Komagataeibacter tool kit (KTK): a modular cloning system for multigene constructs and programmed protein secretion from cellulose producing bacteria

Bacteria proficient at producing cellulose are an attractive synthetic biology host for the emerging field of Engineered Living Materials (ELMs). Species from the Komagataeibacter genus produce high yields of pure cellulose materials in a short time with minimal resources, and pioneering work has shown that genetic engineering in these strains is possible and can be used to modify the material and its production. To accelerate synthetic biology progress in these bacteria, we introduce here the Komagataeibacter tool kit (KTK), a standardised modular cloning system based on Golden Gate DNA assembly that allows DNA parts to be combined to build complex multigene constructs expressed in bacteria from plasmids. Working in Komagataeibacter rhaeticus, we describe basic parts for this system, including promoters, fusion tags and reporter proteins, before showcasing how the assembly system enables more complex designs. Specifically, we use KTK cloning to reformat the Escherichia coli curli amyloid fibre system for functional expression in K. rhaeticus, and go on to modify it as a system for programming protein secretion from the cellulose producing bacteria. With this toolkit, we aim to accelerate modular synthetic biology in these bacteria, and enable more rapid progress in the emerging ELMs community.


29
The emerging field of Engineered Living Materials (ELMs) uses synthetic biology to grow and 30 engineer materials with characteristics that are prominent in nature, such as colour, self-repair, 31 growth, conductivity and inherent sensing (Gilbert and Ellis, 2019;Tang et al., 2020). Being 32 biological, these materials also remain biodegradable and can be grown from sustainable nutrient 33 sources and so have great potential in a circular economy. A significant focus of early ELMs work 34 has been on bacterial biofilms, and in particular, on manipulating proteinaceous amyloid fibres and 35 cellulosic polymers as these lends themselves to a number of ELM-based applications ( both the general cellulose and the more specific cellulose-based ELM fields could greatly benefit 47 4 has been modified to encode Golden Gate cloning sequences. A further possibility when 94 constructing a TU with KTK is to create a fusion protein in the CDS position (E1.3). The system allows 95 this by ligating C-and N-terminal domain CDS parts (E1.3a and E1.3b) (Fig 1b). The basic KTK system 96 includes a library of useful Entry-level Parts (promoters, RBS, terminators, and useful C-and N-97 terminal CDS domains e.g. GFP, His-tag and signal peptides) as well as sequences that can be used 98 as 'spacers' that are useful at subsequent cloning levels. All plasmid construction is designed to be 99 done using E. coli cloning strains, before completed plasmids are then transformed into competent 100 Komagataeibacter for testing.

116
For simple constitutive expression of a protein, constructing a TU via a single round of 117 cloning (level one) is sufficient. However, for multifaceted genetic constructs, such as inducible 118 expression cassettes or multi-part TUs, the KTK adds further cloning levels that enable versatile 119 5 construction. The backbone vectors for each cloning level have an alternating arrangement of 120 flanking Type IIS RE sites (for BsaI and BpiI). These in turn generate specific overhangs that are used 121 for the next level of cloning (illustrated and detailed in Supplementary Materials). The nature of 122 Type IIS RE cloning means that one of set of RE sites are removed during the ligation reaction and 123 the second set then becomes available for the next level of assembly. Furthermore, the KTK system 124 provides two Backbone vectors for each cloning level, each with slight variation in overhangs to 125 allow for the parallel insertion of different ligated DNA parts and TU assemblies. These can then be 126 joined in the next level of ligation, thereby facilitating multiple cycles with multipart constructs (Fig  127   1b) superfolder Green Fluorescent Protein (sfGFP) and mScarlet Red Fluorescent Protein (RFP). These 135 two CDS parts were also cloned as N-terminal (E1.3a) and C-terminal (E1.3b) CDS parts to also allow 136 assembly of a TU expressing sfGFP-mScarlet fusion protein (Fig 2a). Assembly of basic and fusion TU 137 constructs was done using the KTK system with cloning steps in E. coli. The three constructs were 138 then transformed into K. rhaeticus competent cells, and selected transformant colonies were then 139 cultured in liquid growth media in the presence of purified cellulase (to prevent material formation) 140 and measured for green and red fluorescence by flow cytometry (Fig 2b). Destination vectors is important when going to Level 2. A TU constitutively expressing LuxR was 151 6 assembled into a D1.1 vector, while the TU expressing sfGFP from the pLux promoter was assembled 152 into a D1.2 vector (Fig 2c). The multigene Level 2 construct was then assembled from Level 1 153 constructs and transformed into K. rhaeticus. This construct worked as expected showing induction 154 of GFP expression in the presence of increasing AHL concentrations (Fig 2d).

