Identiﬁcation and Functional Analysis of Two Novel Genes—Geranylgeranyl Pyrophosphate Synthase Gene ( AlGGPPS ) and Isopentenyl Pyrophosphate Isomerase Gene ( AlIDI )—from Aurantiochytrium limacinum Signiﬁcantly Enhance De Novo β -Carotene Biosynthesis in Escherichia coli

: Precursor regulation has been an effective strategy to improve carotenoid production and the availability of novel precursor synthases facilitates engineering improvements. In this work, the putative geranylgeranyl pyrophosphate synthase encoding gene ( AlGGPPS ) and isopentenyl pyrophosphate isomerase encoding gene ( AlIDI ) from Aurantiochytrium limacinum MYA-1381 were isolated. We applied the excavated AlGGPPS and AlIDI to the de novo β -carotene biosynthetic pathway in Escherichia coli for functional identiﬁcation and engineering application. Results showed that the two novel genes both functioned in the synthesis of β -carotene. Furthermore, AlGGPPS and AlIDI performed better than the original or endogenous one, with 39.7% and 80.9% increases in β -carotene production, respectively. Due to the coordinated expression of the 2 functional genes, β -carotene content of the modiﬁed carotenoid-producing E. coli accumulated a 2.99-fold yield of the initial EBIY strain in 12 h, reaching 10.99 mg/L in ﬂask culture. This study helped to broaden current understanding of the carotenoid biosynthetic pathway in Aurantiochytrium and provided novel functional elements for carotenoid engineering improvements.


Introduction
Carotenoids, a kind of natural isoprenoid pigments, are distinguished for various biological functions and commercial applications as colorants, antioxidants, pharmaceuticals and nutraceutical agents [1][2][3][4]. However, the rising commercial demand for carotenoids has not been satisfied due to the high cost of natural extraction and mixed stereoisomers of chemical synthesis [5]. Microbially heterologous biosynthesis has exhibited great potentials for efficient carotenoid production [6][7][8][9]. As a typical carotenoid and precursor of vitamin A, β-carotene is also in great demand. The availability of novel key genes and the elucidation of a synthetic pathway facilitate production improvements of β-carotene.
As for the de novo β-carotene biosynthesis pathway, we divided it into the precursor module and the product module bounded with GGPP. Due to the important role of isoprenoid precursors (IPP, DMAPP, GPP, FPP and GGPP) for carotenoid synthesis, genes in the precursor module have attracted increasing attentions [21][22][23]. Among them, IDI catalyzes the reversible conversion of IPP into DMAPP, guiding the carbon flow to the synthesis of GPP, FPP and GGPP. GGPPS is responsible for GGPP synthesis, channeling the FPP flow towards carotenoids (40C) instead of sesquiterpenes (15C) and triterpenes (30C). Thereby, IDI and GGPPS define the availability of IPP, DMAPP and GGPP, functioning as the key regulatory nodes by directing metabolic flux to carotenoid biosynthesis. The two key precursor synthases have been derived from different species [24][25][26][27][28][29], and it has proved to be an effective strategy to strengthen DMAPP and GGPP supply through overexpression of IDI or/and GGPPS, diverting the metabolite flow towards carotenoid biosynthesis [8,[29][30][31].
Interestingly, thraustochytrids were also found to accumulate certain types of carotenoids [42][43][44][45]. High concentrations of butanol, methanol and ethanol were observed to remarkably strengthen the accumulation of carotenoids, especially astaxanthin [46][47][48]. These physiological phenomena indicated that a whole biological pathway for the de novo synthesis of carotenoids existed in Aurantiochytrium and Schizochytrium. Based on the availability of genomic and transcriptome data, researches of gene mining for carotenoid synthesis have been performed. The results were not satisfactory as no impressive carotenoid yields were found. The reported trifunctional β-carotene synthase (CrtIBY) from Schizochytrium did not produce carotenoids in Escherichia coli [49], and the CrtIBY from Aurantiochytrium accumulated a trace amount of β-carotene in Saccharomyces cerevisiae [50]. Furthermore, the excavated CrtO from Schizochytrium was proved to lack the activity of β-carotene ketolase in Escherichia coli [49].
In summary, previous studies have focused particularly on genes in the product module of carotenoid biosynthesis from Aurantiochytrium and Schizochytrium; however, the heterologous expression of these genes may require more complex and delicate regulations or membrane bindings. To date, little research has been conducted on genes in the precursor module of Aurantiochytrium. Given all the evidence above, the current study concentrated on genes in the precursor module of carotenoid biosynthesis. Specifically, the IDI and GGPPS genes from Aurantiochytrium limacinum MYA-1381 were isolated and identified based on the functional verification in the β-carotene biosynthesis pathway. The aim was to construct high-yielding carotenoid strains through the coordinated expression of novel genes. This study will be the first to report on the functional identification and engineering applications of IDI and GGPPS derived from Aurantiochytrium. Overall, the study presents a broader view of current knowledge regarding carotenoid biosynthesis in Aurantiochytrium and provides novel functional elements for carotenoid production.

