Lipids of Clinically Significant Mycobacteria

Abstract

Mycobacterial pathogens, such as Mycobacterium tuberculosis, Mycobacterium leprae, and the “non-tuberculous mycobacteria,” are characterised by the presence of a robust cell envelope that influences pathogenicity but also provides targets for antimycobacterial drugs. A key cell envelope component is a lipid rich mycobacterial outer membrane (MOM) involving “free” lipids interacting with a “bound” layer of high molecular weight (70–90 carbons) long-chain fatty acids, the mycolic acids. The principal free lipids range from the highly hydrophobic phthiocerol dimycocerosate waxes through related glycosyl phenolphthiocerol dimycocerosates (phenolic glycolipids) to strongly antigenic hydrophilic glycopeptidolipids and lipooligosaccharides. The mycolic acids are covalently linked to a special arabinogalactan polysaccharide, which is attached to the basal cell wall structural peptidoglycan. The specialized mycobacterial inner membrane (MIM) is rich in characteristic glycophospholipids, the phosphatidylinositol mannoside family, which are extended into important lipopolysaccharides, the lipomannans, and lipoarabinomanns. The origins and structures of the full range of lipids, from clinically significant mycobacteria, are detailed.

1 Introduction

1.1 Mycobacterial Pathogens

Mycobacterium tuberculosis and Mycobacterium leprae are dominant pathogens, infecting the human race during the past millennia. Recently, a distinct leprosy agent has been characterized and provisionally labelled “Mycobacterium lepromatosis” (Han et al. 2008; Singh et al. 2015), and even more recently, another uncultivable related animal pathogen, “Mycobacterium uberis” has been reported (Benjak et al. 2018). An opportunist human pathogen, Mycobacterium haemophilum, has strong phylogenetic links with leprosy bacilli (Besra et al. 1991; Donoghue et al. 2018). Currently, there are seven recognizable clades of M. tuberculosis sensu stricto, but there are also a number of closely-related mycobacteria that have been grouped together in the “M. tuberculosis complex” (Niemann and Supply 2014). This complex also includes animal pathogens, Mycobacterium bovis being the best documented (Smith et al. 2006). A distinct smooth morphology tubercle bacillus, labelled “Mycobacterium canettii,” is regarded as being on the periphery of the M. tuberculosis complex, but with likely ancestral links (van Soolingen et al. 1997; Supply and Brosch 2017). Other significant mycobacterial pathogens are included in the so-called nontuberculous mycobacteria (NTMs) (Falkinham 2015; Claeys and Robinson 2018). NTMs have been comprehensively reviewed by Turenne (2019) who emphasized the absolute necessity for validating novel taxa, such as “M. lepromatosis,” “M. uberis,” and “M. canettii,” before referring to them as established species.

Mycobacterium kansasii and Mycobacterium gastri are opportunists that produce tuberculosis-like human disease (Philley et al. 2016; Johnston et al. 2017). A very broad taxon, infecting humans and animals, was often referred to as the Mycobacterium avium-Mycobacterium intracellulare-Mycobacterium scrofulaceum (MAIS) complex, but more recently, it has been simply termed the Mycobacterium avium complex (MAC) (Falkinham 2015; Philley et al. 2016; Turenne 2019; Verma 2019). Humans and fish can suffer from Mycobacterium marinum infections (Johnson and Stout 2015) and a closely related pathogen, Mycobacterium ulcerans, causes human “buruli ulcers” (Sizaire et al. 2006; Chany et al. 2013). Mycobacterium fortuitum and related members of the original “M. fortuitum complex,” such as Mycobacterium peregrinum, are opportunist human agents (Falkinham 2015; Prevots and Marras 2015). Cattle diseases result from infection with related Mycobacterium farcinogenes and Mycobacterium senegalense (Chamoiseau 1979; Turenne 2019). The Mycobacterium abscessus complex consists of a group of rapidly growing, inherently drug-resistant, nontuberculous mycobacteria, differentiated into three subspecies, M. abscessus, M. massiliense, and M. bolleti, responsible for a wide spectrum of skin and soft-tissue diseases, central nervous system infections, bacteremia, and ocular and other infections (Medjahed et al. 2010; Tortoli et al. 2018). Mycobacterium abscessus is a serious pathogen for patients with cystic fibrosis (Parkins and Floto 2015; Martiniano and Nick 2015; Skolnik et al. 2016; Ryan and Byrd 2018) and closely related Mycobacterium chelonae can infect heart valve transplants (Wallace 1994). Opportunist pathogens include Mycobacterium gordonae, Mycobacterium malmoense, Mycobacterium mucogenicum, Mycobacterium szulgai, and Mycobacterium xenopi (Rombouts et al. 2009; Prevots and Marras 2015).

