Programmable living assembly of materials by bacterial adhesion

Abstract

The field of engineered living materials aims to construct functional materials with desirable properties of natural living systems. A recent study demonstrated the programmed self-assembly of bacterial populations by engineered adhesion. Here we use this strategy to engineer self-healing living materials with versatile functions. Bacteria displaying outer membrane-anchored nanobody–antigen pairs are cultured separately and, when mixed, adhere to each other to enable processing into functional materials, which we term living assembled material by bacterial adhesion (LAMBA). LAMBA is programmable and can be functionalized with extracellular moieties up to 545 amino acids. Notably, the adhesion between nanobody–antigen pairs in LAMBA leads to fast recovery under stretching or bending. By exploiting this feature, we fabricated wearable LAMBA sensors that can detect bioelectrical or biomechanical signals. Our work establishes a scalable approach to produce genetically editable and self-healable living functional materials that can be applied in biomanufacturing, bioremediation and soft bioelectronics assembly.

Full text

Articles
https://doi.org/10.1038/s41589-021-00934-z
1Materials Synthetic Biology Center, CAS Key Laboratory of Quantitative Engineering Biology, Guangdong Provincial Key Laboratory of Synthetic Genomics,
Shenzhen Institute of Synthetic Biology, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. 2School of
Biomedical Engineering, Sun Yat-sen University, Guangzhou, China. 3Soft Bio-interface Electronics Laboratory, Center of Neural Engineering, CAS Key
Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institute of Artificial Intelligence and Robotics for Society, Shenzhen Institutes
of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. 4Department of Biomedical Engineering, Duke University, Durham, NC, USA.
5These authors contributed equally: Wei Kang, Jing Sun, Runtao Zhu. ✉e-mail: zy.liu1@siat.ac.cn; zj.dai@siat.ac.cn
Natural organisms can produce living materials that can pro-
cess signals, catalyze cascade reactions and heal themselves.
Inspired by nature, the field of engineered living materials
(ELMs) aims to rewire cells to assemble materials that recapitulate
the desirable properties of natural living system1–4. Most efforts
in ELM development have been made in biofilm engineering5–9.
Bacterial biofilms are composed of living cells that generate and
become embedded within an extracellular matrix. During biofilm
formation, certain proteins are secreted and self-assembled into
amyloid fibers as extracellular matrix scaffolds. Escherichai coli curli
is a representative amyloid fiber system that has been extensively
developed to build ELMs5,10. However, the secretion of the curli
system is limited to short peptides or protein domains containing
up to roughly 100 amino acids, restraining its capacity for decora-
tion with functional moieties8,9. Although artificial biofilm systems
are not totally subject to the disadvantages of naturally occurring
biofilm, such as surface attachment during bacterial culture, they
still suffer from limitations in processability and use in large-scale
applications11–13.
One of the most appealing properties of ELMs is self-healing,
since the encapsulated cells of the biofilm can self-regenerate and
heal the material over time. This self-healing process in theory is
caused by cell growth (in hours). However, in applications such as
soft and conformable devices, the healing process has to happen in
minutes, which is much faster than the cell cycle. In polymer chem-
istry, however, numerous studies have reported the engineering of
fast self-healing material by incorporating noncovalent interactions
such as hydrogen bonds14,15. Inspired by this strategy in self-healing
polymer engineering, here we sought to functionalize each
bacterium with noncovalent binding groups to generate macro-
scopic living functional materials by adhering bacteria together.
Previous work developed a synthetic bacterial cell–cell adhesion
toolbox based on outer membrane-anchored nanobody (Nb)–anti-
gen (Ag) pairs and demonstrated the self-assembly of morpholo-
gies and patterns at the microscopic scale16. Using this toolbox, we
engineered cells to display Nb or Ag on their surfaces and cultured
these cells separately in liquid cultures. Mixing the two popula-
tions in large quantities generates processable macroscopic materi-
als, which we term living assembled material by bacterial adhesion
(LAMBA). LAMBA demonstrates better mechanical properties and
processability than individual cells (Nb or Ag cells), and is adapt-
able to versatile engineering methods to fabricate macroscale or
microscale objects. LAMBA is also genetically programmable. By
functionalizing the LAMBA extracellularly and intracellularly, our
material is able to degrade organophosphate pesticides through a
hybrid enzyme–inorganic sequential catalysis or synthesize treha-
lose by harnessing extracellular–intracellular bioconversions.
By design, LAMBA can self-heal. Besides the self-healing pro-
cess led by cell growth, adhesion between Nb–Ag pairs leads to
the recovery of the material property within several minutes. For
example, sliced LAMBA, after recovery, can maintain its conductiv-
ity under multiple cycles of stretching. This property could address
the fatigue issue often occurring in wearable devices. To this end, we
constructed LAMBA sensors that can detect bioelectrical or biome-
chanical signals, for example, electromyography (EMG) and joint
bending signals.
Our study exploited the engineered cell adhesion to prepare a
highly processable, programmable and self-healable living functional
Programmable living assembly of materials by
bacterial adhesion
Baizhu Chen 1,2, Wei Kang1,5, Jing Sun3,5, Runtao Zhu1,5, Yue Yu1, Aiguo Xia1, Mei Yu3, Meng Wang1,
Jinyu Han1, Yixuan Chen1, Lijun Teng3, Qiong Tian3, Yin Yu1, Guanglin Li3, Lingchong You 4,
Zhiyuan Liu 3 ✉ and Zhuojun Dai 1 ✉
The field of engineered living materials aims to construct functional materials with desirable properties of natural living systems.
A recent study demonstrated the programmed self-assembly of bacterial populations by engineered adhesion. Here we use
this strategy to engineer self-healing living materials with versatile functions. Bacteria displaying outer membrane-anchored
nanobody–antigen pairs are cultured separately and, when mixed, adhere to each other to enable processing into functional
materials, which we term living assembled material by bacterial adhesion (LAMBA). LAMBA is programmable and can be func-
tionalized with extracellular moieties up to 545 amino acids. Notably, the adhesion between nanobody–antigen pairs in LAMBA
leads to fast recovery under stretching or bending. By exploiting this feature, we fabricated wearable LAMBA sensors that
can detect bioelectrical or biomechanical signals. Our work establishes a scalable approach to produce genetically editable
and self-healable living functional materials that can be applied in biomanufacturing, bioremediation and soft bioelectronics
assembly.
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles NATure CHeMiCAL BiOLOgy
material, LAMBA, in a fast and scalable fashion. With these advan-
tages, we further explored the applications of LAMBA in biomanu-
facturing, bioremediation and bioelectronics.
Results
Engineering LAMBA by programmable cell–cell adhesion.
We used engineered bacteria with adhesive pairs (selected from a
library of adhesion sequences from the VIB Nanobody Core) as
monomers to generate a suitable mixture for LAMBA (Fig. 1a).
MG1655 cells were first engineered to display Nb or Ag cells on the
outer membrane (Supplementary Tables 1 and 2)16. Mixing Nb cells
with Ag cells led to gradual aggregation (Fig. 1b), with most of the
self-assembly process (roughly 70%) completed within the first 3 h.
