Improving the Production of 9,21-di hydroxy- 20-methyl-pregna- 4- en- 3- one from Phytosterols in Mycobacterium Neoaurum by Modifying Multiple Genes and Improving the Intracellular Environment

Steroid drugs are particularly important for disease prevention and clinical treatment. However, traditional chemical methods are rarely implemented during the whole synthetic process to generate steroid intermediates due to the intricate steroid structure. Novel steroid drug precursors and their ideal bacterial strains for industrial production have yet to be developed. Among these, 9-OH-4-HP is a potential steroid drug precursor for the synthesis of corticosteroids. In this study, a combined strategy of blocking Δ 1 dehydrogenation and the C19 pathway as well as improving the intracellular environment was investigated to construct an effective 9-OH-4-HP-producing strain.

In industrial manufacturing, two major valuable intermediates of sterols, C19 steroids and C22 steroids, can be used to synthesize sex and adrenocortical hormones. However, traditional chemical methods are rarely implemented in the whole synthetic processes of modifying steroid intermediates due to the intricate steroid structure. Thus, the pursuit of novel steroid drug precursors has intrigued researchers.
9α-Hydroxy derivatives, such as 9-hydroxy-androst-4-ene-3, 17-dione (9-OH-AD), are important precursors in the manufacture of several modern glucocorticoid drugs bearing a halogen at the 9α position [11]. An engineered strain of Mycobacterium. neoaurum ATCC 25795 in which kstDs are knocked out can accumulate 6.02 g L − 1 9-OH-AD as the main product from 15 g L − 1 phytosterols [12]. kstd1 and kstd2 have also been reported to be effective for 9-OH-AD degradation in Rhodococcus rhodochrous DSM43269 [13]. By overexpression of certain related genes and knockout of kstDs, the production of 9-OH-AD increased by 45% in M. neoaurum MS136 [14]. A few wild-type strains of Mycobacterium have been reported to be able to produce 9-OH-AD from plant or animal sterols via a single-step microbial conversion [4]. Normally, 9α-hydroxy derivatives can be obtained from industrial strains in which kstDs are deleted.
In contrast, side chain degradation is rather complicated. Some key enzymes involved in the side chain metabolic pathway remain uncertain, limiting the comprehensive understanding of the process. Dual competing pathways, the overwhelming C19 steroid pathway and the C22 steroid pathway, are involved in phytosterol side chain degradation. Recently, the 17-hydroxysteroid/22-OH-BNC-CoA dehydrogenase Hsd4A was found to be relevant during C22 steroid formation [3]. Inactivation of Hsd4A enabled the production of C22 steroids from sterols. For example, M. neoaurum NwIB-XII accumulates androst-4-ene-3,17-dione (AD) and androst-1, 4-diene-3, 17-dione (ADD), which are both C19 steroids, as the main products after culture with cholesterol. The hsd4A knockout strain accumulates 4-HP and 1,4-HP as the main products. Nevertheless, C19 steroids still accumulated in the hsd4A knockout strain from phytosterols, indicating incomplete blockage of the C19 steroid pathway. An Hsd4A-and KstDs-de cient strain of M. neoaurum ATCC 25795 gives a 32% molar yield of 9-OH-4-HP and a 15% molar yield of 9-OH-AD from 40 g L − 1 phytosterols [3]. FadA5, a thiolase, was reported to be essential for the production of AD/ADD from cholesterol by M. tuberculosis H37Rv [15]. A FadA5-de cient strain of M. neoaurum NwIB-XII also accumulates 4-HP and 1,4-HP as the main products. Thus, the deletion of fadA5 may contribute to further blockage of the C19 pathway. Therefore, hsd4A and fadA5 are important targets for modi cation by genetic engineering to develop microorganisms that can transform sterols into the valuable steroidal intermediate 9-OH-4-HP.
