ACP domains exist in 6–13 tandem forms in eukaryotic PUFA synthase, which function as loading acyl groups in the biosynthesis of PUFAs. The serine residue at the active sites of ACP is modified by phosphopantetheinyl transferase (PPTase) to transform apo-ACP into holo-ACP with a 4′-phosphopantetheine (Ppant) arm (Beld et al., 2014; Rullan-Lind et al., 2019). Then, the fatty acyl chain is covalently connected with holo-ACP through thioester bond and catalyzed by different domains to complete the PUFA synthesis process. Three genes encoding PUFA synthase from Schizochytrium sp. ATCC 20888 and a hetI encoding PPTase from Nostoc sp. PCC 7120 were heterologously co-expressed in E. coli (Hayashi et al., 2016). It was found that the number of ACP units in subunit A was positively correlated with the productivity of PUFAs, as the deletion of ACP domains led to a decrease in PUFA productivity and the insertion of ACP domains or inactivating ACP domains led to increased productivity (Hayashi et al., 2016). Each ACP of Shewanella japonica PUFA synthase can be activated by PPTase (Jiang et al., 2008), but this characteristic has not been confirmed in eukaryotic PUFA synthases. PPTases are not present in eukaryotic PUFA synthases but are found elsewhere in the genomes of Thraustochytrium sp. 26185, Schizochytrium sp. ATCC MYA-1381 (S. limacinum SR21) (Meesapyodsuk and Qiu, 2016) and Aurantiochytrium sp. SD116 (Wang et al., 2020a). Previous studies also showed that the overexpression of endogenous PPTase significantly increased the proportion of PUFAs, especially DHA, in Aurantiochytrium (Wang et al., 2020a).

There are two ketoacyl-ACP synthase (KS) domains in eukaryotic PUFA synthase. One is at the N-terminus of subunit A, named KS_A, and the other is KS_B at the N-terminus of subunit B, which is adjacent to a CLF domain. The KS domain catalyzes the Claisen condensation reaction of the extender unit and acyl-ACP, leading to a two-carbon extension in the carbon chain (Hayashi et al., 2016; Yoshida et al., 2016; Xie et al., 2017). It has been found that both the KS_A and KS_B of Thraustochytrium sp. strain ATCC 26185 could complement the defective phenotypes of both ketoacyl-ACP synthase I (FabB) and ketoacyl-ACP synthase II (FabF) mutants in E. coli, suggesting their ketoacyl-ACP synthase activities (Xie et al., 2017). In prokaryotic PUFA synthase, the KS domain adjacent to the CLF domain catalyzes the last elongation step for DHA biosynthesis, and two KS domains have selectivity for the recognition of chain length and the degree of unsaturation of acyl-ACP intermediates (Hayashi et al., 2019a; Santín et al., 2020). It is likely that eukaryotic KS domains have similar characteristics. Hayashi et al. found that mutation of three amino acids located in the KS_B of Aurantiochytrium sp. OH4 could change the major product from DHA (EPA/DHA = 0) to EPA (EPA/DHA=1.07) in E. coli (Hayashi et al., 2019a). However, the substrate selectivity of the two KS domains in eukaryotic PUFA synthase still needs to be further investigated.

In addition, the CLF domain in PUFA synthase shares high sequence similarity with the KS domain, but it lacks conserved active residues (Hayashi et al., 2019a) and cannot form a covalent bond with the acyl group. In prokaryotic PUFA synthase, the CLF domain and its adjacent KS domain are usually studied as a whole (Hayashi et al., 2019a; Santín et al., 2020). It has been demonstrated that the CLF domain is the primary determinant of polyketide chain length in polyketide biosynthesis (Tang et al., 2003). The disruption of the CLF domain in Schizochytrium sp. ATCC MYA-1381 (S. limacinum SR21) affected growth, decreased the content of PUFAs and increased the content of SFAs, and the proportion of C22 PUFA decreased by 57.51% (Li et al., 2018).