Bioinformatic Analysis of AlGGPPS and AlIDI
Putative AlGGPPS and AlIDI genes encoded proteins of 417 and 256 amino acids, respectively, with no transmembrane domains predicted by TMHMM (Figure 1a,b). The

Bioinformatic Analysis of AlGGPPS and AlIDI
Putative AlGGPPS and AlIDI genes encoded proteins of 417 and 256 amino acids, respectively, with no transmembrane domains predicted by TMHMM (Figure 1a,b). The phylogenetic tree indicated that GGPPS and IDI of Aurantiochytrium limacinum MYA-1381 exhibited high homology with the geranylgeranyl pyrophosphate synthase and isopentenyl diphosphate isomerase of Hondaea fermentalgiana ( Figure S1a A reliable 3D model of AlGGPPS was constructed based on the X-ray crystal structure of the human geranylgeranyl pyrophosphate synthase (hGGPPS) mutant (Y246D; PDB ID 6C56) [51], which shared 51.36% sequence identity with AlGGPPS. The structure overlay of the AlGGPPS model and its template hGGPPS mutant (gray) also demonstrated similarity ( Figure 2a). Furthermore, docking conformations of GGPPS-FPP (−6.3 kcal/mol) and GGPPS-IPP (−5.5 kcal/mol) were confirmed with affinity scores. The molecular docking prediction showed that two substrates (IPP and FPP) were both docked into the catalytic pocket of AlGGPPS, and close together (Figure 2b). Three amino acid residues (R95, Q300, K327) were hydrogen-bonded to FPP (Figure 2c). Another three amino acid residues (N141, R143, H172) were also observed to bind IPP through the formation of hydrogen bonds ( Figure 2d). The result indicated the potential catalytic function of AlGGPPS by condensing the IPP with DMAPP. As for AlIDI, the best-scoring protein model and structure overlay were constructed, using the human IPP isomerase I (gray; hIPPI; PDB ID 2i6k) [52] with 52.47% sequence identity as its template (Figure 3a). The docking conformation of IDI-IPP (−6.0 kcal/mol) was also confirmed, indicating that substrate IPP was docked into the active pocket ( Figure 3b). Four amino acid residues (K57, S108, K132, E189) formed hydrogen bonds with IPP, strengthening binding of the substrate (Figure 3c). Results facilitated the revelation of the potential catalytic function and conserved catalytic mechanisms of AlIDI. A reliable 3D model of AlGGPPS was constructed based on the X-ray crystal structure of the human geranylgeranyl pyrophosphate synthase (hGGPPS) mutant (Y246D; PDB ID 6C56) [51], which shared 51.36% sequence identity with AlGGPPS. The structure overlay of the AlGGPPS model and its template hGGPPS mutant (gray) also demonstrated similarity ( Figure 2a). Furthermore, docking conformations of GGPPS-FPP (−6.3 kcal/mol) and GGPPS-IPP (−5.5 kcal/mol) were confirmed with affinity scores. The molecular docking prediction showed that two substrates (IPP and FPP) were both docked into the catalytic pocket of AlGGPPS, and close together ( Figure 2b). Three amino acid residues (R95, Q300, K327) were hydrogen-bonded to FPP (Figure 2c). Another three amino acid residues (N141, R143, H172) were also observed to bind IPP through the formation of hydrogen bonds (Figure 2d). The result indicated the potential catalytic function of AlGGPPS by condensing the IPP with FPP. As for AlIDI, the best-scoring protein model and structure overlay were constructed, using the human IPP isomerase I (gray; hIPPI; PDB ID 2i6k) [52] with 52.47% sequence identity as its template ( Figure 3a). The docking conformation of IDI-IPP (−6.0 kcal/mol) was also confirmed, indicating that substrate IPP was docked into the active pocket ( Figure 3b). Four amino acid residues (K57, S108, K132, E189) formed hydrogen bonds with IPP, strengthening binding of the substrate (Figure 3c). Results facilitated the revelation of the potential catalytic function and conserved catalytic mechanisms of AlIDI.