On the other hand, phytosterols and their metabolites are toxic to cells, as they could inhibit cell growth and biocatalytic activity [16]. A steady-state intracellular environment could be bene cial for phytosterol degradation by Mycobacterium. Toxic steroid intermediates cause cells to produce reactive oxygen species (ROS), including hydrogen peroxide (H 2 O 2 ), during aerobic metabolism. A high level of H 2 O 2 might harm cell growth, hence slowing the rate of phytosterol degradation and decreasing the yield of metabolites [17], and vice versa. For example, elimination of H 2 O 2 from M. neoaurum JC-12 increased the 4-HP yield by 24%. In addition, during the phytosterol degradation process, intracellular nicotinamide adenine dinucleotides (NAD + and NADH) are consumed, which participate in multistep reactions during phytosterol degradation, such as when dehydrogenation occurs. NAD + /NADH regeneration and maintenance of the redox balance are considered the rate-limiting factors in the steroid degradation pathway [17,18]. Manipulation of NAD + /NADH contents could enhance the production of AD and ADD to various degrees [17][18][19]. Overexpression of NADH oxidase in M. neoaurum JC-12 increased ADD production by 43% [17], and an increase in the amounts of the ratio of NAD + /NADH in M. neoaurum TCCC 11978 enhanced the productivity of ADD by 93%. Thus, the elimination of H 2 O 2 and regeneration of NAD + could contribute to higher concentrations of phytosterol metabolites.
Herein, an engineered strain of M. neoaurum DSM 44074, which is a sterol consumer, was constructed for the bioconversion of phytosterols to 9-OH-4-HP. A kstDs knockout strain was constructed based on M. neoaurum DSM 44074, and the C19 steroid pathway was further blocked by knocking out both hsd4A and fadA5. By improving the intracellular environment, an e cient 9-OH-4-HP-producing strain was generated. This strain may contribute to the development of steroid drug precursors.

Accumulation of 9α-hydroxy derivatives
To eliminate Δ 1 -dehydrogenation and accumulate 9α-hydroxy derivatives from phytosterols, kstDs were identi ed and knocked out from the genome of the wild-type strain M. neoaurum DSM 44074, a steroiddegrading Mycobacterium that can completely degrade phytosterols to produce CO 2 and H 2 O [7]. The genome of M. neoaurum DSM 44074 was sequenced as described in the Methods section. Three putative kstD genes (gene 5102 for kstd1, gene 5236 for kstd2, and gene 5233 for kstd3) were identi ed in M. neoaurum DSM 44704. kstDs was successfully knocked out from the genome of M. neoaurum DSM 44704 as described in the Methods section, and the mutant strain ΔKstD was obtained. The cell growth of the ΔkstD strain showed no signi cant difference from that of the wild-type strain (Fig. S1). The wildtype M. neoaurum DSM 44074 strain and the genetically modi ed strain ΔkstD were incubated with phytosterols for 168 h. The resulting metabolites were extracted from the culture supernatants and analysed by HPLC. Compared with the wild-type strain M. neoaurum DSM 44074, which showed no detectable product by HPLC analysis (Fig. 2a), the ΔkstD strain produced 9-OH-AD as the main product with a retention time of 4.2 min (Fig. 2a, peak A), along with 9-OH-4-HP as a by-product with a retention time of 7.1 min (Fig. 2a, peak B). No ADD was detected during phytosterol bioconversion by ΔkstD, which indicated that the phytosterol degradation pathway was interrupted because of the elimination of Δ 1dehydrogenation by kstDs knockout. When MP01 medium plus 1 g L -1 phytosterols was used for incubation of the ΔkstD strain, 0.62 g L -1 9-OH-AD and 0.04 g L -1 9-OH-4-HP were produced within 60 h ( Fig. 3a and 3b). The molar yield of 9-OH-AD reached 84.9%. 9-OH-4-HP was the major by-product duringphytosterol bioconversion by ΔkstD, but the molar yield of 9-OH-4-HP was only 4.8%, and the selectivity of 9-OH-4-HP was 5.6% (Table 2). Thus, 9α -hydroxy derivatives successfully accumulated during phytosterol bioconversion by Mycobacterium, but the purity and yield of 9-OH-4-HP remained unsatisfactory.
24-CTOE-COA could subsequently be catalysed by the thiolase FadA5, leading the phytosterol degradation pathway to the C19 pathway.