Eukaryotic PUFA synthase contains two AT-like domains, the malonyl-CoA: ACP transacylase (MAT) domain in subunit A and the acyltransferase (AT) domain in subunit B. The MAT domain catalyzes the binding of the malonyl group and ACP, and the AT domain has thioesterase activity and catalyzes the release of free fatty acids from long-chain acyl-ACPs. Complementation tests showed that the MAT domain of Thraustochytrium sp. 26185 was able to restore the growth phenotype of a temperature-sensitive E. coli mutant defective in malonyl-CoA: ACP transacylase (FabD) activity, but the AT domain could not (Almendariz-Palacios et al., 2021). In vitro experiments show that the AT domain has thioesterase activity against fatty acyl-ACPs and fatty acyl-CoAs (Hayashi et al., 2020a; Almendariz-Palacios et al., 2021). The AT domain of Schizochytrium tends to catalyze DHA-ACP, EPA-ACP and long saturated acyl-ACPs among fatty acyl-ACP substrates; the AT domain of Thraustochytrium shows higher activity toward DHA-CoA among fatty acyl-CoA substrates, and mutation of two putative active site residues S96 or H220 at the AT domain resulted in reduced catalytic activity toward DHA-CoA (Hayashi et al., 2020a; Almendariz-Palacios et al., 2021). These results suggest the AT domain might be involved in the release of freshly synthesized DHA as free fatty acid from the PUFA synthase. In addition, the overexpression of the AT domain increased the free fatty acid content in the E. coli mutant with the defective β-oxidation-related gene fadD (Almendariz-Palacios et al., 2021). The deletion of the AT domain led to a slow cell growth rate and reduced content of DHA, but the complementary strain with the AT domain from Shewanella exhibited the recovery of growth and increased content of EPA and DHA in Schizochytrium sp. HX-308 (Ren et al., 2018).

There is a KR domain located in subunit A. The KR domain imbedded in prokaryotic PUFA synthase is able to reduce 3-oxoacyl-ACP to (3R)-hydroxyacyl-ACP in the presence of NADPH (Hayashi et al., 2019b), and the function of the KR domain in eukaryotic PUFA synthase is probably the same, although this requires further confirmation. In addition, overexpression of the KR domain led to an increase in biomass, total fatty acid content and DHA content in A. limacinum OUC168 (Liu et al., 2018).

Eukaryotic PUFA synthase contains three DH domains. The sequence of the DH domain at the C-terminus of subunit A is similar to that in PKS synthase, which is named DH_PKS. The other two domains exhibit higher similarity with β-hydroxyacyl-ACP dehydratase (FabA) from E. coli and are named DH_FabA. All three domains can complement the defective phenotype of the E. coli fabA temperature-sensitive mutant, suggesting that they can function as β-hydroxyacyl-ACP dehydratases (Xie et al., 2018). In prokaryotic PUFA synthase, both DH_PKS and DH_FabA catalyze the dehydration of β-hydroxyacyl-ACP but two kinds of DH domains have selectivity toward substrates with different chain lengths and DH_FabA domains also have the ability to catalyze 2,2- and 2,3-isomerization reactions (Hayashi et al., 2019b).

