PNAS PLUS

Engineering control of bacterial cellulose production using a genetic toolkit and a new celluloseproducing strain Michael Floreaa,b,1, Henrik Hagemanna,c, Gabriella Santosaa,b, James Abbottd,e, Chris N. Micklema,b, Xenia Spencer-Milnesa,b, Laura de Arroyo Garciaa,b, Despoina Paschoua,c, Christopher Lazenbatta,b, Deze Konga,c, Haroon Chughtaia,c, Kirsten Jensena,f, Paul S. Freemonta,f, Richard Kitneya,c, Benjamin Reevea,c, and Tom Ellisa,c,2 a Centre for Synthetic Biology and Innovation, Imperial College London, London SW7 2AZ, United Kingdom; bDepartment of Life Sciences, Imperial College London, London SW7 2AZ, United Kingdom; cDepartment of Bioengineering, Imperial College London, London SW7 2AZ, United Kingdom; dBioinformatics Support Service, Department of Surgery and Cancer, Imperial College London, London SW7 2AZ, United Kingdom; eCentre for Integrative Systems Biology and Bioinformatics, Imperial College London, London SW7 2AZ, United Kingdom; and fDepartment of Medicine, Imperial College London, London SW7 2AZ, United Kingdom

Edited by Jef D. Boeke, New York University School of Medicine, New York, NY, and approved April 29, 2016 (received for review November 20, 2015)

synthetic biology genomics

advantage in colonization over other microorganisms (10). In materials science, genetic engineering has been used to create several novel biomaterials, such as strong underwater protein-based adhesives (11), self-assembling, functionalized amyloid-based biofilms (12), biodegradable bacterial cellulose-based tissue engineering scaffolds (13), and many others. Bacterial cellulose has long been a focus of research because, unlike plant-based cellulose, it is pure of other chemical species (lignin and pectin) and is synthesized as a continuous highly interconnected lattice (14). This makes the material mechanically strong [nanocellulose fibers possess tensile stiffness of between 100 and 160 GPa and tensile strength of at least 1 GPa (15, 16)] while still flexible, biocompatible, and highly hydrophilic, capable of storing water over 90% of its total weight (17, 18). Due to these properties, bacterial cellulose is commercially Significance

| bacterial cellulose | genetic engineering | biomaterials |

Bacterial cellulose is a remarkable material that is malleable, biocompatible, and over 10-times stronger than plant-based cellulose. It is currently used to create materials for tissue engineering, medicine, defense, electronics, acoustics, and fabrics. We describe here a bacterial strain that is readily amenable to genetic engineering and produces high quantities of bacterial cellulose in low-cost media. To reprogram this organism for biotechnology applications, we created a set of genetic tools that enables biosynthesis of patterned cellulose, functionalization of the cellulose surface with proteins, and tunable control over cellulose production. This greatly expands our ability to control and engineer new cellulose-based biomaterials, offering numerous applications for basic research, materials science, and biotechnology.

T

he emergence of synthetic biology now enables model microorganisms such as Escherichia coli to be easily reprogrammed with modular DNA code to perform a variety of new tasks for useful purposes (1). However, for many application areas, it is instead preferable to exploit the natural abilities of nonmodel microbes as specialists at consuming or producing molecules or thriving within niche environments (2). Recent work has described adapting common E. coli synthetic biology tools to work across different bacterial phyla (3, 4) and has produced genetic toolkits for new bacteria, where collections of DNA constructs and methods for precise control of heterologous gene expression have been developed for engineering strains naturally specialized for photosynthesis or survival within the gut microbiome (5, 6). An important application area for biotechnology is the production of materials, and bacteria that naturally secrete high yields of cellulose have attracted significant attention not just from people in industry and research (7) but also from those in art, fashion, and citizen science (8). However, despite their widespread use, no toolkit for genetic modification of these cellulose-producing bacteria has previously been described. Komagataeibacter is a genus from the Acetobacteraceae family of which multiple species produce bacterial cellulose. Cellulose nanofibers are synthesized from UDP-glucose by the acs (Acetobacter cellulose synthase) operon proteins AcsA and AcsB (9) and secreted by AcsC and AcsD, forming an interconnected cellulose “pellicle” around cells (7). Although it is still unclear why Acetobacteraceae produce bacterial cellulose in nature (7), it has been shown to confer the host with a high resistance to UV light and a competitive www.pnas.org/cgi/doi/10.1073/pnas.1522985113

Author contributions: M.F., H.H., and B.R., isolated and characterized the iGEM strain; M.F. and B.R. sequenced the genome; M.F. and J.A. scaffolded and analyzed the genome; M.F., H.H., G.S., C.N.M., X.S.-M., L.d.A.G., and D.P. created the genetic engineering toolkit; M.F. created the sRNA construct; H.H., G.S., C.N.M., and X.S.-M. created CBD fusion proteins; M.F., H.H., G.S., C.N.M., and X.S.-M. created patterned and functionalized biomaterials; M.F., H.H., G.S., and T.E. conducted temporal patterning experiments; H.H., G.S., C.N.M., X.S.-M., L.d.A.G., and B.R. created sfGFP-functionalized garments; M.F., C.N.M., D.P., C.L., D.K., and H.C. analyzed data; M.F. and T.E. wrote the manuscript; K.J., P.S.F., R.K., B.R., and T.E. supervised the work; all authors contributed to planning the study. Conflict of interest statement: H.H. and G.S. are researching the industrial uses of cellulose functionalized with cellulose binding domain fusion proteins as part of CustoMem Ltd. This article is a PNAS Direct Submission. Freely available online through the PNAS open access option. Data deposition: The sequence reported in this paper has been deposited in the European Nucleotide Archive, www.ebi.ac.uk/ena (accession no. PRJEB10933). 1

Present address: Department of Biosystems Science and Engineering, Eidgenössische Technische Hochschule Zürich, 4058 Basel, Switzerland.

2

To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1522985113/-/DCSupplemental.

PNAS | Published online May 31, 2016 | E3431–E3440

MICROBIOLOGY

Bacterial cellulose is a strong and ultrapure form of cellulose produced naturally by several species of the Acetobacteraceae. Its high strength, purity, and biocompatibility make it of great interest to materials science; however, precise control of its biosynthesis has remained a challenge for biotechnology. Here we isolate a strain of Komagataeibacter rhaeticus (K. rhaeticus iGEM) that can produce cellulose at high yields, grow in low-nitrogen conditions, and is highly resistant to toxic chemicals. We achieved external control over its bacterial cellulose production through development of a modular genetic toolkit that enables rational reprogramming of the cell. To further its use as an organism for biotechnology, we sequenced its genome and demonstrate genetic circuits that enable functionalization and patterning of heterologous gene expression within the cellulose matrix. This work lays the foundations for using genetic engineering to produce cellulose-based materials, with numerous applications in basic science, materials engineering, and biotechnology.

