All strains and plasmids used or constructed in this study are listed in table 1. The bacterial strains used in this study were Escherichia coli BW-CARO 24 and E. coli DH5α. If not stated otherwise, strains of E. coli were grown at 37°C in LB medium. The minimal medium used had the following composition: KH₂PO₄ 3 g L^-1, K₂HPO₄ 12 g L^-1, (NH₄)₂SO₄ 5 g L^-1, MgSO₄ × 7H₂O 0.3 g L^-1, CaCl₂ × 2H₂O 0.015 g L^-1, NaCl 0.1 g L^-1, Glucose monohydrate 5 g L^-1, FeSO₄ × 7H₂O/sodium citrate 15 ml L^-1 (from the solution of 7.5 g L^-1 FeSO₄ × 7H₂O and 100 g L^-1 sodium citrate), thiamine, trace elements 25 33 ml L^-1. Antibiotics were used at the following final concentrations: ampicillin 100 μg ml^-1, chloramphenicol 50 μg ml^-1. Difco MacConkey agar base was purchased from Nordwald (Hamburg, Germany). Carotenoid standards were provided by the Institute of Technical Biochemistry, University of Stuttgart. All other chemicals and reagents are from Sigma-Aldrich or Roth (Germany) and were of the highest available purity.

The crtW148 reading frame was obtained from plasmid pPQE32-148 20 . After NdeI/HindIII digestion, the fragment was ligated into NdeI/HindIII digested pJF119ΔN 26 and transformed into E. coli DH5α. The resulting plasmid pJF119-crtW148 was further treated with HindIII and ligated with a HindIII digested FRT-cat-FRT fragment derived from plasmid pCAS30-FRT-cat-FRT 27 .

The reading frame of crtZ was isolated from plasmid pJF119-crtZ 24 after NdeI/BamHI treatment and cloned into pAW223 [NdeI/BamHI]. The resulting plasmid pAW-crtZ was further treated with HindIII and ligated with a HindIII digested FRT-cat-FRT fragment. The DNA sequence of idi was amplified by polymerase chain reaction (PCR) from chromosomal DNA of E. coli LJ110 using the following oligonucleotides: a) GTGAGAA

CATATG

GGATCC

The dxs promoter replacement was conducted according to the method used by Yuan et al 28 . Thus, for the construction of pQE31-FRT-cat-FRT, the DNA sequence of FRT-cat-FRT was amplified by PCR from plasmid pCAS30-FRT-cat-FRT by using the following oligonucleotides: a) CACAGACTGAGGATC

CTCGAG

CTCGAG

The individual expression cassettes of idi, crtW148, and crtZ from plasmid pJF-idi- FRT-cat-FRT, pJF-crtW-FRT-cat-FRT, and pAW-crtZ-FRT-cat-FRT, respectively, were amplified by PCR. The oligonucleotide primers for the P_tac-crtW expression cassette were: 1) CGGCGCATATTGCCCTGATGGACATTGACCCCACCCGCCTGGAAGAGTCGCATATTGTTCAAGGCGCACTCCCGTTCTGG and 2) AGCGCAACGATGGCTTTAAGTGTCAGATGGCTTCCTTCAGCAGACGGTTGATTGTCTGCAGGGTTATTGTCTCATGAGCG. Primers for the P_rha-crtZ expression cassette were: ACCGTTCTTAATTCTGATATTTCATCGGTGATCTCCCGTCTGGGACATACCGATATGCATGCATCGATCACCACAATT and ATTCACGCTAGCCCATACACCACGACTTCCTAAAGTAATCAGTACAGTACGGATACC CAGGGTTATTGTCTCATGAGCGGATAC. Primers for the P_tac-idi expression cassette were: TGCAGGCATGAAACCGCGTCTTTTTTCAGATAAAAAGCGCAATCAGTCGCTCAA GGCGCACTCCCGTTCTGG and TAACATTACTCAGCAATAAACTGATATTCCGTCAGGCTGGAATAAGGATGGCCT TCTGCTTAATTTGATGCCTG.