The DH domains in eukaryotic PUFA synthase may have the same characteristics, which still needs to be confirmed. Man et al. showed that heterologous expression of two DH_FabA domains can increase the proportion of unsaturated fatty acids, while heterologous expression of DH_PKS helps increase saturated fatty acids (Man et al., 2021) in E. coli. Overexpression of DH_PKS in A. limacinum OUC168 showed increased contents of both C20:4 and total fatty acids (Liu et al., 2018). Heterologous expression of DH_PKS and DH_FabA domains in E. coli increased the biomass and weakened the inhibitory effect of mid-chain fatty acids on cell growth (Man et al., 2021). In addition, Xie et al. (Xie et al., 2020) pointed out that there are some differences in terms of the functions of these two DH domains, DH_FabA-1 and DH_FabA-2. The domain swapping analysis showed that only the DH_FabA-1 domain can functionally replace the DH domain of a type I fatty acid synthase in yeast. The heterologous expression of PUFA synthase in E. coli showed that the mutation or deletion of DH_FabA-1 or substitution of DH_FabA-1 with DH_FabA-2 led to the loss of PUFA production capacity, and the mutation or deletion of DH_FabA-2 led to a small amount of DPA production and no DHA production (Xie et al., 2020). The overexpression of the DH_FabA-1 domain in S. limacinum SR21 increased the content of PUFAs and total lipids (Shi et al., 2021). Disruption of the DH_FabA-2 domain of Schizochytrium sp. ATCC MYA-1381 (S. limacinum SR21) affected cell growth, and the proportion of PUFAs was decreased, while the proportion of saturated fatty acids (SFAs) was increased (Li et al., 2018). In a heterologous expression of protist PUFA synthase in canola seeds, Walsh et al. mentioned that the ratio of DHA to DPA was increased by replacing DH_FabA-2 of Schizochytrium with DH_FabA-2 of Thraustochytrium sp. 23B (Walsh et al., 2016).

Enoyl-ACP reductases catalyze the reduction of 2-trans enoyl-ACPs to form α, β-saturated acyl-ACPs in fatty acid biosynthesis (Massengo-Tiassé and Cronan, 2009). There are two ER domains in eukaryotic PUFA synthase. The ER_B or ER_C domain is located at the C-terminus of subunit B or C, respectively. Knockout of ER_B significantly decreased the content of PUFAs in S. limacinum SR21. Knockout of ER_C reduced the content of SFAs, while overexpression of ER_C led to a decrease in PUFA content and an increase in SFA content in S. limacinum SR21 (Ling et al., 2020; Shi et al., 2021). In addition, several lines of evidence have demonstrated that the addition of triclosan can inhibit the expression of ER_C and lead to an increase in PUFA production (Ling et al., 2020). These results suggest that ER_B plays an important role in the synthesis of PUFAs, and ER_C is more likely to be related to the synthesis of SFAs in S. limacinum SR21. When ER_B of S. limacinum SR21 was replaced by the ER domain of Shewanella SCRC2738, the content of EPA was increased by 85.7%, and the transcriptional level of several genes of PUFA synthase was also upregulated. However, there was little effect on the synthesis of PUFAs when ER_C was replaced by ER of Shewanella SCRC2738 (Yang et al., 2021).

The initiation of PUFA synthesis involves the synthesis of acetyl-ACP and malonyl-ACP. MAT, KS, tandem ACP domains and PPTase participate in this process. The tandem ACPs are activated by PPTase from apo-ACPs to holo-ACPs. In the prokaryotic PUFA synthase pathway, the MAT domain specifically recognizes and catalyzes the loading of malonyl-CoA onto holo-ACP to form malonyl-ACP (Santin and Moncalian, 2018), and then, the KS_A domain catalyzes the decarboxylation of malonyl-ACP to form acetyl-ACP (Hayashi et al., 2019a; Santín et al., 2020). Similarly, the MAT domain of eukaryotic PUFA synthase has the function of malonyl-CoA: ACP transacylase (Almendariz-Palacios et al., 2021), which transfers the malonyl moiety from malonyl-CoA onto ACP, but its substrate specificity has not been evaluated. In vitro experiments also suggested that the starter unit acetyl-ACP could be derived from malonyl-ACP in the PUFA synthesis process of Schizochytrium (Metz et al., 2009). Malonyl-ACP is used as the carbon chain extender unit during the subsequent carbon chain extension process.

The carbon chain extension step of PUFA synthesis involves the elongation of fatty acyl-ACPs and the formation of cis double bonds at the appropriate position. KS, CLF, tandem ACPs, KR, DH_PKS, DH_FabA and ER domains are involved. It is widely accepted that the carbon chain extension of the PUFA synthase pathway is an iterative process, which is the same as the type I fatty acid synthesis pathway or iterative PKS pathway (Hayashi et al., 2020b; Qiu et al., 2020; Remize et al., 2021). Each iterative process elongates two carbon chains and forms saturated or cis double bonds at the appropriate positions, but the details have not been well elucidated until now.