Identification of AlGGPPS for β-Carotene Biosynthesis
The constructions of pEBIY, pBIY and pGBIY were verified by PCR amplification with primer pairs P1/P2, P1/P3 and P4/P3, respectively (Figures 4 and S2). As mentioned, three stains bearing expression plasmids of pEBIY, pBIY and pGBIY were successively constructed for positive control, negative control and functional identification, respectively. Results showed that all 3 stains exhibited similar OD 600 values at 12 h ( Figure 5a). β-carotene was not detected in the BIY stain with CrtE deletion by HPLC analysis (Figure 6b), and the strain harboring the pEBIY or pGBIY plasmid accumulated β-carotene with an orange pigmented colony (Figures 5b and 6a), indicating that the CrtE gene was crucial for β-carotene production. Furthermore, AlGGPPS could compensate for the lack of the original CrtE gene and replace its function in β-carotene biosynthesis.  tively. Results showed that all 3 stains exhibited similar OD600 values at 12 h (Figure 5a) β-carotene was not detected in the BIY stain with CrtE deletion by HPLC analysis (Figure  6b), and the strain harboring the pEBIY or pGBIY plasmid accumulated β-carotene with an orange pigmented colony (Figures 5b and 6a), indicating that the CrtE gene was crucial for β-carotene production. Furthermore, AlGGPPS could compensate for the lack of the original CrtE gene and replace its function in β-carotene biosynthesis.   (c) (d) As for the β-carotene content, the GBIY stain harboring the AlGGPPS-CrtB-CrtI-CrtY gene cluster performed better in β-carotene accumulation than the initial EBIY stain during cultivation. The GBIY strain produced 5.14 mg/L of β-carotene in 12 h, which was a 39.7% increase in production compared with strain EBIY. In other words, in situ replacement of the original CrtE with AlGGPPS enhanced β-carotene accumulation with a 1.40fold production of the initial EBIY strain. In summary, the result of genetic complementation expression indicated that AlGGPPS functioned in β-carotene biosynthesis and performed better than the original CrtE gene of Pantoea ananatis, with higher β-carotene yield, As for the β-carotene content, the GBIY stain harboring the AlGGPPS-CrtB-CrtI-CrtY gene cluster performed better in β-carotene accumulation than the initial EBIY stain during cultivation. The GBIY strain produced 5.14 mg/L of β-carotene in 12 h, which was a 39.7% increase in production compared with strain EBIY. In other words, in situ replacement of the original CrtE with AlGGPPS enhanced β-carotene accumulation with a 1.40-fold production of the initial EBIY strain. In summary, the result of genetic complementation expression indicated that AlGGPPS functioned in β-carotene biosynthesis and performed better than the original CrtE gene of Pantoea ananatis, with higher β-carotene yield, implying a superior catalytic property. Thus, the GBIY stain harboring the AlGGPPS-CrtB-CrtI-CrtY gene cluster was chosen for further genetic manipulation and engineering improvement.

Overexpression of AlIDI for Improving β-Carotene Production
PCR amplification with primer pairs P6/P8 and P6/P10 confirmed the constructions of pGBIY-EcIDI and pGBIY-AlIDI, respectively (Figures 4 and S2). As expected, the strain bearing the pGBIY-EcIDI or pGBIY-AlIDI plasmid both enhanced the β-carotene accumulation compared with the control stain harboring the pGBIY plasmid (Figure 5d). By contrast, the GBIY-AlIDI stain harboring the AlGGPPS-CrtB-CrtI-CrtY-AlIDI genes performed better in β-carotene production than the GBIY-EcIDI stain bearing the AlGGPPS-CrtB-CrtI-CrtY-EcIDI cluster. Specifically, the GBIY-AlIDI strain produced 10.99 mg/L of β-carotene, which was a 80.9%, 113.9% and 198.8% increase in production compared with the GBIY-EcIDI, GBIY and EBIY stains in 12 h, respectively. The higher β-carotene yield implied the superior catalytic property of AlIDI compared with the endogenous IDI of Escherichia coli. In conclusion, the coordinate expression of the AlGGPPS and AlIDI genes enhanced β-carotene accumulation with a 2.99-fold production of the initial EBIY strain in 12 h, which explained the intensity of pigmentation (Figure 6a). Notably, the GBIY-AlIDI strain exhibited a similar OD 600 value compared with other strains (Figure 5c), indicating the increase in β-carotene production did not result in the potential growth inhibition.