Thus, to construct a 9-OH-4-HP-producing strain, hsd4A and fadA5 were identi ed in the genome of M. neoaurum DSM 44074 and separately knocked out in ΔkstD, resulting in the strains ΔkstDΔhsd4A and ΔkstDΔfadA5.
The cell growth of ΔkstDΔhsd4A and ΔkstDΔfadA5 showed no signi cant difference from that of the wild-type strain M. neoaurum DSM 44074 (Fig. S1). The strains ΔkstDΔhsd4A and ΔkstDΔfadA5 were cultured with phytosterols for 168 h, and the metabolites were analysed by HPLC (Fig. 2b). As shown in respectively. Nevertheless, both strains still showed only a small amount of 9-OH-AD accumulation (Fig.  3a). The selectivity of 9-OH-AD from strains ΔkstDΔhsd4A and ΔkstDΔfadA5 were 7.2% and 8.0%, respectively. After culture with 1 g L -1 phytosterols in MP01 medium, the strain ΔkstDΔhsd4A accumulated 0.59 g L -1 9-OH-4-HP and 0.13 g L -1 9-OH-AD, while 0.47 g L -1 9-OH-4-HP and 0.08 g L -1 9-OH-AD were obtained from strain ΔkstDΔfadA5. The molar yields of 9-OH-4-HP and 9-OH-AD from strain ΔkstDΔfadA5 were both lower than those from strain ΔkstDΔhsd4A. The molar yield of 9-OH-4-HP from strain ΔkstDΔfadA5 was 20.3% lower than that from strain ΔkstDΔhsd4A, and the molar yield of 9-OH-AD from strain ΔkstDΔfadA5 which was 38.5% lower than that from strain ΔkstDΔhsd4A.
Considering that 9-OH-AD still accumulated in both the ΔkstDΔhsd4A and ΔkstDΔfadA5 strains, the C19 steroid pathway of the phytosterol degradation pathway was not completely blocked in either strain.
Thus, to enhance the purity and productivity of 9-OH-4-HP and obstruct the yield of 9-OH-AD, hsd4A and fadA5 were both knocked out in strain ΔkstD, and strain ΔkstDΔhsd4AΔfadA5 was obtained. The cell growth of the ΔkstDΔhsd4AΔfadA5 strain showed a trend similar to that of the wild-type strain M. neoaurum DSM 44074 (Fig. S1). As shown in Fig. 2b, after culture with phytosterols and metabolites being analysed by HPLC, the strain ΔkstDΔhsd4AΔfadA5 accumulated 9-OH-4-HP as the main product, while the accumulation of 9-OH-AD was signi cantly decreased compared with the strains ΔkstDΔhsd4A and ΔkstDΔfadA5. The selectivity of 9-OH-4-HP and 9-OH-AD from strain ΔkstDΔhsd4AΔfadA5 were 94.9% and 2.0%, respectively. The purity of 9-OH-4-HP from strain ΔkstDΔhsd4AΔfadA5 was higher than those from strains ΔkstdΔhsd4A and ΔkstDΔfadA5. After culture with 1 g L -1 phytosterols, 0.66 g L -1 9-OH-4-HP was obtained from strain ΔkstDΔhsd4AΔfadA5, which is 11.9% more than that from strain ΔkstDΔhsd4A and 40.4% more than that from strain ΔkstDΔfadA5 (Fig. 3b). The selectivity and production of 9-OH-4-HP from strain ΔkstDΔhsd4AΔfadA5 were both higher than those from strains ΔkstDΔhsd4A and ΔkstDΔfadA5, indicating that double knockout of hsd4A and fadA5 could effectively block the accumulation of AD homologues.
Moreover, to verify the functions of hsd4A and fadA5 during phytosterol degradation, ΔkstDΔhsd4A-hsd4A, the hsd4A complementation strain, and ΔkstDΔfadA5-fadA5, the fadA5 complementation strain, were also constructed. As shown in Figure 2c, when the two complementation strains were cultured with phytosterols and metabolites being analysed by HPLC, the accumulation of 9-OH-AD was recovered. The purities of 9-OH-AD from ΔkstDΔhsd4A-hsd4A and ΔkstDΔfadA5-fadA5 were 90.0% and 88.5%, respectively, which is nearly consistent with those of strain ΔkstD. These results indicated that Hsd4A and FadA5 were key enzymes in the C19 steroid pathway during phytosterol degradation. Phylogenetic trees of Hsd4A and FadA5 were constructed to elucidate evolutionary relationship of the two enzymes (Fig. S2).