In the prokaryotic PUFA synthase pathway, the iterative process is speculated to be as follows. First, fatty acyl-ACP is condensed with malonyl-ACP to form β-ketoacyl-ACP through catalysis of the KS domain. Two KS domains have selectivity for the chain length and the degree of unsaturation of acyl-ACPs. The KS_A domain is responsible for the elongation of short chain (C2, C4, C6, C10) and long chain (C16, C18) substrates, and the KS-CLF didomain is responsible for the elongation of medium chain (C10) and very long chain C20 substrates (Hayashi et al., 2019a; Santín et al., 2020). Second, the KR domain is responsible for reducing the extended β-ketoacyl-ACP to (3R)-hydroxyacyl-ACP (Hayashi et al., 2019b). Then, (3R)-hydroxyacyl-ACP is dehydrated by the DH domain to form an α, β-trans double bond, but the difference between DH_PKS and DH_FabA leads to distinct subsequent reactions. When the trans double bond is formed under the dehydration of DH_PKS, the next reaction will be the reduction by the ER domain and form an α, β-saturated bond. When the trans double bond is formed under the dehydration of DH_FabA, it will be further catalyzed to form a cis double bond by 2,2- or 2,3-isomerization (Hayashi et al., 2019b). After several iterations, the extension of fatty acyl-ACP was finished, forming PUFA-ACP.

In eukaryotes, the iterative process is likely the same as that in prokaryotes, which comprises condensation of acyl-ACPs and extender units, reduction of β-ketoacyl-ACPs, dehydration of β-hydroxyacyl-ACPs and selective reduction of enyl-ACPs. However, as the sequence, number and location of KS, DH and ER are obviously different from those in prokaryotes, it is speculated that the details of carbon chain elongation and double bond formation could also be different from those in prokaryotes, although details still need further investigation.

The release process of PUFAs involves the offloading of the fatty acyl group from ACP, which is finished by the AT domain in subunit B. Catalyzed by the AT domain, a specific long chain fatty acyl group linked with the thioester bond is hydrolyzed from the Ppant arm of holo-ACP, and the final PUFA is released as a free fatty acid (Metz et al., 2009; Hayashi et al., 2020a).

First, both are independent of the FAS pathway, and PUFAs can be synthesized de novo from malonyl-CoA using NADPH as a cofactor (Metz et al., 2009) and released in the form of free fatty acids. Second, the PUFA synthases of eukaryotes and prokaryotes contain the same types of domains, and the basic functions of these domains are also similar (Qiu et al., 2020). Last, the main methodologies used in studies on the PUFA synthase pathway of prokaryotes and eukaryotes are basically the same, including in vitro enzyme reactions and in vivo experiments such as heterologous expression, deletion, addition and replacement of subunits or domains.