Fig. 1. Characterization of K. rhaeticus iGEM. (A) Morphology of a typical cellulose pellicle produced by K. rhaeticus iGEM. White patches are light reflected by the pellicle surface. (B) Cellulose productivity of K. rhaeticus iGEM and G. hansenii ATCC 53582 on different growth media, shown as pellicle dry weight after a 10-d incubation in 20 mL liquid Hestrin–Schramm (HS) media. Cellulose production of K. rhaeticus exceeds that of G. hansenii in sucrose-containing media (HS sucrose and Kombucha tea, adjusted P = 0.0128 and P = 0.039, respectively), but is lower than that of G. hansenii in HS glucose (adjusted P = 0.027). (C and D) Growth and production of a cellulose pellicle (denoted by an arrow) by K. rhaeticus iGEM in nitrogen-free LGI medium. K. rhaeticus iGEM shows significant growth compared with negative controls G. hansenii and E. coli (adjusted P = 0.026 and P = 0.011, respectively), whereas G. hansenii and E. coli do not differ (adjusted P = 0.742). ns, not significant. (E) Comparison of K. rhaeticus and E. coli survival after incubation for 5, 30, or 90 min with toxic chemicals [0.1 M HCl, 0.1 M NaOH, 70% (vol/vol) EtOH, and 10% (vol/vol) bleach]. Survival is defined as the fraction of survived cells compared with PBS-treated cells. (F) Scanning electron micrographs of K. rhaeticus iGEM encased in bacterial cellulose, taken after 8 d of growth, at 6,000× magnification. *P < 0.05. n = 3 biological replicates for all experiments; error bars indicate SD. Statistical significance was determined for B with two-way ANOVA and Bonferroni’s multiple comparisons test and for D with one-way ANOVA with Tukey’s multiple comparisons tests. For A and C, images were taken 9 d postinoculation. Images were cropped, and contrast was adjusted to improve clarity, without affecting details. See Methods for experimental details.

used in medical wound dressings, high-end acoustics, and many other products (7), and in the laboratory has been used to create biodegradable tissue scaffolds (13), nanoreinforcements (19), and artificial blood vessels (18), as well as sensors (20), flexible electrodes (21), organic light-emitting diode (OLED) displays, and other materials (22). Functionalization or modification of bacterial cellulose has mainly been achieved by chemical or mechanical modifications of the cellulose matrix or via changing culturing conditions (7, 22), whereas only a few attempts at genetic engineering have been made (13, 23). However, genetic engineering may allow a greater range of materials to be produced, by enabling fine control over cellulose synthesis genes and production of proteincellulose composite biomaterials. Here we isolate a strain of Komagataeibacter rhaeticus (previously Gluconacetobacter rhaeticus) E3432 | www.pnas.org/cgi/doi/10.1073/pnas.1522985113

(24) that can grow in low-nitrogen conditions and produce cellulose at high yields, sequence its genome, and develop a synthetic biology toolkit for its genetic engineering. This toolkit provides the characterization data necessary for the engineering of K. rhaeticus iGEM, and enables transformation, controlled expression of constitutive and inducible transgenes, and control over endogenous gene expression of this strain. We use these tools to engineer a system that allows tunable control over native cellulose production, and produce novel patterned and functionalized cellulose-based biomaterials. Results Isolation, Characterization, and Genome Sequencing of K. rhaeticus iGEM. As part of the International Genetically Engineered Ma-

chine Competition (iGEM) (25), we evaluated Gluconacetobacter Florea et al.

Florea et al.

Genetic Engineering Toolkit for Komagataeibacter. Very few genetic tools are available for the engineering of Acetobacteraceae. We therefore developed a complete set of tools for its engineering, consisting of protocols, modular plasmids, promoters, reporter proteins, and inducible constructs that enable external control of gene expression (Fig. 3A). Protocols and plasmid backbones. We first used the plasmid pBlaVhb-122 [previously described to replicate in Acetobacteraceae (38)] to develop protocols for the preparation of electrocompetent cells, transformation, plasmid purification, and genomic DNA extraction of K. rhaeticus iGEM (SI Appendix, Supplementary Protocols). Using these protocols, we then assessed eight plasmids for propagation in K. rhaeticus iGEM: pSEVA311, pSEVA321, pSEVA331, pSEVA341, pSEVA351, pBAV1K-T5-sfGFP, pSB1C3, and pBca1020 (see SI Appendix, Table S2 for details). From these, pSEVA321, 331, 351, pBAV1K-T5-sfGFP, and pBla-Vhb-122 showed replication in iGEM (SI Appendix, Fig. S5), giving a total of five different plasmids to act as vectors. We further engineered pSEVA321 and pSEVA331 into pSEVA321Bb and pSEVA331Bb, making them compatible with the widely used BioBrick standard (25), to enable rapid cloning of publicly available DNA parts. We then used pSEVA331Bb for all subsequent studies, due to its likely higher copy number. Reporter proteins and constitutive and inducible promoters. We next tested expression of seven reporter proteins (mRFP1, GFPmut3, sfGFP, and chromoproteins tsPurple, aeBlue, gfasPurple, and spisPink) (see SI Appendix, Tables S3 and S4 for details), of which mRFP1, GFPmut3, and sfGFP showed visually detectable expression. We then chose 10 promoters from an open-access collection of synthetic minimal E. coli promoters and, using mRFP1 as the reporter, characterized these in K. rhaeticus iGEM (Fig. 3B and SI Appendix, Table S4; also see SI Appendix, Fig. S6 for a comparison with promoter strengths in E. coli). For inducible promoters, we engineered four constructs allowing gene expression to be induced externally by anhydrotetracycline (ATc) or N-acyl homoserine lactone (AHL) (see SI Appendix, Fig. S7 for an overview of constructs). From these, the AHL-inducible constructs (pLux01 and pLux02) showed higher induction and lower leakiness than the ATc-induced constructs (pTet01 and pTet02) (Fig. 3C) and, contrary to our initial expectations, they also gave robust induction of mRFP1 expression when cells were encased in the pellicle, showing visible fluorescence (Fig. 3 D and E and SI Appendix, Fig. S8). This is notable, as it shows that cells in the pellicle can effectively receive signals from their environment despite their cellulose encasing. Because K. rhaeticus is highly resistant to various environmental hazards within cellulose (Fig. 1E), the ability to receive signals while being protected by cellulose makes it a potentially suitable host for applications requiring tolerance to toxic chemicals and long-term survival. Engineering Control over Cellulose Production. Because wild-type species produce cellulose constitutively, a major goal of genetic engineering of Acetobacteraceae has been to achieve control over cellulose production. Constitutive cellulose production complicates genetic engineering techniques and is not always desirable for industrial applications as it imparts a high metabolic cost, which in well-aerated conditions typically leads to the emergence of cellulose-nonproducing mutants (39). It is therefore desirable to inhibit cellulose production during periods when it is not required, to PNAS | Published online May 31, 2016 | E3433

PNAS PLUS

kpsS, and rfaB) and may play a role in cellulose productivity (35, 36). Finally, to determine whether the iGEM strain can fix atmospheric nitrogen similar to G. diazotrophicus, we searched its genome for genes associated with nitrogen fixation. We located five genes (ntrB, -C, -X, and -Y and nifU) (SI Appendix, Table S1) associated with nitrogen fixation in Acetobacteraceae (37); however, interestingly, we did not find the genes homologous to nifHDK, which form the main nitrogenase subunits in G. diazotrophicus.