Evaluation of the 9-OH-4-HP producer
After culture with 1 g L -1 phytosterols, the molar yield of 9-OH-4-HP from ΔkstDΔhsd4AΔfadA5 was 78.9%. To evaluate the ability of ΔkstDΔhsd4AΔfadA5 to transform phytosterols into 9-OH-4-HP, different concentrations of phytosterols were incubated with ΔkstDΔhsd4AΔfadA5.
As shown in Fig. 4b, the yields of 9-OH-4-HP from the bioconversion of 2 g L -1 , 5 g L -1 , 8 g L -1 , and 10 g L -1 phytosterols by ΔkstDΔhsd4AΔfadA5 were 1.43 g L -1 , 2.78 g L -1 , 1.98 g L -1 , and 1.73 g L -1 , respectively. 9-OH-AD was also obtained during the incubation, showing yields of 0.06 g L -1 , 0.10 g L -1 , 0.03 g L -1 , and 0.04 g L -1 , respectively (Fig. 4a). The productivity of 9-OH-4-HP was enhanced as the concentration of phytosterols increased up to 5 g L -1 . However, at phytosterol concentrations of 8 g L -1 and 10 g L -1 , the productivity of 9-OH-4-HP decreased, showing results only slightly higher than that with 2 g L -1 phytosterols, and was obviously lower than that with 5 g L -1 phytosterols. The molar yields of 9-OH-4-HP from different concentrations of phytosterols are listed in Table 3. A downward trend in the molar yield of 9-OH-4-HP appeared as the concentration of phytosterols increased from 2 g L -1 to 10 g L -1 . However, the purity of 9-OH-4-HP remained stable. Previous research has reported that phytosterols and their metabolites could be noxious to cells during bioconversion [16, [20][21][22], which might account for the poor performance of ΔkstDΔhsd4AΔfadA5 during phytosterol bioconversion as the concentration of phytosterols increased. Intracellular environmental balance contributes to higher 9-OH-4-HP production The 9-OH-4-HP-producing strain ΔkstDΔhsd4AΔfadA5 did not perform well during the bioconversion of phytosterols when the concentration of phytosterols was higher than 2 g L -1 . This might be due to multiple factors that in uence the bioconversion of phytosterols.
A series of redox reactions occur during phytosterol degradation, which use oxygen as an electron acceptor, and cholesterol dehydrogenases/isomerases require intracellular nicotinamide adenine dinucleotides (NAD + and NADH) as cofactors [23,24]. NAD + and NADH play important roles during phytosterol transformation. They act in many oxidation-reduction reactions and regulate various enzymatic activities and genetic processes. The intracellular NAD + concentration decreased due to its consumption. Therefore, NAD + and NADH have critical effects on the maintenance of the intracellular redox balance. Regeneration of NAD + and enhancement of the NAD + /NADH ratio may be of great assistance during phytosterol transformation.
In addition, hydrogen peroxide (H2O2) is produced due to incomplete oxidation during aerobic metabolism and the regeneration of avin adenine dinucleotide (FAD) during the phytosterol transformation process [17]. A high level of H2O2 can damage proteins, DNA, and lipids in cells, resulting in inhibition of cell growth and metabolite yield [25].
To enhance the ability of strain ΔkstDΔhsd4AΔfadA5 to transform phytosterols into 9-OH-4-HP, the catalase katE from DSM 44074 and the NADH oxidase nox from Bacillus subtilis [17] were co-expressed in strain ΔkstDΔhsd4AΔfadA5 to construct strain ΔkstDΔhsd4AΔfadA5-NK.