166
Error bars represent SD of three replicates.

Expanding and characterising the KTK parts library 169
The utility of GG-based cloning toolkits are dependent on the size, characterization and availability 170 its Parts. To populate the KTK system we prepared a library of shareable Entry-Level Parts as well as 171 useful Destination-Level plasmids (Table S1). This library includes a panel of promoters, RBS 172 sequences, terminators, and basic CDS parts (selection cassettes, fluorescent proteins) as well as 173 spacer sequences that enable more complex assembly steps. Our basic library will be made available 174 to researchers and is intended to become a resource for the wider community.

7
As tuning gene expression is a key goal in engineering cells, we characterised a panel of 176 promoters including heterologous promoters and promoters native to K. rhaeticus, with the latter 177 identified from genes and operons known to express from the strain genome. KTK enabled 178 straightforward and quick assembly of modular constructs expressing sfGFP in basic TUs, each with 179 a different promoter. These promoters were then characterised for expression strength in liquid 180 growth phase using flow cytometry quantification of green fluorescence. J23104 and pTac 181 promoters were the strongest heterologous promoters (Fig 3a), while the native promoters pTtcA 182 and pHxu were the strongest of the ones taken from the genome sequence (Fig 3b).   We further sought to add reporter proteins visible to the naked eye that can be expressed 196 within cellulose materials. To this end, we cloned 3 chromoproteins previously characterised in E. 197 coli (Liljeruhm et al., 2018) that are visibly blue (cjBlue), red (eforRed) and purple (gfasPurple). When 198 expressed in K. rhaeticus as a level 1 construct with J23104 promoter, each chromoprotein was able 199 to change the colour of the bacterial cell pellet (Fig 3b), although cjBlue gave a more greenish hue 200 than expected. The resulting bacterial cellulose pellicles grown from these strains showed visible 201 colouration of the material, with the red and purple colours being particularly striking. However, on 202 dehydration of the cellulose, pigmentation was lost, although the final material was still notably a 203 different shade to cellulose made from unmodified cells (Fig 3b, Fig S3). 204 The basic KTK parts library contains several reporter proteins, terminators, RBS sequences and 205 constitutive promoters, as well as the AHL-inducible promoter (Fig 2). We aim to expand this soon 206 to include more inducible expression systems and constructs that allow for CRISPR-based gene 207 regulation, and share the most relevant plasmids as a distributable collection. To port the E. coli curli system into BC-producing bacteria, we chose to maintain the linear 225 operon format successfully employed in past ELMs work. The region encoding the 6 genes from the 226 synthetic E. coli operon was therefore cloned into a KTK Entry-level vector as if it was a single CDS 227 9 (E1.3) part (Fig 4B). This was then assembled into a Level 1 curli-containing plasmid (D1.2_Curli), 228 with the AHL-inducible pLux promoter (E1.1), a well-characterised terminator (E1.4I) and an RBS 229 (E1.2) part, selected by the RBS Calculator (Salis et al., 2009) to give a similar predicted strength for 230 CsgB translation as that seen in the native E. coli system. D1.2_Curli was then assembled with a 231 Level 1 LuxR-expressing plasmid (D1.1_LuxR) to create the multigene Level 2 plasmid construct 232 giving AHL-inducible Curli expression (D2.2_Curli). In this design LuxR is expressed by a medium 233 strength promoter (pLac) to reduce any potential burden. Transmission electron microscopy (TEM) was next performed to visualise curli fibre 244 production from transformed K. rhaeticus. As control, TEM analysis of AHL-induced E. coli showed 245 clear filamentous Curli structures (Fig S1a) confirming the plasmid function in this bacterium. Direct 246 visualisation was more challenging with K. rhaeticus due to cellulose fibres being present in large 247 amounts, even after 20% (w/v) cellulase treatment (Fig S1b). Therefore, to distinguished curli from 248 cellulose, TEM samples were treated with CsgA-polyclonal antibody and immunogold-labelled. Gold 249 nanoparticle staining was evident in a tangle of fibres outside the K. rhaeticus cell, demonstrating 250 that the plasmid is functional in the cellulose-producing bacterium (Fig 4D and Fig S1b). The effect 251 of induced co-production of curli on the material properties of grown bacterial cellulose was then 252 investigated to see if this cellulose-amyloid composite had greater strength than cellulose alone. 253 Strength tests were performed on pellicles grown from AHL-induced K. rhaeticus with the D2.2_Curli 254 construct (Fig S1c). Pellicles grown from cells containing the D2.2_GFP construct were used as a 255

