and more
Figure 1. a) The shape of a WGA dimer, one monomer was
colored with orange and another monomer was shown as yellow
green (structure adapted from reported crystal structure, PDB
code: 2X52). b) The 4 strong carbohydrate-binding sites at two
positions (white triangle, four strong carbohydrate-binding sites
were labeled as ‘1’ on the top of WGA). c) The 4 weak
carbohydrate-binding sites at another two positions (white triangle,
labeled as ‘2’ at the bottom of WGA). The four black circles in b
and c represent the four distribution positions of binding sites. d)
The molecular structure of inducing ligand (R-SL).
shape and carbohydrate-binding sites of the protein used to
construct Pascal triangle are preferred to be anisotropic.
After careful selection, it was found that wheat germ agglutinin
(WGA) appeares to be an ideal candidate for construction of a
Pascal triangle. WGA is a homodimer, with each monomer
containing four domains and the two monomers dimerize in a
“head-to-tail” fashion forming a horseshoe-shaped complex[12a]
(Fig. 1a). This anisotropic shape seems to favor the Pascal
triangle
construction.
Meanwhile,
eight
independent
carbohydrate-binding sites are concentrated at four positions on
the monomer interface of WGA. The four binding sites on the top
of WGA (Fig. 1b) exhibit a higher affinity than those at the bottom
(Fig. 1c), with a stronger binding constant of at least a factor of
Figure 2. The results of molecular dynamics (MD) simulations and Brownian dynamics (BD) simulations. (a) The all-atom MD result of two ligands binding to the
strong sites of WGA protein. The possible packing way of (b) two WGA proteins and (c) three WGA proteins via RhB dimerization. (d) Schematic illustration of the
coarse-grained (CG) model for the WGA proteins (with two ligands at the strong binding sites). (e) The final snapshot for the assembly of twenty-five proteins in BD
simulation. (f) The probability of different states of proteins in the simulations (averaged by seven independent runs, i.e., Fig. 2e and Fig. S7a-f). Open trimer refers
to trimer with non-closed geometry, while closed trimer means those trimers with closed-loop geometry. For dimer and tetramer, details are shown in Fig S7.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 3. a, b) Cryo-EM image of the WGA 2D lattice (inset of a: enlarged Cryo-EM image of the WGA 2D lattice; inset of b: Fourier transform of selected area). c)
2D class average of 414 sub-areas from Cryo-EM images of the 2D lattice (one white dot represents one WGA protein dimer). d) Overlay of WGA crystal structures
(PDB:2X52) with the 2D class average. e) Pascal- triangle pattern model (three types Pascal triangle pattern coexist, containing: small triangle void (pink, C3
symmetry, 3 nm length side), medium triangle void (blue, C3 symmetry, 5 nm length side), big triangle void (green, C2 symmetry, 7 nm length side)).
triangular trimers can be found in the simulation box (Fig. 2e, 2f,
S3e). However, here the closed dimer was not observed during
the whole simulation period (Fig. 2e, 2f), which was consistent
with the result from all-atom MD simulations. Besides, we also
observed closed tetramers in the simulation, but it was a rare
event, because this type of closing loop would sharply decrease
with increased monomers (Fig. 2f, S7). In addition, due to the
limited time scale of the simulation, some non-closed oligomers
such as open dimers, trimers and tetramers still remained (Fig.
S7). Whereas due to their flexibility, these oligomers were not able
to further assemble into large and ordered structures. In general,
these simulation results suggested the feasibility of our design,
namely, WGA could initially form triangular trimers, which would
be used as basic building block in the subsequent Pascal-triangle
2D lattice construction.
by atomic force microscopy (AFM) (Fig. S6), which was consistent
with the width of WGA, indicating that the 2D lattice was mono-
layered.
The successful construction of Pascal-triangle 2D lattice was
further confirmed by reference-free 2D class averaging shown in
Fig. 3c, 3d. Fourier transformation suggested the crystal nature of
this 2D lattice (Fig. 3b inset). In addition, it was notable that planar
crystals of such large size could grow without support within the
confines and be dispersed in solution, which was supported by
DLS data (Fig. S8).