Discussion
Due to its high growth rate, excellent product yield and mature gene manipulation, Escherichia coli was chosen as the host strain for functional verification of novel genes and efficient production of β-carotene in this study. It was previously reported that multiphasic transformation inhibited the growth and biomass of carotenoid producing stains due to the expression burden [49]. In this work, five functional genes were integrated into one expression vector (Figure 4), preventing potential growth inhibition (Figure 5a,c). This indicated that the strategy balanced well between biomass and high product yield. However, some challenges still needed to be considered for heterologous expression in E. coli, especially the post-translational modifications and availability of membrane binding sites. As reported, the transmembrane trifunctional synthase (CrtIBY) from Schizochytrium lacked the activity to produce β-carotene in E. coli [49]. In this work, no transmembrane domain was predicted in AlIDI and AlGGPPS (Figure 1), which might help them to function efficiently in E. coli. Furthermore, the application of these two novel functional genes could be extended to other prokaryotes, such as Synechocystis, which is regarded as a potential cell factory for carotenoid synthesis [21,53,54].
Adequate supply of isoprenoid precursors was necessary for β-carotene accumulation. GGPPS was selected for high frequency because of its key role in FPP distribution from the sterol synthesis towards the carotenoid pathway. In this study, the carotenoid analysis of the CrtE/GGPPS knockout mutant suggested its importance in β-carotene biosynthesis. AlGGPPS functioned in β-carotene biosynthesis and performed better than the original CrtE gene of Pantoea ananatis with a higher β-carotene yield. This was consistent with the previously reported GGPPS gene of Haematococcus pluvialis with higher synthase activity, leading to higher carotenoid level [26]. It was proposed that high GGPPS activity may lead to a shortage of DMAPP and narrow the chance of carotenoid biosynthesis [24]. Overexpression of IDI was used to equilibrate the concentrations of IPP and DMAPP, overcoming the restriction of carotenoid overproduction [24]. Thus, IDI was another key target of metabolic engineering for carotenoid production.
In this study, AlIDI also displayed a superior catalytic property compared to the endogenous IDI gene of Escherichia coli with higher β-carotene production. In conclusion, the expression of AlGGPPS and/or AlIDI, two key precursor synthase encoding genes in Aurantiochytrium, both improved the production of β-carotene (Figure 5b,d). The results were consistent with previous studies [8,[29][30][31]. Furthermore, the highest β-carotene yield was observed in the stain coordinately expressing AlGGPPS and AlIDI, indicating the additive effects of AlGGPPS and AlIDI on β-carotene overproduction. Notably, the highest β-carotene production of the GBIY-AlIDI strain did not result in growth inhibition, indicating the excellent potential for further production improvement.
Previous work showed that an engineered E. coli strain produced 39.03 mg/L β-carotene under a small-scale culture through combined engineering with MEP (endogenous DXS and IDI genes), β-carotene synthesis and central metabolic moules [55]. Furthermore, the maximum titer was raised to 2.1 g/L (53.8-fold) in fed-batch fermentation (7L). Thus, the scale of cultivation and introduced gene had a marked impact on β-carotene production. In this work, the accumulation of 10.99 mg/L β-carotene was also conducted under flask culture on a small scale. The β-carotene yield of modified E. coli, theoretically, would be enhanced with large-scale fed-batch fermentation and more introduced genes. As β-carotene is a direct precursor of astaxanthin, the engineered strain could be used as a promising starting strain for astaxanthin synthesis.