MICROBIOLOGY

hansenii ATCC 53582 [one of the highest-reported celluloseproducing strains (26), recently reclassified as Komagataeibacter hansenii ATCC 53582 (27)] and a strain isolated from Kombucha tea as potential new synthetic biology hosts (Fig. 1A). We chose the latter strain (hereafter called “iGEM”) for further work, as preliminary experiments showed that it can be transformed more readily with plasmid DNA than G. hansenii ATCC 53582. Furthermore, the iGEM strain produced more cellulose than G. hansenii ATCC 53582 on sucrose in small-scale tests (Fig. 1B) and also produced cellulose at high yields on cheap, low-nitrogen Kombucha tea medium (Fig. 1B). Surprisingly, it could also grow on the defined nitrogen-free LGI medium (Fig. 1 C and D and SI Appendix, Figs. S1 and S2). Because several species of Acetobacteraceae, notably Gluconacetobacter diazotrophicus, have been confirmed to fix atmospheric nitrogen (28, 29), this suggested possible nitrogen fixation (see below). Finally, because cellulose has been reported to increase resistance toward UV light and environmental stresses in closely related species (10), we tested whether cellulose could also confer resistance toward chemical stresses, which may occur in unrefined feedstocks or during industrial production. We tested the susceptibility of the iGEM strain to 70% (vol/vol) ethanol, 10% (vol/vol) bleach, 0.1 M NaOH, and 0.1 M HCl (Fig. 1E), and found that when encased in cellulose it is highly tolerant to chemical stressors, being over 1,000-fold more resistant than E. coli to all treatments (Fig. 1E). Finally, scanning electron microscopy of a cellulose pellicle confirmed that, as with other closely related species, cells of K. rhaeticus iGEM are rod-like, ∼2 μm in length, and heavily encased in cellulose during normal growth (Fig. 1F and SI Appendix, Fig. S3). To determine the genetic basis of high cellulose productivity and to provide background information for genetic engineering of the iGEM strain, we sequenced its genome to 400× coverage and assembled the genome using the genome of Gluconacetobacter xylinus NBRC 3288 (30) as the reference genome for scaffolding (European Nucleotide Archive accession no. PRJEB10933). Sequencing showed that the genome totals 3.87 Mbp with a GC% of 62.7 and contains a predicted 3,573 genes, with an N50 (shortest sequence length at 50% of the genome) of 3.16 Mbp. The genome is divided between a chromosome of 3.16 Mbp, at least two plasmids: pKRi01 (238 kbp) and pKRi02 (3 kbp), and 37 unplaced contigs (in total 460 kbp), which may be part of the chromosome or additional plasmids, and could not be confidently assigned due to being flanked by repetitive sequences (Fig. 2A). The genome sequence revealed several interesting aspects about the biology of K. rhaeticus iGEM. First, a 16S rRNA phylogeny suggests the iGEM strain to be a previously unidentified strain of K. rhaeticus (SI Appendix, Fig. S4), rather than G. xylinus, which is normally thought to be associated with Kombucha tea. Furthermore, the sequence shows the presence of four acs operons on the chromosome, sharing 40–65% amino acid identity (Fig. 2). Up to three acs operons have been reported in other bacterial cellulose-producing species [G. xylinus ATCC 23769 and G. hansenii ATCC 53582 (31, 32)], indicating that the high cellulose synthase copy number may be a possible contributor to the high cellulose productivity observed here (Fig. 1B). These operons also differ in structure. The acs1 operon contains separate acsA and acsB, whereas they are fused in the other operons, and the only genomic copy of acsD is found in acs1. Operon acs4 uniquely contains only acsAB, and phylogenetic analysis indicates that acs4 is most closely related to the acs2 operon (Fig. 2B), and possibly arose via duplication and subsequent translocation. From the genes flanking acs operons, cmcAX, ccpAX, bglxA, bcsX, and bcsY have been previously shown to contribute to cellulose production in closely related species (26, 33, 34). We found two other, stand-alone copies of bglxA from the genome (genomic positions 517401– 519440 and 3029825–3032221), and also identified genes close to acs2 that are associated with extracellular matrix formation (kpsC,

Fig. 2. K. rhaeticus iGEM genome. (A) Overview of the K. rhaeticus iGEM genome. The iGEM genome totals 3.87 Mbp with a GC% of 62.7, and contains a predicted 3,505 protein-coding genes, 3 rRNAs, 52 tRNAs, and 13 other noncoding RNAs. The genome consists of a chromosome of ∼3.16 Mbp and at least two plasmids, pKRi01 (238 kbp) and pKRi02 (3 kbp). The chromosome contains 2,899 predicted genes with 63% GC content and four copies of acs operons. For the chromosome, consecutive rings show (from the outside in): (i) read coverage, (ii and iii), genes on (ii) forward and (iii) reverse strands, (iv) acs operons involved in cellulose synthesis, (v) GC%, and (vi) GC skew. Additionally, the genome contains 37 scaffolds (totaling 460 kbp) that could not be confidently placed due to repetitive sequences. These scaffolds may be part of the chromosome or plasmids, or may belong to additional plasmids (the closely related species G. xylinus NBRC 3288 and K. xylinus E25 contain five and seven plasmids, respectively). (B and C) Phylogenetic relationships (B) and amino acid sequence identity (C) of acs operons. Phylogeny and sequence identity indicate that acs2 and acs4 operons are most closely related. Amino acid sequences were aligned and percent identity was calculated using MUSCLE (54) and the tree was generated using the neighbor-joining method (55). The tree is drawn to scale, with bootstrap values (56) from 1,000 replicates shown next to the branches. All positions containing gaps were eliminated from analysis. See Methods for further details on sequencing and analysis.

prevent the proliferation of these mutants. Furthermore, fine control over cellulose production levels may allow control over the density of cellulose fibrils, and thus the macroscale properties of the cellulose. To achieve controlled cellulose production, we engineered a system in which an E. coli Hfq and an sRNA targeting UGPase mRNA (UDP-glucose pyrophosphorylase) are coexpressed from a plasmid in response to AHL (plasmid J-sRNA-331Bb) (Fig. 4A; also see SI Appendix, Fig. S9 for a detailed overview). The sRNA contains a 24-base region complementary to UGPase mRNA and an E. coli Hfq-binding region. When expressed, it binds to the target UGPase mRNA and recruits E. coli Hfq, inhibiting UGPase translation. We targeted the UGPase gene, as it catalyzes the production of UDP-glucose critical for cellulose synthesis (40) and is present in single copy in the genome, allowing knockdown by a single sRNA. We found this system to be highly efficient, as cellulose production was suppressed completely upon full induction and could be fine-tuned using different concentrations of AHL (Fig. 4B). The observed reduction in cellulose E3434 | www.pnas.org/cgi/doi/10.1073/pnas.1522985113

production was not related to toxicity, as growth rate did not decrease compared with wild-type levels (Fig. 4C and SI Appendix, Fig. S10). This system was engineered to be a general platform for targeted knockdowns in Komagataeibacter and other bacterial species, as expression of E. coli Hfq makes it independent from the host Hfq, and the broad host range pSEVA331Bb backbone enables replication in a wide range of species. Furthermore, new sRNAs can be added to the plasmid, and the 24-base sRNA region can be recoded rapidly by site-directed mutagenesis, making the construct easily modifiable for other targets. Genetic Engineering of Patterned and Functionalized Biomaterials.

Owing to the discovery that K. rhaeticus gene expression can be induced even when inside a cellulose pellicle, we hypothesized that it may be possible to generate spatially and temporally patterned biomaterials that are controlled by the diffusion of the inducer AHL and timing of exposure to induction during pellicle growth. To test this, we induced growing cellulose pellicles with Florea et al.