1.2 Mycobacterial Lipids

As a consequence of having two distinct membrane organelles in the cell envelope (Minnikin 1982; Brennan and Nikaido 1995; Hoffmann et al. 2008; Zuber et al. 2008; Daffé and Reyrat 2008; Daffé and Zuber 2014; Minnikin et al. 2015), mycobacterial lipids are remarkably diverse. The cytoplasmic membrane is a bilayer based on polar lipids, but including high proportions of characteristic glycophospholipids known as phosphatidylinositol mannosides (PIMs). This distinct composition warrants recognition of this organelle as a special mycobacterial inner membrane (MIM) (Bansal-Mutalik and Nikaido 2014; Minnikin et al. 2015). The inner leaflet of the mycobacterial outer membrane (MOM) is based on a monolayer of long-chain mycolic acids (MAs), covalently bound to terminal arabinose units presented by an arabinogalactan-peptidoglycan macromolecule. A wide variety of free lipid types are considered to interact with the MOM platform, mainly contributing to the MOM outer leaflet. The origins and structures of the individual lipid classes are detailed in the following sections.

The special character of mycobacterial lipids was revealed by Rudolf J. Anderson and colleagues in the 1920s and 1930s (see Anderson 1939, 1940, 1941, 1943). Methods were not available to enable the determination of precise chemical structures, though significant progress was made by Edgar Lederer and coworkers (see Asselineau 1966; Lederer 1967, 1971). Technological advances in infra-red spectrometry and chromatography allowed discriminatory examination of mycobacterial lipid fractions (Smith et al. 1954, 1957, 1960a, b; Smith and Randall 1965). The mainly glycolipid fractions were designated as characteristic “Mycosides” (Smith et al. 1960b). Mycosides A, B, and G are phenolic glycolipids from M. kansasii, M. bovis, and M. marinum, respectively (Smith and Randall 1965; Navalkar et al. 1965). The labels mycosides C and D were applied to families of glycopeptidolipids, expressed by the MAIS complex (Smith et al. 1957, 1960a, b; Smith and Randall 1965). Mycosides F, from M. fortuitum (Fregnan et al. 1961), were later shown to be acyl trehaloses (Gautier et al. 1992; Hamid et al. 1993; Sempere et al. 1993).

A number of substantial early reviews included general surveys of mycobacterial lipids (Anderson 1939, 1940, 1941, 1943; Asselineau 1966; Lederer 1967, 1971; Goren 1972; Ratledge 1976; Barksdale and Kim 1977; Asselineau and Asselineau 1978; Goren and Brennan 1979; Minnikin and Goodfellow 1980; Minnikin 1982). Later, highly informative reviews appeared (McNeil et al. 1989; Minnikin 1991; McNeil and Brennan 1991; Brennan and Nikaido 1995; Daffé and Draper 1998; Daffé and Lemassu 2000; Minnikin et al. 2002; Draper and Daffé 2005; Kremer and Besra 2005). Recent general surveys are available (Daffé and Reyrat 2008; Daffé and Zuber 2014; Jackson 2014; Minnikin et al. 2015; Lowary 2016; Chiaradia et al. 2017; Singh et al. 2018).

2 Mycolic Acids (MAs) and Conjugates

Mycolic acids (MAs) are 3-hydroxy-2-alkyl-branched long-chain fatty acids produced by mycobacteria and related taxa, the so-called “mycolata” (Chun et al. 1996). Anderson and co-workers identified mycolic acids, in pioneering studies (Stodola et al. 1938; Anderson 1941). The essential 3-hydroxy-2-alkyl-branched nature of TB mycolic acids was defined by Asselineau and Lederer (1950), explaining how, on pyrolysis, cleavage produces straight-chain hexacosanoate and long-chain aldehydes. In accord with accepted chemical usage, the long-chain aldehydes were designated as “meromycolic aldehydes” from the Ancient Greek méros, denoting a part or portion (Morgan and Polgar 1957). Distinct structural types are found in mycobacteria with variations in the meromycolate chain (Etémadi 1967a, b; Minnikin 1982; Daffé et al. 1983; Marrakchi et al. 2008, 2014; Verschoor et al. 2012; Minnikin et al. 2015; Daffé et al. 2017).

