A review on the upstream production and downstream purification of mannosylerythritol lipids

Biosurfactants are natural compounds with remarkable surface‐active properties that may offer an eco‐friendly alternative to conventional surfactants. Among them, mannosylerythritol lipids (MELs) stand out as an intriguing example of a glycolipid biosurfactant. MELs have been used in a variety of sectors for various applications, and are currently commercially produced. Industrially, they are used in the pharmaceutical, cosmetic, food, and agricultural industries, based on their ability to reduce surface tension and enhance emulsification. However, despite their utility, their production is comparatively limited industrially. From a bioprocessing standpoint, two areas of interest to improve the production process are upstream production and downstream (separation and purification) product recovery. The former has seen a significant amount of research, with researchers investigating several production factors: the microbial species or strain employed, the producing media composition, and the production strategy implemented. Improvement and optimization of these are key to scale‐up the production of MELs. On the other hand, the latter has seen comparatively limited work presented in the literature. For the most part traditional separation techniques have been employed. This systematic review presents the production and purification methodologies used by researchers by comprehensively analyzing the current state‐of‐the‐art with regards the production, separation, and purification of MELs. By doing so, the review presents different possible approaches, and highlights some potential areas for future work by identifying opportunities for the commercialization of MELs.

and pour-point depressants (Shaban et al., 2020).In addition, their intermolecular interactions have permitted various biomedical applications, such as pharmaceutical formulations and drug delivery systems (Shaban et al., 2020).This has corresponded to a steady increase in the global surfactant market size, as shown in Figure 1a.Surfactants now constitute one of the largest categories of synthetic chemicals manufactured worldwide, with a global market size that is estimated to increase to $81.7 billion by 2030 (Surfactantsmarket, 2022).
Two primary categories of surfactants can be considered, that is, chemical and natural surfactants.These surfactants differ in their origin and properties.For instance, chemical surfactants are referred to as synthetic surfactants, since they are artificially synthesized from nonrenewable sources in chemical processes.This permits precise control of their structure and properties (Jesus et al., 2021).On the other hand, microbial biosurfactants, are produced by various microbial species, including various strains of bacteria, fungi, and yeasts.There is a further class of biosurfactants produced from biological materials, such as saponins.Microbial biosurfactants can be more environmentally friendly, biocompatible, and biodegradable in comparison to their synthetic chemical counterparts.Hence, there has been a significant research focus in the last two decades on the production and application of biosurfactants, since they can be produced from renewable sources and they may have useful properties (Soberón-Chávez, 2011).
Focussing on biosurfactants, different classes have been identified, primarily categorized based on the structure of their hydrophilic head group (Muthusamy et al., 2008).This includes low molecular weight (phospholipids, glycolipids, lipopeptides), as well as high molecular weight (polymeric and particulate) biosurfactants.Glycolipids have gained particular interest due to their particular properties and high production yields.Within this class one biosurfactant of particular interest are mannosylerythritol lipids (MELs).MELs are of interest due to their applicability in the biomedical and cosmetic industries (Jezierska et al., 2018).
The usage of MELs in the biomedical and cosmetic industries can be attributed to their potential for application as a moisturizing agent for dry and damaged skin.(Kitamoto et al., 2021;Yamamoto et al., 2012).Also, MELs have shown remarkable protective and reparative activity toward the development of human hair (Morita, Kitagawa, Yamamoto, Sogabe, et al., 2010;Morita, Kitagawa, Yamamoto, Suzuki, et al., 2010).
Due to such applications, research has been directed toward understanding the production and purification of MELs.This review intends to pull together and summarize the relevant work on MEL production and purification.
The first section of this review delves into the current state of research aimed at the upstream production of MELs.As shown in Figure 2, recent research has been primarily focussed on the upstream production of MELs.This section covers the production of MELs using various microbial strains, while exploring the primary factors that influence production.To understand the industrial scalability of these processes, a critical review is provided on research focussed on scaling MEL production from shake flasks to bioreactors.The limitations associated with the scalability of these processes are also discussed.
Finally, this section then builds on the production of MELs from various industrial waste streams as a route toward a circular economy.
Although upstream production is vital, significant downstream processing is required to produce a purified MEL product, should a purified product be required for the particular application in question.
Hence, the second section of this review focuses on the downstream processing-recovery and purification-of MELs.As shown in Figure 2, limited research has focussed on improving the downstream processing of MELs (only approximately 17% of papers reviewed).As a result, the efficient separation and purification of MELs remains a significant challenge limiting the scalability of MEL production.The current state of the downstream processing of MELs is evaluated in this review in an attempt to highlight the most significant challenges associated with the purification of MELs.This will help identify a clear F I G U R E 1 (a) The estimated total global surfactant market size (Surfactantsmarket, 2022).(b) The current biosurfactant market share of various industries (Foley et al., 2012).
focus for future research through the challenges and opportunities presented in the future perspective sections.

| MICROBIAL BIOSURFACTANTS
Biosurfactants are defined as surface-active compounds of biological origin and more specifically, this paper will consider microbial biosurfactants which are produced by various microbial strains.These compounds are produced as secondary metabolites by various bacteria, fungi, and yeasts (Sharma et al., 2021), and some can be produced from plants or animals.Five main categories of biosurfactants have been defined, presented in Figure 3, and are based on the structures of the hydrophilic head groups found in the compounds (Mulligan & Gibbs, 2004).While these main classifications are useful, there is additional complexity in biosurfactant structures-specifically, the hydrophobic fatty acid tail can be of variable length, giving rise to a set of congeners: structurally similar biosurfactants, with slightly differing properties.Further, additional congeners can occur when the hydrophilic head group's sugar moiety has slight structural variations-such as degree and position of acetylation, in the case of MELs.Regardless of these structural complexities, biosurfactants have found application in a number of products.| 855 Among these five categories, glycolipids are one of the most widely used categories.Glycolipids consists of a hydrophilic carbohydrate moiety covalently bonded to a hydrophobic lipid residue.They stand out with the demonstration of relatively high production yields, and their useful properties (Jezierska et al., 2018).
MELs are a type of glycolipid biosurfactant which have attracted significant attention, mainly due to their desirable surface-active and wetting properties which have made them ideal for various biomedical and cosmetic applications (Kitamoto et al., 2002).