PNAS PLUS

A

B

K. rhaeticus synthetic biology toolkit v1.0

Plasmid backbones: pSEVA321, 331, 351 pBla-Vhb-122 pBAV1k

Reporter genes: GFPmut3

sfGFP

mRFP1

C

Constitutive promoters: 9 promoters of different strength

Inducible promoters:

2x AHL-inducible promoters

2x ATc-inducible promoters

D

E

F Uninduced

Microscope

1 cm

Induced in pellicle

1 cm

1 cm

200 μm

cells containing the AHL-inducible construct pLux01 with different concentrations of AHL, at different locations and time points (Fig. 5 A and B). We found that both spatial and temporal control were possible. When a limited amount of inducer was added to one side of the pellicle, cells produced mRFP1 following the diffusion gradient of AHL (Fig. 5A). Furthermore, because cells are active only in the top layer of the cellulose pellicle (7), when inducer was added at different times midway through pellicle growth, only cells at the growing top layer produced mRFP1, capturing the temporal difference between uninduced cells in the bottom layers and induced cells at the top (Fig. 5B). To produce functionalized cellulose materials where the nanocellulose matrix is coated by proteins of interest, we considered two strategies: genetic engineering of K. rhaeticus to produce these proteins in situ, or separate expression of proteins in E. coli, which are then purified and applied directly to bacterial cellulose (Fig. 5C). Although the latter requires a three-step process (protein and cellulose production separately, followed by combining the two), it may be preferred for medical and other applications where very high purity of material is required, as it would allow defined Florea et al.

and purified components to be used for functionalization. To test for the possibility of post hoc functionalization, we produced mRFP1 in E. coli, extracted and added it to bacterial cellulose, and compared it by fluorescence microscopy with cellulose produced by K. rhaeticus with in vivo constitutive mRFP1 expression. We found that extracted proteins can diffuse well throughout the pellicle and functionalize the cellulose evenly (Fig. 5D), whereas the granular fluorescence exhibited by expression from pelliclebased cells (Fig. 3E; also see SI Appendix, Fig. S11 for a full-size comparison) indicates that mRFP1 remains largely in the K. rhaeticus cells and would likely require active secretion or lysis of cell membranes to access the extracellular cellulose. To further increase the efficiency of functionalization, we engineered expression vectors that allow easy fusion of proteins to one of four different cellulose-binding domains (CBDs): CBDclos (41), CBDCex (42), dCBD (43), and CBDcipA (44). CBDs are short peptides that bind tightly to cellulose fibrils, thus increasing protein adhesion to cellulose (45). In these constructs, proteins can be modularly fused to CBDs via restriction enzyme cloning. We assessed the cellulose binding strengths of these CBDs PNAS | Published online May 31, 2016 | E3435

MICROBIOLOGY

Fig. 3. K. rhaeticus synthetic biology tookit. (A) Overview of the toolkit contents. (B and C) Constitutive promoter strengths (B) and AHL (pLux01, pLux02) and ATc (pTet01, pTet02) inducible construct expression strengths (C) in K. rhaeticus iGEM, as measured by total mRFP1 fluorescence per cell (fluorescence at 630 nm divided by OD600). Promoter strengths in B and C were assayed in liquid HS with cellulase to remove formation of cellulose fibrils interfering with measurements. (D) Total mRFP1 fluorescence expressed from the pLux01 construct when K. rhaeticus cells were induced with AHL inside the cellulose pellicle (induced in pellicle) compared with uninduced or wild-type (WT) cells. Induction caused a significant increase in fluorescence compared with uninduced or wild-type cells (P < 0.001 for both induced vs. uninduced, and induced vs. wild-type, determined with one-way ANOVA and Tukey’s post hoc tests). Cells were grown in HS (without cellulase), and fluorescence was quantified by fluorescence microscopy image analysis. (E and F) Induction in the pellicle results in visible mRFP1 production compared with uninduced cells (E) (indicated by the arrow; also see SI Appendix, Fig. S8), and results in granular fluorescence due to localization within cells (F). n = 3 for all experiments; error bars indicate SD. For E and F, images were cropped and contrast was adjusted to improve clarity. See SI Appendix, Fig. S7 for a detailed overview of constructs and Methods for details of characterization assays.

Fig. 4. sRNA construct (J-sRNA-331Bb) for control of cellulose production. (A) Overview of the sRNA silencing construct. Constitutively produced LuxR binds to pLux in the presence of AHL and up-regulates production of E. coli Hfq and an sRNA targeting UGPase mRNA. The 5′ end of the sRNA contains a 24-base sequence complementary to UGPase mRNA, and binds to it in the presence of E. coli Hfq (53), leading to silencing of the UGPase gene. (B) Cellulose production of induced and uninduced cultures shown as cellulose dry weight measured 40 h postinoculation. “No cellulose” indicates empty weighing boats; “pSEVA331Bb 100 nM AHL” and “pSEVA331Bb” indicate iGEM strain with empty pSEVA331Bb vector with and without 100 nM AHL, respectively. Full induction with 100 and 500 nM AHL results in complete suppression of cellulose synthesis (adjusted P < 0.0001 for uninduced vs. 100 nM AHL and 500 nM AHL). Addition of AHL itself does not decrease cellulose productivity (adjusted P > 0.999 for pSEVA331Bb vs. pSEVA331Bb with 100 nM AHL), and uninduced cells are not different from negative controls (adjusted P = 0.12 for uninduced vs. pSEVA331Bb 100 nM AHL). (C) OD600 of cultures 3 h postinoculation. Differences between OD600 are not significant for any comparisons (adjusted P > 0.05), showing that induction of the sRNA silencing construct does not reduce growth rate. n = 5 for all samples (except n = 3 for pSEVA331Bb and pSEVA331Bb with 100 nM AHL in B); error bars indicate SD. Statistical significance was determined with one-way ANOVA and Tukey’s multiple comparisons tests for B and C. See Methods for further details of all assays.

by washing four different E. coli-extracted CBD-sfGFP fusion proteins with different solvents (dH2O, EtOH, BSA, and PBS) and measuring the fluorescence that remained bound. Addition of CBDs to sfGFP gave up to a fivefold increase in binding to cellulose compared with GFP alone (SI Appendix, Fig. S12). Finally, because bacterial cellulose is a candidate for new textile materials and of high interest to the fashion industry, we used our approach to demonstrate production of functionalized garments. Using the CBDcipA-sfGFP fusion protein extract and dried pellicle material from K. rhaeticus cultures, we created bacterial cellulose fashion accessories by functionalization of cellulose fibrils with green fluorescent protein (SI Appendix, Fig. S13), indicating that this approach is scalable to produce macroscale objects. Discussion Here we isolated a strain of bacterial cellulose-producing K. rhaeticus (K. rhaeticus iGEM), which can grow on acidic and low-nitrogen media, is highly resistant to damaging chemical agents when encased in cellulose, and is notable for its high cellulose production. Although it is clear that K. rhaeticus iGEM can effectively produce cellulose in low-nitrogen media, whether it fixes nitrogen is still an open question. Nitrogen fixation seems to be prevalent in Acetobacteraceae, as multiple species (most notably G. diazotrophicus) have been recorded to be nitrogenfixing (46). However, in the case of K. rhaeticus iGEM, although it can grow in nitrogen-free LGI medium (Fig. 1 C and D and SI Appendix, Figs. S1 and S2) and its genome contains a set of genes associated with nitrogen fixation (SI Appendix, Table S1), we were unable to find genes homologous to G. diazotrophicus nifHDK, which form the structural subunits of the nitrogenase complex (47). This either suggests that despite careful handling, very low level nitrogen contamination may have been present, allowing K. rhaeticus to produce cellulose in LGI medium or, alternatively, that K. rhaeticus uses a different set of nitrogenase E3436 | www.pnas.org/cgi/doi/10.1073/pnas.1522985113