To identify the driving force of the formation of Pascal-triangle
2D lattice, several control experiments have been conducted.
Firstly, DLS measurement showed that the WGA was not able to
self-assemble into such 2D lattice in the absence of R-SL (Fig.
S9), which was consistent with previous literatures.[13] Secondly,
the UV-vis and circular dichroism (CD) spectra results confirmed
the RhB dimerization between R-SL (Fig. S10). The red-shift in
UV-vis absorption spectrum could be ascribed to the dimerization
of RhB. Meanwhile, the obvious CD signal agreed with the peak
found in UV-vis, further confirmed the occurrence of RhB
dimerization in the chiral environment provided by proteins.
Moreover, the addition of either free 2,3-sialyllactose, which
competitively bound with WGA; or -CD, which inhibited the
dimerization of RhB, leads to the dissociation of the assemblies
(Fig. S11). This again confirmed that the formation of such
Pascal-triangle 2D lattice was mainly driven by the dimerization of
RhB.
Preparation and characterization of the Pascal-triangle 2D
lattice of WGA. The corresponding synthesis details of R-SL
were shown in supporting information (Scheme S1). The
procedure of generating a 2D protein lattice is as follows: equal
molar concentrations of R-SL and WGA were mixed in 4-(2-
hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer
([HEPES] =20 mM, [NaCl] = 40 mM, [CaCl2] = 5 mM); the final
concentration of R-SL and WGA was 2.0 × 10-4 M. The mixture
was incubated at 4 °C for 48 h, then the Pascal-triangle 2D lattice
was obtained as observed under Cryo-EM. The results shown in
Fig. 3a and Fig. S4 demonstrate a clear lattice structure, with
representative areas marked with white boxes. It can be clearly
seen that a triangle structure, rather than the trapezoid shape of
WGA, constituted the 2D lattice (Fig. 3a inset). The self-assembly
of WGA induced by R-SL was also confirmed by dynamic light
scattering (DLS) (Fig. S5). The height of this 2D lattice was about
4.1 nm as revealed
We used cryo-EM to trace the formation process of the Pascal-
triangle 2D lattice. Some triangular aggregates with about 7 nm
side length were firstly observed after incubation for 8 h (Fig.
S12a). When increasing the incubation time to 16 h,
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
Scheme 1. The proposed self-assembly mechanism underlying the formation of Pascal-triangle 2D lattice with p3 symmetry. It includes two sets of RhB dimerization
(RhB dimerization at strong binding sites results in triangular trimer with C3 symmetry, RhB dimerization at weak binding sites gives rise to trimer with C3 symmetry,
combination of two sets of RhB dimerization leads to hexamer with C2 symmetry.
clusters composed of three triangular aggregates were also
detected (Fig. S12b). In addition, the DLS analysis confirmed the
sequential formation from triangular trimers to clusters (Fig. S8).
In short, the self-assembly process observed here was consistent
to our design.
the remaining two strong binding sites and two weak binding sites
having adjacent pre-binding ligands were finally occupied.
Notably, the experiment and theoretical calculation results
suggested that about four ligands bound to each protein as the
ratio of inducing ligand R-SL/WGA was set at 1:1 (Fig. S15, Table
S1). Consequently, based on above discussion, for the four
ligands, two of them would firstly bind to the strong sites
distributed at different positions, allowing the formation of
triangular trimers, then another two ligands may occupy two weak
binding sites located at different positions, permitting the
triangular trimers to assemble into clusters and further into
Pascal-triangle 2D lattice. On the contrary, when the ratio of R-SL
to WGA was decreased to 0.5:1 (in this case, the average number
of ligands binding to each protein was about two, see Table S1
and Fig. S15), only triangular trimers but no 2D lattice appeared
(Fig. S16). In other words, the dimerization of RhB at the strong
binding sites can only induce the formation of triangular trimers
but that at the weak binding sites is essential for the further
assembly of WGA into Pascal-triangle 2D lattice.