For the integration of the expression cassettes, the purified PCR products were each transformed into E. coli BW-CARO and its variant strain carrying plasmid pKD46. Expression of the λ-Red enzymes and the preparation of competent cells were carried out as described previously 29 . Competent cells were electroporated with 0.2-0.4 μg of PCR product. After electroporation, the cells were resuspended in 1 ml LB-medium and incubated at 30°C for 12 h with shaking. Subsequently, the cell suspension was spread onto MacConkey-agar plates containing chloramphenicol and 1% of the respective sugar (D-galactose, D-ribose, or melibiose), corresponding to the targeted genes responsible for sugar degradation. MacConkey agar plates were incubated over-night at 37°C. 5-10 pale colonies per transformation were picked 24 and checked regarding correct recombination by control PCR. The chloramphenicol resistance cassette was eliminated using the plasmid pCP20 as previously described 30 .

E. coli strains carrying carotenoid biosynthetic genes were cultivated in shake flasks in 50 ml LB medium or minimal medium. Cultivations were carried out at 30°C. At an optical density (OD₆₀₀) of 0.3 - 0.4 the cultures were induced, if required, by addition of IPTG (0.5 mM, final conc.) and/or L-rhamnose (0.2% w/v, final conc.). Samples (1 ml) were withdrawn from cultures after different time points within a period of 48 h. The cells were harvested by centrifugation, washed with cold water, and subsequently extracted by vigorous shaking with acetone (500 μl) for 15 min at 50°C. Insoluble components of the extract were removed by centrifugation (20,000 × g). HPLC analysis was performed on Dionex HPLC Instrument (Idstein, Germany), installed with a Chromeleon Software, Gina autosampler, P580 pumps, and a diode array detector. Products were analyzed by loading 50 μl of the supernatant onto a C30-reverse-phase HPLC column (250 mm × 4.6 mm, 5 μm, YMC-Europa GmbH, Dinslaken, Germany) attached with a guard column containing matrix of the same material as the column. A solvent flow rate of 1.0 ml/min was used. The solvents used were Solvent A, consisting of MTBE (methyl-tert-butylether)/methanol/water (19/80/1) and Solvent B, consisting of MTBE/methanol/water (90/9/1). Gradient conditions: equilibration conditions at 2% B; 0 to 22 min linear gradient from 2% B to 100% B. The spectra of the eluted carotenoids were recorded online with a Dionex UVD340 diode array detector (200 - 600 nm).

Carotenoid compounds were identified by co-chromatography using authentic standard compounds and by analysis of its UV-Vis spectra.

For the quantification of the carotenoid compounds the integrated peak areas were compared to HPLC standard curves of authentic standards.

Although isoprenoid products of CrtW148 and CrtZ were detected, neither enzyme activity tests nor SDS-PAGE were able to verify the presence of these proteins. Therefore, mRNA from crtW148 and crtZ was detected by means of absolute RT-qPCR using an internal standard.

Recently our group developed a method for the integration of heterologous expression cassettes into the chromosome of E. coli and the subsequent screening 24 . From this work resulted a plasmid-free strain (E. coli BW-CARO) carrying the four genes crtEBIY from P. ananatis in individual expression cassettes under the control of IPTG inducible tac-promoters. To enhance the expression level of the native 1-deoxyxylulose-5-phosphate synthase gene (dxs), a rate-limiting enzyme of the isoprenoid pathway 32 , the native dxs promoter was exchanged by the strong bacteriophage promoter T5 28 . First, a chloramphenicol resistance gene, flanked by FRT sites, was cloned upstream of the T5-promoter into pQE31. The resistance gene and promoter region of the plasmid gained (pQE31-FRT-cat-FRT) was PCR amplified and subsequently inserted upstream of the native dxs in E. coli BW-CARO using the λ-Red recombineering technique. After elimination of the resistance gene, the resulting strain E. coli BW-CARO-dxs was further modified by the insertion of an additional copy of the native isopentenyl diphosphate isomerase gene (idi), a further rate-limiting enzyme of the isoprenoid pathway 33 . The E. coli idi gene was cloned into the expression vector pJF119ΔN and to allow transient selection for chromosomal integration, the cloned gene was hooked up to a downstream chloramphenicol resistance cassette. The expression cassette of the resulting plasmid pJF119-FRT-cat-FRT, including resistance gene, was PCR amplified and transformed into E. coli BW-CARO-dxs harboring the λ-Red recombinase expression plasmid pKD46. The selected PCR primers led to the homologous recombination of the expression cassette into the galactose locus (galETKM). The engineered E. coli strains gained (E. coli BW-CARO, E. coli BW-CARO-dxs, and E. coli BW-CARO-dxs-idi) were cultivated in complex medium and the carotenoid content was analyzed. All strains produce β-carotene as the sole carotenoid (Figure 2A). As expected, the amount of β-carotene increased due to the enhanced expression of dxs and idi. The exchange of the native dxs promoter by the T5-promoter in E. coli BW-CARO increased the amount of β-carotene per cell by a factor of 2.6. After insertion of the additional idi copy into E. coli BW-CARO-dxs the amount of β-carotene per cell increased by the factor of 1.9 (β-carotene content in E. coli BW-CARO 1.24 ± 0.17 mg/g cellular dry weight (cdw); E. coli BW-CARO-dxs 3.254 ± 0.24 mg/g cdw; E. coli BW-CARO-dxs-idi 6.18 ± 0.78 mg/g cdw).