genes for nitrogen fixation, as alternative nitrogenases have been isolated in other species (48). However, in either case, from the perspective of manufacturing, low nitrogen tolerance allows the use of cheap, low-nitrogen media, potentially reducing cellulosemanufacturing costs. To use this strain for biotechnology applications, it was first necessary to develop methods and tools for its genetic manipulation. For a genetic toolkit to be useful, it should minimally allow introduction of foreign DNA, plasmid propagation, and a degree of control over heterologous gene expression levels. Although the required tools can differ between species and the applications in which they are used, they should ideally also allow tunable regulation of transgene expression through inducible or repressible systems and regulation over native gene expression. A toolkit should further provide relevant characterization data about the host, such as a genome sequence and growth and productivity rates in different conditions. Indeed, many of these elements have been part of toolkits created for other species (5, 6, 49). The toolkit described here for K. rhaeticus iGEM contains all of these features, providing protocols and DNA constructs to allow transformation, precisely controlled constitutive and inducible expression of heterologous genes, control over endogenous gene expression, and a genome sequence with characterization data. Using this toolkit, we created two approaches for engineering new cellulose-based materials, first through genetic engineering of K. rhaeticus itself, and second through application of extracted proteins to the bacterial cellulose. Genetic engineering offers simple, one-step in situ production of materials, and would be ideal for applications where the cellulose matrix is modified as it is being made. The physical and biochemical properties of cellulose are largely dependent on the microscale morphology and structure of cellulose fibrils, so modification of these during production by expressed heterologous enzymes could result in new material properties. For example, by engineering cells to Florea et al.

C

Spaal paerning

Post-hoc funconalizaon

PNAS PLUS

A

D

Transform E. coli

10 cm

Extract protein

Apply to cellulose

10 cm

B Temporal

Not induced

paerning

Induced on day 0 only

Induced on days 4 -9

Induced on days 2-9

100 μm

Induced on days 0 - 9

i

3 cm

ii 1 cm

iii

1 cm

incorporate N-acetylglucosamine residues into cellulose, Yadav et al. (13) created biodegradable, low-immunogenicity cellulose tissue scaffolds. Our second approach (addition of purified proteins to already-produced cellulose) is likely to be preferable for applications where bulk material properties are not changed but the material is functionalized with new properties. In our experiments, post hoc functionalization resulted in an even, thorough distribution of functionalizing proteins (Fig. 5D), and may be preferable in medical and tissue engineering applications where highly pure materials are required, as both the cellulose and functionalizing proteins can be purified before formation of the composite material. Together, these two approaches complement each other, as the ability to modify cellulose via genetic engineering as well as functionalize it with purified proteins allows for a wider range of materials to be engineered. Here, genetic engineering allowed us to control cellulose production by sRNA-mediated knockdown and create spatial and temporal patterning through induction with N-acyl homoserine lactone. We used an AHL-inducible system here; however, for industrial applications, it is likely straightforward to exchange it for a control system using cheaper chemical or physical inducers if required. Although we did not test for this specifically, it is important to note that as knockdown of cellulose production did Florea et al.

not significantly change growth rate, control over cellulose production levels may also allow control of the specific cell-vs.-cellulose and thus protein-vs.-cellulose concentrations in the material. In the case of functionalization of cellulose with CBD fusion proteins, in principle any protein stable enough to function in the intended extracellular environment could be used for functionalization. By providing genetic engineering tools that allow control of production of bacterial cellulose and modification of it as a material, this work enables a variety of potential applications. The physical properties and pore size of the cellulose might be tunable by altering gene expression of the K. rhaeticus acs operons or by complexing cellulose in situ with secreted structural proteins such as curli fimbriae (12). Cellulose can be functionalized with specific proteins (Fig. 5C and SI Appendix, Figs. S12 and S13), which may be used to create advanced materials such as bacterial cellulose wound dressings coated with antimicrobial peptides or novel high-specificity water filters coated with proteins binding specific contaminants. In another area, the toolkit allows creating patterned cellulose structures in three dimensions (Fig. 5 A and B). In combination with functionalization, this may be used in tissue engineering for one-step production of cellulosebased tissue engineering scaffolds that contain specifically patterned PNAS | Published online May 31, 2016 | E3437

MICROBIOLOGY

Fig. 5. Engineering of patterned and functionalized cellulose materials on a macroscale. (A) Spatial patterning. Cells were induced for mRFP1 expression by addition of 100 nM AHL to one side of a 1-L culture and the pellicle was imaged for red fluorescence 3 d postinduction. (Inset) Pellicle imaged in white light. (B) Temporal patterning. Cells were induced with AHL daily at different times through pellicle growth (0 d only, or starting at 0, 2, or 4 d postinoculation) and imaged 9 d postinoculation, with overview (i), white light (ii), and fluorescence (iii) images of the pellicles and pellicle cross-sections shown. Note that pellicles induced on day 0 only show low fluorescence due to natural degradation of AHL in the media over time. (C) Overview of the cellulose functionalization strategy via post hoc addition of mRFP1 extracted from E. coli. (D) Fluorescence microscopy of a cross-section of cellulose functionalized with mRFP1 through addition of mRFP1 extracted from E. coli. Smooth fluorescence is seen throughout the pellicle cross-section, compared with the granular fluorescence seen in Fig. 3E (also see SI Appendix, Fig. S11). For A and B, iii, computationally determined and averaged brightness (gray value) along the pellicle crosssection is shown (Right). Images were cropped and contrast was adjusted to improve clarity for all images. See Methods for fluorescence imaging conditions and other details.