Based on the simulation and experimental results, the
mechanism underlying the formation of the Pascal-triangle 2D
lattice was proposed as illustrated in Scheme 1. At first, the RhB
dimerization at strong binding sites gave rise to triangular trimers;
then the RhB dimerization at weak binding sites tended to trigger
the assembly of the triangular trimers into clusters and further into
a Pascal-triangle 2D lattice.
More interestingly, along with WGA assembled into a Pascal-
triangle pattern with p3 symmetry, three types equilateral triangle
voids were also formed. Detailly, three WGAs made up an
equilateral triangle voids with 3 nm side-length and C3 symmetry
by RhB dimerization at strong binding sites. Similarly, equilateral
triangle voids with 5 nm side-length, C3 symmetry by RhB
dimerization at weak binding sites, and six WGAs made up an
How did the triangular trimers form and further assemble into a
Pascal-triangle 2D lattice? Given the presence of two sets of
binding sites on a WGA (Fig. S1) and the proximity of two binding
sites at each position (Figure 1b, 1c), the steered all-atom
molecular dynamics (MD) simulation was employed to test the
probability of the ligand binding to the different sites, where one
free ligand was pulled to the strong binding site of the WGA with
a very slow relative velocity (~0.2 nm/ns) in the absence (case 1)
and presence (case 2) of the adjacent pre-binding ligand, and to
the weak binding site in the absence (case 3) and presence (case
4) of the adjacent pre-binding ligand respectively (Fig. S13a). As
shown in Fig. S13b, the pulling forces exhibited a sharp decrease
when the ligand began to pack into the binding site in case 1 and
case 3, but the decrease in case 1 was more sharp than that in
case 3, suggesting that it was easier for the ligand to be occupied
to the strong binding site. More importantly, the pulling forces both
increased monotonously in case 2 and case 4, possibly since the
steric effect between the free ligand and the pre-binding ligand
(Figure S13c, d). As a result, the probability of the ligand binding
to the strong site with adjacent pre-binding ligand (i.e., case 2)
became lower than that of the ligand binding to the weak site
without adjacent pre-binding ligand. Moreover, we also compared
the binding energy of the WGA and the ligand, and the binding
energy in case 1 was the lowest among the four cases (Figure
S14). On the basis of above discussion, the order of binding
priority appears to be as follows: case 1 > case 3 > case 2 > case
4, namely, two strong binding sites distributed at different
positions tended to be firstly occupied, and then two weak binding
sites distributed at different positions subsequently were occupied, equilateral triangle voids with 7 nm side-length, C2 symmetry
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
Figure 4. a) the AFM height image of R-SL/WGA (1.5:1) after incubation 6 h. b) The AFM height image and c) corresponding height profiles of (R-
SL/WGA (1.5:1) after incubation 12 h. d) Negative stain TEM (inset: 2D class average image of selected area), e) Cryo-EM (inset: Fourier transform of
selected area) and f) AFM height image of 3D crystals (R-SL/WGA (1.5:1) after incubation 48h.
tended to promote the 2D lattice stacking and 3D crystal
growth.[18] Therefore, it can be envisioned that more inducing
ligands may give rise to the stacking of 2D lattice and 3D crystal
growth.
results from RhB dimerization at both strong and weak binding
sites (Fig. 3e, Scheme 1).
Although some transmembrane proteins can self-assemble into
2D lattice in a membrane environment, they exist as the free-
monomer state in solution.[14] Additionally, some researchers
found that a 2D lattice could be generated by designed protein
interacting interface.[7h, 15] However, we checked all the 200 types
(up to now) of 2D lattices from Reticular Chemistry Structure
Resource (RCSR),[16] no similar Pascal-triangle 2D lattice with
three types of voids could be found.