Cyclopropane rings were identified in mycobacterial MAs (Gastambide-Odier et al. 1964), with later distinction between cis- and trans-cyclopropane rings (Minnikin 1966; Minnikin and Polgar 1967a, b, c). MAs with no meromycolate oxygen functions are termed α-mycolic acids (~C₇₀–C₉₀); smaller (C₆₆ and C₆₈) α′-mycolates are also encountered. Ketomycolates are widespread, often accompanied by methoxy- or wax-ester mycolates, all in the C₈₀–C₉₀ size range. These oxygenated MAs have either cis- or trans- double bonds or cyclopropane rings, the trans- unsaturations being accompanied by an adjacent methyl branch. Minor amounts of extended oxygenated MAs are encountered (Qureshi et al. 1978; Takayama et al. 1979; Watanabe et al. 2002), currently being labelled XL-MAs (Slama et al. 2016). Epoxymycolates (Daffé et al. 1981b; Minnikin et al. 1980, 1982b) and ω – 1 methoxymycolates (Luquín et al. 1987, 1990, 1991) have a more limited distribution in certain rapid-growing mycobacteria. MAs are mainly covalently bound structural MOM components, but are also found in free lipid conjugates (Asselineau 1966; Minnikin 1982; Brennan and Nikaido 1995; Daffé and Zuber 2014; Marrakchi et al. 2008, 2014; Minnikin et al. 2015; Daffé et al. 2017).

2.1 Mycolic Acids (MAs)

The major MAs, typical of the M. tuberculosis complex, are shown in Fig. 1A. The di-cis-cyclopropyl α-MAs of M. tuberculosis have a characteristic short 11- and 13- carbon chain between the 3-hydroxy group and the proximal cis-cyclopropane (Fig. 1A), contrasting with other mycobacteria that usually have 17-carbon chains (Fig. 1B–D). Additionally, TB α-MAs have a 20 carbon terminal chain, with the others having 18 carbons, and the chain in 2-postion has 24 rather than 22 carbons (Fig. 1B–D). Such subtle differences in functional group location influence conformational behavior, with TB α-MAs packing in a more extended manner than α-MAs from M. kansasii (Villeneuve et al. 2010; Minnikin et al. 2015). It is appropriate, therefore, to assign a distinguishing label “TB-α” to the α-MAs from members of the M. tuberculosis complex (Fig. 1A: a, b). In many publications, such importance of MA alkyl chain length is ignored; these errors are often perpeptuated, as exemplified by Marrakchi et al. (2014) and Daffé et al. (2017).

It is notable that in the major trans-cyclopropyl methoxy and keto MAs, from the M. tuberculosis complex, the adjacent methyl branch is distal (Fig. 1A: e, i). In the minor trans-alkene methoxy and keto MAs (Fig. 1A: f, j), however, the adjacent methyl branch is proximal (Fig. 1A: e, i). Similar proximal methyl trans-alkene keto MAs are found in M. avium (Fig. 1D) and M. marinum, where they are accompanied by corresponding methoxy MAs (Fig. 1E). The MAs of M. kansasii (Fig. 1B), M. leprae (Fig. 1C), and M. avium (Fig. 1D) have mainly a 22-carbon chain in 2-position. In M. kansasii, α-, methoxy-, and keto MAs provide an overall pattern similar to M. tuberculosis, with the above mentioned subtle differences in α-MA structure (Fig. 1B).

The M. leprae MAs are similar to those of M. kansasii, but lack a methoxy component (Fig. 1C) (Minnikin et al. 1985a); an earlier report indicated a minor methoxymycolate (Daffé et al. 1981a). The M. avium complex has α- and keto-MAs similar to those of M. kansasii, but methoxy-MAs are replaced by wax-ester MAs (Fig. 1D). The MAs from M. marinum and M. ulcerans (Fig. 1E) lack cyclopropanation in the oxygenated MAs, the methyl branch, adjacent to trans-alkenes in methoxy- and keto-MAs, being on the proximal side (Daffé et al. 1991a). It is also notable that, as in TB α-MAs (Fig. 1A), the dicyclopropyl α-MAs of M. marinum have a short 13-carbon chain between the 3-OH group and the proximal cyclopropane (Fig. 1E). Conformational investigations should be made on all the M. marinum MA classes as in previous studies (Villeneuve et al. 2005, 2007, 2010, 2013). It would be interesting to see if the α-MAs of M. marinum perform in an analogous manner to the TB α-MAs and if an adjacent proximal methyl group facilitates folding of M. marinum trans-alkene methoxy and keto MAs in the same way that a distal methyl branch assists folding of a trans-cyclopropane oxygenated MAs from the M. tuberculosis complex (Villeneuve et al. 2013). Several isolates of M. ulcerans appear to lack methoxymycolates (Daffé et al. 1984).