growth factors. Furthermore, the remarkable robustness of K. rhaeticus within cellulose (Fig. 1E) offers potential uses in biosensing. The growing pellicle can store changes in environmental signals by writing the conditions into different cellulose layers as they grow (Fig. 5B), acting as a record akin to ice cores. Finally, because cellulose synthesis and nitrogen fixation in closely related Acetobacteraceae are areas of active research, this toolkit is also a valuable resource for basic studies aiming to dissect the molecular mechanisms of these processes. As there are many possible applications enabled by this work, we are making this strain and the genetic engineering toolkit available through ATCC and AddGene. Methods Isolation, Characterization, and Culturing of K. rhaeticus iGEM. K. rhaeticus iGEM was isolated from a Kombucha SCOBY (symbiotic colony of bacteria and yeast) of Czech origin (Happy Kombucha) by streaking homogenized SCOBY material on Hestrin–Schramm (HS) agarose (SI Appendix, Table S5), verifying cell morphology under a light microscope, and restreaking isolated colonies on HS agarose twice. Two percent (wt/vol) glucose was used in HS unless stated otherwise. Glycerol stocks were prepared by culturing the iGEM strain statically in HS medium for 6 d, followed by addition of 0.2% (vol/vol) cellulase (Trichoderma reesei cellulase; C2730; Sigma), incubation with 230-rpm shaking, 30 °C for 1 d, addition of glycerol to 25% (vol/vol), and storage at −80 °C. For cellulose production, seed cultures of K. rhaeticus iGEM were inoculated from glycerol stocks and grown statically at 30 °C in Kombucha tea, HS, or LGI medium (see SI Appendix, Supplementary Protocols and Table S5 for details). When grown with shaking without cellulose production, cultures were grown in HS cellulase [0.4% (vol/vol)] at 30 °C, 230-rpm shaking at 45° tube angle. Unless otherwise stated, 50-mL Corning tubes (CLS430829; Sigma) with 5–20 mL media were used for culturing. When grown on HS agar, plates were incubated inverted at 30 °C. For culturing of K. rhaeticus transformed with plasmids encoding kanamycin or chloramphenicol resistance genes, kanamycin was added to 500 μg/mL for HS agar and 50–100 μg/mL for liquid HS, and chloramphenicol was added to 340 μg/mL for HS agar and 34–68 μg/mL for liquid HS. Cellulose productivity on different media was measured by culturing in 20 mL HS glucose [2% (wt/vol)], HS sucrose [2% (wt/vol)], HS without a carbon source (negative control), and Kombucha tea in 50-mL Corning tubes at 30 °C for 10 d, with loose caps for increased air diffusion, and kept at 4 °C until measurement of cellulose weight (SI Appendix, Supplementary Protocols). Productivity of K. rhaeticus iGEM was reported in comparison with productivity of the high-producing G. hansenii ATCC 53582 instead of maximal cellulose yield per volume of media, as maximal total productivity is highly dependent on specific culturing conditions and may not be a good measure of genetically determined production capabilities. To test cellulose productivity on nitrogen-free LGI medium, 80 μL OD600 1 E. coli or K. rhaeticus iGEM seed culture was inoculated into 25 mL LGI medium in 50-mL glass tubes (3119-0050; Thermo Fischer) and imaged 9 d postinoculation. For scanning electron micrographs, cellulose was gold-coated under vacuum and imaged at 2,000–6,000× magnification and 20 kV. See SI Appendix, Supplementary Methods for details. Chemical Tolerance Assays. For both E. coli and K. rhaeticus, cells were treated with 0.5 mL 0.1 M NaOH, 0.1 M HCl, 70% (vol/vol) ethanol, 10% (vol/vol) bleach, or PBS for 5, 30, or 90 min. Treatments were then plated; colonies were photographed in white light and counted to determine the fraction of surviving cells compared with PBS-treated cells. For details of the assay, see SI Appendix, Supplementary Methods. Genome Sequencing, Assembly, Bioinformatics, and Statistics. The K. rhaeticus iGEM genome was sequenced with an Illumina MiSeq using 250-bp pairedend reads, to a coverage of ∼400×. Reads were then downsampled to 100× coverage, assembled using the BugBuilder pipeline (50), quality-controlled, and annotated using Prokka with the full read set (51). All statistical tests were performed with Prism 6 (GraphPad Software). For details of genome sequencing and bioinformatics, see SI Appendix, Supplementary Methods. Engineering of Constitutive Promoters, Inducible Promoters, CBD Fusions, and sRNA Constructs. pSEVA331Bb and pSEVA321Bb were constructed from pSEVA331 and pSEVA321, respectively, via substituting the native polylinker with BioBrick prefix and suffix. This was done by PCR mutagenesis with Q5 polymerase (M0491S; NEB), primers i75 and i76 (SI Appendix, Table S6), di-

E3438 | www.pnas.org/cgi/doi/10.1073/pnas.1522985113

gestion with SpeI (R3133S; NEB), and subsequent religation. Constitutive promoter-mRFP1 constructs (BBa_J23100–Bba_J23117 by iGEM 2006 Berkeley) were received from the iGEM Registry of Standard Biological Parts (52) and subcloned into pSEVA331Bb. ATc-inducible constructs were kindly provided by Francesca Ceroni at Imperial College London, and AHL-inducible constructs BBa_J09855 (Jon Badalamenti, iGEM 2005) and BBa_F2620 (iGEM 2004 MIT) were received from the Registry of Standard Biological Parts. Inducible constructs were then cloned into J23100-mRFP1-pSEVA331Bb, replacing the constitutive J23100 promoter (thus controlling downstream mRFP1 expression) to create pTet01, pTet02, pLux01, and pLux02. CBDclos and CBDcex were received as BBa_K863111 and BBa_K863101 (iGEM 2012 Bielefeld) from the Registry of Standard Biological Parts, and CBDcipA and dCBD were synthesized as a GeneArt String (Life Technologies). CBDs were then fused to sfGFP (BBa_I746909, iGEM 2007 Cambridge) and cloned into an expression vector (BBa_J04500, Kristen DeCelle, iGEM 2005) downstream of the pLacI promoter. An sRNA construct was also synthesized as a GeneArt String (Life Technologies) based on descriptions of Na et al. (53) and subcloned downstream of the pLux promoter in pLux01, replacing mRFP1. All constructs were first transformed into E. coli Turbo, and colonies were screened using colony PCR with GoTaq Green (M7122; Promega) and the primers i53 and i54 (SI Appendix, Table S6). Plasmid DNA from positive colonies was extracted with a QIAprep Spin Miniprep Kit (27104; Qiagen), DNA was sequenced (Source BioScience), and correct sequences were transformed into K. rhaeticus iGEM (SI Appendix, Supplementary Protocols). DNA sequences of all constructs created in this work are accessible through the Registry of Standard Biological Parts (25) (see SI Appendix, Table S4 for accession numbers). Because many of the constructs characterized in K. rhaeticus iGEM are widely used and accessible through the Registry of Standard Biological Parts, no constructs were codon-optimized, to allow them to be used in K. rhaeticus iGEM without modification by the user. Characterization of Constitutive Promoters, Inducible Promoters, CBD Fusions, and sRNA Constructs. For characterization of constitutive promoters and inducible promoters without cellulose formation, 1–5 mL liquid HS cellulase [8–10% (vol/vol)] containing 34 μg/mL chloramphenicol in 50-mL Corning tubes was inoculated from seed cultures to OD600 0.02 and grown at 30 °C, 230 rpm for 3 h, and 200 μL culture was then pipetted into 96-well plates (Corning Costar). Inducer was added to inducible cultures to 1 μg/mL for ATc and 1 μM for AHL. High concentrations of cellulase were used to remove any interference by cellulose in spectroscopy measurements. Plates were covered with Breathe-Easy membrane (Z380059; Sigma), and OD600 and mRFP1 intensity (excitation 590 nm, emission 630 nm) were measured every 15 min using a Synergy HT microplate reader (BioTek) at 29 °C and high shaking speed. For characterization of inducible promoters in pellicle form (Fig. 3 D and F and SI Appendix, Fig. S8), K. rhaeticus iGEM containing pLux01 was inoculated from glycerol stocks into HS with 34 μg/mL chloramphenicol and grown statically for 8 d at 30 °C. For cultures induced during growth, AHL was added to 1 μM before inoculation, and for cultures induced in pellicle, AHL was added to 1 μM 6 d postinoculation. Pellicles were washed with PBS 8 d after inoculation, microscopy samples were prepared with a sterile razor, and fluorescence was quantified using fluorescence microscopy (Nikon Eclipse Ti) at mCherry preset (590-nm emission, 0.2-s exposure, 675 V EM gain, 150× magnification). A single-layer cellulose sheet was placed on the sample and fluorescence intensity was determined as intensity/exposure time for quantitative measurements. For white-light images (Fig. 3E), 250-mL beakers (CLS1000250; Sigma) containing 100 mL HS media were inoculated with 5 mL OD600 1 seed culture, and the beakers were covered with Breathe-Easy membrane and incubated statically at 30 °C. For the induced pellicle, 1,000 μL 100 mM AHL was pipetted daily 4 d after inoculation along the edges of the pellicle, and the pellicle was imaged 9 d postinoculation. For characterization of CBD binding strengths, 200 μL CBD-sfGFP fusion proteins extracted from E. coli (SI Appendix, Supplementary Protocols) were added to a 96-well plate containing homogenized bacterial cellulose, incubated overnight at 4 °C, and washed thrice with treatment [dH2O, PBS, 5% (vol/vol) BSA or 70% (vol/vol) EtOH] (SI Appendix, Supplementary Protocols). GFP fluorescence was measured on a 96-well plate reader (Synergy HT; BioTek) (see SI Appendix, Supplementary Protocols for details). For characterization of sRNA constructs, HS with 34 μg/mL chloramphenicol was inoculated to OD600 0.04 with K. rhaeticus iGEM containing plasmid J09855sRNA-331Bb and induced with 10–500 nM AHL. Forty hours after inoculation, pellicles were washed, dried at 60 °C for 16 h, and weighed (see SI Appendix, Supplementary Protocols for details). For characterization of growth rate, OD600 was measured in [3% (vol/vol)] HS cellulase with 34 μg/mL chloramphenicol using the protocol used for characterization of constitutive promoters.

