Dynamic decoupling of biomass and lipid biosynthesis by autonomously regulated switch

For improving the microbial production of fuels and chemicals, gene knock-outs and overexpression are routinely applied to intensify the carbon flow from substrate to product. However, their possibilities in dynamic control of the flux between the biomass and product synthesis are limited, whereas dynamic metabolic switches can be used for optimizing the distribution of carbon and resources. The production of single cell oils is especially challenging, as the synthesis is strongly regulated, competes directly with biomass, and requires defined conditions, such as nitrogen limitation. Here, we engineered a metabolic switch for redirecting carbon flow from biomass to wax ester production in Acinetobacter baylyi ADP1 using acetate as a carbon source. Isocitrate lyase, an essential enzyme for growth on acetate, was expressed under an arabinose inducible promoter. The autonomous downregulation of the expression is based on the gradual oxidation of the arabinose inducer by a glucose dehydrogenase gcd. The depletion of the inducer, occurring simultaneously to acetate consumption, switches the cells from a biomass mode to a lipid synthesis mode, enabling the efficient channelling of carbon to wax esters in a simple batch culture. In the engineered strain, the yield and titer of wax esters were improved by 3.8 and 3.1 folds, respectively, over the control strain. In addition, the engineered strain accumulated wax esters 19% of cell dry weight, being the highest reported among microbes. The study provides important insights into the dynamic engineering of the biomass-dependent synthesis pathways for the improved production of biocompounds from low-cost, sustainable substrates. Significance statement In the biological production, one of the greatest challenges is to find ways for optimal distribution of resources between cell growth, maintenance, and product synthesis. Robust and reliable circuits are required to allow autonomous switching of cells from biomass mode to lipid synthesis mode. Dynamic production of single cell oils such as triacylglycerols and wax esters is especially challenging due to the strong regulation. We present a dynamic genetic circuit based on conditional knockdown of a glyoxylate shunt enzyme, which is essential for cell growth. By gradual repression of the gene, the cells autonomously switch from biomass mode to product synthesis mode. We demonstrate the functionality of the circuit by using bacterium Acinetobacter baylyi ADP1 for the production of long chain alkyl esters, namely wax esters, with titer and yield improved by over 3-fold using acetate as the carbon source.

different time-points to quantitatively determine WE (by NMR) and biomass 175 production (as cell dry weight). 176 Lipid and end-metabolite analyses 177 The amount of total lipids and WEs were estimated by TLC or quantified by NMR. For 178 were extracted using 'miniscale' chloroform-methanol extraction as described 180 previously (17). Thirty µl of the chloroform phase was applied on 20 × 10 cm Silica Gel 181 60 F254 HPTLC glass plates with 2.5 × 10 cm concentrating zone (Merck, USA). Mobile 182 phase used was n-hexane: diethyl ether: acetic acid 90: 15: 1 and iodine was used for 183 visualization. Jojoba oil was used as the standard for WEs. For comparative evaluation 184 of the intensities of the WE bands on TLC, the Gel analysis method of ImageJ software 185 (rsb.info.nih.gov/ij/index.html) was applied as described in the ImageJ 186 documentation. 187 188 For NMR analyses, the 40-ml biomass samples were freeze-dried and the cell dry 189 weight (CDW) was determined gravimetrically. The lipid extraction and the 190 quantitative 1 H NMR analysis of WEs was carried out as described earlier (26). The 191 amount of total lipids was determined gravimetrically. The areas of the peaks in the 192 NMR spectrum are directly proportional to the molar concentration of each functional 193 group, yielding specific concentration for WEs in total biomass. The concentration of 194 WEs was calculated from the integrated signal at 4.05 ppm which is characteristic for 195 protons of α-alkoxy-methylene group of esters (-CH2-COO-CH2-). For the calculation 196 of the WE titer in grams per liter, an average molar mass of 506 g/mol was used (17).