To test this possibility, the ratio of R-SL to WGA was directly
increased to 1.5:1. CD and Cryo-EM were firstly employed to
trace the assembly process. During the initial 6 h of incubation, a
pronounced RhB dimerization signal increase was observed in
CD spectra (Fig. S19), which often appeared as the RhB
dimerization transformed from a free state into a confined state.[18]
Indeed, as observed by Cryo-EM, some small 3D aggregates
appeared after 6 h incubation, despite many Pascal triangles
remained (Fig. S20a). This indicated that more inducing ligands
participated in interlayer RhB dimerization, promoting the growth
of 3D aggregates[18]. When increasing the incubation time to 12 h,
the CD signal assigned to RhB dimerization increased
significantly (Fig. S19), indicating that most of the Pascal triangles
assembled into larger 3D aggregates (Fig. S20b). Therefore, it
can be deduced that more inducing ligands gave rise to interlayer
RhB dimerization compared to the case of 2D lattice, which
enabled the formation of 3D aggregates.
AFM allowed further detection of the assembly process. After
incubation of 6 h, both single-layer 2D lattice and multi-layer small
aggregates were observed (Fig. 4a). Upon increasing the
incubation time to 12 h, some larger aggregates appeared (Fig.
4b). A common feature shared by these aggregates was that their
thicknesses were integral multiple of that of a single-layer 2D
lattice (4.1 nm). Typically, even in the same aggregate, there were
two different thicknesses: 4.1 nm and 8.0 nm (Fig. 4c), implying
that certain parts of this aggregate already grown in the third
dimension, while the other areas still retained a single-layer
Fabrication and characterization of 3D crystals. After
elucidating the self-assembly mechanism of the Pascal-triangle
2D lattice, we next investigated whether the 2D lattice could be
further packed into 3D crystals. As an important driving force of
protein self-assembly, protein-protein interaction was firstly
considered to promote the stacking of 2D lattices. Thus, we
attempted to extend incubation time. However, as the ratio of R-
SL/WGA was set at 1:1, no obvious 3D crystals came to emerge
even incubated for 120 h (Fig. S17). Further, more NaCl was
added into the assembly solution to enhance the protein
aggregation tendency, whereas only a few small 3D crystals can
be observed (Fig. S18). This suggested that here the protein-
protein interaction could indeed, to some extent, promote the
stacking of 2D lattice but was not strong enough to generate large
3D crystals. To strengthen the interaction between layers, given
the carbohydrate-binding sites of WGA were still unsaturated as
the ratio of R-SL to WGA was set at 1:1, increasing this ratio may
be a practical way to permit more binding sites to be occupied (or
even become saturated).[11a,11d,17] In addition, it was reported that
the interlayer dimerization of RhB
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
structure. In other words, a hierarchical self-assembly occured in
two and three dimensions simultaneously.
[1]
(a) Newkome, G. R.; Wang, P.; Moorefield, C. N.; Cho, T. J.; Mohapatra,
P. P.; Li, S.; Iancu, V. Science. 2006, 312(5781), 1782-1785. (b) Sarkar,
R.; Guo, K.; Moorefield, C. N.; Saunders, M. J.; Wesdemiotis, C.;
Newkome, G. R. Angew. Chem. Int. Ed. 2014, 126(45), 12378-12381. (c)
Newkome, G. R.; Moorefield, C. N. Chem. Soc. Rev. 2015, 44(12), 3954-
3967. (d) Zhang, X.; Li, N.; Gu, G. C.; Wang, H.; Nieckarz, D.; Szabelski,
P.; Lu, J. T; Wu, K; Wang,Y,F. ACS Nano. 2015, 9(12), 11909-11915. (e)
Li, N.; Zhang, X.; Gu, G. C.; Wang, H.; Nieckarz, D.; Szabelski, P.; Peng,
L. M. Chinese. Chem. Lett. 2015, 26(10), 1198-1202. (f) Zhang, X.; Li,
N.; Liu, L.; Gu, G.; Li, C.; Tang, H.; Wang, Y. Chem. Commun.
2016, 52(69), 10578-10581. (g) Xing, L.; Peng, Z.; Li, W.; Wu, K. Acc.