Further incubation for 12 h recruited more than 90% of the cells into
the living material (Fig. 1b). We performed dynamic shear rheology
experiments in which storage and loss modulus were monitored
and compared for the assembled mixture and individual cells at the
same cell density. The storage modulus of the mixture increased by
roughly twofold, suggesting the strengthening of mechanical rigid-
ity after cell–cell adhesion (Fig. 1c and Supplementary Fig. 1a–c).
The viscosity of the mixture increased by roughly tenfold compared
with the individual cells (Fig. 1d and Supplementary Fig. 1d–f).
The resultant viscoelasticity of the mixture is similar to a hydrogel
(Supplementary Fig. 2), making it more suitable as a bioink (inks
composed of the cells) for three-dimensional (3D) printing than the
individual cells (Fig. 1d and Supplementary Fig. 3a).
We then adapted the assembled cell mixture to multiple process-
ing techniques and shaped it into either macroscale or microscale
LAMBA structures (Fig. 1e–j). We first labeled Ag cells and Nb cells
with green fluorescent protein (GFP) and mCherry, respectively,
then generated and collected the mixture as a bioink for 3D print-
ing. We were able to fabricate multiple features with well-defined
shapes, including square with grids and letters (Fig. 1e–g). The
green (Ag cells) and red (Nb cells) fluorescence of these struc-
tures suggested the composition of the LAMBA. The system can
be printed in layers. We printed 30 layers of LAMBA to create a
designed cube at roughly 5 mm height, suggesting that the LAMBA
can be processed into 3D objects (Fig. 1f and Supplementary
Fig. 3b,c). Notably, LAMBA is regenerable due to its living nature.
We first printed the pattern onto a glass slide, and placed the printed
pattern in direct contact with a nutrient-rich agar plate. The pat-
tern regenerated after overnight culture at room temperature
(Fig. 1g and Supplementary Fig. 3d). The regenerated pattern could
be used for a second print with high fidelity, underscoring the
robust self-regenerative capacity of our system.
LAMBA can also be processed into microscale fibers. Briefly,
the Nb/Ag cell mixture, sodium alginate and calcium chloride were
injected sequentially into separate channels on the microfluidic
As printed First regeneration Second regeneration
Green channel
Merged
Red channel
0 h
1 h
2 h
3 h
4 h
5 h
e
g
h
i j
d
c
ab
Merged
Red
Green
Bright field
BF
f
CaCl2
LAMBA
Sodium alginate
Core-flow channel
Sample-flow channel
Sheath-flow
channel
0 3 6 9 12
0
50
100
Cell density/ (%)
Time (h)
**
102
103
G' (Pa)
Strain per %
104
10–2 100102
10–1 100101
100
102
104
η per Pa.s
Shear rate per s–1
Living assembled materials
by bacterial adhesion
Nanobody
Antigen
Fig. 1 | LAMBA can be flexibly processed into diverse structures. a, Engineered adhesion between bacterial populations leads to autonomous assembly.
b, Self-assembly was mostly completed within 3 h. OD600 (normalized by the value at time zero) of the mixture (mixing Nb and Ag cells at a 1:1 volume
ratio) is significantly lower than that of the individual cells (** indicates P < 0.01 (P = 1 × 10−8 or P = 1 × 10−5 for LAMBA compared with Ag or Nb cells)
by a two-sided Student’s two-sample t-test assuming unequal variances; error bars = s.d. (n = 3)). Insert photographs show that mixing two cells (top
and middle tube) underwent self-assembly (aggregation in the bottom tube). c, LAMBA had a stronger storage modulus. The strain varied from 10 to
0.1% (frequency, 1 rad s−1). d, LAMBA was more viscous. The sheer rate ranged from 0.01 per s to 100 per s. Experiments in c and d were repeated more
than three times independently (Supplementary Fig. 1). e, Construction of macroscale LAMBA objects via 3D printing. The green and red fluorescence
indicated the GFP labeled Ag cells and mCherry labeled Nb cells. Scale bar, 5 mm. f, LAMBA being printed in 30 layers. The final object attained a height
of roughly 4.8 mm. Scale bar, 2 mm. g, The LAMBA pattern was self-regenerative. We printed the pattern onto a glass slide and stamped it onto a LB agar
plate. The pattern regenerated after culturing overnight (room temperature). The regenerated pattern could print again with reasonable fidelity. Scale bar,
5 mm. h, Generation of microscale fibers by chip-based microfluidics. The Nb/Ag cell mixture, sodium alginate and CaCl2 were injected sequentially into
the different channels. The resultant fiber was further solidified in the CaCl2 and then cultured in LB for cell growth. i, LAMBA grew into living fibers. The
fluorescence indicated that the fiber was composed of Ag cells (GFP) and Nb cells (mCherry). Scale bar, 200 μm. j, The diameter of the fiber was flexibly
modulated. Tuning the device (Methods) allowed the flexible modulation of the fiber size. Scale bar, 500 μm. Experiments in i and j were repeated more
than three times independently with similar results.
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles
NATure CHeMiCAL BiOLOgy
device. The gelation of alginate (triggered by calcium ions) formed
the fiber sheath, with the LAMBA encapsulated inside (Fig. 1h and
Supplementary Fig. 4a,b). We then cultured the fiber in Luria–
Bertani (LB) medium. The nutrients in the medium could diffuse
through the crosslinked alginate, and the LAMBA grew and filled
the hollow space of the sheath (Fig. 1i and Supplementary Video
1). The diameter of the living fiber could be modulated flexibly
from roughly 100 to 400 μm by adjusting the structure of the chan-
nel (Fig. 1j and Supplementary Fig. 4c). The fiber could be hung
by tweezers and persevered in water with shape fidelity for over
6 months (Supplementary Fig. 5).
Programmable LAMBA enabled diverse bioconversions.
LAMBA uses multiple cell populations, and these individual cells
can be functionalized extracellularly or intracellularly. In nature,
executing a function through division of labor can be advantageous
in reducing engineering complexity and improving production
efficiency17,18. Inspired by division of labor, we allocated differ-
ent missions to individual populations inside LAMBA to degrade
organophosphate pesticides through a hybrid enzyme–inorganic
sequential catalysis, or synthesize trehalose by harnessing extracel-
lular–intracellular bioconversions.
We first engineered LAMBA that could degrade paraoxon
(PAR) (Fig. 2a). To functionalize LAMBA extracellularly, we fused
organophosphate hydrolase (OPH, 337 amino acids) with an outer
membrane expression protein YiaT19. Our results showed that
OPH was successfully decorated onto the membrane of Ag cells
(Supplementary Fig. 6a). We next engineered Nb cells to immobi-
lize gold nanoparticles (AuNPs) through coordination the between
His-tag and Ni-NTA (Fig. 2a and Supplementary Fig. 7). PAR was
first converted to less harmful paranitrophenol (PNP) by Ag cells
(displaying OPH) of the LAMBA, and was further reduced to harm-
less p-aminophenol (PAP) by the AuNPs immobilized on Nb cells
of the LAMBA with addition of NaBH4 (Fig. 2b and Supplementary
Fig. 8). Cells without synthetic adhesion (either free floating or
spun down) could convert PAR to PAP, with a smaller conversion
ratio (roughly 89 and 26% for PAR to PNP) compared with LAMBA
(94%) (Supplementary Fig. 9). When AuNPs were not bound to the
Ag cells, the reduction from PNP to PAP was less efficient, possibly
due to an enhanced local AuNP concentration when immobilized
on the surface of the bacteria (Supplementary Fig. 10a). The mate-
rial was recyclable to process multiple rounds of pollutant degrada-
tion (Fig. 2c). The gradual loss of performance was partially caused
by cell loss during the recycling process (Supplementary Fig. 10b).
