A red-pigmented bacterium NS-17 was isolated from the soil of the Nanchang area with a nutrient agar plate. NS-17 was incubated in liquid medium containing 4 g of glycerol, 16 g of peptone, 4 g of NaCl and 7 g of KCl at 28˚C and shaken at 160 rpm for 24 h in a shaking incubator.

Morphological and biochemical tests were done using standard procedures [15] . Ultrastructural studies were performed by scanning electron microscopy.

The 16S rDNA of NS-17 was isolated and sequenced. The sequence was compared with available 16S sequences in GenBank at the National Center for Biotechnology Information (NCBI). A phylogenetic tree was constructed with the Mega program, and 1000 bootstrap resampling was applied to evaluate the robustness of the phylogenetic tree.

After fermentation, NS-17 cells were collected by centrifugation and washed twice with distilled water. Acidified ethanol (pH 3.0) was added to the cell mass to extract pigment, and the mixture was centrifuged. The pigment in ethanol was isolated using phase separation with water and chloroform. The chloroform phase containing red pigment was concentrated by rotary evaporation. The red pigment was separated by silica gel column chromatography and eluted with a mixture of hexane: ethyl acetate (2:1, v/v). The concentrated pigment was purified by development in a 1: 8: 1 (v/v) mixture of ligroin: chloroform: methanol using thin-layer chromatography (TLC) and then repurified with further TLC (chloroform: acetidin = 2: 1 [v/v]). The isolated pigment was then collected and dissolved in acidified ethanol. The pigments were dried using a vacuum freeze dryer.

The purified pigment was dissolved in ethanol, with pH = 5.0 or 9.0 (adjusted by HCl/NaOH) to analyze the absorption spectra under acidic and basic conditions, and the absorbance at wavelengths 300 - 700 nm of the pigment was scanned under those two pH values.

The molecular mass of the pigment was determined using an Agilent LC/MS (HPLC: 1200, eluted with a mixture of methanol: 0.1% formic acid [70:30, v/v]; TOFMS: G 6220A , ESI-MS). The ¹H- and ¹³C -NMR spectra were recorded on a Bruker Avance III 600 spectrometer (300 MHz), and CDCl₃ was used as the solvent. FT-IR spectra of the pigments were recorded with a Nicolet 5700 FT -IR spectrometer.

The purified prodigiosin was used to establish a formula between OD₅₃₅ and mass concentration.

The culture broth was mixed with acidic methanol (pH 3.0). The volume of methanol was adjusted according to the OD₅₃₅ of the mixture to ensure the optical density was in the range of 0. 2- 1.0.

Because NS-17 produced minimal pigment in LB broth (the most widely used medium for production of prodigiosin), we tested several media reported for fermentation of prodigiosin [1] [9] [16] [17] [18] and chose the initial medium as follows: glycerol 4 g/L, peptone 16 g/L, NaCl 3 g/L, and KCl 4 g/L; pH 7.0. The initial culture conditions were as follows: inoculum size: 1.0%, 100 mL of medium in flask (250 mL), culture temperature 28˚C, and rotation speed 160 rpm. The production of prodigiosin was measured after 24 h culture.

A 50 L bioreactor was used to perform the continuous fermentation under the optimized conditions of the flask culture with no additional defoamer, and the foam was removed as soon as it reached the top of the bioreactor. A medium with a nutrient concentration 5-fold higher than that of the fermentation medium (with half the inorganic salt concentration) was supplied as a feed medium according to the flow rate of efflux foam. The pH of the broth was automatically adjusted with HCl and NH₄・OH.

NS-17 was a gram-negative, subspheroidal or rod-shaped bacterium strain (Figure 1). The colonies of NS-17 on nutrient agar were moist, sticky and spreading, with a metallic luster and irregular margins. The organism could liquefy gelatin and use citrate or acetate as the only carbon source; the phenylalanine deaminase, L-arabinose fermentation, pectate decomposition and H₂S tests were negative, while the results were positive for DNAase, V. P, lipase, lysine decarboxylase, ornithine decarboxylase and aesculin hydrolysis tests. Moreover, NS-17 could use trehalose and D-sorbitol but not malonate or alginate. The physiological and biochemical identification tests suggested bacterium NS-17 to be S. marcescens.

The 16S rDNA showed the highest similarity to that of S. marcescens and 99% similarity to both S. marcescens L1 (EF 208031) and S. marcescens SA1 (HM 136580). A close evolutionary relationship between NS-17 and some other strains of S. marcescens was revealed (Figure 2).

All the results showed that NS-17 was a member of S. marcescens, and the strain was therefore named S. marcescens NS-17.