Fig. 1A

Mycolic acids (MAs) of the Mycobacterium tuberculosis complex

Fig. 1B

Mycolic acids (MAs) of Mycobacterium kansasii and Mycobacterium gastri

Fig. 1C

Mycolic acids (MAs) of Mycobacterium leprae

Fig. 1D

Mycolic acids (MAs) of the Mycobacterium avium complex (MAC)

Fig. 1E

Mycolic acids (MAs) of Mycobacterium marinum and Mycobacterium ulcerans

Rapidly growing mycobacteria, such as M. chelonae, M. abscessus, and members of the so-called “M. fortuitum complex,” have characteristic MA patterns (Fig. 1F, G). The MAs of M. chelonae and M. abscessus are limited to comparable amounts of α- and α′-MAs (Fig. 1F) (Minnikin et al. 1982a). The α′-MAs are relatively simple with a single cis-alkene; however, there are two varieties of α-MAs, one having two cis-double bonds (α₁) and the other (α₂) containing a trans-alkene with an adjacent proximal methyl branch (Fig. 1F). The main MAs of M. fortuitum, and related mycobacteria, are also α₁- and α₂-MAs (Fig. 1G), accompanied occasionally by minor proportions of α′-MAs in M. farcinogenes and M. senegalense (Ridell et al. 1982), but rarely in M. fortuitum (Minnikin et al. 1980, 1984a; Daffé et al. 1983). Epoxy-MAs (Fig. 1G) in the M. fortuitum complex were first recognized as acid methanolysis products (Minnikin et al. 1980; Ridell et al. 1982) and later characterized as having epoxide rings (Daffé et al. 1981b; Minnikin et al. 1982b, 1984a; Lacave et al. 1987). An unknown minor mycolate, labelled “K,” was recorded in M. fortuitum (Minnikin et al. 1980, 1984a) and in M. senegalense (Ridell et al. 1982), M. peregrinum (Minnikin et al. 1984a), and Mycobacterium porcinum (Luquín et al. 1987). Using an environmental organism, it was shown that mycolate “K” is a novel ω – 1 methoxymycolate (Fig. 1G) (Luquín et al. 1990), and its presence was confirmed in M. fortuitum, M. porcinum, M. peregrinum, and M. senegalense (Luquín et al. 1991).

Fig. 1F

Mycolic acids (MAs) of Mycobacterium chelonae and Mycobacterium abscessus

Fig. 1G

Mycolic acids (MAs) of the Mycobacterium fortuitum complex

2.2 Trehalose Mono- and Dimycolates (TMMs and TDMs)

Trehalose dimycolates (TDMs) were isolated from tubercle bacilli and an association with cording morphology resulted in the loose term “cord factors” being assigned to these lipids (Bloch 1950; Asselineau 1966; Goren 1975; Asselineau and Asselineau 1978). The essential structures of TDMs were established (Noll, 1956; Noll et al. 1956) and early detailed analyses undertaken by Adam et al. (1967) and Strain et al. (1977). These two latter studies demonstrated, for the first time, that certain M. bovis BCG strains lacked methoxymycolates, a phenomenon only clarified later (Minnikin et al. 1984b). In mass spectrometric studies, the complexity of trehalose monomycolates (TMMs) (Fujita et al. 2005a) and TDMs (Fujita et al. 2005b) from a range of mycobacteria was displayed, including M. leprae (Kai et al. 2007). Representative structures of TDMs and TMMs from M. tuberculosis are shown in Fig. 2. Trehalose mycolates have a clear metabolic role in the transfer of mycolic acids into the mycobacterial cell wall (Belisle et al. 1997; Kremer and Besra 2005; Takayama et al. 2005).