| Structure of MELs
Analytical techniques used to analyze the structure of MELs Over the years, various analytical techniques have been employed to determine the structures variants of MELs when produced by different microbes under varying operating conditions.For example, nuclear magnetic resonance (NMR) is used to determine the molecular structure of the compounds.Using this technique, purified MEL extracts are dissolved in dimethyl sulfoxide-d6, and both 1 H NMR and 13 C NMR spectra are recorded (Fan et al., 2014).Also, LC-MS analysis is commonly employed to determine the molecular weight of the various structural variants which have been identified.
Finally, GC-MS analysis is typically employed to analyze the fatty acid composition of these compounds.Before the GC-MS analysis to obtain a fatty acid profile, the extracted compounds are methanolized by an alkaline method to synthesize the methyl ester derivatives, as Fan et al. (2014) described.

Structural diversity of MELs
Using the techniques described in the previous section, four naturally occurring structural variants of MELs (A, B, C, and D) have been identified.These variants, which all consist of a hydrophilic 4-O-β-Dmannopyranosyl-D-erythritol core, are distinguished based on the acylation patterns occurring at the C 4 and C 6 positions of the mannose (Bhattacharjee et al., 1970).MEL-A represents the diacylated form of the compound, while MEL-B and -C are monoacylated at the C 4 and C 6 positions, respectively.Finally, MEL-D represents the completely deacylated structure of the compound (Rau, Nguyen, Schulz, et al., 2005).Furthermore, hydrophobic fatty acid chains of variable length are covalently bonded to the hydrophilic sugar core (Arutchelvi et al., 2008;Rau, Nguyen, Schulz, et al., 2005).Di-acetylated MELs, containing two fatty acid chains at the C 2 and C 3 positions of the mannose in the core of the compound, are the most common naturally produced structural variant of MELs, however, the production of mono-and tri-acetylated MELs have been observed (Fukuoka, Morita, Konishi, Imura, Sakai, et al., 2007;Fukuoka et al., 2007b).
These different structural variants are produced in varying amounts by different microorganisms.However, the carbon source used during MEL production significantly affects the degree of saturation of the fatty acid chains found in the compounds (Arutchelvi et al., 2008).Table 1 summarizes the differences in the fatty acid profiles based on microbial species, carbon sources and the produced variant of MEL as reported in the literature.
MELs have shown to be effective for the treatment of dry or damaged skin and hair (Yamamoto et al., 2012).Yamamoto et al. (2012) observed that MELs exhibited high recovery rates (>80%) of sodium dodecyl sulfate damaged skin cells, comparable to that of natural ceramide, a known skin moisturizing agent.Takahashi et al. (2012) observed that MEL-C achieved a higher protective activity against oxidative stress than arbutin, known for its antistress and antiaging activity on skin.These ceramide-like properties also prove the applicability of MELs in the realm of hair care (Morita, Fukuoka, Imura, et al., 2015).
Another interesting property of MELs, is their potential as an antimicrobial agent which arises as a result of their antimicrobial activity toward certain gram-positive bacteria and fungi (Shu et al., 2019;Yoshida et al., 2015).Shu et al. (2019) observed that MELs inhibited the proliferation of the gram-positive bacteria B. cereus by damaging their bilayer cell membranes.The minimum inhibitory concentration and minimum bactericidal concentration of MELs toward these microbes were 1.25 and 2.5 mg/mL 1 , respectively (Shu et al., 2019).Furthermore, MELs have shown similar antimicrobial activity toward other food-borne and plant pathogens (Liu et al., 2020;Yoshida et al., 2015).These results suggest the potential of utilizing MELs as antimicrobial agents for food preservation and as green pesticides in agriculture.

| PRODUCTION OF MELS
This section details the methods employed in the literature for the production of MELs.Much of the literature is based on small scale experiments in shake flasks, considering media composition and conditions (carbon source, nitrogen source, trace elements, pH, temperature, oxygenation, and agitation).Some of the details of small-scale experiments can be used to inform larger volume ferments, which is discussed in Section 3.2.As has been highlighted, a significant expense in MEL production is the carbon source, and many researchers have therefore investigated the use of low-value or waste streams as substrates; this is discussed in Section 3.3.Finally, Section 3.4 considers how the structure of the produced MEL could be altered, post-fermentation, to produce structurally different MELs with different characteristics.
T A B L E 1 The type and characteristics of the mannosylerythritol lipids produced by various microorganisms (adapted from Arutchelvi et al., 2008).(Kitamoto, Akiba, et al., 1990).While most vegetable oils served as good carbon sources for the production of MELs, soybean oil led to a high reported MEL titer of 40 g/L.Following that work, even higher MEL concentrations (100 g/L) were obtain with hydrophobic carbon sources (either soybean or olive oil) with different strains of the genus Pseudozyma, as shown in Table 2 (Fukuoka, Morita, Konishi, Imura, Sakai, et al., 2007;Konishi et al., 2011;Morita et al., 2006).These differences in titers may be attributed to different species and strains, as well as carbon source and feeding strategies.
Nitrogen source Hewald et al. (2005) performed an analysis of the genes involved in the biosynthetic pathway for the production of MELs by Ustilago maydis.Although the organism did not grow effectively in the absence of nitrogen, it was observed that the expression of the glycosyl transferase (EMT1), a gene that is essential for the production of MELs in all strains, was enhanced under nitrogen limitation.These results indicated that the presence of nitrogen is required for microbial growth.Conversely, its absence is required for the production of MELs.
Various nitrogen sources have been used for the growth of MELproducing microbes, as shown in Table 2.After investigating the efficiency of many different nitrogen source for the growth, and the consequent production of MELs, of P. antarctica T-34, Kitamoto, Akiba, et al. (1990) concluded that NaNO 3 at a concentration of 2 g/L performed the best in terms of microbial growth.This led to a maximum product concentration of 40 g/L.Ammonium salts were identified as less suitable nitrogen sources, since their consumption decreased the pH of the culture medium, inhibiting the production of MELs by this strain (Kitamoto, Haneishi, et al., 1990;Rau, Nguyen, Schulz, et al., 2005).These results have led to the widespread utilization of NaNO 3 as a nitrogen source for the growth of MEL producing microbes.