Chem. Res, 2019, 52(4): 1048-1058. (h) Sun, Q.; Cai, L.; Ma, H.; Yuan,
C.; Xu, W. Chem.Commun. 2015, 51(75), 14164-14166. (i) Jiang, Z.; Li,
Y.; Wang, M.; Liu, D.; Yuan, J.; Chen, M.; Li, X; Wang, P. Angew. Chem.
Int. Ed. 2017, 129(38), 11608-11613.
Notably, regular 3D crystals appeared after 48 h incubation (Fig.
4d-e). The AFM height image suggested that the thickness of 3D
crystals was about 50 nm (Fig. 4f), which corresponded to about
twelve layers of WGA. Significantly, both the results from 2D class
average and Fourier-transform for this 3D crystal (Fig. 4d inset,
4e inset) were similar to those of 2D lattices (Fig. 3c, 3b inset),
suggesting that the 3D crystals shared analogous protein packing
with 2D lattices.
To summarize, a possible self-assembly mechanism of 3D
crystals was proposed as illustrated in Fig. S21. Initially, the WGA
still assembled into triangular trimers through RhB dimerization at
strong binding sites. Then, the RhB dimerization at weak binding
sites enabled the triangular trimers assemble into 2D clusters; at
the same time, the interlayer RhB dimerization contributed by
more inducing ligands allowed the clusters to further grow in the
third dimension, namely, more inducing ligands permitted the
simultaneous occurrence of self-assembly in two dimensions and
three dimensions. As a result, these clusters gradually evolved
into 3D crystals through self-assembly.
[2]
[3]
(a) Khire, T. S.; Kundu, J.; Kundu, S. C.; Yadavalli, V. K. Soft Matter.
2010, 6(9), 2066-2071. (b) Shin, S.; Gu, M. L.; Yu, C. Y.; Jeon, J.; Lee,
E.; Choi, T. L. J. Am. Chem. Soc. 2018, 140(1), 475-482. (c) Hernández,
N. E.; Hansen, W. A.; Zhu, D.; Shea, M. E; Khalid. M.; Manichev, V.;
Putnins, M.; Chen, M. Y.; Dodge, A. G.; Yang, L.; Berríos, I. M.; Banal,
M.; Rechani, P.; Gustafsson, T.; Feldman, L. C.; Lee, S. H.; Wackett, L.
P.; Dai, W.; Khare, S. D. Nat. Chem. 2019, 11(3): 605-614.
(a) Wang, M.; Wang, C.; Hao, X. Q.; Liu, J.; Li, X.; Xu, C.; Li, X. J. Am.
Chem. Soc. 2014, 136(18), 6664-6671. (b) Jiang, Z.; Li, Y.; Wang, M.;
Song, B.; Wang, K.; Sun, M.; Li, X; Wang, P. Nat. Commun. 2017, 8,
15476-15484. (c) Wang, L.; Liu, R.; Gu, J.; Song, B.; Wang, H.; Jiang,
X.; Wang, M; Li, X. J. Am. Chem. Soc. 2018, 140(43), 14087-14096. (d)
Yin, G. Q.; Wang, H.; Wang, X. Q.; Song, B.; Chen, L. J.; Wang, L.; Li,
X. Nat. Commun. 2018, 9(1), 567-576.
Conclusion
In this work, we constructed Pascal-triangle 2D lattices using well-
selected protein building blocks with anisotropic shapes and two
sets of carbohydrate binding sites. Moreover, the dynamic and
exchangeable nature of non-covalent interactions make the
further manipulation on 2D lattices to 3D crystals possible. Such
results not only demonstrate the first construction of Pascal-
triangle 2D lattice from native protein, but also drive the protein
lattices fabricated by our inducing ligand strategy to an
unprecedented level. We believe that this work opens a new
avenue on fabrication of novel protein assemblies with desired
shapes, sizes and functionalities, which might help us to
understand the design principles underlying living organism better.
[4]
[5]
Barnsley, M. F. Fractals Everywhere. Morgan Kaufmann Publishers:
Burlington, MA, 1993.