As for the precursor synthesis for carotenoids, Aurantiochytrium showed excellent potential to provide novel and valuable genes. Thus, the transformation of whole genes involved in the upper MVA pathway of Aurantiochytrium and rebuilding the high-efficiency MVA pathway in E. coli would be promising. Due to the lack of functional characterization of key genes, the pathway of de novo β-carotene biosynthesis in Aurantiochytrium has not been well elaborated. Based on the results of this work, we proposed a simplified pathway model for β-carotene biosynthesis in Aurantiochytrium (Figure 7). A similar pathway has also been suggested in Schizochytrium based on transcriptome data [46][47][48]. However, the annotated genes from transcriptome data without functional identification may be unconvincing. For instance, another two putative GGPPS genes were also isolated from Aurantiochytrium limacinum MYA-1381 together with AlGGPPS, based on the genomic and transcriptome data. The result of genetic complementation expression indicated that the other two putative GGPPS genes did not function in β-carotene biosynthesis. Although the route taken in this work was not unique, it was the first to be proposed based on the functional identification of genes. As shown in the pathway, AlGGPPS and AlIDI tightly connected the upstream MVA pathway and the downstream carotenogenesis enzymes. Thus, the gene mining and functional identification of AlGGPPS and AlIDI in this work enriched the de novo β-carotene biosynthesis pathway in Aurantiochytrium. two putative GGPPS genes did not function in β-carotene biosynthesis. Although the route taken in this work was not unique, it was the first to be proposed based on the functional identification of genes. As shown in the pathway, AlGGPPS and AlIDI tightly connected the upstream MVA pathway and the downstream carotenogenesis enzymes. Thus, the gene mining and functional identification of AlGGPPS and AlIDI in this work enriched the de novo β-carotene biosynthesis pathway in Aurantiochytrium.  Recent studies provided guidance for further modifications of AlGGPPS and AlIDI. As reported, the rationally designed GGPPS with multisite mutations from Nicotiana tabacum exhibited significantly higher efficiency for GGPP synthesis than the wild-type gene, increasing carotenoid levels greatly [29]. A directed evolution method was applied to improve the enzyme activity, half-life and substrate affinity of IDI from Saccharomyces cerevisiae [56]. Additionally, multienzyme assembly using docking domains to increase cascade catalytic efficiency was also a potential strategy for enhancing carotenoid production [23]. As previously reported, the 3D model of thraustochytrid squalene synthase (SQS) provided important insights into possible binding and active sites [57]. Thus, results of protein structure and substrate molecular docking in this work (Figures 2 and 3) facilitated further engineering modifications of AlGGPPS and AlIDI for higher catalytic efficiency and activity. Notably, GGPP also functioned as the common precursor of various medicinal terpenoids, including sclareol [58], taxadiene [59][60][61][62], miltiradiene [63] and rographolide [27]. Thus, the excavated AlGGPPS and AlIDI with superior catalytic properties could also be applied for the biosynthesis of valuable GGPP-derived isoprenoid drugs.