Whole-cell biocatalysis with native or rewired pathways presents a
green, scalable and easy-to-operate biomanufacturing approach20,21.
However, poor intracellular transport of the substrate often leads
to low transformation efficiency22,23. Therefore, we implemented a
two-step trehalose synthesis by allocating substrate digestion and
enzymatic conversion to different units of the LAMBA (Fig. 2d).
We programed Ag cells to display β-amylase (BA, 545 amino acids)
extracellularly by the same strategy mentioned above, and Nb cells
to express trehalose synthase (TreS) intracellularly (Fig. 2d and
Supplementary Fig. 6). The resultant LAMBA was then fed with
starch. The starch was first digested into maltose extracellularly by
Ag cells (displaying BA). The lower molecular weight of the maltose
allowed effective intracellular transportation and was subsequently
converted into trehalose by Nb cells (expressing TreS) (Fig. 2e and
His
PAR PNPPAP
Ag
Yiat-
OPH
His-Nb
Au
12345
0.6
0.8
1.0
Conc-R/conc-O
Cycles
PAR PNP PAP
a cb
fd e
10 15 20 25
0
25
50
Intensity
Time (min)
Free cells
LAMBA
Glucose
Maltose
Trehalose
0 100 200
0
0.05
0.10
Conc. (mM)
Time (min)
Starch Maltose Trehalose
BA
TreS
Ag
Yiat
-BA
Nb
TreS
PAR to PNP
PNP to PAP
0 4 8 12
0
5
10
Trehalose mg ml
–1
Time (h)
OPH
Fig. 2 | Programmable LAMBA enables diverse sequential bioconversions. a, Programmed LAMBA enabled PAR degradation by a hybrid enzyme–
inorganic sequential catalysis. We programed Ag cells to display OPH and Nb cells to display His-tags for AuNP decoration. PAR was first converted
to less harmful PNP by OPH, and reduced to harmless PAP by AuNPs. The inserts represented samples collected during the bioremediation. b, Kinetics
study confirmed that PAR was sequentially converted to PNP and PAP. PAR was first hydrolyzed to PNP. NaBH4 (40 mM) was supplemented at 160 min
to further reduce PNP to PAP. Error bars, s.d. (n = 4). c, LAMBA was recycled multiple times without substantial loss in performance. The LAMBA kept
roughly 70% catalysis efficiency for both conversions even after being recycled five times. The performance was reflected as the ratio of the concentration
in the current cycle (Conc-R) and the original cycle (Conc-O) for generated PNP or PAP. Error bars, s.d. (n = 4). d, Programmed LAMBA enabled trehalose
synthesis by integrating extracellular and intracellular transformations. We programed Ag cells to display β-amylase (BA), and Nb cells to express TreS.
Starch was converted into maltose by BA extracellularly, and the maltose was transported and further catalyzed into trehalose by TreS intracellularly.
e, HPLC results confirmed the conversion from starch to trehalose. Accumulation of trehalose (roughly 11 mg ml−1, feeding LAMBA with 5% w/v starch
solution and incubating for 13 h) was confirmed by HPLC (equipped with a refractive index detector) analysis. Maltose and glucose were the intermediate
product and by-product, respectively. The y axis values are in arbitrary units. f, LAMBA was more efficient in trehalose synthesis compared to cells without
adhesion. LAMBA enabled a higher trehalose yield compared a with mixture of two cells (expressing BA and TreS but not the adhesion pairs), possibly due
to more efficient maltose transport between cells after adhesion.
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles NATure CHeMiCAL BiOLOgy
Supplementary Fig. 11). Notably, functionalized LAMBA catalyzed
the conversion of starch to trehalose with an increased yield com-
pared with cells without adhesion (cells displaying β-amylase and
cells expressing TreS but without Nb or Ag expression, at the same
cell density of LAMBA) (Fig. 2f).
LAMBA is self-healing. Soft and stretchable electronics are envi-
sioned as one of the foundations for next-generation electronic
devices. They possess a similar Young’s modulus to human tissues,
and establish a conformal contact with human tissues to moni-
tor biosignals or apply proper stimuli in a real-time manner24–26.
The conventional active materials used by those devices are usu-
ally made of metal or conductive polymer. These active materials
are not self-healable and tend to lose function due to the fatigue,
deformation or damage incurred by repeated operations (such as
stretching and bending)27,28. Therefore, materials that are conduc-
tive, stretchable and self-healing can potentially alleviate these
issues. So far, there has been limited research exploiting the electri-
cal properties, such as conductivity of ELMs. Although ELMs have
the ability to self-heal in principle due to cell growth, no study has
explored the fast self-healing property of ELMs or used this prop-
erty in bioelectronics.
We examined the conductivity of LAMBA and monitored the
change in conductivity during the self-healing process. We first
printed a LAMBA disk (full circle or two semicircles, diameter
roughly 2 cm, height 1.1 mm) onto an inert substrate (polydimeth-
ylsiloxane, PDMS), and connected the disk to a light emitting diode
(LED) and a DC power supply (10 V). The LED lit up only when
it was connected to the full-circle LAMBA disk, instead of the two
semicircles, indicating that the LAMBA was conductive (Fig. 3a).
We next compared the conductivity of individual cells and LAMBA
(labeled or not labeled with AuNPs) before and after self-healing.
Due to its living nature, LAMBA can self-heal through continual
bacterial growth when supplemented with nutrients. Therefore, we
dropped the cell culture (Nb and Ag cells mixture (1:1 volume ratio)
for LAMBA, or the same cell density of Nb cells in 70 μl of LB) onto
a LB agar plate and let it grow into a disk after overnight culture
at room temperature (diameter roughly 2 cm). We made four dif-
ferent disks: Nb cells, Nb cells labeled with AuNPs, LAMBA and
LAMBA labeled with AuNPs, then connected these individual disks
into a circuit and recorded the current (Fig. 3b and Supplementary
Fig. 12a). The LAMBA disk (with or without AuNPs) had a better
conductivity compared with the disk made from individual cells (Nb
cells with or without AuNPs), possibly due to better connectivity
between cells after adhesion (Fig. 3b). Slicing the disk from the mid-
dle (slicing size, 0.4 mm) resulted in the disconnection of the circuit.