The extracellular H 2 O 2 concentrations of the two strains ΔkstDΔhsd4AΔfadA5 and ΔkstDΔhsd4AΔfadA5-NK were measured when they were cultured with 5 g L -1 phytosterols for 168 h. As shown in Figure 5b, the extracellular H 2 O 2 concentration of strain ΔkstDΔhsd4AΔfadA5 showed an upward trend during the bioconversion process. The extracellular H 2 O 2 concentration increased from 0.59 μmol L -1 at the beginning to 1.05 μmol L -1 after 168 h and reached a peak of 1.10 μmol L -1 at 120 h. In contrast, the extracellular H 2 O 2 concentration of strain ΔkstDΔhsd4AΔfadA5-NK remained nearly stable during the bioconversion process at approximately 0.51 μmol L -1 . Therefore, the overexpression of katE eliminated excessive extracellular H 2 O 2 . Moreover, to verify the toxicity of H 2 O 2 , cell growth of the strains ΔkstDΔhsd4AΔfadA5 and ΔkstDΔhsd4AΔfadA5-NK was also measured. As shown in Figure  5a, the biomass of strain ΔkstDΔhsd4AΔfadA5-NK was higher than that of strain ΔkstDΔhsd4AΔfadA5, indicating that the elimination of extracellular H 2 O 2 could help with cell growth.
Likewise, the NAD + /NADH ratios of the strains ΔkstDΔhsd4AΔfadA and ΔkstDΔhsd4AΔfadA-NK were also measured after they were cultured with 5 g L -1 phytosterols for 168 h. As shown in Figure 5c, the NAD + /NADH ratio of strain ΔkstDΔhsd4AΔfadA-NK was consistently higher than that of strain ΔkstDΔhsd4AΔfadA by at least 10.9%. At 96 h, the NAD + /NADH ratio was enhanced by 25.4% in strain ΔkstDΔhsd4AΔfadA-NK compared with strain ΔkstDΔhsd4AΔfadA. The overexpression of nox could signi cantly in uence the NAD + /NADH ratio during phytosterol bioconversion. 9-OH-4-HP productivity was also measured to test whether overexpression of katE and nox could enhance the ability of ΔkstDΔhsd4AΔfadA to transform phytosterols into 9-OH-4-HP. The recombinant strain ΔkstDΔhsd4AΔfadA-NK was cultured with 1 g L -1 , 2 g L -1 , 5 g L -1 , 8 g L -1 , and 10 g L -1 phytosterols for 168 h, and the productivity of 9-OH-4-HP was measured every 24 h. As shown in Figure 5d, the nal productivities of 9-OH-4-HP from 1 g L -1 , 2 g L -1 , 5 g L -1 , 8 g L -1 , and 10 g L -1 phytosterols were 0.68 g L -1 , 1.53 g L -1 , 3.58 g L -1 , 2.51 g L -1 , and 2.73 g L -1 , respectively. Compared with ΔkstDΔhsd4AΔfadA cultured with the same concentrations of phytosterols, the productivities of 9-OH-4-HP were enhanced by 3.03%, 6.99%, 28.7%, 26.8%, and 57.8%. The highest yield of 9-OH-4-HP was obtained when strain ΔkstDΔhsd4AΔfadA-NK was cultured with 5 g L -1 phytosterols, with a molar yield that reached 85.5%, which was 28.8% higher than that of ΔkstDΔhsd4AΔfadA. Moreover, no signi cant difference in the productivity of 9-OH-AD was observed between the two strains ΔkstDΔhsd4AΔfadA5 and ΔkstDΔhsd4AΔfadA5-NK (Fig. 5e), indicating that the purity of 9-OH-4-HP was also enhanced during phytosterol bioconversion by the strain ΔkstDΔhsd4AΔfadA5-NK. All of the results above con rm that regulation of the intracellular NAD + /NADH ratio and H 2 O 2 levels could be an effective way to improve sterol transformation e ciency and the production of steroid intermediates.