(a) Lakhtakia, A.; Messier, R.; Varadan, V. K.; Varadan, V. V. J. Phys. A:
Math. Gen. 1987, 20(11), 735-738. (b) Reiter, A. M. Fibonacci. Quart.
1991, 31(2), 112-120.
[6]
[7]
(a) Weibel, E. R. Am. J. Physiol-Lung.C. 1991, 261(6): 361-369. (b) Losa,
G. A. Fractals in biology and medicine. 2006. (c) Reznikov, N; Bilton, M.;
Lari, L.; Stevens, M. M.; Kröger, R. Science. 2018, 360, No. eaao2189.
(a) Salgado, E. N.; Radford, R. J.; Tezcan, F. A. Acc. Chem. Res. 2010,
43, 661 – 672. (b) Koji, O.; Akira, O.; Takashi, H. Chem.Commun. 2012,
48, 11714 – 11726; (c) Yeates, T. O.; Liu, Y.; Laniado, J. Curr. Opin.
Struct. Biol. 2016, 39, 134 – 143. (d) Luo, Q.; Hou, C.; Bai, Y.; Wang, R.;
Liu, J. Chem. Rev. 2016, 116(22), 13571-13632. (e) Bai, Y.; Luo, Q.; Liu,
J. Chem. Soc. Rev. 2016, 45(10), 2756-2767. (f) Sun, H.; Luo, Q.; Hou,
C.; Liu, J. Nano Today. 2017, 14, 16-41. (g) Baneyx, F.; Matthaei, J.
F. Curr. Opin. Biotech. 2014, 28, 39-45. (h) Norn, C. H.; André, I. Curr.
Opin. Struc. Biol. 2016, 39, 39-45. (i) King, N. P.; Bale, J. B.; Sheffler,
W.; McNamara, D. E.; Gonen, S., Gonen; Baker, D. Nature. 2014, 510,
103-108. (j) Huang, P. S.; Boyken, S. E.; Baker, D. Nature. 2016, 537,
320-327. (k) Klinker, K; Schafer, O; Barz, M. Angew. Chem. Int. Ed.
2017, 56(32), 9608-9613.
Acknowledgements
G.C. thanks NSFC/China (No. 51721002, 21861132012 and
91527305) for financial support. Y.L. thanks the Deutsche
Forschungsgemeinschaft (DFG, German Research Foundation) -
Project number 410871749 for financial support. H.D. thanks
NSFC/China (No. 21604060 and 11874045) for financial support.
The authors would like to thank the Joint Lab for Structural
Research at the Integrative Research Institute for the Sciences
(IRIS Adlershof, Berlin) for Cryo-EM imaging. We thank Dr. Xiao
Xu and Dr. Qidi Ran for helpful discussions. This work is
supported by Shanghai Municipal Science and Technology Major
Project (No.2018SHZDZX01) and ZJ Lab.
[8]
[9]
(a) Brodin, J. D.; Smith, S. J.; Carr, J. R.; Tezcan, F. A. J. Am. Chem.Soc.
2015, 137, 10468 – 10471. (b) Biswas, S.; Kinbara, K.; Oya, N. I.;
Taguchi, H.; Aida, T. J. Am. Chem. Soc. 2009, 131, 7556 –7557. (c) Miao,
L.; Fan, Q.; Zhao, L.; Qiao, Q.; Zhang, X.; Hou, C.; Liu, J. Chem.
Commun. 2016, 52(21), 4092-4095.
(a) Chou, T. F.; So, C.; White, B. R.; Carlson, J. C.; Sarikaya, M.; Wagner,
C. R. ACS Nano. 2008, 2(12), 2519 – 2525. (b) Bai, Y.; Luo, Q.; Zhang,
W.; Miao, L.; Xu, J.; Li, H.; Liu, J. J. Am. Chem. Soc. 2013,135(30), 10966
– 10969.
[10] (a) Li, X.; Qiao, S.; Zhao, L.; Liu, S.; Li, F.; Yang, F.; Liu, J. ACS Nano.