Trace elements
Due to the important role played by trace elements in the activation of the enzymes involved in the metabolic synthesis of microorganisms, it can be deduced that trace elements are an important factor affecting the production of MELs (He et al., 2011;Kiran et al., 2011).
As reported by Kitamoto, Akiba, et al. (1990), the production of MELs During the study, different trace elements, including Fe 2+ , Fe 3+ , Ca 2+ , Mn 2+ , and Cr 2+ , were added individually to the substrate of a P.
aphidis DSM 70725 culture at a concentration of 0.1 mM.It was observed that MEL production could be enhanced with the addition of Fe 2+ and Fe 3+ to the production medium, while MEL production by this strain was inhibited in the presence of Ca 2+ , Mn 2+ , and Cr 2+ (Yang et al., 2023).These results indicate the important effects of trace elements on MEL production and supports further investigation into the topic, particularly since these effects may be species specific.

Hydrophilic carbon sources
Although the use of hydrophobic carbon source has led to extremely high product concentrations, often exceeding 100 g/L, their utilization significantly complicates the downstream processing required to produce a purified MEL product.This is due to the structural similarities between MELs and residual oils and fatty acids, resulting in the requirement for a complex downstream processing train to separate MELs from the residual oils.Furthermore, hydrophilic carbon sources have proven to be comparatively expensive substrates, hindering the economic performance of MEL production (Hubert et al., 2012;Morita et al., 2011).
Certain smut fungi, belonging to the genera Ustilago and Sporisorium, possess the ability to produce MELs from either hydrophobic or hydrophilic carbon sources.Sporisorium scitamineum NBRC 32730 showed efficient MEL production from sucrose, achieving a product concentration of 12.8 g/L.Although it was an order of magnitude lower than the best concentrations achieved by strains of the genus Pseudozyma cultivated on oils, is still a reasonable product titer (Alimadadi et al., 2018;Morita, Ishibashi, et al., 2009;Spoeckner et al., 1999).Although the utilization of hydrophilic carbon sources for the production of MELs demands significant sacrifices in product concentrations, their utilization both improves the production costs associated with substrates and simplifies the downstream processes required to produce a purified product.Furthermore, the successful application of hydrophilic carbon sources for the production of MELs has opened the door to the utilization of industrial The production of mannosylerythritol lipids in shake flasks.(Konishi et al., 2008;Morita, Ishibashi, et al., 2009).

Microorganism
Finally, Kim et al. (1999) showed that MELs can be produced from fatty acids as alternatives to vegetable oils.This production route is protected by a patent (Hee-Sik, 2011).

Fed-batch production
Other studies investigated the fed-batch supplementation of days with the addition of 40 g/L soybean oil, 20 g/L glucose, and 2 g/L yeast extract every 4 days (Konishi et al., 2008).In another study, it was shown that P. parantarctica JCM 11752 achieved maximum MEL titer on the 28th day of incubation, despite showing no cell growth from the 7th day.This culture was continuously feed with soybean oil.The results from this study resulted in a patented fed-batch process for the production of MELs by yeasts of the genus Pseudozyma from a medium consisting of oils and fats (Morita, Konishi, Fukuoka, Imura, Sakai, et al., 2008).These studies demonstrate that the productivity of MEL producing strains might be maintained for many weeks after cell growth has ceased, through the implementation of fed-batch systems.

Two-stage growth systems
Two-stage growth systems have been identified as a potential route of maximizing the yield of vegetable oils to product (Kitamoto et al., 1992).This production strategy, illustrated in Figure 4, includes an initial growth stage during which MEL-producing microbial biomass is grown to high cell concentrations on comparatively cheap hydrophilic carbon sources.In the subsequent production stage, the biomass can be fed on a substrate which contains vegetable oils, The potential set-up of a two-stage growth system which could achieve enhanced mannosylerythritol lipid production and optimal substrate utilization.
a growth and nitrogen limiting medium which consisted of 8% (v/v) soybean oil in distilled water.A final MEL concentration of 47 g/L was reported after 6 days.Importantly, this represented an 18% increase in the conversion of soybean oil to MELs compared to standard single-batch production (Kitamoto, Akiba, et al., 1990;Kitamoto et al., 1992).The time required to harvest and wash the cells after the growth stage proved to be the main challenge associated with this process.However, this washing process is not necessarily required.Instead, vegetable oil could just be added on top of the depleted growth medium.The implementation of two-stage growth systems has not been investigated for other MEL producing microbes.reported by Saika et al. (2018).Furthermore, it was determined that the MELs were acylated at the C 2 position, indicating that MAC2 is responsible for acylation at the C 3 position.This was the first evidence which supported the regioselectivity of the acyl transferase genes (Saika et al., 2018).Konishi and Makino (2018) showed that the disruption of the acetyl transferase (MAT1) in P. hubeiensis SY62 led exclusively to the formation of the deacetylated MEL variant MEL-D.