Discussion
By genome sequencing, three kstDs were found in M. neoaurum DSM 44074. kstD2 and kstD3 in M. neoaurum DSM 44074 showed 100% similarity with those in M. neoaurum ATCC 25795, a strain that was deemed to be the same strain as M. neoaurum DSM 44074. However, kstD1 in M. neoaurum DSM 44074 showed 5 mismatches with that in M. neoaurum DSM 44074, causing 3 amino acid changes. When the kstds knockout strain of M. neoaurum DSM 44074 was cultured with phytosterols, AD and 4-HP were nearly undetectable in the nal products. The kstd knockout strain accumulated 9-OH-AD as the main product and 9-OH-4-HP as a by-product. Certain other strains have also been reported to accumulate 9-OH-AD from phytosterols. Mycobacterium sp. 2-4M [26] showed a 50% molar yield of 9-OH-AD, a 22% molar yield of AD and a 2% molar yield of 4-HP from 5 g L − 1 sitosterol [27]. In a kstds knockout strain of M. neoaurum ATCC 25795, a 55% molar yield of 9-OH-AD and a 15% molar yield of AD were obtained from 15 g L − 1 phytosterols. Compared with other 9-OH-AD-producing strains, the purity and molar yield of 9-OH-AD from ΔkstD after culture with phytosterols were notably higher. The accumulation of AD from kstDs knockout strains might be due to residual Δ 1 -dehydrogenation activity. The genome of R. ruber contains at least two other possible ORFs other than kstD1, kstD2, and kstD3 with certain identity to kstDs (approximately 38%) [28]. The existence of more than 3 KstDs has also been reported for other Rhodococcus species, such as R. jostii Rha1 [29]. Thus, inactivation of all KstD activities ought to be the fundamental premise to develop promising 9α-hydroxy derivatives.
Due to dual competing pathways, the dominant C19 steroid pathway and the C22 steroid pathway exist in the phytosterol degradation pathway. 9-OH-4-HP is usually produced as a by-product in 9-OH-ADproducing strains. M. V. Donova reported a wild-type strain Mycobacterium sp. 2-4M, which produces 9-OH-AD as the major product with a 1.5%-1.6% molar yield of 9-OH-4-HP [26]. Mycobacterium sp. VKM Ac-1815D, Mycobacterium sp. VKM Ac-1817D, and Mycobacterium fortuitum ATCC-6842 have also been reported to accumulate a small amount of 9-OH-4-HP during the 9-OH-AD production process [4][5][6]. The enzymes that catalyse 22-hydroxy-3,24-dioxo-4-ene-cholest-COA into 4-HP homologues remain unidenti ed. Thus, Hsd4A is normally chosen to manipulate the metabolic ux to generate AD homologues or 4-HP homologues. Xu reported the characterization of Hsd4A in vivo and in vitro, testifying that deletion of hsd4A resulted in blockage of the C19 steroid pathway and enhanced the accumulation of 4-HP homologues. During the Hsd4A investigation, Xu constructed a 9-OH-4-HPproducing strain by knocking out hsd4A in the kstD-de cient strain of M. neoaurum ATCC 25795. This mutant strain displayed 32% molar yield of 9-OH-4-HP and 15% molar yield of 9-OH-AD from 40 g L − 1 phytosterols [3]. Here, in this research, it was con rmed that double knockout of hsd4A and fadA5 could further block the C19 steroid pathway. The purity and molar yield of 9-OH-4-HP of strain ΔkstDsΔhsd4AΔfadA5 were notably higher than those of Xu's strain. Although strain ΔkstDsΔhsd4AΔfadA5 did not perform well when cultured with higher concentrations of phytosterols, the purity of 9-OH-4-HP was not in uenced, indicating its potency as a promising 9-OH-4-HP producer.