204
In order to investigate the effect of the knock-out of isocitrate lyase AceA on the 205 growth and WE production in A. baylyi ADP1, we employed a knock-out mutant strain 206 A. baylyi ADP1 ΔaceA::Tdk/kanr (25) for preliminary test cultivations. We observed 207 that when grown on glucose, the cells grow more slowly, but produce WE titers 208 comparable to those of the wild type (wt) strain; after 48 hours of cultivation, the wild 209 type had produced 470±150 mg/l WEs compared to 460±40 mg/l WEs produced by 210 the knock-out strain. In opposite to the wt strain, however, the mutant strain did not 211 exhibit growth on minimal medium supplied with acetate as the sole carbon source. 212 This is due to the lack of route for acetyl-CoA to be directed in biosynthetic pathways 213 via malate. Thus, as acetyl-CoA represents the key precursor in both the biomass 214 production through the glyoxylate shunt and the wax ester biosynthesis, we 215 hypothesized that by dynamically regulating the isocitrate lyase, the state of the cells 216 could be switched between biomass and lipid synthesis modes (Fig 1). In order to make 217 the shift dynamic, we introduced an approach for autonomous regulation of the 218 isocitrate lyase AceA; by expressing the enzyme under an arabinose-inducible 219 promoter AraC-pBAD, the induction is gradually repressed due to the depletion of 220 arabinose by the glucose dehydrogenase activity of ADP1. The arabinose inducible 221 promoter has been previously shown to function in A. baylyi ADP1 (18,27). In order 222 to establish a system with maximal linearity and controllability, we constructed a gene 223 cassette for genomic expression of aceA (Fig 1c, d). Exploiting the natural 224 transformation machinery of ADP1, the gene cassette was integrated in the genome 225 to replace a gene poxB (ACIAD3381), which has been previously shown to be a neutral 226 target site in terms of growth and wax ester production (26, 28). The resulting strain 227 A. baylyi ADP1ΔaceA::tdk/Kan r ΔpoxB::araC-pBAD-aceA-Cm r was designated as ADP1-228 ara-aceA. The strain A. baylyi ADP1ΔpoxB::Cm r was used as the reference strain, from 229 now on designated as the ADP1 wt. acetate. The ADP1 wt grew to slightly lower biomass (OD ~2.4) and consumed all the 240 acetate; arabinose supplementation had no effect on the ADP1 wt growth. After 68 241 hours of cultivation, approximately 85 % of the 1% arabinose had been oxidized. The 242 control knockout strains ADP1ΔaceA::tdk/Kan r and ADP1ΔaceA::tdk/Kan r ΔpoxB::Cm r 243 did not exhibit growth nor acetate consumption with or without the presence of 244 arabinose (OD 0 at 0-68 h). We also confirmed, that the AceA expression is repressed 245 due to arabinose oxidation, i.e. the conversion of arabinose to non-inducive form, 246 arabino-lactone and further to arabonate ( Figure S1). 247