Genetic engineering for enhanced and targeted MEL production
Additional and ongoing work in this space will continue to improve strain characteristics.

| Scalability of microbial MEL production processes
The production of MELs in shake flasks has been widely reported.
This has led to significant progress in understanding the factors SY 16 in a 5 L jar fermenter.Initially, the substrate consisted of 15 g/L glucose and 15 g/L soybean oil as the carbon sources.The cultures were then supplemented with 70 and 100 g/L soybean oil after 1.8 and 4.3 days.This led to a maximum MEL concentration to 95 g/L (Kim et al., 2006).
These studies demonstrated that various MEL-producing strains could achieve product yields in the range of laboratory-scale cultivations up to a scale of 1000 L. However, to maintain productivity, the cultures had to be supplemented with large amounts of vegetable oils.This has significant negative effects on the economic performance of the process at these scales.Therefore, Mawani et al. (2021) investigated the production of MELs by P.
antarctica MTCC 2706 from an alternative carbon source in a 5 L jar fermenter.The optimized substrate consisted of 22 g/L of sweetwater, a by-product of the fat-splitting industry, and 7 g/L soybean oil as the carbon sources.A maximum MEL concentration of 21.5 g/L was achieved after 7 days.This represented a 185% increase in the MEL concentration compared to that achieved when the production was performed in shake flasks (Mawani et al., 2021), although is still significantly less than the best performance of this organism using other substrates.Regardless, this cultivation utilized significantly less soybean oil when compared to the other industrial scale cultivations.
T A B L E 3 The production of mannosylerythritol lipids in bioreactors.This demonstrated that the utilization of mixtures of hydrophilic and hydrophobic carbon source could significantly improve the production costs associated with the substrates used during the production of MELs.However, the utilization of these carbon sources may demand a sacrifice in the final product concentration.

| Utilizing industrial waste streams for the production of MELs
The potential of utilizing various industrial waste streams as feedstocks for the microbial production of MELs has been studied with the aim of improving their scalability and economic performance (Banat & Thavasi, 2019).This includes the use of waste streams with high concentrations of lipids as alternatives to vegetable oils, as well as waste streams with high concentrations of hydrophilic carbon sources, such as sugars, as presented in Table 4. Bednarski et al. (2004) investigated the potential of utilizing oil refinery waste streams as alternatives to vegetable oils to produce MELs.These wastes are consistently produced in large amounts at industrial lipid refineries and has therefore been identified as a reliable substrate which can potentially be utilized for the production of MELs (Lad et al., 2022).During the study, P. antarctica ATCC producing MELs from pentose sugars, achieving a maximum titer of 5.8 g/L (Faria, Santos, Fernandes, Fonseca, et al., 2014).In a followup study, Faria, Santos, Fernandes, Fonseca, et al. (2014) investigated the production of MELs by cellulosic materials, which were broken down into digestible sugars by commercial cellulolytic enzymes.
Maximum MEL concentrations of 4.5 and 2.5 g/L were obtained by implementing a fed-batch fermentation with a prehydrolysis step by P. antarctica PYCC 5048 (Faria, Santos, Fernandes, Fonseca, et al., 2014) (Faria et al., 2015).These studies showed the potential of producing MELs exclusively from cellulosic materials, although the final product titers are low in comparison to fermentations based on oils.This is alternative route to MEL production, and such feedstocks are abundant.However, pretreatment with cellulolytic enzymes is required to enhance MEL production, which is a significant expense.
Therefore, consequent research was aimed at the production of MELs from waste streams containing high concentrations of simple sugars rather than ones requiring pretreatment.respectively (Faria et al., 2023).These studies showed the potential of utilizing a combination of hydrophobic and hydrophilic waste streams to achieve enhanced microbial MEL production.Although the utilization of industrial waste streams containing exclusively hydrophilic carbon sources address the challenges associated with the downstream processing of MELs, these substrates are associated with significant decreases in both substrate yield and production rate, as shown in  | 867 determine if these sacrifices can be justified from an economical perspective.
As expected, the production of MELs from industrial waste streams required a significant sacrifice in product yield and quality when compared to pure carbon sources.However, the utilization of these wastes as substrates dramatically improve the costs associated with the substrates used for the production of MELs.As discussed previously, the high costs of vegetable oils have limited the scalability of MEL production.The utilization of waste streams as substrates could make the industrial scale production of MELs economically feasible.However, the variability in the compositions of these waste streams, as well as the decrease in the product quality makes the MELs produced from these wastes unsuitable for medical or cosmetic applications due to strict regulations within these industries.This would limit the applicability of MELs derived from waste streams, ultimately decreasing their value.

| Enzymatic post-modification for the production of particular structural variants of MELs
The implementation of enzymatic post-modification has been identified as a promising method for the production of unique  et al., 2013).
tsukubaensis, was deacylated after being reacted with immobilized lipase Novozyme 435 to yield deacylated MEL-D.The experiment was repeated for di-acylated MEL-A to yield mono-acylated MEL-C.
Therefore, it was concluded that the deacylation could only be achieved at the C 6 position using this methodology.However, in both cases, deacylation led to the formation of a MEL variant with increased hydrophilicity (Fukuoka et al., 2011).Similar results were obtained when these experiments were repeated on mono-acylated MEL-B, produced by S. scitamineum (Fukuoka et al., 2012).These findings indicate the potential of using enzymatic post-modification to alter the acylation patterns of existing MEL variants to produce new structural variants with unique properties and applications.
Furthermore, enzymatic post-modification provides greater control over the structural variants which are produced as opposed to microbial production, which usually leads to the formation of mixtures consisting of unique structural variants of MELs.