Fig. 2

Trehalose mono- and dimycolates (TMMs, TDMs) of Mycobacterium tuberculosis

2.3 Other Mycolic Acid Esters

2.3.1 Glucose Monomycolates (GMMs)

Glucose monomycolates (GMMs) (Fig. 3) were reported by Brennan et al. (1970) in mycobacteria when glucose is included in growth media. Similarly, M. avium and M. tuberculosis grown on fructose gave glycolipids based on fructose (Itoh and Suzuki 1974).

Fig. 3

Glucose monomycolates (GMMs) of Mycobacterium tuberculosis

2.3.2 Monomycoloyl Glycerols (MMGs, GroMMs)

Monomycoloyl glycerols (MMGs, GroMMs) (Fig. 4) are expressed in members of the M. tuberculosis complex, particularly M. bovis (Tsumita 1956; Bloch et al. 1957; Noll 1957; Asselineau 1966; Dobson et al. 1985; Andersen et al. 2009), with the mycolic acid composition reflecting that of whole cells. The natural MMGs from M. bovis BCG included both glycerol isomers, with the R-isomer showing enhanced antigenic properties (Layre et al. 2009).

Fig. 4

Monomycoloyl glycerols (MMGs) of Mycobacterium bovis

2.3.3 Mono- and Dimycoloyl Diarabinoglycerols (MMAGs, DMAGs)

Mono- and dimycoloyl diarabinoglycerol glycolipids (MMAG, DMAG) (Fig. 5) were initially characterized from the MAIS complex (Watanabe et al. 1992, 1999), but they are more widespread including M. tuberculosis (Rombouts et al. 2012) and M. marinum (Elass-Rochard et al. 2012). Synthetic studies indicated that these lipids are based on substituted glycerols with L-configuration (Ali et al. 2019).

Fig. 5

Mycoloyl mono- and diarabinoglycerols (MMAGs, DMAGs) of Mycobacterium avium

2.3.4 Mycenyl Mycolate Ester Waxes (MEWs)

Investigations of mmpL11 lipid transporter proteins in M. smegmatis revealed the presence of partially characterized mycolate ester waxes (MEWs), based on long-chain alcohols (Pacheco et al. 2013; Melly and Purdy 2019). MEWs, based principally on methoxy mycolates, were characterized from M. tuberculosis (Fig. 6) (Wright et al. 2017). The unsaturated long-chain alcohols are a new class of mycobacterial lipids, deserving of a recognizable label, so the term “mycenols” is proposed. Related MEWs from M. smegmatis have been analyzed in more detail (Llorens-Fons et al. 2018).

Fig. 6

Mycenyl mycolate ester wax (MEW) of Mycobacterium tuberculosis

3 Multimethyl-Branched and Polyunsaturated Fatty Acid Esters of Trehalose

Trehalose is an effective scaffold for a range of glycolipids, esterified with fatty acids other than mycolic acids (Asselineau 1966; Asselineau and Asselineau 1978; Khan et al. 2012). In M. tuberculosis, related di-, tri-, and pentaacyl trehaloses (DATs, TATs, and PATs) include dextrorotatory fatty acids whose methyl branches have S-configuration (Minnikin et al. 1985b, 2002; Daffé et al. 1988a; Muñoz et al. 1997; Jackson et al. 2007). Acyl trehaloses (DATs and TATs) are also produced by M. fortuitum (Gautier et al. 1992; Hamid et al. 1993; Sempere et al. 1993). Sulfoglycolipds (SGLs), based on trehalose, have very long multimethyl branched phthioceranic and hydroxyphthioceranic acyl chains. The trehalose polyphleates (TPPs) have similarly long polyunsaturated fatty acids. In contrast, glycosylated acylated trehaloses (lipooligosaccharides, LOSs) have relatively short methyl-branched fatty acids, but usually longer oligosaccharide units.