The reading frame of the β-carotene ketolase gene (crtW148) from N. punctiforme PCC73102 was amplified from plasmid pQE32-148 and cloned into the expression vector pJF119ΔN, resulting in plasmid pJF-crtW148. In transformants of β-carotene-producing E. coli BW-CARO-dxs-idi with pJF-crtW148 the di-keto carotenoid canthaxanthin represented 85% (0.896 ± 0.12 mg/g cdw) of the total carotenoid content with about 10% β-carotene and small amounts of echinenone (Figure 2C). The ratio of the three carotenoids did not vary during the 48 h cultivation. The addition of the inducer IPTG had neither influence on the carotenoid formation nor on growth (data not shown).

After cloning of a FRT-cat-FRT cassette into pJF-crtW148, chromosomal integration of the P_tac-crtW148 expression cassette into the melibiose locus (melAB) of E. coli BW-CARO-dxs-idi, and elimination of the cat resistance cassette, the engineered strain E. coli BW-CANT showed an about 20% higher formation of canthaxanthin (1.085 ± 0.15 mg/g cdw) but the percentage of canthaxanthin of all carotenoids was 85% as in the plasmid carrying strain E. coli BW-CARO-dxs-idi pJF-crtW148. The addition of IPTG up to 1 mM into LB-medium cultures of E. coli BW-CARO-dxs-idi pJF-crtW148 and E. coli BW-CANT, respectively, did not change the ratio of canthaxanthin, β-carotene, and echinenone as well as the total carotenoid yield, which we take as evidence of the leakiness of the tac-promoter in LB-medium.

In order to engineer an astaxanthin-producing strain, the β-carotene hydroxylase gene (crtZ) was inserted into the chromosome of E. coli BW-CANT. To allow a variable expression of crtZ compared to the tac-promoter controlled biosynthetic genes, crtZ was expressed under control of the rhamnose-promoter (P _rhaBAD ). For this purpose crtZ was cloned from plasmid pJF119-crtZ into pAW223. To verify the functional expression of this construct, pAW-crtZ was introduced into β-carotene-producing E. coli. Cultivation of E. coli BW-CARO-dxs-idi pAW-crtZ and subsequent carotenoid analysis showed that zeaxanthin was produced as the only carotenoid product (1.48 mg/g cdw).

Cloning of a FRT-cat-FRT cassette into pAW-crtZ and chromosomal integration of the P_rhaBAD-crtZ expression cassette into the ribose locus (rbsDABCK) of E. coli BW-CANT and elimination of the resistance cassette resulted in the strain E. coli BW-ASTA, containing all required biosynthetic genes for the formation of astaxanthin as single expression units, respectively, on the chromosome. Cultivation of E. coli BW-ASTA in LB-medium showed that astaxanthin is the predominant carotenoid product (71%). Besides astaxanthin, zeaxanthin (25%) was the only other detectable carotenoid (Figure 2B, Figure 3). The addition of the inducer L-rhamnose to E. coli BW-ASTA cultures led to an increase of zeaxanthin (69%) compared to astaxanthin (20%) (Figure 2D, Figure 3). The enhanced expression of crtZ by L-rhamnose induction therefore led to a significant increase of zeaxanthin, obviously due to a low activity of CrtW148 for the substrate zeaxanthin. The induction of E. coli BW-ASTA LB medium cultures with IPTG in the presence of 0.2% L-rhamnose or without had neither effect on the amount of carotenoid nor on the ratio of astaxanthin and zeaxanthin.