| Downstream processing of MELs
As elaborated in Section 3, significant efforts have been devoted toward improving the upstream production of MELs, leading to exceptional product concentrations of over 100 g/L (Konishi et al., 2011;Morita et al., 2006).Presently, TOYOBO (a MEL producing company based in Japan) has successfully commercialized the production of MEL-B (marketed as Ceramela™), while Nippon Fine Chemical Co.Ltd, (another Japanese company) has commercialized Phytopresome™ MEL which is a premix of MEL-B, hydrogenated lecithin and polyhydric alcohol (Kitamoto et al., 2021;Morita et al., 2014).However, downstream processing remains a challenge for industrial-scale production of MELs, since economic viability is tightly linked to the extent of downstream purification required (Shen et al., 2019).Given that over 60%−80% of the costs associated with the production of MELs are attributed to downstream processing, identifying efficient and cost-effective processes for the purification of MELs is crucial (Najmi et al., 2018) Several downstream processing techniques have been investigated for the laboratory-scale purification of MELs, including solvent extraction, adsorption, membrane filtration, column chromatography, foam fractionation, and decantation (Arutchelvi et al., 2008).The following sections will explore these downstream processing techniques, highlighting their strengths and weaknesses in the context of MEL recovery as has been reported in the literature.

| Solvent extraction
Solvent extraction is the most widely reported downstream process for the recovery of MELs, followed by column chromatography and HPLC for further purification (de Andrade et al., 2022;Rau et al., 2005a).A number of solvents have been used, including polar organic solvents such as chloroform, ethyl acetate, methanol, and butanol (Nascimento et al., 2023).Additional substrate extraction with non-polar solvents such as n-hexane, pentane, and methyl tertbutyl ether (MTBE) have been employed to remove residual oils and FFA from crude MEL extracts if a hydrophobic substrate such as soybean oil is used for MEL production (Hubert et al., 2012).Table 5 presents some of the solvent extraction systems that have been reported and these extraction techniques are shown in Figure 5.
Solvent extraction with an equal volume of ethyl acetate to culture media is the most widely utilized laboratory-scale downstream process for the extraction of MELs from culture suspensions.
After extraction, the organic layer is separated and evaporated, forming a MEL-rich concentrate (Morita et al., 2006).A similar extraction system, utilizing MTBE, was developed by Rau et al. (2005a), with the aim of isolating MELs produced by Pseudozyma aphidis DSM 14930, shown in Figure 5c.However, this procedure resulted in product losses of more than 90% and the formation of large amounts of toxic waste solvents (Rau et al., 2005b).Shen et al.
(2019) developed an extraction system that utilized a solvent mixture of methanol, water (pH 2), and n-hexane shown in Figure 5d.In this study, 90% of the MELs present in the culture suspension were recovered.
Although organic solvent extraction is an effective technique, and commonly applied for the recovery of MELs, there are some drawbacks associated with this process.For instance, this technique often involves the use of organic solvent mixtures which require complicated recycling processes as a result of the formation of stable azeotropes (Nascimento et al., 2023).Furthermore, these processes are associated with high solvent consumption rates which can be costly and potentially harmful to the environment (Shen et al., 2019).
For these reasons, the focus of research has been shifted toward integrating solvent extraction techniques with complementary separation methods such as membrane filtration or acid The downstream processing of mannosylerythritol lipids with the different solvent extraction techniques reported in literature (all the studies used soybean oil as the carbon source).F I G U R E 5 An overview of mannosylerythritol lipid separation and purification block flow diagrams: (a) Separation using an equal volume of ethyl acetate modified from Kim et al. (1999).(b) Separation and purification using ethyl acetate, methanol, and chromatography modified from Morita, Konishi, Fukuoka, and Imura, Kitamoto (2007).(c) Separation and purification through a series of extractions modified from Rau et al. (2005a).(d) Separation and purification through a series of extractions using a methanol/water/n-hexane extraction system modified from Shen et al. (2019).
precipitation.This circumvents the need for the large volumes of toxic solvent mixtures required during the downstream processing of MELs and mitigates the associated recycling and disposal challenges.
Section 4.1.2will delve further into exploring this integrated approach.
A significant advancement in the downstream processing of MELs was highlighted which introduces a method for purifying hydrophilic MELs through acid hydrolysis and subsequent n-hexane extraction (Long et al., 2021).However, acid hydrolysis introduces challenges related to handling and disposal of the acid reagents and careful control of the acid hydrolysis reaction to prevent undesirable side reactions or degradation of the desired MELs.Therefore, the scalability, cost-effectiveness, and environmental impact of the acid hydrolysis step should be thoroughly evaluated in the context of large-scale industrial production.
Another patent presents an alternative approach to the recovery and purification of MELs produced by Pseudomonas microorganism cultures (Rau et al., 2002).This method employs specialized equipment, such as a large continuous centrifugal separator, thus obviating the requirement for organic solvent extraction.The technique involves the creation of a water-insoluble aggregation through the coexistence of MELs and fatty acids within an aqueous medium.Subsequent separation of the MELs and the fatty acid occurs from this aggregation.
Unlike the current approaches, an integrated fermentation methods could be vital for MEL separation and purification especially in cases where pretreatment of lignocellulosic biomass releases byproduct that inhibits microbial growth and limits productivity (Faria, Santos, Ferreira, Marques, et al., 2014;Santos et al., 2019;Zhang et al., 2015).Teke et al. (2022a) highlighted the efficacy of solvent extraction (using experimental and computational modeling) with a semi-partitioning bioreactor (SPB) configuration that merges fermentation and separation within one unit (Teke et al., 2022a(Teke et al., , 2022b;;Teke et al., 2023;Teke & Pott, 2021).This departure from the conventional setup incorporates solvent extraction within the SPB offering a novel approach with potential application in MEL production, streamlining the recovery process while enhancing yield and recovery.