After overnight culture, the disks reconnected due to cell growth
and the circuit was partially restored (Fig. 3c and Supplementary
Figs. 12 and 13). The disks made of LAMBA and LAMBA with
AuNPs could restore their conductivity back to 81 and 86% of the
unsliced state, underscoring its self-healing capability (Fig. 3c and
Supplementary Fig. 12c,d). The disks made of Nb cells or Nb cells
with AuNPs also showed self-healing capability due to their living
nature. However, the overall conductivity after self-healing was lower
than that of the LAMBA or LAMBA with AuNPs (Supplementary
Fig. 12b).
Adhesion between Nb–Ag pairs in LAMBA can lead to fast
self-healing. To evaluate this property, we used a mechanical tester
0 100 200
100
300
500
Strain (%)
0 100 200
100
300
500
R (k
Ω
)
R (k
Ω
)
Strain (%)
d
ca b
f
e
*
Individual cells or LAMBA
Stretching repeatedly
Original
Sliced
Recovered
Ω
**
0
5
10
I (mA)
Original
Recovered
Sliced
*
**
0
5
10
I (mA)
Fig. 3 | LAMBA functioned as self-healing materials. a, LAMBA was conductive. LED was lit up only it was connected to the full-circle LAMBA disk
(diameter roughly 2 cm, height roughly 1.14 mm) printed on an inert PDMS surface (voltage 10 V). b, LAMBA disks functioned as bioelectronics. We
grew four disks by Nb cells, Nb cells-AuNPs, LAMBA and LAMBA-AuNPs (Methods). The adhesion between Nb and Ag cells enhanced the conductivity
of LAMBA (voltage, 10 V; *P < 0.05, **P < 0.01 (P = 0.016 for LAMBA compared with Nb cells, P = 0.0001 for LAMBA-AuNPs compared with Nb
cells-AuNPs), see the statistical method in Fig. 1b; error bars = s.d. (n = 4)). c, LAMBA disk was self-healing due to cell growth. The recovered disk resumed
conductivity back to roughly 81% (LAMBA) or roughly 86% (LAMBA-AuNPs) of the uncut state (*P < 0.05, **P < 0.01 (P = 0.015 or 0.0002 for the current
after recovery compared with the current instantly after slicing of LAMBA or LAMBA-AuNPs), see statistical method in Fig. 1b; error bars = s.d. (n = 4)).
d, Testing self-healing of the LAMBA-based wire. We assembled a wire by individual cells or LAMBA onto a SEBS substrate, and monitored the resistance
change during cyclic stretching before slicing and after healing (5 min). e, LAMBA wire maintained a stable conductivity under cyclic stretching. The
LAMBA wire kept a stable resistance up to 250% strain within a single cycle and among ten cycles. f, The recovered LAMBA wire maintained a stable
conductivity compared with the uncut condition. The resistance of wires (individual cells) increased more drastically during cyclic stretching compared
with the uncut status. In contrast, the self-healed LAMBA wire retained approximately equivalent resistance. Experiments in e and f were repeated more
than three times (Supplementary Fig. 15).
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles
NATure CHeMiCAL BiOLOgy
to imitate the repeated stretching condition and monitored the
resistance change of the individual cells or LAMBA. We used
a mold to assemble a wire of individual cells or LAMBA (length
10 mm, width 2 mm and height 0.8 mm) onto a poly(styrene-b-(eth
ylene-co-butylene-b)-styrene) (SEBS) substrate, and applied ten
cycles of tensile test by holding the edges of the substrate (Fig. 3d
and Supplementary Fig. 14). Wire made of LAMBA maintained
a stable resistance within a single cycle as well as between differ-
ent cycles compared with wire of the individual cells (Fig. 3e and
Supplementary Fig. 15a–c), and the resistance of LAMBA wire was
comparable with some published hydrogel conductors29,30. We then
sliced the wire in the middle with a poly(ethylene terephthalate)
(PET) board (slicing size 0.8 mm) and then pinched to recover it
within 5 min. The recovered wire was then subject to the same set
of tests immediately. Again, the resistance of the wire made of indi-
vidual cells increased obviously within each cycle and between dif-
ferent cycles, and the increase was more drastic compared with the
unsliced status. In contrast, the self-healed wire made of LAMBA
remained approximately equivalent in conductivity compared with
the uncut status (Fig. 3f and Supplementary Fig. 15d–f).
Self-healing LAMBA enabled wearable bioelectronics assembly.
We next assembled wearable LAMBA sensors to detect bioelectri-
cal or biomechanical signals, for example, EMG and joint bending.
EMG is electrical signals generated by muscle fiber deformation
and can be detected on skin by appropriate stretchable conductors.
Multi-channel EMG signals are essential for the monitoring of mus-
cle functions and motor units inversion, which can facilitate muscle
disease diagnosis and the human–machine interface31,32. We con-
structed polymeric multi-channels in the elastic thin-film substrate
and then injected LAMBA as the stretchable interconnect between
the carbon paste (interface materials with human skin) and the
rigid wires (connected to the amplifier) (Fig. 4a and Supplementary
Fig. 16). The device was attached to a human forearm to detect the
EMG signals with different hand gestures (Fig. 4b). The energy of the
detected bioelectrical signals was mainly distributed below 500 Hz,
the typical frequency range for on-skin EMG as confirmed by the
frequency spectrum analysis (Fig. 4c). The sensor made by individual
cells or gold film could also capture the EMG signal. Nevertheless, the
signal-to-noise ratio of the LAMBA EMG sensor was significantly
higher, underscoring its better performance (Supplementary Fig. 17).
–0.2
0
0.2
–0.2
0
0.2
–0.2
0
0.2
–0.2
0
0.2
Amplitude (mV)
Time (s)
0 10 20
140
210
Time (s)
①
②
0 10 20
0
60
120
R (kΩ)
R (kΩ)
Time (s)
①
②
–60
–100
–140
00 35
Frequency (Hz)
(dB)
Time (s)
500
035
Carbon paste c
a b
d e f
LAMBA
Fist Relaxation
LAMBA
SEBS
Copper foil
Package
②Stretching
①Original
Fig. 4 | LAMBA as stretchable sensors for wearable devices. a, Schematic diagram to show the EMG signal detection by LAMBA EMG sensor.
Multi-channel EMG electrodes made of LAMBA and carbon paste were attached to the forearm to detect EMG signals for different hand gestures. Scale
bar, 2 mm. b, The EMG signals were captured by LAMBA EMG sensor. The EMG data reflected that the hand gestures switched between clenching fist
and relaxation. The EMG signals at different channels were recorded by the LAMBA EMG sensor, respectively. c, The energy spectrum reflected the EMG
signal by the frequency amplitude with the variation of time as captured by LAMBA EMG sensor. The signal frequency changed regularly from clenching
fist to relaxation. d, Fabrication of LAMBA strain sensor to detect biomechanical signals. The sensor was properly packaged and attached to the finger
joint. The bending angle of the finger was reported by the change in the resistance of the LAMBA strain sensor. e, LAMBA strain sensor was able to
monitor the cyclic finger joint motion. The finger was straight at first and then bended, with the resistance increasing accordingly. The bending duration
was indicated by the peak width. f, The traditional sensor made of stretchable gold film failed to monitor the finger joint bending due to the limitation of
the stretchability. When the finger joint bent, the stretchable gold film could not detect the signal due to the limitation of the stretchability.