First, the major product synthesized by the PUFA synthase pathway in eukaryotes is DHA, while it is EPA, DHA, ARA or LA in prokaryotes (Gemperlein et al., 2014; Morabito et al., 2019). Second, the number of ER domains and the domain organizations of PUFA synthases of eukaryotes and prokaryotes are different (Qiu et al., 2020). Third, there are some differences in the properties and catalytic activities of the domains. For example, two KS domains in prokaryotic PUFA synthase could not be expressed in the form of a single enzyme and were studied in the form of KS-AT and KS-CLF didomains (Hayashi et al., 2019a; Santín et al., 2020). However, in Thraustochytrium sp., both KS domains were expressed as stand-alone enzymes in E. coli (Xie et al., 2017). The AT domain of Schizochytrium sp. exhibited strong substrate selectivity, while the AT domain in P. profundum showed promiscuous substrate specificity against short to long acyl chains (Hayashi et al., 2020a). Moreover, since a special 1-acylglycerol-3-phosphate O-acyltransferase (AGPAT) domain exists in the PUFA synthase of myxobacteria (Gemperlein et al., 2014), it is speculated that there might be other PUFA release mechanisms by directly forming glycerol phospholipids in prokaryotes, while it has never been found in eukaryotes. Last, studies on these two pathways typically have different focuses. The diversity of major products synthesized by prokaryotic PUFA synthases suggests that there are differences in catalytic activity of some domains (Figure 3), which makes it relatively easy to study the catalytic mechanism. Thus, studies of prokaryotic PUFA synthases mainly focus on mechanism research, such as elucidation of the structure and function of domains. In contrast, eukaryotic microorganisms such as Schizochytrium have already been used as industrial strains for the production of DHA-rich oils (Chi et al., 2021), therefore most studies of eukaryotic PUFA synthase pathways focus on application research, such as changes in the profiles and contents of PUFA products (Table 2). Similarities and differences between prokaryotic and eukaryotic PUFA synthase pathways.

Different from those in eukaryotes, prokaryotic PUFA synthases have been found in many marine bacteria and even myxobacteria, and the main PUFA products produced by these PUFA synthases are more diverse. Table 3 summarizes the prokaryotes with PUFA synthase that have been heterogeneously expressed to date.

In addition to the diverse PUFA profiles, the domain organizations between some prokaryotic PUFA synthases are also quite different (Figure 3), which provides good candidates for the studies of prokaryotic PUFA synthases. At present, more PUFA-producing eukaryotic microorganisms are being discovered (Colonia et al., 2021; Kalidasan et al., 2021; Wang et al., 2021; Bai et al., 2022), and species with differentiated PUFA profiles are likely to be discovered in the future. Analyzing the genomes of these species and expressing the possible genes in E. coli will be an efficient way to identify novel PUFA synthases. The discovery and identification of distinct PUFA synthases in more eukaryotic species is of great value for both mechanistic research and engineering of eukaryotic PUFA synthases.

As mentioned above, functional research on each domain can help elucidate the biosynthesis process of PUFAs; for example, the study of the MAT domain can reveal the initiation of PUFA synthesis, the study of the KS, KR, DH, and ER domains can reveal the iterative process of PUFA synthesis, and the study of the AT domain helps us understand the release of PUFAs. Since AT-like, KS, ACP, DH, and ER domains have multiple copies in eukaryotic PUFA synthases, it is necessary to determine whether there are functional differences between homologous domains. Referring to the existing research of PUFA synthase and other multidomain enzymes, the functional research of eukaryotic PUFA synthase domains can be expanded through in vivo, in vitro and structural analysis.

To increase the content of specific PUFA, the modification of eukaryotic PUFA synthases should be gradually carried out while elucidating the mechanisms of eukaryotic PUFA synthases. Some strategies, such as domain swapping and mutagenesis, can be used to modify eukaryotic PUFA synthases.

At present, the genetic manipulation of eukaryotes that possess PUFA synthases is still challenging, which makes it difficult to modify and apply the eukaryotic PUFA synthase pathway. Expanding the molecular biology techniques of these eukaryotes may start from at least two aspects: developing efficient genetic transformation methods and establishing efficient genetic manipulation tools.

Until now, there has been no genetic transformation method that is universally applicable for PUFA-producing eukaryotes. In thraustochytrids, electroporation is the most common strategy for transformation (Wang et al., 2019; Han et al., 2020; Han et al., 2021; Rau and Ertesvag, 2021; Shi et al., 2021), and particle bombardment (Metz et al., 2009) and Agrobacterium-mediated transformation (Huang et al., 2021) have also been reported. To make the transformation more efficient, it is important to optimize the operating conditions according to different strains; in addition, other methods, such as bacterial conjugation (Sreenikethanam et al., 2022) and cell-penetrating peptide (CPP) transport (Rau and Ertesvag, 2021), should also be considered. For screening transformants, it has been reported that resistance genes for bleomycin, neomycin, hygromycin, cycloheximide and paromomycin (Wang et al., 2019; Du et al., 2021) can be used as selectable markers. However, the resistance to corresponding antibiotics varies greatly among different strains; thus, it is necessary to determine the inhibitory concentration specifically for different strains.