| Membrane filtration
Membrane filtration is another approach that has been explored for the extraction of MELs from culture suspensions.Different membrane types have been investigated such as polysulfone (PS) and polyethersulfone (PES) membranes and regenerated cellulose membranes (Jauregi & Kourmentza, 2018).However, de Andrade et al.
(2022) reported that ceramic membranes hold greater promise as an alternative membrane material to supplant the prevailing polymeric membranes due to the enhanced resistance of ceramic membranes to organic solvents and their extended operational lifespan.
In the usage of ultrafiltration units, Andrade et al. (2017) reported that 100 KDa MWCO ultrafiltration membranes in a 20 mL centrifugal device led to the recovery of about 80% of the MELs produced by P. tsukubaensis, cultivated on cassava wastewater.
The results remained unchanged after the ultrafiltration procedure was scaled up (up to 500 mL) using a crossflow filtration unit (Andrade et al., 2017).The complexities of scaling up such a separation have not been fully investigated (Table 6).
Membrane filtration is particularly appealing because no harmful residues are produced, and it may be relatively simple to scale up because of its simple operation and low energy consumption.
However, membrane processes are limited by membrane fouling which results in loss of productivity due to reduced equipment efficiency, increased cost of cleaning, and microbial contamination concerns (Charcosset, 2006, Vicente et al., 2021).Moreover, the process is not very selective as separation is solely based on compound size (Jauregi & Kourmentza, 2018), such that all small molecules will be collected with the MELs.
A study by (Nascimento et al., 2023)    methods can be challenging due to the interference of biomass with downstream separation processes-biomass will also be collected along with precipitated MELs.Consequently, additional cell separation techniques may be required before implementing an in situ precipitation process (Teke, Tai, et al., 2022).In the context of the study by Kitamoto, Akiba, et al. (1990), observed coprecipitation of MELs with yeast cells and oil at reduced pH levels may be linked to their association with other components like proteins, oil, and FFA in the fermentation broth, rather than a direct response to pH changes.
While acid precipitation aids in coprecipitation and subsequent separation of MELs alongside these constituents, its direct influence on precipitating MELs due to their non-ionic nature may be limited.

| Foam fractionization
Foam fractionation has also emerged as a promising method for the purification of biosurfactants due to its cost-efficiency and the possibility of continuous product removal.Foam fractionation is performed by employing the tendency of biosurfactants to adsorb onto air bubbles that rise through fermentation vessels (Venkataraman et al., 2021).The biosurfactant-rich foam can then be collected from the top of the vessel and directed toward a foam collection vessel.Furthermore, these collection vessels can be fitted with different types of inserts onto which biosurfactants can adsorb and be collected (Oraby et al., 2023).Figure 7 illustrates a setup of a foam fractionation column for the continuous purification of cellobiose lipids, a glycolipid biosurfactant produced from hydrophilic carbon sources by S. scitamineum.
The feasibility of employing foam fractionation hinges upon several critical factors.Primarily, the surface activity of MELs plays a pivotal role, as their ability to adsorb at the air-liquid interface dictates the efficacy of separation.As such, the implementation of foam fractionization is only feasible in cultures grown on hydrophilic carbon sources due to the antifoaming properties of hydrophobic carbon sources such as vegetable oils (Rau et al., 2005a).Therefore, to implement continuous foam fractionization for the purification of MELs, processes that utilize exclusively hydrophilic carbon sources would need to be developed.Moreover, biomass may also be withdrawn during foam fractionation, reducing the effective biomass loading and therefore the volumetric productivity of the reactor over time (Jia et al., 2020).
The only study which has investigated the purification of MELs by foam fractionation was performed by Andrade et al. (2017), although many studies have examined the technique for other biosurfactants such as lipopeptides.In the Andrade study, it was determined that the foam at the top of a reaction vessel of a P. tsukubaensis culture cultivated on sweetwater contained MELs at a comparatively low concentration of 1.3 g/L (Andrade et al., 2017).

| FUTURE PERSPECTIVES
The first phase of MEL-related research was focused on identifying microbial strains capable of efficiently producing MELs, as well as understanding the factors which affect their production.These studies led to a good understanding of the optimal conditions for labscale MEL production in shake flasks.It has been observed that vegetable oils, most notably soybean oil, serve as the optimal carbon F I G U R E 6 An illustration of the proposed downstream process which integrates solvent selective dissolution and organic solvent nanofiltration (OSN) for the purification of mannosylerythritol lipids modified from Nascimento et al. (2023).
source for the production of MELs from microbial strains of the genus Pseudozyma.However, several problems arise from the utilization of vegetable oils as substrates for the production of MELs.
This includes the high substrate-related production costs, as well as increased complexity in the downstream processes required to produce a purified product due to the structural similarities between MELs and residual oils (Hubert et al., 2012;Morita et al., 2011).
Therefore, to develop a cost-effective bioprocess for the industrial scale production of MELs, the focus of future research should be aimed at finding solutions to these challenges.
As a first approach to solving these problems, research could be aimed at identifying alternative carbon sources for the production of MELs.It has been observed that certain smut fungi, including U. maydis and S. scitamineum, possess the ability to produce MELs in the absence of vegetable oils.This could lead to the identification of abundant and cost-effective substrates which do not inhibit the downstream processing of MELs in the same way as vegetable oils.
Alternatively, to decrease the high operating costs associated production of MELs from vegetable oils, the research could be aimed at maximizing the microbial conversion of vegetable oils to MELs.
This could be achieved by optimizing the operating conditions of microbial MEL production processes, or by implementing more advanced fermentation protocols or microbiology techniques such as fed-batch fermentation systems or genetic engineering.
An estimated 60%−80% of the overall processing costs associated with the production of MELs can be attributed to downstream processing (Shen et al., 2019).Consequently, to

F
I G U R E 2 (a) The focus of MEL-related research studies which were published from 2010 to 2022.(b) The percentage of research articles focused on either upstream or downstream process development which were published from 2010 to 2022.This data was obtained with a scopus search on the main research data bases.MEL, mannosylerythritol lipids.F I G U R E 3 The classification of microbial biosurfactants according to their molecular weight and the structure of their hydrophilic head groups.VALKENBURG ET AL.