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles NATure CHeMiCAL BiOLOgy
The living stretchable LAMBA conductor can also be used to
detect biomechanical signals during joint bending as a strain sen-
sor. LAMBA was scraped onto the SEBS film and encapsulated with
another SEBS film. Two copper foils were placed on both ends of
LAMBA as the interconnector to the external electronic system,
and the film was attached to the finger joint (Fig. 4d). Our results
showed that the stretchable LAMBA strain sensor could report the
finger joint bending by an increase in the resistance, and the bend-
ing duration could be determined from the peak width (Fig. 4e and
Supplementary Video 2). In comparison, the traditional stretchable
sensor made of gold films more easily lost function due to the large
deformation of the finger joint (Fig. 4f and Supplementary Video
3). The LAMBA strain sensor could function properly to monitor
the finger joint bending with repeated usage (25–150 finger bending
cycles) on each day for 1 month (Supplementary Fig. 18).
Discussion
In this study, we explored the assembly of a living material (LAMBA)
by a cell–cell adhesion strategy using a previously developed adhe-
sion toolbox. The system is scalable. We prepared individual cells
using a 5-l shaking flask (1–2 l culturing volume) and generated
LAMBA in 3 h after mixing these cells (Supplementary Fig. 19). We
proved that LAMBA was either printable to construct macroscale
objects, or injectable to generate microscale fibers. Besides, LAMBA
could potentially be used to generate a series of microscale living
materials with other desired morphologies, such as seeded inside a
hydrogel microcapsule to grow into microspheres.
By using an export machinery, we were able to functionalize
LAMBA with proteins up to 545 amino acids. The export of adhe-
sion pairs (Nb–Ag) and functional moieties (enzymes) were pro-
grammed independently, and thus had limited effects on both sides
(Supplementary Fig. 20). By leveraging the tools of genetic engi-
neering, we have endowed the living material with diverse function-
alities by integrating the intracellular and extracellular biochemical
transformations. This strategy can be readily extended to reconsti-
tute various metabolic pathways to enable versatile biomanufactur-
ing and bioremediation.
Our system demonstrates the integration of synthetic biology,
protein engineering and material science to achieve living func-
tional material assembly. In particular, it represents an example
of engineering self-healing ELMs for wearable device fabrication,
where the unique features of the living materials (self-healing) are
critical for performance of the devices. Variants of our system can
be adopted for other types of application including energy transfor-
mation to living therapeutics.
Online content
Any methods, additional references, Nature Research report-
ing summaries, source data, extended data, supplementary infor-
mation, acknowledgements, peer review information; details of
author contributions and competing interests; and statements of
data and code availability are available at https://doi.org/10.1038/
s41589-021-00934-z.
Received: 28 February 2021; Accepted: 22 October 2021;
Published: xx xx xxxx
References
1. Chen, A. Y., Zhong, C. & Lu, T. K. Engineering living functional materials.
ACS Synth. Biol. 4, 8–11 (2015).
2. Nguyen, P. Q., Courchesne, N. M. D., Duraj-atte, A., Praveschotinunt, P. &
Joshi, N. S. Engineered living materials: prospects and challenges for using
biological systems to direct the assembly of smart materials. Adv. Mater. 30,
e1704847 (2018).
3. Gilbert, C. & Ellis, T. Biological engineered living materials: growing
functional materials with genetically programmable properties. ACS Synth.
Biol. 8, 1–15 (2019).
4. Tang, T. C. et al. Materials design by synthetic biology. Nat. Rev. Mater. 6,
332–350 (2021).
5. Chen, A. Y. et al. Synthesis and patterning of tunable multiscale materials
with engineered cells. Nat. Mater. 13, 515–523 (2014).
6. Praveschotinunt, P. et al. Engineered E. coli Nissle 1917 for the delivery of matrix-
tethered therapeutic domains to the gut. Nat. Commun. 10, 5580 (2019).
7. Moser, F., am, E., González, L. M., Lu, T. K. & Voigt, C. A.
Light-controlled, high-resolution patterning of living engineered bacteria onto
textiles, ceramics, and plastic. Adv. Funct. Mater. 29, 11 (2019).
8. Nguyen, P. Q., Botyanszki, Z., Tay, P. K. & Joshi, N. S. Programmable
biolm-based materials from engineered curli nanobres. Nat. Commun. 5,
4945 (2014).
9. Huang, J. et al. Programmable and printable Bacillus subtilis biolms as
engineered living materials. Nat. Chem. Biol. 15, 34–41 (2019).
10. Zhong, C. et al. Strong underwater adhesives made by self-assembling
multi-protein nanobres. Nat. Nanotechnol. 9, 858–866 (2014).
11. Evans, M. L. & Chapman, M. R. Curli biogenesis: order out of disorder.
Biochim. Biophys. Acta 1843, 1551–1558 (2014).
12. Barnhart, M. M. & Chapman, M. R. Curli biogenesis and function. Annu.
Rev. Microbiol. 60, 131–147 (2006).
13. Petrova, O. E. & Sauer, K. Sticky situations: key components that control
bacterial surface attachment. J. Bacteriol. 194, 2413–2425 (2012).
14. Ahn, B. K., Lee, D. W., Israelachvili, J. N. & Waite, J. H. Surface-initiated
self-healing of polymers in aqueous media. Nat. Mater. 13, 867–872 (2014).
15. Huang, Y. et al. A self-healable and highly stretchable supercapacitor based
on a dual crosslinked polyelectrolyte. Nat. Commun. 6, 10310 (2015).
16. Glass, D. S. & Riedel-Kruse, I. H. A synthetic bacterial cell–cell adhesion
toolbox for programming multicellular morphologies and patterns. Cell 174,
649–658.e616 (2018).
17. Tsoi, R., Dai, Z. & You, L. Emerging strategies for engineering microbial
communities. Biotechnol. Adv. 37, 107372 (2019).
18. Hays, S. G., Patrick, W. G., Ziesack, M., Oxman, N. & Silver, P. A. Better
together: engineering and application of microbial symbioses. Curr. Opin.
Biotechnol. 36, 40–49 (2015).
19. Han, M. J. & Lee, S. H. An ecient bacterial surface display system based on
a novel outer membrane anchoring element from the Escherichia coli protein
YiaT. FEMS Microbiol. Lett. 362, 1–7 (2015).
20. Rollié, S., Mangold, M. & Sundmacher, K. Designing biological systems:
systems engineering meets synthetic biology. Chem. Eng. Sci. 69, 1–29 (2012).
21. Behrendor, J. B. & Gillam, E. M. Prospects for applying synthetic biology to
toxicology: future opportunities and current limitations for the repurposing
of cytochrome P450 systems. Chem. Res. Toxicol. 30, 453–468 (2017).
22. Tsoi, R. et al. Metabolic division of labor in microbial systems. Proc. Natl
Acad. Sci. USA 115, 2526–2531 (2018).