2019, 13(2), 1861-1869. (b) Der, B. S.; Kuhlman, B. Curr. Opin. Struc.
Biol. 2013, 23(4), 639-646.
Keywords: inducing ligand • pascal triangle • self-assembly •
[11] (a) Sakai, F.; Yang, G.; Weiss, M. S.; Liu, Y.; Chen, G.; Jiang, M. Nat.
Commun. 2014, 5, 4634-4641. (b) Yang, G.; Zhang, X.; Kochovski, Z.;
Zhang, Y.; Dai, B.; Sakai, F.; Liu, C; Chen, G.; Jiang, M. J. Am. Chem.
Soc. 2016, 138(6), 1932-1937. (c) Yang, G.; Ding, H. M.; Kochovski, Z.;
WGA
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
Hu, R.; Lu, Y.; Ma, Y. Q.; Chen, G.; Jiang, M. Angew. Chem. Int. Ed.
2017, 56(36), 10691-10695. (d) Qi, W.; Zhang, Y.; Kochovski, Z.; Wang,
J.; Lu, Y.; Chen, G.; Jiang, M. Nano Research. 2018, 11(10), 5566-5572.
[12] (a) Wright, C. S.; Kellogg, G. E. Protein. Sci. 1996, 5(8), 1466-1476. (b)
Lis, H.; Sharon, N. Chem. Rev. 1998, 98(2), 637-674. (c) Wright, C. S. J.
Biol. Chem. 1992, 267(20): 14345-14352.
[13] (a) Muraki, M.; Ishimura, M.;
& Harata, K. BBA. Gen. Subjects.
2002, 1569(1-3), 10-20. (b) Wright, C. S.; Jaeger, J. J. Mol. Biol. 1993,
232(2): 620-638.
[14] (a) Reviakine, I.; Bergsma, S. W.; Mazeres, D. C.; Govorukhina, N.;
Brisson, A. J. Struc. Biol. 2000, 131(3), 234-239. (b) Richter, R. P.; Him,
J. L. K.; Tessier, B.; Tessier, C.; Brisson, A. R. Biophys. J. 2005, 89(5),
3372-3385.
[15] (a) Gonen, S.; DiMaio, F.; Gonen, T.; Baker, D. Science. 2015, 348,
1365-1368. (b) Suzuki, Y.; Cardone, G.; Restrepo, D.; Zavattieri, P. D.;
Baker, T. S.; Tezcan, F. A. Nature. 2016, 533, 369-373.
[16] O’Keeffe, M.; Peskov, M. A.; Ramsden, S. J.; Yaghi, O. M. Acc. Chem.
Res. 2008, 41(12), 1782-1789.
[17] Dotan, N.; Arad, D., Frolow; F.; Freeman.A. Angew. Chem. Int. Ed.
1999, 38(16), 2363-2366.
[18] Brodin, J. D.; Ambroggio, X. I.; Tang, C.; Parent, K. N.; Baker, T. S.;
Tezcan, F. A. Nat. Chem. 2012, 4(5), 375-382.
This article is protected by copyright. All rights reserved.

10.1002/anie.202000771
Angewandte Chemie International Edition
RESEARCH ARTICLE
Entry for the Table of Contents
RESEARCH ARTICLE
Rongying Liu, Zdravko
We
constructed
Pascal-
Kochovski, Long Li, Yue-wen
Yin, Jing Yang, Guang Yang,
Guoqing Tao, Anqiu Xu,
Ensong Zhang, Hong-ming
Ding*, Yan Lu*, Guosong
Chen*, Ming Jiang
triangle lattice by using
a
careful selection of protein
building blocks with anisotropic
shapes and two sets of
carbohydrate binding sites.
Moreover, the dynamic and
exchangeable nature of non-
covalent interactions make the
further manipulation on 2D
lattices to 3D crystals possible.
Page No. – Page No.
Fabrication of Pascal-triangle
Lattice of Proteins by
Inducing Ligand Strategy
This article is protected by copyright. All rights reserved.