In order to avoid the formation of zeaxanthin by E. coli BW-ASTA, an increase of the ketolase activity or a decrease of the hydroxylase activity would be necessary. To better control the expression by the tac- and, in particular, by the rha-promoter that are both not tightly controlled in LB-medium (see Figure 4), minimal medium with glucose as C- and energy source was used. The cultivation of E. coli BW-ASTA in minimal medium without the addition of inducer molecules resulted in the formation of only small amounts (0.11 mg/g cdw) of astaxanthin, which we interpret as the result of the higher repression of the tac-promoter in minimal medium in contrast to LB-medium. On the other hand, the addition of IPTG to the cultures, which induces the P_tac controlled gene expression of dxs, idi, crtE, crtB, crtI, crtY, and crtW, resulted after 48 h in a 12-fold increase in the formation of carotenoids. After 24 h of incubation in minimal medium, E. coli BW-ASTA synthesized 0.96 ± 0.14 mg/g cdw astaxanthin and the by-products adonirubin (13%), canthaxanthin (12%), and β-carotene (1%) (Figure 2E). In contrast to the cultivation in LB medium no zeaxanthin formation was observed. After an incubation time of 48 h astaxanthin was produced as exclusive carotenoid (>95%) in a concentration of (2.07 ± 0.15 mg/l; 1.41 ± 0.11 mg/g cdw) (Figure 3). This result shows that during the cultivation of E. coli BW-ASTA in minimal medium with glucose and IPTG the β-carotene produced is predominantly converted by the ketolase (CrtW148) into cantaxanthin, which is subsequently slowly hydroxylated by CrtZ. The concurrent addition of both inducers, IPTG and L-rhamnose, also led after 24 h to the formation of astaxanthin (68%), adonirubin (14%), cantaxanthin (12%), and traces of β-carotene (1%). In the late stationary phase (48 h) of the culture astaxanthin was still the predominant carotenoid with >90%, however, the cells also contained about 5% of zeaxanthin as a by-product (Figure 2F).

In order to ascertain whether the assumed expression levels of CrtW148 and CrtZ based on carotenoid formation by E. coli BW-ASTA were in accordance with the mRNA levels of crtW148 and crtZ, we performed absolute RT-qPCR studies. A method previously described by Schuhmacher et al. 31 that uses an external standard for absolute transcript number quantification, was adapted to the Roche UPL system allowing for absolute transcript number determination per cell.

From the cultivation of E. coli BW-ASTA in LB medium and in minimal medium, with or without induction of the tac- and rha-promoter, respectively, cells were withdrawn in the late exponential phase and were analyzed for the crtZ and crtW148 transcript level by RT-qPCR. The number of transcripts of crtW148 and crtZ under the different conditions are shown in Figure 4. In LB medium, comparable transcript numbers per cell could be detected for crtW148 with or without IPTG induction (between 22 ± 4 and 25 ± 1 copies/cell). A 15% higher copy number was detected for crtZ (41 ± 3 copies/cell compared to 48 ± 5 to 53 ± 3 copies/cell) after induction with L-rhamnose. The transcript level of crtZ reflects the formation of zeaxanthin when cultivated in complex medium. An increased expression of crtZ correlated with an increased formation of zeaxanthin by E. coli BW-ASTA. Under all cultivation conditions in LB medium, zeaxanthin and astaxanthin had been detected. By using glucose-containing minimal medium the transcription of the heterologous genes was more tightly regulated than in LB medium. Transcript numbers per cell of crtW148 without the addition of IPTG were 12 ± 1. For crtZ in average, less than one copy number per cell was detected in minimal medium without induction. This reflects the tight regulation of the rhamnose-promoter under these conditions. These results are in good accordance with the product formation; in minimal medium without induction, only small amounts of astaxanthin as the sole carotenoid were detected (Figure 3). The induction by IPTG led to a 2.3 to 2.6 fold increase in mRNA level of crtW148, and resulted in a 20-fold increase of the carotenoid concentration. Here, astaxanthin was the only carotenoid that was accumulated by E. coli BW-ASTA after 48 h (Figure 2). This shows that the low mRNA level of crtZ yielded in a sufficient amount of hydroxylase activity converting all the produced precursors into astaxanthin. The addition of L-rhamnose led to a significant increase in transcript levels of crtZ (8.6 fold). In these cases, both astaxanthin (90%) and zeaxanthin (5%) had been detected as products after 48 h.