Programming protein secretion via the Type VIII secretion system 274
For an example of a Level 3 KTK multigene assembly, we next constructed a plasmid that exploits 275 the curli-specific T8SS system for heterologous protein secretion (Fig 5a) part and the N22 signal used as a tag to be fused to proteins for secretion (Fig 5a). 280 11 A panel of three T8SS modules were cloned into KTK as CDS parts; the minimal T8SS (just the 281 secretion pore CsgG), CsgG plus the CsgE 'adapter protein' (CsgGE), and both of these proteins plus 282 secretion chaperon CsgF (CsgEFG) (Costa et al., 2015;Hospenthal et al., 2017). These were 283 assembled to Level 2 with a LuxR TU, so that expression of these T8SS modules was AHL-dependent 284 (Fig 5b). In parallel, a small panel of heterologous cargo targets were prepared using the 2-part CDS 285 fusion approach of KTK to bring together the N-terminal N22 region of CsgA (E1.3a part) with C-286 terminal CDS parts (E1.3b) encoding three targets: a 98 amino acid elastin-like protein (ELP), the 287 enzyme Beta-lactamase (Bla) and the Gram-positive amyloid protein TasA (Fig 5a-b). ELPs are small 288 unstructured versatile proteins that are highly hydrophobic and an attractive ELM component 289 (Roberts et al., 2015), Beta-lactamase is a classic enzyme easily detected by the colorimetic 290 nitrocefin assay (Gilbert et al., 2021), and TasA is an amyloid protein from B. subtilis that could be 291 an alternative to curli (Erskine et al., 2018). 292 The 3 heterologous cargo targets were assembled into a Level 1 TU construct with the pLux 293 promoter, before then being each combined into a Level 2 construct with a short spacer part 294 (D1.2_Spacer, 23 bp) designed to assist cloning of an odd number of TUs in a multigene assembly 295 (Fig 5b). With all modules now in place in Level 2 plasmids, a combinatorial set of nine possible Level 296 3 constructs could be assembled with the 3 different T8SS versions and 3 cargo proteins (Fig 5a). 297 To test the function of these constructs, we again performed experiments in parallel in E. 298 coli and K. rhaeticus transformed with constructs, culturing with AHL induction and then harvesting 299 in mid-log growth phase and separating supernatant and cell fractions. We first assessed ELP 300 expression and secretion, using SDS-PAGE of both K. rhaeticus (Fig 5c) and E. coli (Fig S2a) samples 301 to confirm the presence of processed and secreted ELP when any one of three T8SS versions were 302 present. As ELP proteins are difficult to transfer by Western Blot due to their hydrophobic nature, 303 the constructs were designed to include a C-terminal his-tag. This enabled us to identify that the 10 304 kDa protein observed in the secreted fraction was indeed ELP-his when either K. rhaeticus (Fig 5d) 305 or E. coli (Fig S2b) were expressing the CsgEFG T8SS constructs. Similar experiments with the TasA 306 cargo were not as successful, only showing a small amount of TasA in cell fractions of E. coli (Fig  307   S2c), suggesting that TasA cannot use the curli T8SS for export. 308 Finally, we assessed enzyme secretion from the beta-lactamase (Bla) encoding Level 3 309 constructs, as before testing in both E. coli and K. rhaeticus. Beta-lactamase converts nitrocefin from 310 a yellow to red colour, which is quantifiable by spectroscopy at 450nm and initial E. coli data showed 311 that in the presence of either T8SS version, Bla enzymatic activity in the supernatant was higher 312 than when no T8SS was present (Fig. S2d). In this case the minimal T8SS (CsgG) gave the highest 313 values, so the Level 3 construct expressing this with N22-Blac was then assessed in K. rhaeticus. 314 Supernatant from an AHL-induced culture of these bacteria showed beta-lactamase activity nearly 315 2-fold higher than when no T8SS is expressed in the cells (Fig 5d). Altogether the results with ELP 316 and Bla cargos show that the modular T8SS system generated here for the KTK system can be used 317 to program protein secretion from K. rhaeticus. Further work is needed to understand the optimal 318 To demonstrate how multigene assembly with the KTK system can advance engineered living 353 materials, we imported the well-studied E. coli curli fibre production system into KTK and used it to 354 engineer a construct that instructs K. rhaeticus to co-secrete curli fibres alongside cellulose as 355 material is produced. Although KTK-based cloning was successful in enabling the production of 356 amyloid curli outside the cell, this did not result in any marked change in material properties for the 357 cellulose-curli composite pellicle. This is likely due to relatively very low production of protein 358 compared to cellulose from these bacetria, or could be related to curli fibre formation being 359 impaired chemically by the low pH of K. rhaeticus cultures, or physically by the large amounts of 360 cellulose being extruded from the cell surface. In future work, curli production could be improved 361 by increasing gene expression with different promoter/RBS parts and by using alternative gene 362 arrangements, e.g. by using strong promoters and a two-module design (Bongers et al., 2005;363 Mierau et al., 2005;Tabor, 1990). 364 The potential of the curli system was further exploited here by taking advantage of its Type 365 VII secretion system (T8SS) to export heterologous cargo proteins. An N-terminal fused N22 region 366 from CsgA enabled secretion of a small unstructured ELP proteins when expressed in cells co-367 expressing the CsgEFG T8SS. The tag also gave promising results with Bla for the secretion of active 368 globular enzymes from K. rhaeticus strains expressing CsgG. This offers the first described route to 369 getting enzymes expressed in BC-producing bacteria to be secreted extracellularly, opening up the 370 14 possibility of having cellulose-binding or cellulose-modifying enzymes co-produced as the cellulose 371 material is grown. However, the relatively low levels of protein and enzyme activity observed, and 372 the failure to secrete TasA here suggests that achieving high-level programmed secretion of a 373 desired target protein will more often than not be a challenge and will likely to require significant 374 optimisation. We hope that many people will use and contribute to the KTK system in the future, to 375 add more genetic parts that enable rapid optimisation of constructs in BC-producing bacteria. The 376 parts and toolkit described here and in the supplementary materials currently remain untested in 377 other BC-producing strains but we are optimistic that they will work well in all Komagataeibacter 378 and other acetobacter. Indeed modular DNA parts developed previously by us and others for K. The E. coli Turbo (NEB) was used throughout this study. Cultures were grown at 37 °C in shaking 390 liquid Lysogeny Broth (LB) (10 g/l Tryptone, 5 g/l Yeast Extract, 5 g/l NaCl) or on LB agar (1% agar), 391 and when appropriate supplemented with ampicillin (100 μg/ml), chloramphenicol (34 μg/ml) or 392 spectinomycin (100 μg/ml). Transformation was done using chemically competent cells. 393 K. rhaeticus iGEM cultures were grown at 30°C in liquid Hestrin-Schramm media (HS) (2% 394 glucose, 10 g/l yeast extract, 10 g/l peptone, 2.7 g/l Na2HPO4 and 1.3 g/l citric acid, pH 5.6-5.8) or 395 on HS agar plates (1.5% agar). When growing shaking cultures the media was supplemented with 396 2% cellulase (Sigma Alrich, C2730) and, when appropriate, supplemented with chloramphenicol (34 397 μg/ml) or spectinomycin (100 μg/ml). Electroporation of K. rhaeticus strains was performed as 398 described previously (Florea et al., 2016), and transformants screened on 10x Chloramphenicol (340 399 μg/ml) or 5x spectinomycin HS plates (500 μg/ml). When preparing pellicles for strength tests, K. 400 rhaeticus was strains were firstly grown to high density in liquid in the presence of 1x antibiotic. This 401 was then washed twice with HS media, resuspended in fresh cellulose-free HS media to a density of 402 0.5 OD600. This density was also used as standard for pellicle inoculums. Pellicles were grown 403 stationary at 30°C in fresh cellulase free HS media (75-100 ml), supplemented with antibiotic, and 404 when appropriate 50 mM AHL. Thick pellicles were observed and harvested after 2-4 weeks. 405 406