AD homologues accumulated in the 9-OH-4-HP producer strains ΔkstDsΔhsd4A, ΔkstDsΔfadA5, and ΔkstDsΔhsd4AΔfadA5, indicating incomplete blockage of the C19 steroid pathway. Similar results have been previously reported. Analysis of the M. neoaurum CCTCC AB2019054 genome revealed that there were 6 proteins with > 38% identity and 10 proteins with 31-38% identity to Hsd4A, which may compensate for its function [30]. The most identical gene, hsd4A2 (45% identity to hsd4A), was deleted, and fermentation analysis revealed that it can indeed produce 4-HBC at a signi cantly increased ratio. A similar result was found in M. neoaurum DSM 44074. Eight proteins showed certain identities to Hsd4A, indicating the presence of isoenzymes of Hsd4A in the genome of M. neoaurum DSM 44074. Some attempts have been made to enhance the ability of microorganisms to transform phytosterols. Considering the toxicity of phytosterols and their derivatives to cells, the balance of the intracellular environment could improve the ability of microorganisms to transform phytosterols. Intracellular factors such as NAD + and NADH have drawn increasing attention in recent years. NAD + and NADH participate in multiple steps during steroid bioconversion, and the intracellular NAD + concentration decreases as it is consumed. Regeneration of NAD + and enhancement of the NAD + /NADH ratio have been proven to be able to enhance the ability of microorganisms to transform phytosterols. Overexpression of NADH oxidase in M. neoaurum JC-12 increased ADD production by 43% [17]. Overexpression of avin oxidoreductase and NADH oxidase from Lactobacillus brevis in M. neoaurum TCCC 11978 increased the NAD + /NADH ratio by 113% and 192%, respectively, and signi cantly enhanced the conversion ratio of AD(D) [18]. The type II NADH dehydrogenases ndhN and ndhF were overexpressed in M. neoaurum MNR, resulting in an increase in the NAD + /NADH ratio from 3.93 to 5.91 and 10.96, respectively. The highest molar biotransformation rates of AD with 5 g L − 1 phytosterol feed were 5.32% and 12.38% higher than those of the original strain, respectively.
In addition, reactive oxygen species are generated during the conversion process of sterols, which impair cell viability and hinder the conversion of sterols to steroid synthons. Elimination of ROS and H 2 O 2 have been reported to be an effective method to improve cell growth under phytosterol feed and enhance phytosterol bioconversion. Combinatorial augmentation with catalase, mycothiol, and ergothioneine increased 4-HP productivity by 47.5% in M. neoaurum WIII-egt&msh&cat [31]. Elimination of H 2 O 2 in M. neoaurum JC-12 increased the 4-HP yield by 24%. In this study, when combining the abilities of NAD + regeneration and H 2 O 2 elimination, the performance of the new mutant strain ΔkstDsΔhsd4AΔfadA5-NK to transform phytosterols into 9-OH-4-HP improved. The extracellular H 2 O 2 concentration of ΔkstDsΔhsd4AΔfadA5-NK after culture with 5 g L − 1 phytosterols remained at a low level, which was only 49.4% that of the extracellular H 2 O 2 concentration of ΔkstDsΔhsd4AΔfadA5 culture under the same conditions. The NAD + /NADH ratio was also enhanced by 25.4% after 96 h. Although the molar yield of 9-OH-4-HP decreased as the phytosterol concentration increased, the molar yield of 9-OH-4-HP from strain ΔkstDsΔhsd4AΔfadA5-NK was signi cantly higher than that from strain ΔkstDsΔhsd4AΔfadA5 at the same phytosterol concentration. The highest yield of 9-OH-4-HP was 3.58 g L − 1 , which was achieved when ΔkstDsΔhsd4AΔfadA5-NK was fed 5 g L − 1 phytosterols. The molar yield of 9-OH-4-HP from strain ΔkstDsΔhsd4AΔfadA5-NK was 28.7% higher than that from strain ΔkstDsΔhsd4AΔfadA5 with 5 g L − 1 phytosterol feeding. All of these results proved that the elimination of H 2 O 2 and regeneration of NAD + could be an effective method to improve phytosterol bioconversion. When 10 g L − 1 phytosterols were cultured with ΔkstDsΔhsd4AΔfadA5-NK, the molar yield of 9-OH-4-HP was 32.6%, indicating further manipulation to enhance its molar yield. A number of different aqueousorganic two-phase systems have been studied, such as water-vegetable oils and water-ionic liquids [32]. The use of cyclodextrins could enhance the uptake of phytosterols by microorganisms [16,33]. Recently, some novel methods have been developed to improve phytosterol bioconversion. Increasing cell permeability could improve the production of phytosterol metabolites [34,35]. Deletion of the transmembrane transporter trehalose monomycolate mmpL3 in a 4-HP-producing strain derived from M. neoaurum ATCC 25795 increased 4-HP production by 24.7%. A "resting cell-cyclodextrin" system has also been widely used in industry and research to improve bioconversion ability.