248
In order to find the optimal arabinose concentration in terms of both biomass and wax 249 ester production, the strain ADP1-ara-aceA was cultivated in several different 250 arabinose concentrations in minimal salts medium supplemented with acetate ( Figure  251 3) for 21 hours. ADP1 wt was cultured as the reference strain. Casein amino acids 252 (0.1%) were added to the culture in order to promote the growth and to prevent 253 nitrogen limitation. As indicated by the previous growth experiment, we found that 254 1% arabinose was sufficient to allow the engineered strain to reach the same biomass 255 as ADP1 wt, albeit the cells grew slower. Within the concentration range 0 -0.2%, only 256 small differences in growth pattern or biomass production were observed. The slight 257 increase in OD of uninduced cells is due to the utilization of the casein amino acids 258 present in the growth medium; a same amount of biomass is achieved without acetate 259 supplementation with the wild type strain and the knock-out strain 260 ADP1ΔaceA::tdk/Kan r ΔpoxB::Cm r with both acetate and casein amino acid 261 supplementation (data not shown). For ADP1 wt, all the growth curves were similar 262 regardless of the arabinose concentration used (data not shown). 263 264 Next, we determined which initial arabinose concentration most optimally distributes 265 the carbon between the biomass and the WE production. The strain ADP1-ara-aceA 7, 9 and 10 h time-points) were determined ( Figure 5). ADP1 wt consumed all the 290 acetate in approximately 9-10 hours; the amount of biomass increased until the 10-291 hour time-point, the CDW being 1.3 g/l. The highest amount of WEs was measured at 292 the 9-hour time-point, being 47 mg/g CDW and 60 mg/l. In ADP1 wt, the WEs 293 accounted for 44 % of total lipids. The WE yield was found to be 0.02 g WE/g consumed 294 acetate. The strain ADP1-ara-aceA utilized acetate more steadily and produced less 295 biomass compared to ADP1 wt; the growth ceased after 16 hours along with the 296 arabinose depletion: the biomass remained at 0.6-0.7 g/l CDW between the 16-24 297 hours of cultivation. Thereafter, the WE content of the cells strongly increased, being 298 highest at 38 h time-point, which also increased the amount of total biomass to 1.0 299 g/l. The WE titer was found to be 184 mg/l representing 19 % of CDW, which was 3.8-300 fold higher compared to the ADP1 wt. In addition, the WEs accounted for 80 % of all 301 cellular lipids in ADP1-ara-aceA. The WE yield was 0.08 g WE/g consumed acetate, also 302 being 4-fold higher over the ADP1 wt. 303

304
Sugars, mainly glucose, have been the major carbon source for the heterotrophic 305 microbial production of fatty acid derived compounds, such as TAGs and WEs, which 306 can be used for the production of biofuels, biochemicals, and other biocommodities 307 (29, 30). However, in order to increase the feasibility and sustainability of the 308 processes, the possibility to utilize alternative carbon sources is of high interest. In this study, we developed an autonomously regulated circuit for programmable 335 synthesis of WEs in a native production host, A. baylyi ADP1. The circuit allows the 336 cells to shift from the biomass mode to the WE synthesis mode independent from the 337 carbon/nitrogen ratio or the growth phase of the culture. In practice, we replaced the 338 native isocitrate lyase aceA with an arabinose inducible system, which allows a 339 conditional and timed knockdown of the expression of aceA. This enzyme is essential 340 for the biomass production when the cells grow on acetate. The timed repression of 341 aceA expression is achieved by gradually eliminating the inducer, namely arabinose; 342 the native enzyme activity of glucose dehydrogenase Gcd of A. baylyi oxidizes 343 arabinose to arabino-lactone and further to arabonate, which in turn cannot serve as 344 inducers. Importantly, and in contrast to other auto-induction-based systems, 345 arabinose oxidation does not interfere with the utilization of the carbon source, here 346 acetate, and can be thus considered as an orthogonal system. By adjusting the 347 arabinose concentration, a predefined and optimal amount of biomass can be 348 produced. When the inducer concentration is oxidized below the 'threshold' 349 concentration, the cells shift from the biomass producing mode to the synthesis mode, 350 efficiently directing carbon to product synthesis. 351 352 First, we confirmed that the engineered strain ADP1-ara-aceA with complemented 353 isocitrate lyase was able to grow on acetate as the sole carbon source. We observed 354 that without the presence of arabinose, the cells did not exhibit growth, showing 355 phenotype similar to the knockout strain A. baylyi ADP1ΔaceA::tdk/Kan r ΔpoxB::Cm r .
We also observed that arabinose concentration 1% is sufficient to allow the growth of cultures, whereas 0.2% culture had clearly lower volumetric WE production due to low 380 biomass production. Thus, we considered 0.5% arabinose as the most effective 381 inducer concentration in terms of optimal distribution of carbon between biomass and 382 products. 383 strengthens the status of A. baylyi ADP1 as a convenient host for metabolic 469 engineering and high-value lipid production from sustainable substrates.