Molecular techniques, primers and plasmids 407
Modular DNA parts and plasmids used and constructed in this study are listed in Supplementary 408 Table S1 and S2. Standard PCR or primer joining was used to generate the DNA fragments for the 409 entry-level parts. PCR was performed by standard protocol using a high-fidelity polymerase as per 410 manufacturer's instructions. Primers are designed to include BsmBI and BsaI required for cloning as 411 detailed in sequences shown in Table G1 and Figure G1. If Entry-level parts were small (<60 bp), 412 instead of a PCR reaction, a primer joining protocol was used. Each primer would cover the target 413 sequence and including the overhang and RE sites detailed in Table G1 and Figure G1. In order to 414 allow primers to anneal together a sequence of overlap of at least 15 bp is required. Oligos (100 μM) 415 are separately phosphorylated with T4 PNK (NEB) before combined and heated to 96˚C for 6 min. 416 Samples are annealed by ramping down to 0.1˚C per second till 23˚C. 10 μl of the mixture was used 417 to clone into the Entry-Level Backbone vectors. 418 Golden Gate (GG) assembly followed standard GG protocols published (Lee et al., 2015). 419 Briefly T4 DNA Ligase (Promega, C1263), T4 ligase (0.5 μl; NEB, M0202), Type IIs RE (0.5 μl, NEB), 420