Strains, Media and Growth Conditions
Aurantiochytrium limacinum MYA-1381 was obtained from the American Type Culture Collection (ATCC). Escherichia coli DH5α and BL21(DE3) were acquired from Takara (Dalian, China). The plasmids and strains constructed in this work were listed in Table 1.
Aurantiochytrium was grown at 30 • C and 200 rpm in ATCC 790 By+ medium (1 g/L peptone, 1 g/L yeast extract, 5 g/L glucose in seawater). As for the recombinant Escherichia coli for β-carotene production, bacterial culture was performed as previously described [55]. Single colonies were inoculated into 4 mL Luria-Bertani (LB) medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) with 34 mg/L chloramphenicol; the stains were cultured at 37 • C and 200 rpm overnight. Seed culture was then inoculated into fresh LB containing antibiotic with an initial OD 600 of 0.05. The bacteria were incubated at 37 • C and 200 rpm for 12 h. Cells were collected for further analysis with three biological replicates.

Bioinformatic Analysis
Candidate genes of isopentenyl pyrophosphate isomerase (AlIDI) and geranyl-geranyl pyrophosphate synthase (AlGGPPS) were excavated based on the genome data of Aurantiochytrium limacinum MYA-1381 (https://genome.jgi.doe.gov/portal/AurlimATCMYA1 381_FD/AurlimATCMYA1381_FD.info.html) (accessed on 1 January 2023). The transcriptome annotation of Schizochytrium treated with butanol [46], methanol [47] and ethanol [48] facilitated further determination of putative genes. The transmembrane prediction program DeepTMHMM (https://dtu.biolib.com/DeepTMHMM) (accessed on 1 February 2023) was used to identify transmembrane regions. Homologous sequences from other species were retrieved from the GenBank using BLAST with the amino acid sequences of AlGGPPS and AlIDI. Multiple sequence alignment was conducted with ClustalW. Phylogenetic trees of AlGGPPS and AlIDI were built using MEGA XI by the maximum likelihood method with WAG model and 1000 bootstrap replications.
Furthermore, proteins were modeled using the I-TASSER online server (https:// zhanggroup.org/) (accessed on 1 March 2023). The structure was selected with the highest score of confidence coefficient (C-score). Meanwhile, the confidence of each amino acid residue by Robetta modeling was referred to ensure that the RMSD value of key sites was within 1.5 Å. An in silico molecular docking analysis was conducted to investigate potential binding modes between the protein and substrate molecules using Autodock vina 1.2.0. The search grid for substrate molecules (IPP and FPP from the Pubchem database) was determined with dimensions size_x: 40 Å, size_y: 40 Å and size_z: 40 Å. Docking conformations were confirmed with affinity scores. Visualization and analysis of the docking results were performed with ChimeraX 1.15. Pymol 2.5.2 was used to complete the final drawing.

RNA Extraction, cDNA Synthesis and Genomic DNA Extraction
For the preparation of RNA extraction, the single colony of Aurantiochytrium was inoculated into 20 mL ATCC 790 By+ medium and cultivated at 30 • C and 200 rpm for 24 h. Cell pellets were harvested and crushed by the liquid nitrogen grounding method. The total RNA of Aurantiochytrium limacinum MYA-1381 was then isolated with the Rapid Fungal RNA Extraction Kit (Coolaber Biotechnology, Beijing, China). Thermo Scientific NanoDrop 2000 spectrophotometer was applied to determine the quantity and quality of RNA extracts. With the removal of the residual genomic DNA, the cDNA was subsequently generated using the HiScript cDNA Synthesis Kit (Vazyme Biotechnology, Nanjing, China). Briefly, a 16 µL reaction consisting of 1 µg total RNA, 4 µL 4× gDNA wiper mix and optimal amount of RNase-free ddH 2 O was incubated at 42 • C for 2 min. Then, an additional 4 µL 5× HiScript III qRT SuperMix was added and the final 20 µL reaction mixture was incubated at 37 • C for 15 min to synthesize the cDNA. After cultivation at 37 • C and 200 rpm for 12 h, cell pellets of Escherichia coli BL21 (DE3) were harvested and crushed.
Genomic DNA extraction from Escherichia coli was conducted with the TIANamp Bacteria DNA Kit (TIANGEN Biotechnology, Beijing, China).

Extraction and Measurement of β-Carotene
Pigment extraction and HPLC analysis followed the reported method [64]. Cells were harvested after 12 h cultivation and centrifuged at 10,000 rpm for 5 min. Subsequently, the pellets were resuspended in 1 mL of acetone and incubated at 55 • C for 15 min in the dark to extract carotenoids. The supernatant was then collected at 10,000 rpm for 10 min. The accurate β-carotene yield of each strain was determined and analyzed using the HPLC (Shimadzu A20 system, Shimadzu, Kyoto, Japan) configured with a C18 analytical column (5 µm, 4.6 mm × 250 mm, Waters, Wexford, Ireland). Methanol and isopropanol (8:2, V:V) were chosen as the mobile phase at a flow rate of 1 mL/min at 35 • C. A β-carotene standard sample (Cat. No. C4582, Sigma, St. Louis, MI, USA) was used for carotenoid identification and quantitation with the absorbance signal recorded at 450 nm. Results represented the means ± S.D. of three independent experiments.

Conclusions
In this work, two isoprenoid precursor synthetases coding genes AlGGPPS and AlIDI were functionally characterized. The encoded proteins both exhibited superior catalytic performance with higher carotenoid yield. Furthermore, the redesigned β-carotene synthesis module with coordinated expression of the two genes demonstrated the best production capacity. It was proven to be an effective strategy to divert and enhance metabolite flow towards carotenogenesis through overexpression of AlGGPPS and AlIDI, strengthening DMAPP and GGPP supply. Thus, this study enriched the de novo β-carotene biosynthesis pathway in Aurantiochytrium and provided the availability of two effective function elements to produce β-carotene and other valuable isoprenoids.