3.1 Di-, Tri-, and Pentaacyl Trehaloses (DATs, TATs, and PATs)

Early extraction of mycobacterial lipids used ethanol-diethyl ether to both kill the bacteria and provide a lipid extract that included phosphatides and other low-melting “fats” that were easily saponified (see Anderson 1939, 1940; Aebi et al. 1953; Asselineau 1966). Chloroform extraction of the bacterial residue yielded higher-melting “waxes,” which required extended saponification to release long-chain components (see Sect. 4.2). Facile saponification of M. tuberculosis “fats” gave a crude dextrorotatory fatty acid, labelled “phthioic acid” (Anderson 1929). These multimethyl-branched fatty acids were characterized as mycolipenic (phthienoic) (Chanley and Polgar 1950), mycolipodienoic (Coles and Polgar 1969), mycolipanolic (Coles and Polgar 1968), and mycosanoic (Cason et al. 1964) acids (Minnikin 1982). A “3-hydroxy-C₂₇-acid,” corresponding to mycolipanolic acid, was recorded by Asselineau (1966), citing unpublished work by C Asselineau, J-C Promé, and J Asselineau. The nature of the parent lipids was revealed as diacyl and pentaacyl trehaloses (DATs and PATs), in a focused systematic search (Minnikin et al. 1985b, 2002). Structures of the DATs were elaborated (Fig. 7A) (Besra et al. 1992a; Baer 1993), along with those of the triacyl trehaloses (TATs) (Muñoz et al. 1997) and PATs (Daffé et al. 1988a) (Fig. 7B). More than 30 molecular species of DATs from M. tuberculosis H37Rv were recently revealed (Frankfater et al. 2019).

Fig. 7A

Diacyl trehaloses (DATs) of Mycobacterium tuberculosis

2-Methyloctadec-2-enoic acid was found in M. fortuitum (Valero-Guillén et al. 1987) and its location in specific DAT and TATs (Fig. 8) established (Gautier et al. 1992; Hamid et al. 1993; Sempere et al. 1993). These acyl trehaloses are the mycoside F, detected by Fregnan et al. (1961). Further detailed studies of the M. fortuitum DATs (Ariza et al. 1994; Ariza and Valero-Guillén 1994) showed the presence of four distinct types with additional branched fatty acids, particularly a partially characterized 2-methyl-octadecadienoic acid.

Fig. 7B

Tri- and pentacyl trehaloses (TATs, PATs) of Mycobacterium tuberculosis

Fig. 8

Di- and triacyl trehaloses (DATs, TATs) of Mycobacterium fortuitum

3.2 Sulfoglycolipids (SGLs)

Sulfur-containing lipids were detected by Middlebrook et al. (1959) in extracts of virulent tubercle bacilli and characterized as families of trehalose 2′-sulfates acylated with phthioceranic (PA) and hydroxyphthioceranic acids (HPA) (Fig. 9) (Goren 1970a, b; Goren et al. 1971, 1976; Goren 1984; Goren 1990). The SGLs were assigned to families, labelled SL-I to SL-III (Goren 1984, 1990), with a later addition of SL-IV (Gilleron et al. 2004; Layre et al. 2011a). The trehalose 2′-sulfates are acylated at positions 2, 3, 6, and 6′, the 2-position being invariably occupied by straight-chain C₁₆ and C₁₈ fatty acids. Early studies recognized an SL-II′ family with acylation at position 4 (Goren 1984; Goren 1990); however, detailed analyses failed to pinpoint this feature (Layre et al. 2011a), so SL-II′ has not been included in Fig. 9. In studies on the SL-IV antigenic diacylated sulfoglycolipids (Gilleron et al. 2004), the nomenclature Ac₂SGL was recommended. It is proposed to make SGL labels more informative and incorporate the original SL nomenclature, as exemplified for Ac₂SGL: SL-IV (Fig. 9h). Layre et al. (2011a) confirmed the SGLs as Ac₄SGL: SL-I′, SL-I, SL-II; Ac₃SGL: SL-III; Ac₂SGL: SL-IV′, SL-IV) (Fig. 9c–h). The illustrative data in Fig. 9 include the common C₃₇-PA and C₄₀-HPA acyl chains, but PA and HPA fatty acids from 25 to 54 carbons are present (Layre et al. 2011a). The structure of Ac₂SGL: SL-IV (Fig. 9h) was confirmed by chemical synthesis, including the absolute stereochemistry of the principal HPA (Fig. 9b) (Geerdink et al. 2013); similarly, the structure of Ac₄SGL: SL-I (Fig. 9d) was validated by synthesis (Geerdink and Minnaard 2014).