Purified Serratia marcescens strain NS-17 was fermented in a shaken flask to produce pigments. After separation with column chromatography and purification by TLC, the red pigment was obtained.

The UV-Vis spectrum of the pigment (Figure 3) was similar to the typical spectrum of prodigiosin [2] . The TOFMS analysis result is shown in Figure 4. The [M + H] of the pigment was 324.2072, corresponding to C₂₀H₂₅N₃O, which is the molecular formula of prodigiosin.

FT-IR absorption of the pigment in KBr was at ν_max 3406.91, 2928.02, 2852.17, 1624.86, 1539.83, 1454.07, 1240.98, 1117.75, 1071.29, 960.33 and 772.16. The absorption corresponded to the groups as follows: aromatic NH (3406.91), CH₂ (2928.02 and 2852.17), C = C (1624.86, 960.33 and 772.16), pyrrole skeleton (1539.83 and 1454.07), C-O-C (1240.98, 1117.75 and 1071.29).

The NMR chemical shifts of the pigment were as follows: ¹H-NMR (CDCl₃, 600 MHz, ppm) δ 0.905 (3H, m), 1.294 (2H, s), 1.311 (2H, m), 1.522 (2H, t), 2.395 (2H, m), 2.565 (3H, s), 3.996 (3H, s), 6.066 (1H, d), 6.355 (1H, brd), 6.688

(1H, brd), 6.921 (1H, s), 6.965 (1H, s), 7.091 (1H, d), 12.579 (1H, brs) and 12.731 (1H, brs); and ¹³C -NMR (CDCl₃, 151 MHz, ppm) δ 13.74, 16.96, 22.54, 25.48, 29.71, 31.52, 52.47, 85.86, 108.69, 109.75, 112.32, 120.26, 122.77, 123.62, 124.09, 130.82, 136.22, 138.01, 157.63 and 162.83.

UV-Vis, TOFMS, FT-IR and NMR analyses suggested that the red pigment isolated from NS-17 was PG. The pigment structure is shown in Figure 5.

The strain NS-17 produced minimal red pigment in LB broth, weakly responded to supplementation with Tween-80 and oil and showed a high production of prodigiosin under the temperature of 30˚C - 32˚C, different from most reported S. marcescens strains [2] [3] [19] . The results of single-factor experiments are shown in Table 1, with the maximum values presented in bold. Experiments were carried out according to the order in Table 1 with the factors optimized in the previous steps.

In the investigated factors and concentrations, the optimal conditions for NS-17 to produce prodigiosin were as follows: maltose 6 g /L; peptone 12 g /L; NaCl 1 g /L; KCl 2 g/L; Tween-80 1% (w/v); soybean oil 4% (w/v); pH of medium 8; inoculum size ≥ 1.5%; volume of medium in a 250 mL flask ≤ 50 mL; rotational rate ≥ 190 rpm; and cultivation temperature 32˚C. Figure 6 shows the yield of prodigiosin according to time for an inoculum size of 1.5%, volume of medium of 50 mL and rotational rate of 190 rpm. The concentration of prodigiosin reached a relatively high value (60.51 mg/L) within 32 h of incubation, and the pH of the culture broth was stable at approximately 8.0, which is supposed to be good for the production of prodigiosin [19] [20] [21] .

Oxygen supply significantly affected the concentration of prodigiosin in investigations of the volume of medium and rotational rate. Sufficient oxygen supplementation was the essential condition for prodigiosin fermentation.

With a 250 rpm rotor speed and 1:1 ventilation ratio in a bioreactor (GUJS-50L, EASTBIO Ltd., China), a control growth culture was repeated three times under the optimized conditions of the flask culture with 5 g/L of propoxylated glycerin as a defoamer. Group 1 was carried out three times under the same conditions as those of the control group but with no defoamer, and Group 2 was carried out

three times under the same conditions as those of the control group without a defoamer and with a feed medium that had a 5-fold higher nutrient concentration than that of the fermentation medium (with half the inorganic salt concentration) supplied at a flow rate of 8 mL/min according to the flow rate of the efflux foam. After 56 h of fermentation, the production of cells and PG was measured.

As the results show, the efflux foam contained a mass of cells and PG, and continuous fermentation could be successfully achieved with additional supplementation of medium. The efflux foam liquefied automatically in the collector, and the weight of the efflux broth is shown in Figure 6. The PG concentrations in the bioreactor and efflux broth are shown in Figure 7 and Figure 8.

Because the process of foam collection was subject to contamination, the continuous fermentation stage of S. marcescens was restricted to less than 72 h, and no contamination occurred in 56 h.