Bioinformatic analysis
The genome of M. neoaurum DSM 44074 was sequenced by Shanghai Majorbio Co., Ltd. The DNA sample was extracted and sheared into 400-500 bp fragments using a Covaris M220 Focused Acoustic Shearer (Covaris, USA). Illumina sequencing libraries were prepared from the sheared fragments using a NEXT ex™ Rapid DNS-Seq Kit (Bioo Scienti c, USA). The sequencing data were assembled using SOAPdenovo2(GitHub -aquaskyline/SOAPdenovo2: Next generation sequencing reads de novo assembler.). Further prediction and annotation were produced by Glimmer (Glimmer (jhu.edu)) and BLAST (blast.ncbi.nlm.nih.gov). The putative genes for kstD, hsd4A, and fadA5 were identi ed by comparison with known gene sequences taken from the NCBI database. MEGA-X software (Home (megasoftware.net)) was used to construct a phylogenetic tree of hsd4A and fadA5 with the known amino acid sequences taken from the NCBI.

Mutant strain construction
A CRISPR-assisted nonhomologous end-joining strategy was used to delete the target gene in M. neoaurum DSM 44074 based on previous reports. The PSBY1 plasmid harbouring cpf1 was obtained from Jiang [36], and the PCR-Hyg plasmid harbouring sgRNA was obtained from Sun [37].

ClonExpress One
Step Cloning Kit mutated spacers were used to construct different plasmids harbouring target sgRNA. The plasmid harbouring target sgRNA was transfected into M. neoaurum, and the PSBY1 plasmid was transfected beforehand by electroporation. The recombinant clones were sequenced using speci c primers to determine the deletion.
The vector P38Mu (pMV306 with the Psmyc promoter) with kanamycin resistance was used to overexpress the target gene. The genes hsd4A, fadA5, katE from M.neoaurum DSM 44074, and nox from Bacillus subtilis were recombined on P38Mu. Speci c primers were used to amplify the corresponding gene, and the PCR product was inserted into the Nde site (and Hind site, if two genes were inserted) of P38Mu using the ClonExpress One Step Cloning Kit.

Bioconversion and analysis
The transformation capability of the mutant strains was identi ed in MP01 medium with an initial phytosterol concentration of 1 g L -1 . A concentration gradient was later tested to further determine the capability of phytosterol bioconversion. Phytosterols were prepared in (2-hydroxypropyl)-β-cyclodextrin (HP-β-CD) at a ratio of 1:1.5. The recombinant cells were inoculated into 30 mL of MYD medium in a 250 mL shaker ask and cultured at 30°C and 200 rpm. Three millilitres of seed medium was transferred to 30 mL of MP01 medium in a 250 mL shaker ask with a ba e when the optical density reached the mid-log exponential phase. The fermentation of M. neoaurum DSM 44074 and recombinant strains was sampled every 12 or 24 h, and three replicates were used to measure the steroids. The bioconversion mixture was extracted with 3 volumes of ethyl acetate, and the solvent was removed to give a residue that was redissolved in methanol. The resulting solution was used for HPLC analysis. HPLC was performed on a Shimadzu Separations module connected to a Shimadzu SPD-M20A detector equipped with a C18 column (250 mm × 4.6 mm, 5 µm) and detected at a wavelength of 254 nm. A mixture of methanol and water (80:20, v/v) was used as the mobile phase at a ow rate of 0.8 mL min -1 . The accumulation of 9-OH-4-HP was achieved by blocking the C19 steroid pathway and 3-ketosteroid-Δ 1dehydrogenation. Compared with single deletion of hsd4A and single deletion of fadA5, double deletion of hsd4A and fadA5 could further block the C19 steroid pathway. By eliminating H 2 O 2 and regenerating NAD + in the triple hsd4A, fadA5, and kstDs knockout strain, the highest 9-OH-4-HP productivity was 3.58 g L -1 with 5 g L -1 phytosterol feed.