23. ommes, M., Wang, T., Zhao, Q., Paschalidis, I. C. & Segrè, D. Designing
metabolic division of labor in microbial communities. mSystems 4, e00263–18
(2019).
24. Kim, D. H. et al. Epidermal electronics. Science 333, 838–843 (2011).
25. Schiavone, G. & Lacour, S. P. Conformable bioelectronic interfaces: mapping
the road ahead. Sci. Transl. Med. 11, https://doi.org/10.1126/scitranslmed.
aaw5858 (2019).
26. Liu, Y. et al. So and elastic hydrogel-based microelectronics for localized
low-voltage neuromodulation. Nat. Biomed. Eng. 3, 58–68 (2019).
27. Chortos, A., Liu, J. & Bao, Z. Pursuing prosthetic electronic skin. Nat. Mater.
15, 937–950 (2016).
28. Jiheong Kang, J. B.-H. T. Z. B. Self-healing so electronics. Nat. Electron. 2,
144–150 (2019).
29. Cai, G. et al. Extremely stretchable strain sensors based on conductive
self-healing dynamic cross-links hydrogels for human-motion detection. Adv.
Sci. 4, 1600190 (2017).
30. Wu, J. et al. Highly stretchable and transparent thermistor based on
self-healing double network hydrogel. ACS Appl. Mater. Interfaces 10,
19097–19105 (2018).
31. Zhu, M., et al. Towards optimizing electrode congurations for silent speech
recognition based on high-density surface electromyography. J. Neural Eng.
18, 016005 (2020).
32. Jiang, N., Wang, L., Huang, Z. & Li, G. Mapping responses of lumbar
paravertebral muscles to single-pulse cortical TMS using high-density surface
electromyography. IEEE Trans. Neural Syst. Rehabil. Eng. 29, 831–840 (2021).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature America, Inc. 2021
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles
NATure CHeMiCAL BiOLOgy
Methods
Strains, circuits and media. Bacterial strains. E. coli strain MG1655 constitutively
expressing GFP or mCherry was used to generate LAMBA for characterization,
macroscale objects and microscale ber assembly. MG1655(DE3) transformed with
the plasmid (with T7 promoter) was used for protein expression in bioremediation.
BW25113ΔtreAΔtreCΔtreF transformed with the plasmid (with T5 promoter) was
used for trehalose synthesis.
Circuit and plasmids.
• Circuit programing cells to display Nb (Nb cells) or Ag (Ag cells) was
published previously16. Briey, Nb or Ag was anchored via a Neae surface
display system that included a short N-terminal export tag, a LysM domain for
peptide glycan binding and a β−barrel for transmembrane insertion. Protein
expression was controlled by a TetR (tetracycline repressor) operator (p15A
origin).
• Plasmid encoding surface expression of OPH (YiaT-OPH) was synthesized
and subcloned into pET28a vector between BamHI and HindIII by the
Genewiz Company.
• Plasmids encoding TreS and surface expression of β-amylase (YiaT-ΒΑ) were
synthesized and subcloned into pET28a vector between BamHI and HindIII
by the Genewiz Company. e T7 promoter in the vector was later changed to
a T5 promoter. Details of the plasmids and protein sequences can be found in
Supplementary Tables 1 and 2.
Growth media. LB medium. Here, 25 g LB medium powder (Aladdin) was added
into 1 l of deionized H2O. After autoclaving for 45 min, the LB medium was
stored at room temperature. The medium was supplemented with appropriate
antibiotics (50 μg ml−1 kanamycin, 100 μg ml−1 ampicillin) when applicable.
Isopropyl-β--thiogalactoside at 0.1 mM or ATc at 100 ng ml−1 was used to induce
gene expression when applicable.
Methods. LAMBA formation. A single colony was picked and inoculated to
generate overnight culture of Nb cells or Ag cells on day 0. On day 1, overnight
cultures were diluted to fresh LB medium (1:100) supplemented with 100 ng ml−1
ATc and induced for 24 h. On day 2, Nb cells and Ag cells cultures were mixed at
an equal volume and settled down for 3 h for LAMBA formation. To thoroughly
remove the remaining culture medium, LAMBA was collected by centrifugation at
4,000g for 8 min.
Bacteria affinity assay. Nb and Ag cells were induced as mentioned above. After
mixing a culture of Nb and Ag cells at equal volume, 100 µl of supernatant was
taken at a series of time points for optical density (OD600) evaluation. We used
(Tn − T0)/T0 × 100% to evaluate the percentage of cells recruited into the LAMBA,
where Tn and T0 were the OD600 at n hours and the starting point, respectively.
Rheological measurement. The rheological properties of individual cells and
LAMBA were evaluated on a strain-controlled rheometer (Anton par MCR101)
equipped with a 40-mm diameter cone plate (200 µm gap). Individual cells or
LAMBA (same cell density with the individual cells) were collected as mentioned
above. Strain sweep experiments from 0.1 to 10% strain amplitudes were
performed at a fixed frequency of 1 rad s−1. Frequency sweep experiments
from 10 to 0.1 rad s−1 were performed at a 1% strain amplitude. The temperature
(25 °C) was kept at a constant throughout the experiments by a Peltier
thermoelectric device.
Fabrication of microscale fiber by LAMBA via microfluidic chips. Microfluidic chips
were fabricated in PDMS by rapid prototyping method. The templates consisting
of silicon wafer substrate and photoresist (SU-8 3035, MicroChem Corporation)
microstructure were used for casting the PDMS chips. The fabrication of templates
was achieved via spin coating, soft baking, exposure with mask, post exposure
baking, development and hard baking. To get the multilayer microstructure,
four procedures, including spin coating, soft baking, exposure with mask, post
exposure baking were repeated several times sequentially with different masks
during each exposure33. To generate our design chips, two pieces of slices with the
same structure were sealed face to face under the stereomicroscope after plasma
treatment with oxygen. Ethanol was applied as lubricant during the adjustment of
the two slices. The chips were then transferred to an oven and baked at 80 °C for
1 h. In our experiment, three chips were used for the preparation of microfiber.
The structure and size of each chip are shown in Supplementary Fig. 4 and
Supplementary Table 3.
Both Nb cells and Ag cells (induced) were concentrated at 100-fold and
resuspended in LB medium containing 1% (w/v) methylcellulose (1,500 mPa.s,
Aladdin), 100 μg ml−1 ampicillin and 100 ng ml−1 ATc. Nb and Ag cells were mixed
at the same volume and incubated at room temperature for LAMBA formation.
After 3 h, LAMBA, sodium alginate (1% (w/v), 200 mPa.s) and calcium chloride
(1% (w/v)) were injected to microfluidic chips sequentially using Harvard Syringe
pumps. The assembled fiber was first solidified in calcium chloride solution (1%
(w/v)) for 1 h, and further transferred to LB medium for fiber growth. The flow
rate of the LAMBA, sodium alginate and calcium chloride solutions are shown in
Supplementary Table 4.