|
Microbial production of MELs 3.1.1| Factors affecting microbial MEL production by P. antarctica T-34 was improved with the addition of 1 g/L yeast extract as a source of minerals and trace elements.Therefore, yeast extract has been consistently supplemented to MEL production mediums in all subsequent publications, with limited research aimed at understanding the effects of individual trace elements on MEL production.Yang et al. (2023) undertook the only investigation on the effect of individual trace elements on the production of MELs.
L E 2 (Continued) additional hydrophilic carbon sources on MEL production.Morita et al. (2006) observed that MEL production could be increased to 142 g/L by supplementing a P. rugulosa NBRC 10877 culture with 80 g/L soybean oil and 20 g/L erythritol every 7 days for 28 days.This remains the highest MEL production yield yet achieved in the literature.Similar results were obtained for P. hubeiensis KM 59 byKonishi et al. (2008).During batch fermentation, a maximum MEL concentration of 21.8 g/L was achieved after 4 days.But MEL concentration could be increased to a maximum of 74.3 g/L after 16 which have been shown to give high concentration of MELs.This may be an approach to improving the economic viability of MEL production processes by achieving improved conversion of vegetable oils to MELs.This was shown by Kitamoto et al. (1992), who produced MELs from resting cells of P. antarctica T-34.During the growth stage, biomass was produced by utilizing glucose as the exclusive carbon source.During the production stage, the biomass was transferred to Recent advancements in understanding the genes involved in the biosynthetic pathway of MELs have opened the door to the development of mutant strains with enhanced MEL production abilities.Tran et al. (2022) managed to increase the MEL yield achieved by Candida sp.SY-16 by 31.6% compared to the wild strain by overexpressing the two acyl transferases (MAC1 and MAC2) and the acetyl transferase (MAT1)(Tran et al., 2022).Saika et al. (2017) successfully enhanced the production of MELs by Pseudozyma tsukubaensis by transferring two lipase genes (PaLIPA and PaLIPB) from P. antarctica T-34 into the modified strain.These manipulations lead to a significant increase in the achieved substrate yield.The increased conversion of oil into MELs was attributed to an increased lipase activity, indicating that MEL production in some microbial strains might be limited by their ability to synthesize fatty acids from vegetable oils(Saika et al., 2017).These molecular biological techniques have been employed to produce strains possessing targeted MEL production abilities.The disruption of the acyl transferase gene, MAC2, in P. tsukubaensis JCM 16987 led to the production of a monoacylated MEL variant, as affecting microbial MEL production, as discussed in the preceding sections.However, less research has been aimed at producing MELs in bioreactors, or at larger volumes.The studies that are available which investigated the scalability of MELs production are summarized in Table3.Rau et al. (2005b) reported on the production of MELs in a stirred bioreactor.The study investigated the fed-batch cultivation of two P. aphidis strains in a 72 L jar fermenter.The strains were initially grown on 67.8 g/L soybean oil before being supplemented with 37 g/L soybean oil at a rate of 0.3 g/L/h.This led to maximum MEL concentrations of 70 g/L for P. aphidis DSM 70725 and 90 g/L for P. aphidis DSM 14930.Uncontrolled foaming was identified as a challenge during all cultivations.To minimize foaming, the stirrer speed and rate of aeration were continuously adjusted(Rau et al., 2005a).Rau et al. (2005b) then aimed to increase the maximum cell concentration of P. aphidis DSM 14930 by including an initial growth-stage with 23 g/L glucose and 17 g/L soybean oil as the carbon sources.To further promote cell growth, the culture was supplemented with solutions of 285 g/L glucose, 16 g/L nitrate, and 14 g/L yeast extract at a rate of 125 mL/h between 1.75 and 2.8 days.By including this growth stage, an 81% increase in the maximum cell concentration was achieved.After supplementation with 126 g/L soybean oil, a maximum MEL concentration of 165 g/L was achieved(Rau et al., 2005a).Goossens et al. (2016) investigated the fed-batch production MELs by P. aphidis DSM 70725 in a 22 L stirred bioreactor.Rapeseed oil at a concentration of 74.3 g/L was initially provided as the sole carbon source.A final MEL concentration of 69 g/L was achieved by supplementing the culture with 55.4 g/L rapeseed oil after 8.9 days(Goossens et al., 2016).Yang et al. (2023) then managed to improve MEL production by this strain by optimizing the levels of Fe 2+ and Fe 3+ in the culture medium and utilizing the continuous supplementation of soybean oil to the 1000 L bioreactor both as an additional substrate and antifoam agent.Ultimately, a maximum MEL concentration of 76.6 g/L was demonstrated(Yang et al., 2023).Kim et al. (2006) investigated the batch and fed-batch cultivation of Candida sp.
20509 cultures were supplemented with either 50−120 g/L soap stock or 20−50 g/L post-refinery fatty acids.It was observed that maximum MEL concentrations of 13.4 and 10.4 g/L were achieved by adding 100 g/L soap stock and 40 g/L post-refinery fatty acids, respectively.These titers were an order of magnitude less than the best titers achieved on pure carbon sources.In a different study,Dzięgielewska and Adamczak (2013) investigated the production of MELs from different hydrophobic waste streams.This included three different glycerol fractions which were derived from biodiesel at 100 g/L, as well as free fatty acids (FFA), post-refining fatty acids from plant oils, and soap stock at 200 g/L by P. antarctica ATCC 28323 and P. aphidis DSM 70725.The P. antarctica ATCC 28323 and P. aphidis DSM 70725 cultures reached maximum MEL concentrations of 107.2 and 77.7 g/L with the addition of post-refining fatty acids and soap stock, respectively(Dzięgielewska & Adamczak, 2013).These findings showed the potential of replacing vegetable oils with industrial lipid wastes.Niu et al. (2019) then investigated the production of MELs fromwaste cooking oil by P. aphidis ZJUDM34.It was reported that a maximum MEL concentration of 55 g/L was achieved, comparing well with the maximum MEL concentration of 61 g/L achieved by growing this strain on soybean oil(Niu et al., 2019).However, waste cooking oils do not address the challenges associated with the downstream processing of MELs but rather