Florea et al.

professional fashion designer. See SI Appendix, Supplementary Protocols for further details.

1. Cameron DE, Bashor CJ, Collins JJ (2014) A brief history of synthetic biology. Nat Rev Microbiol 12(5):381–390. 2. Nikel PI, Martínez-García E, de Lorenzo V (2014) Biotechnological domestication of pseudomonads using synthetic biology. Nat Rev Microbiol 12(5):368–379. 3. Silva-Rocha R, et al. (2013) The Standard European Vector Architecture (SEVA): A coherent platform for the analysis and deployment of complex prokaryotic phenotypes. Nucleic Acids Res 41(Database issue):D666–D675. 4. Kushwaha M, Salis HM (2015) A portable expression resource for engineering crossspecies genetic circuits and pathways. Nat Commun 6:7832. 5. Markley AL, Begemann MB, Clarke RE, Gordon GC, Pfleger BF (2015) Synthetic biology toolbox for controlling gene expression in the cyanobacterium Synechococcus sp. strain PCC 7002. ACS Synth Biol 4(5):595–603. 6. Mimee M, Tucker AC, Voigt CA, Lu TK (2015) Programming a human commensal bacterium, Bacteroides thetaiotaomicron, to sense and respond to stimuli in the murine gut microbiota. Cell Syst 1(1):62–71. 7. Lee K-Y, Buldum G, Mantalaris A, Bismarck A (2014) More than meets the eye in bacterial cellulose: Biosynthesis, bioprocessing, and applications in advanced fiber composites. Macromol Biosci 14(1):10–32. 8. Rognoli V, Bianchini M, Maffei S, Karana E (2015) DIY materials. Mater Des 86: 692–702. 9. Wong HC, et al. (1990) Genetic organization of the cellulose synthase operon in Acetobacter xylinum. Proc Natl Acad Sci USA 87(20):8130–8134. 10. Williams WS, Cannon RE (1989) Alternative environmental roles for cellulose produced by Acetobacter xylinum. Appl Environ Microbiol 55(10):2448–2452. 11. Zhong C, et al. (2014) Strong underwater adhesives made by self-assembling multiprotein nanofibres. Nat Nanotechnol 9(10):858–866. 12. Chen AY, et al. (2014) Synthesis and patterning of tunable multiscale materials with engineered cells. Nat Mater 13(5):515–523. 13. Yadav V, et al. (2010) Novel in vivo-degradable cellulose-chitin copolymer from metabolically engineered Gluconacetobacter xylinus. Appl Environ Microbiol 76(18): 6257–6265. 14. Mohite BV, Patil SV (2014) Physical, structural, mechanical and thermal characterization of bacterial cellulose by G. hansenii NCIM 2529. Carbohydr Polym 106:132–141. 15. Hsieh YC, Yano H, Nogi M, Eichhorn SJ (2008) An estimation of the Young’s modulus of bacterial cellulose filaments. Cellulose 15(4):507–513. 16. Rusli R, Eichhorn SJ (2008) Determination of the stiffness of cellulose nanowhiskers and the fiber-matrix interface in a nanocomposite using Raman spectroscopy. Appl Phys Lett 93(3):033111. 17. Sun S, et al. (2010) Comparison of the mechanical properties of cellulose and starch films. Biomacromolecules 11(1):126–132. 18. Klemm D, Schumann D, Udhardt U, Marsch S (2001) Bacterial synthesized cellulose— Artificial blood vessels for microsurgery. Prog Polym Sci 26(9):1561–1603. 19. Lee K-Y, Wong LLC, Blaker JJ, Hodgkinson JM, Bismarck A (2011) Bio-based macroporous polymer nanocomposites made by mechanical frothing of acrylated epoxidised soybean oil. Green Chem 13(11):3117–3123. 20. Hu W, et al. (2011) Highly stable and sensitive humidity sensors based on quartz crystal microbalance coated with bacterial cellulose membrane. Sens Actuators B Chem 159(1):301–306. 21. Yoon SH, Jin H-J, Kook M-C, Pyun YR (2006) Electrically conductive bacterial cellulose by incorporation of carbon nanotubes. Biomacromolecules 7(4):1280–1284. 22. Hu W, Chen S, Yang J, Li Z, Wang H (2014) Functionalized bacterial cellulose derivatives and nanocomposites. Carbohydr Polym 101:1043–1060. 23. Ishida T, Sugano Y, Nakai T, Shoda M (2002) Effects of acetan on production of bacterial cellulose by Acetobacter xylinum. Biosci Biotechnol Biochem 66(8): 1677–1681. 24. Yamada Y, et al. (2012) Description of Komagataeibacter gen. nov., with proposals of new combinations (Acetobacteraceae). J Gen Appl Microbiol 58(5):397–404. 25. Smolke CD (2009) Building outside of the box: iGEM and the BioBricks Foundation. Nat Biotechnol 27(12):1099–1102. 26. Kawano S, et al. (2002) Cloning of cellulose synthesis related genes from Acetobacter xylinum ATCC23769 and ATCC53582: Comparison of cellulose synthetic ability between strains. DNA Res 9(5):149–156.