Additionally, other gene elements, such as promoters and terminators, for gene manipulation of PUFA-producing eukaryotes are limited. Rau et al. introduced some endogenous and exogenous promoters and terminators used in thraustochytrids (Rau and Ertesvag, 2021). To perform more complex genetic manipulations, more available gene elements need to be mined in the future. To achieve multigene expression under genetic element-restricted conditions, single multi-cistronic transcription units were constructed using 2A self-cleavage peptides in several thraustochytrid strains (Wang et al., 2020b; Rau and Ertesvag, 2021). Sun et al. reported the application of Cre/loxP site-specific recombination (Sun et al., 2015) to eliminate the antibiotic resistance gene in A. limacinum OUC88, which can reduce the number of different resistance genes needed in multistep genetic manipulation. In addition, most studies utilize homologous recombination for the gene integration of thraustochytrids, but the efficiency is low, and random integration still exists (Sakaguchi et al., 2012; Rau and Ertesvag, 2021). To achieve genetic manipulation at the nucleotide level, methods for site-directed integration and mutagenesis need to be developed. For example, the efficiency of specific gene knock-in by homologous recombination increased more than 10-fold by combining the clustered regularly interspaced short palindromic repeats (CRISPR/Cas9) system in Aurantiochytrium sp (Watanabe et al., 2021). Other gene-editing tools, such as zinc-finger nucleases (ZFNs), meganucleases (MNs), and transcription activator-like effector nucleases (TALEN)-mediated genome editing, have been used in many microalgae (Kumar et al., 2020), these tools can also be tested in PUFA-producing eukaryotes.

Although the native hosts of eukaryotic PUFA synthase have high industrial value, their genetic engineering remains a challenge. Therefore, the heterologous expression of eukaryotic PUFA synthase in an appropriate host is also an option for PUFA production. At present, eukaryotic PUFA synthases have been expressed in E. coli, Brassica napus (Walsh et al., 2016) and Arabidopsis, and corresponding PUFAs have been detected in product (Table 1). In addition to these strains, some industrial microorganisms can also be used as chassis. For example, Lactococcus lactis, Myxococcus xanthus, Pseudomonas putida and Yarrowia lipolytica are used as chassis for the heterologous expression of prokaryotic PUFA synthase. (Table 3).

Heterologous expression of PUFA synthase and the production of PUFAs can be improved by using synthetic biology approaches such as sequence design, promoter selection and engineering, and operon design. Meesapyodsuk et al. expressed Thraustochytrium PUFA synthase in E. coli and found that the organization of three genes as one operon produced higher DHA and DPA than the expression of genes in three plasmids (Meesapyodsuk and Qiu, 2016). Gemperlein et al. applied many strategies such as gene optimization and promoter engineering to better express myxobacterial PUFA synthases in Pseudomonas putida (Gemperlein et al., 2014; Gemperlein et al., 2016) and Y. lipolytica (Gemperlein et al., 2019), which improved the production of PUFAs and provided references for heterologous expression of eukaryotic PUFA synthase.

In addition, many metabolic engineering strategies can also be used to improve the PUFA production, for example, optimization of the precursor supply and reduction of PUFA catabolism. The key precursor of PUFA synthesis is malonyl-CoA, improving the synthesis of malonyl-CoA and weakening competitive pathways such as the FAS pathway can make more malonyl-CoA direct to the PUFA synthase pathway. β-oxidation of fatty acids will lead to a decrease in the production of major PUFA. Some researchers chose to knock out the fadE gene when expressing PUFA synthase from Schizochytrium in E. coli. (Hayashi et al., 2016). These metabolic engineering strategies can also be applied to the natural hosts of PUFA synthase (Chi et al., 2021).