contaminate the product with additional compounds which need to be purified.Finally, there are significant challenges associated with the collection of waste cooking oils to ensure a steady supply of substrate for the process.The microbial production of MELs from industrial waste streams containing exclusively hydrophilic carbon sources has also been investigated with the aim of simplifying the downstream processes T A B L E 3 (Continued) final MEL concentrations reported in each study.T A B L E 4 The production of mannosylerythritol lipids from industrial waste streams or alternative substrates.a purified product.Faria, Santos, Fernandes, Fonseca, et al. (2014) investigated the growth and MEL production of Pseudozyma sp. with pentose sugars.It was observed that P. antarctica PYCC 5048 showed the highest potential for Bhangale et al. (2013) produced MELs at a concentration of 5.61 g/L by growing P. antarctica ATCC 32657 on 14% (w/v) honey, whileMadihalli et al. (2020) produced MELs at a concentration of 3.85 g/L by growing P. antarctica JCM 10317 on coconut water.Finally,Andrade et al. (2017) produced MELs by growing P. tsukubaensis on cassava wastewater in a 3 L bioreactor.Finally, recent research has been aimed at the production of MEL from a combination of hydrophobic and hydrophilic waste streams.Nascimento et al. (2022) successfully produced MELs at a concentration of 13.98 and 13.1 g/L by cultivating P. antarctica PYCC 5048 and P. aphidis PYCC 5535 on a combination of 30% (v/v) cheese whey and 20 g/L waste cooking oil(Nascimento et al., 2022).Similarly,Faria et al. (2023) investigated the production of MELs by P. antarctica PYCC 5048 and P. aphidis PYCC 5535, initially grown on 40 g/L glucose before being supplemented with 20 g/L waste cooking oil.It was found that P. antarctica PYCC 5048 and P. aphidis PYCC 5535 achieved competitive MEL titers of 12.6 and 10 g/L, final MEL concentrations reported in each study.VALKENBURG ET AL.
structural variants of MELs.Enzymatic post-modification focusses on the production of new MEL homologs by treating existing homologs with commercial enzymes.Fukuoka et al. (2007b) successfully synthesized a tri-acylated MEL variant by treating di-acylated MEL-A, produced by P. antarctica T-34, with Novozyme 435, a commercial lipase enzyme.Similarly, Recke et al. (2013) managed to modify diacylated MEL-A and mono-acylated MEL-B with fatty acids isolated from other glycolipids, including rhamnolipids and sophorolipids.This led to the formation of tri-acylated MEL variants with unusual fatty acid chains at the C 1 position.These structural variants exhibited enhanced surface-active and antimicrobial properties (Recke addressed the challenge of separating molecules with comparable molar masses by developing a recovery and purification process which integrates solvent-selective dissolution to separate triacylglycerols (TAGs) from MEL, followed by organic solvent nanofiltration to remove smaller molar mass lipid derivatives such as FFA and mono-or diacylglycerols.The proposed process, depicted in Figure6, offers a dual benefit for the purification of MELs produced from lipid-based substrates by effectively T A B L E 6 The downstream processing of mannosylerythritol lipids with the different membrane filtration techniques reported in literature.
4.1.3| Acid precipitationAcid precipitation has also been proposed to improve the performance of solvent extraction techniques by reducing the volume of solvents required during the extraction of MELs.Kitamoto, Akiba, et al. (1990) the observed that MELs precipitated, along with proteins, residual fatty acids, and residual oil when the fermentation broth of Pseudozyma antarctica T-34 was reduced to a pH of 3.This precipitate could be easily recovered by solid−liquid separation methods, such as centrifugation.It should be noted that precipitation efficiently produce MELs at an industrial-scale, efforts should be aimed at improving the downstream processing of MELs by developing more efficient separation techniques.Furthermore, efficient production of MELs at an industrial scale necessitates not only effective recovery of MELs from the fermentation broth, but also a robust focus on advanced purification techniques to meet the stringent quality requirements for the various industrial applications of MELs.Hence alongside integrating efficient separation techniques like in situ product recover, prioritizing and refining purification strategies remains essential in ensuring the quality and purity of MELs in large-scale production.6 | CONCLUSIONS This review has summarized the research in MEL production.From this study, it is evident that a substantial amount of work has been demonstrated in the upstream production, using strains from the genera Pseudozyma, Ustilago, Candida, and Sporisorium.Different factors affecting MEL productivity have been investigated in line with upstream production, including media optimization, choice of carbon source, and nitrogen source.Although improved yields and product concentration have been demonstrated, more research is needed to develop a scalable upstream production strategy.In terms of downstream separation and purification, further research would be beneficial; although, currently, there is successful commercial production of MELs with current methodologies.Solvent extraction is widely used, often with two solvents-to initially separate the MELs from the media and then to clean the crude MEL extract of residual oil and fatty acids.There have been studies examining alternative downstream process options, such as membrane separation, or foam F I G U R E 7 The setup of a foam fractionation column for the continuous separation of cellobiose lipids.The foam which forms during fermentation flows through the foam column to a foam collecting vessel with inserts on which biosurfactants can accumulate.The foamate can then be recirculated back to the fermenter with fresh growth medium.Modified from Oraby et al. (2023).fractionation.Regardless, the field of MEL development is a growing one, and one which warrants further work toward growing industrial application.
The first phase of MEL-related research was aimed at identifying MEL producing microbes.Consequently, it has been established that MELs are produced by a variety of basidiomycetous yeasts belonging to the genera Pseudozyma and Candida as well as the smut fungi of the genera . The results from these studies then prompted Faria

Table 4 .
Therefore, further research is required to T A B L E 4 (Continued)