The time-lapse fluorescence microscopy was used to capture the growth of
the living fiber. To image the process, the bright-field or fluorescent images (GFP
and mCherry) were collected at 600-s frame rates by a micro zoom fluorescence
microscope (Olympus, MVX10) equipped with a ×1 objective (numerical aperture
0.25) and a sCMOS camera (Photometrics, Prime BSI). When acquiring images,
the zoom of microscope was set to 6.3. GFP or mCherry was excited using 488 or
565 nm LED lights (Lumencor, Spectra X) and imaged using single-band emission
filters (Semrock). Image acquisition was implemented through a custom protocol,
written in MATLAB to control hardware through the μManager core.
3D printing using LAMBA. Nb and Ag cells were cultured and induced as
mentioned above. The Nb and Ag cells were mixed and loaded into a syringe to
form LAMBA. LAMBA (together with syringe) was first centrifuged (4,000g) for
8 min before transferred to a 3 ml cartridge (Cellink) for 3D printing (printing
pressure 120 kPa, inner diameter of the nozzle 0.41 mm and printing speed
7 mm s−1). We printed a square with grids with a 20% filling percentage, and an
iSynbio logo with a 90% filling percentage at a height of 0.25 mm.
To explore the printing height of the final objects, we used LAMBA as bioink
to fabricate rectangular constructs (10 × 10 mm) with a 20% filling percentage in
various of layers (from 1 to 30 layers). The printing height of the first layer was set
to 0.1 mm and the height of each following layer was 0.15 mm (printing pressure
150 kPa, inner diameter of the nozzle 0.41 mm and the printing speed 3 mm s−1).
Fractionation of outer membrane proteins. Fractionation of bacterial outer
membrane proteins was performed as previously described19. Briefly, the cell
culture was pelleted first and then washed with 1 ml of Na2HPO4 (10 mM, pH 7.2),
centrifuged at 3,500g (5 min, 4˚C) and resuspended in 0.5 ml of Na2HPO4
(10 mM, pH 7.2). Cells were then lysed by ultra-sonication for six cycles
(each for 10 s of lysis and 10 s of interval). Partially lysed cells were centrifuged
at the speed of 12,000g for 2 min at room temperature. Membrane associated
samples were collected by another centrifugation at 12,000g for 30 min at 4 °C.
After removal of supernatant, each precipitant was resuspended in 0.5 ml of
Na2HPO4 (10 mM, pH 7.2) containing 0.5% (w/v) sarcosine. The samples were
then incubated at 37 °C for 30 min and the insoluble pellet containing membrane
proteins were collected by centrifugation at 12,000g for 30 min at 4 °C. Membrane
proteins were obtained by washing the insoluble pellet with Na2HPO4 (10 mM,
pH 7.2) and resuspended in 50 μl of TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0).
SDS–PAGE was performed together with Coomassie Brilliant Blue (R250) to
characterize each protein sample.
Biomanufacturing of trehalose by LAMBA. Ag cells (BW25113ΔtreAΔtreCΔtreF
(YiaT-BA)) and Nb cells (BW25113ΔtreAΔtreC ΔtreF (TreS)) were cultured
and induced as mentioned above. Here, 10 ml of culture of Nb cells and Ag cells
were mixed to form LAMBA. The LAMBA was washed twice with 20 ml of
PBS, resuspended in 20 ml of 5% starch and incubated at 37 °C for 13 h. Then
1 ml of supernatant was taken at different time points for quantitative analysis
of the products (maltose, trehalose and glucose) by a high-performance liquid
chromatography system (HPLC) equipped with a refractive index detector.
Samples were injected into a HPLC system (DIONEX) equipped with an NH2
column (250 × 4.6 mm, Sepax). The column was maintained at 35 °C and was eluted
isostatically with a mobile phase of acetonitrile and Milli-Q water (v:v = 75:25) at a
flow rate of 0.6 ml min−1.
Transmission electron microscopy (TEM) observation. Here, 20 μl of bacteria liquid
culture was deposited onto carbon coated TEM grids (Linxia Tainuo Technology)
and incubated for 5 min at room temperature. The grid was then washed with 20 μl
of PBS and 20 μl of deionized water. The excess liquid was removed by a filter paper
(9 cm, Newstar Paper Co. Ltd). The sample was negatively stained by 10 μl of 2%
(w/v) phosphotungstic acid solution, followed by air drying and then examined
by a FEI Spirit T12 transmission electron microscope at an accelerating voltage of
120 kV. Images were taken with an Orius SC200B 200 kV camera.
To image AuNPs (Ni-NTA-AuNPs, 5 nm, Nanoprobes) labeled Nb cells, we
loaded the bacterial culture on the TEM grids and rinsed with 30 μl of deionized
water three times. The grids were then treated with 30 μl of selective binding buffer
(20 mM Na2HPO4, 500 mM NaCl, 20 mM imidazole, pH 7.4), placed upside down
on a droplet containing Ni-NTA-AuNPs (5 nM) and incubated for 90 min at room
temperature. The grid was then washed twice with selective binding buffer and
twice with water. The thoroughly washed grid was then negatively stained tested
by a FEI Spirit T12 transmission electron microscope with the same protocols
described above.
Bioremediation of PAR by LAMBA. The degradation of PAR is a two-step
sequential reaction, catalyzed by LAMBA assembled by Ag cells (MG1655(DE3)
(YiaT-OPH)) and His-Nb cells (MG1655(DE3)). To label LAMBA with AuNP, one
volume of LAMBA was mixed with two volumes of AuNP solutions to attain the
mixture (0.5 M NaCl, 30 mM imidazole, 5 nM AuNPs, pH 7.4), incubated at room
temperature for 90 min and washed with 1 ml of PBS five times. The kinetics of the
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology

Articles NATure CHeMiCAL BiOLOgy
two-step reaction was characterized by the formation and degradation process of
the intermediate PNP.
(1) Degradation of PAR into PNP by LAMBA. LAMBA was collected and
resuspended in CHES buer (2-(N-cyclohexylamino) ethane sulfonic acid,
Sigma C2885, 0.1 M, pH 7.5) with 50 mM CoCl2. Aer 1 h of incubation at
30 °C, LAMBA was harvested by centrifugation at 4,000g for 10 min and
resuspended in CHES buer (0.1 M, pH 10). For each assay, 9.9 ml of LAMBA
was mixed with 100 μl of 10 mM PAR (in 10% methanol). e reaction
mixtures were incubated at 37 °C. Supernatants were taken every 20 min and
the concentrations of PNP were evaluated by measuring the absorbance at
405 nm using Tecan Innite Pro 200 Plate Reader. e consumption of PAR
was calculated based on the generation of PNP.
(2) Reduction of PNP into PAP by LAMBA. Aer completing the degrada-
tion from PAR to PNP, the reaction of PNP to PAP was initiated by mixing
4.9 ml of reaction mixture in step 1 and 0.1 ml of 2 M NaBH4. Supernatants
were collected every 20 min, and the PNP concentrations were evaluated by
measuring its absorbance at 405 nm by Tecan Innite Pro 200 Plate Reader.
e generation of PAP was calculated based on the consumption of PNP.