27. Lisdiyanti P, Navarro RR, Uchimura T, Komagata K (2006) Reclassification of Gluconacetobacter hansenii strains and proposals of Gluconacetobacter saccharivorans sp. nov. and Gluconacetobacter nataicola sp. nov. Int J Syst Evol Microbiol 56(Pt 9): 2101–2111. 28. James EK, Reis VM, Olivares FL, Baldani JI, Döbereiner J (1994) Infection of sugar cane by the nitrogen-fixing bacterium Acetobacter diazotrophicus. J Exp Bot 45(6): 757–766. 29. Dutta D, Gachhui R (2007) Nitrogen-fixing and cellulose-producing Gluconacetobacter kombuchae sp. nov., isolated from Kombucha tea. Int J Syst Evol Microbiol 57(Pt 2):353–357. 30. Ogino H, et al. (2011) Complete genome sequence of NBRC 3288, a unique cellulosenonproducing strain of Gluconacetobacter xylinus isolated from vinegar. J Bacteriol 193(24):6997–6998. 31. Iyer PR, Geib SM, Catchmark J, Kao TH, Tien M (2010) Genome sequence of a cellulose-producing bacterium, Gluconacetobacter hansenii ATCC 23769. J Bacteriol 192(16):4256–4257. 32. Florea M, Reeve B, Abbott J, Freemont PS, Ellis T (2016) Genome sequence and plasmid transformation of the model high-yield bacterial cellulose producer Gluconacetobacter hansenii ATCC 53582. Sci Rep 6:23635. 33. Koo HM, Song SH, Pyun YR, Kim YS (1998) Evidence that a beta-1,4-endoglucanase secreted by Acetobacter xylinum plays an essential role for the formation of cellulose fiber. Biosci Biotechnol Biochem 62(11):2257–2259. 34. Umeda Y, et al. (1999) Cloning of cellulose synthase genes from Acetobacter xylinum JCM 7664: Implication of a novel set of cellulose synthase genes. DNA Res 6(2): 109–115. 35. Willis LM, Whitfield C (2013) KpsC and KpsS are retaining 3-deoxy-D-manno-oct-2ulosonic acid (Kdo) transferases involved in synthesis of bacterial capsules. Proc Natl Acad Sci USA 110(51):20753–20758. 36. Pradel E, Parker CT, Schnaitman CA (1992) Structures of the rfaB, rfaI, rfaJ, and rfaS genes of Escherichia coli K-12 and their roles in assembly of the lipopolysaccharide core. J Bacteriol 174(14):4736–4745. 37. Lee S, et al. (1998) Analysis of nitrogen fixation and regulatory genes in the sugar cane endophyte Acetobacter diazotrophicus. Nitrogen Fixation with Non-Legumes, Developments in Plant and Soil Sciences, eds Malik KA, Sajjad Mirza M, Ladha JK (Kluwer Academic, Dordrecht, The Netherlands), Vol 79, pp 11–19. 38. Chien L-J, Chen H-T, Yang P-F, Lee C-K (2006) Enhancement of cellulose pellicle production by constitutively expressing vitreoscilla hemoglobin in Acetobacter xylinum. Biotechnol Prog 22(6):1598–1603. 39. Valla S, Kjosbakken J (1982) Cellulose-negative mutants of Acetobacter xylinum. Microbiology 128(7):1401–1408. 40. Koo HM, Yim SW, Lee CS, Pyun YR, Kim YS (2000) Cloning, sequencing, and expression of UDP-glucose pyrophosphorylase gene from Acetobacter xylinum BRC5. Biosci Biotechnol Biochem 64(3):523–529. 41. Goldstein MA, et al. (1993) Characterization of the cellulose-binding domain of the Clostridium cellulovorans cellulose-binding protein A. J Bacteriol 175(18):5762–5768. 42. Ong E, Gilkes NR, Miller RC, Jr, Warren RA, Kilburn DG (1993) The cellulose-binding domain (CBD(Cex)) of an exoglucanase from Cellulomonas fimi: Production in Escherichia coli and characterization of the polypeptide. Biotechnol Bioeng 42(4): 401–409. 43. Linder M, Salovuori I, Ruohonen L, Teeri TT (1996) Characterization of a double cellulose-binding domain. Synergistic high affinity binding to crystalline cellulose. J Biol Chem 271(35):21268–21272. 44. Andrade FK, Costa R, Domingues L, Soares R, Gama M (2010) Improving bacterial cellulose for blood vessel replacement: Functionalization with a chimeric protein containing a cellulose-binding module and an adhesion peptide. Acta Biomater 6(10): 4034–4041. 45. Linder M, Teeri TT (1997) The roles and function of cellulose-binding domains. J Biotechnol 57(1–3):15–28. 46. Pedraza RO (2008) Recent advances in nitrogen-fixing acetic acid bacteria. Int J Food Microbiol 125(1):25–35. 47. Franke IH, Fegan M, Hayward AC, Sly LI (1998) Nucleotide sequence of the nifH gene coding for nitrogen reductase in the acetic acid bacterium Acetobacter diazotrophicus. Lett Appl Microbiol 26(1):12–16.

Florea et al.

PNAS PLUS

Engineering of Functionalized Biomaterials. For spatial patterning, K. rhaeticus containing pLux01 was inoculated into 1 L HS chloramphenicol in 2-L Erlenmeyer flasks, 500 μL 100 nM AHL was added to one side of the pellicle 2 d later, and pellicle was imaged 4 d after induction. For temporal patterning, 250-mL beakers with 100 mL HS media were inoculated with 5 mL seed culture at OD600 1, and the beakers were covered with Breathe-Easy membrane and incubated statically at 30 °C. After this, 1,000 μL 100 mM AHL was pipetted along the edges of the pellicle daily with membrane replacement, starting at different days postinoculation as shown in Fig. 5B, and the pellicle was imaged 9 d postinoculation. For functionalization of cellulose with mRFP1 and CBDcipA-sfGFP, proteins were first produced in E. coli and extracted via sonication, and extracts were applied to purified wet or dried bacterial cellulose, respectively. For mRFP1-functionalized cellulose, fluorescence intensity was determined from a single cellulose sheet using fluorescence microscopy. For CBDcipAsfGFP, extracts were applied to dried bacterial cellulose with a paintbrush and dried. The resulting material was then crafted into accessories by a

PNAS | Published online May 31, 2016 | E3439

MICROBIOLOGY

ACKNOWLEDGMENTS. We thank Dr. Cheng-Kang Lee from the National Taiwan University of Science and Technology and Dr. Jyh-Ming Wu from Chinese Culture University for sharing plasmid pBla-Vhb-122; Ms. Victoria Geaney from the Royal College of Art, London, for help with functionalized garments; Dr. Koon-Yang Lee (Imperial College London) for helpful advice and feedback on the manuscript; Dr. Takayuki Homma (Imperial College London) for help with genome library preparation; Dr. Laurence Game (Imperial College London) for help and advice with genome sequencing; Mr. Geraint Barton (Imperial College London) for help and advice with genome assembly and analysis; Dr. Francesca Ceroni (Imperial College London) for providing ATc-inducible constructs and advice; Ms. Catherine Ainsworth, Mr. Nicolas Kral, Dr. Geoff Baldwin, and Dr. Guy-Bart Stan of Imperial College London for instruction and helpful discussions throughout the project; Dr. Mahmoud Ardakani (Imperial College London) for taking SEM images; and Dr. Carlos Bricio Garberi (Imperial College London) for taking pellicle photographs.

48. Chisnell JR, Premakumar R, Bishop PE (1988) Purification of a second alternative nitrogenase from a nifHDK deletion strain of Azotobacter vinelandii. J Bacteriol 170(1): 27–33. 49. Lee ME, DeLoache WC, Cervantes B, Dueber JE (2015) A highly characterized yeast toolkit for modular, multipart assembly. ACS Synth Biol 4(9):975–986. 50. Abbott J (2015) BugBuilder: Microbial Genome Assembly Pipeline. Available at www.imperial.ac.uk/bioinformatics-support-service/resources/software/bugbuilder/. Accessed September 7, 2015. 51. Seemann T (2014) Prokka: Rapid prokaryotic genome annotation. Bioinformatics 30(14):2068–2069.

E3440 | www.pnas.org/cgi/doi/10.1073/pnas.1522985113

52. Shetty RP, Endy D, Knight TF, Jr (2008) Engineering BioBrick vectors from BioBrick parts. J Biol Eng 2:5. 53. Na D, et al. (2013) Metabolic engineering of Escherichia coli using synthetic small regulatory RNAs. Nat Biotechnol 31(2):170–174. 54. Edgar RC (2004) MUSCLE: Multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32(5):1792–1797. 55. Saitou N, Nei M (1987) The neighbor-joining method: A new method for reconstructing phylogenetic trees. Mol Biol Evol 4(4):406–425. 56. Felsenstein J (1985) Confidence limits on phylogenies: An approach using the bootstrap. Evolution (N Y) 39(4):783–791.

Florea et al.