(3) Evaluation of cell loss during the recycling. For the conversion of PAR to
PNP, cells were suspended in CHES buer (0.1 M, pH 10). In each cycle, 1 μl
of suspension was taken for OD600 measurement. For the conversion of PNP
to PAP, in each cycle, 1 μl of suspension was taken for OD600 measurement
before addition of NaBH4 solution.
Evaluation of self-healing capacity (caused by cell growth) of LAMBA. Nb and Ag
cells were prepared and induced as mentioned above. For LAMBA, 1 ml of Nb cell
culture was mixed with 1 ml of Ag cell culture (both induced) and settled down for
3 h. For Nb cells, 2 ml of bacterial culture was collected. Decoration with AuNPs
of LAMBA or individual cells were described above. After centrifugation, LAMBA
or individual cells was resuspended in 70 μl of LB medium and dropped onto a LB
agar dish for overnight culture at the room temperature (for generation of roughly
2 cm diameter disk). Four disks made of Nb cells, Nb cells with AuNPs, LAMBA
and LAMBA with AuNPs, were connected with a SIAT LED light and d.c. power
(10 V) individually for current measurement. For the self-healing experiment, PET
boards with different thicknesses were used to slice the disk in the middle. The
disks were placed at room temperature for recovery after overnight culture.
Evaluation of fast recovery capacity of LAMBA by cyclic stretching. LAMBA was first
assembled from shaking culture. We then covered the SEBS surface with a mold
(length, width and height were 10, 2 and 0.8 mm, respectively), transferred the
LAMBA into the mold and removed the residue by scraping with a glass coverslip
for wire preparation (Supplementary Fig. 14). We connected two ends of the
LAMBA wire to a digital meter (KEITHLEY 2000 multimeter) using copper wires
to record the resistance change during the cyclic stretching (ten cycles) applied by
a mechanical tester (Shimadazu, AG-X plus 100N). We used a 0.8mm thickness
PET board to slice the LAMBA wire in the middle, pinched the sliced LAMBA to
heal within 5 min and started the cyclic stretching (ten cycles). Resistance of wires
made of LAMBA or individual cells was measured and compared before slicing
and recovery after slicing.
Preparation of the stretchable electronic devices and biosignals detection. Assembly
of the sensors for EMG monitoring. Eight PDMS channels were fabricated by a
template method (Supplementary Fig. 16). Channels were cylindrical in shape,
with roughly 0.35 mm in bottom-circle radius, roughly 18 or 25 mm each in length.
LAMBA was injected into PDMS channels and connected by the female header at
the end to the external home-made amplifier circuit. Silicone adhesive was used to
encapsulate the interconnection to prevent the leakage. Carbon paste was used as
the interface electrode with the skin and connected to the LAMBA through thin
copper wires. For fabrication of the sensor made of gold film, magnetron sputtering
coating apparatus (China, JS4S-75G) was used to deposit stretchable gold film
onto PDMS. A mask was used to generate the gold film in the shape equivalent to
the eight channels described above. The devices were attached to human forearm
(pronator teres) to collect EMG signals for different hand gestures. EMG data were
further analyzed in MATLAB (R2018a). To evaluate signal-to-noise ratio (SNR),
ten sets of values for ASignal and ANoise were randomly selected for EMG measured
by each sensor, where ASignal denotes the average amplitude of the signal and ANoise
denotes the average amplitude of the noise. The data were brought into the SNR
equation to calculate the SNR.
SNR =20log
(A
Signal
ANoise )
Joint bending monitoring by strain sensors. SEBS film was prepared by drop
casting method with a thickness of roughly 200 μm, and used as the substrate of
the biomechanical sensors. Two pieces of copper foils were attached to the two
ends of the film leaving a gap in between, then the LAMBA wire was assembled
in the gap using the method described in Supplementary Fig. 14. Briefly, LAMBA
was first assembled from shaking culture. We then covered the SEBS surface with
a mold (length, width and height of 10, 2 and 0.8 mm, respectively), transferred
the LAMBA into the mold and removed the residue by scraping with a glass
coverslip for wire preparation. Finally, the device was encapsulated by another
layer thin SEBS film of roughly 200 μm in thickness fabricated by drop casting
method (Fig. 4d). The skin adhesive was used to attach the device to the finger
joint and the resistance was recorded by a multimeter. The control sample was
fabricated by sputtering stretchable gold film onto the SEBS substrate. The
data of EMG and strain sensor were obtained with the informed consent of all
participants. The institutional review board of the Shenzhen Institutes of Advanced
Technology, Chinese Academy of Sciences approved this study, with the code
SIAT-IRB-180315-H0242.
Reporting Summary. Further information on research design is available in the
Nature Research Reporting Summary linked to this article.
Data availability
The authors declare that all the source data processed for figures generation in this
study are available within the paper, source data file and the supplementary data
files. Any additional information is available upon reasonable request. Source data
are provided with this paper.
Code availability
The code that supports the findings of this study are available from the
corresponding author upon reasonable request.
References
33. Yu, Y. et al. Simple spinning of heterogeneous hollow microbers on chip.
Adv. Mater. 28, 6649–6655 (2016).
Acknowledgements
We thank D.S. Glass for sharing plasmids; X. Shen for sharing strains and instructive
comments; L. Jiang for insightful comments on trehalose synthesis experiments; F. Jin
and S. Huang for assistance in microscopy and microfluidic instruments setup; C. Liu
and W. Liu for help in strain engineering; the Testing Technology Center of Materials and
Devices and the Tsinghua Shenzhen International Graduate School for TEM instrument
usage. This study was partially supported by National Key Research and Development
Program of China (grant nos. 2018YFA0903000 and 2020YFA0908100 to Z.D.: these two
grants provide equal support), National Natural Science Foundation of China National
Natural Science Foundation of China grant nos. 81927804 (Z.L.) and 32071427 (Z.D.).
Shenzhen Science and Technology Program grant no. KQTD20180413181837372 (Z.D.).
Author contributions
B.C. designed and performed the experiments, interpreted the results and wrote the
paper. W.K. assisted in experimental setup and data interpretation. J.S. designed and
performed the experiments, interpreted the results and revised the paper. R.Z. performed
the experiments, interpreted the results and revised the paper. Yue Y. and A.X. assisted
in experimental setup and data interpretation of microfluidic and microscopy. M.Y.,
M.W. and J.H. assisted in performing experiments and experimental setup in electronic
circuit assembly, 3D printing and rheology measurement. Y.C., L.T. and Q.T. assisted
in performing experiments and experimental setup in living fiber generation and EMG
measurement. Yin Y., G.L. and L.Y. assisted in research design, experimental setup,
data interpretation and paper revisions. Z.L. conceived the research, assisted in the
experimental design, results interpretation and paper revisions. Z.D. conceived the
research, designed the experiments, interpreted the results and wrote the paper.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information The online version contains supplementary material
available at https://doi.org/10.1038/s41589-021-00934-z.
Correspondence and requests for materials should be addressed to
Zhiyuan Liu or Zhuojun Dai.
Peer review information Nature Chemical Biology thanks Neel Joshi and the other,
anonymous, reviewer(s) for their contribution to the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
NATURE CHEMICAL BIOLOGY | www.nature.com/naturechemicalbiology