Fig. 9

Sulfoglycolipids (SGLs) of Mycobacterium tuberculosis

Certain MmpL8 lipid transporter mutants of M. tuberculosis are perturbed in SGL synthesis (Converse et al. 2003; Domenech et al. 2004; Bertozzi and Schelle 2008; Seeliger et al. 2012), and it was found that Ac₂SGL lipids, in particular, elaborated oxophthioceranic (OPA) acids, analogous to the HPAs (Layre et al. 2011a). It is apparent that Ac₄SGL: SL-II (Fig. 9e), and not Ac₄SGL: SL-I (Fig. 9d), is usually the most abundant SGL family (Layre et al. 2011a; Rhoades et al. 2011). A minor PA-containing component of the Ac₂SGL lipids has been assigned the label Ac₂SGL: SL-IV′ (Fig. 9g). Certain Ac₄SGL fractions apparently contained mycolipanolates, as found in DAT-II glycolipids (Fig. 7A) (Alugupalli et al. 1995).

3.3 Trehalose Polyphleates (TPPs)

“Phleic acids” occur as trehalose esters in the saprophyte Mycobacterium phlei (Asselineau et al. 1969b, 1972; Asselineau and Asselineau 1978). Trehalose polyphleates (TPPs) are now found in a range of mycobacteria, including M. abscessus and M. chelonae (Burbaud et al. 2016). The principal M. abscessus heptaacyl trehalose phleate is shown in Fig. 10 (Llorens-Fons et al. 2017). These very nonpolar lipids are structurally analogous to the pentaacyl trehaloses (PATs) from the M. tuberculosis complex (Sect. 3.1) (Fig. 7B).

Fig. 10

Trehalose polyphleates (TPPs) of Mycobacterium abscessus

3.4 Glycosylated Acylated Trehaloses (Lipooligosaccharides, LOSs)

Characteristic polar lipids were pinpointed by thin-layer chromatography (TLC) in clinically significant NTMs (Jenkins et al. 1972; Jenkins 1981). Brennan et al. (1978) observed that such lipids were alkali labile, unlike the alkali stable glycopeptidolipids (GPLs) (see Sect. 6) from the M. avium complex (see Aspinall et al. 1995). These trehalose-based lipids have fatty acyl substituents, and sometimes a methoxy group, on one α-D-glucopyranose (Fig. 11) (Camphausen et al. 1987). The other α-D-glucopyranose residue carries one or more sugars to form distinct oligoglycosyl attachments, responsible for distinct serological activity (Fig. 11) (Hunter et al. 1988).

Fig. 11A

Lipooligosaccharides (LOSs) of Mycobacterium kansasii and Mycobacterium gastri

The nature of alkali-labile “lipooligosaccharides” (LOSs) was revealed for M. kansasii (Hunter et al. 1983; Aspinall et al. 1995) (Fig. 11A). Eight such glycolipids (Fig. 11A, LOSs I–VIII) were isolated (Hunter et al. 1984, 1985) and the oligosaccharides were acylated by 2,4-dimethyltetradecanoyl functions. The structure of the N-acylamido sugar, labelled N-acylkansosamine, was established by Hunter et al. (1984, 1985) (Fig. 11A) and this moiety was the basis of the specific serological identity of M. kansasii (Hunter et al. 1985). Subsequent chemical synthesis confirmed the absolute configuration of the sugar (Yoshimura et al. 1987). Four specific polar lipids in eight strains of M. kansasii had been previously identified by TLC (Szulga et al. 1966), and they correspond well with four major LOSs displayed later by 2D-TLC in Fig. 10 of Dobson et al. (1985).

M. gastri is closely related to M. kansasii, but it has distinct LOSs that are based on essentially common triacyl trehaloses and poly-L-xylose backbones (Fig. 11A) (Gilleron and Puzo 1995). An unusual terminal moiety is found in M. gastri LOSs III-IV (see Fig. 11A) (Gilleron et al. 1993, 1994; Longépé et al. 1997).

Mainstream tubercle bacilli do not express LOSs, but the ancestral relative, M. tuberculosis Canetti (“M. canettii”) did express polar antigenic glycolipids resembling LOSs (Papa et al. 1989; Minnikin et al. 1990). This diagnosis was confirmed and the main glycolipids were labelled LOS I and LOS II (Daffé et al. 1991b). The specific terminus of LOS II was initially only partially defined, with the presence of 4-amino-4,6-dideoxy-Gal-pyranose being suspected (Daffé et al. 1991b; Gilleron and Puzo 1995). Current opinion indicates 3-amino-3,6-dideoxy-Gal-pyranose and the LOS I, II structures shown in Fig. 11B (Daffé et al. 2014; Angala et al. 2014).

Antigenic LOSs were detected in M. marinum by Minnikin et al. (1989) and structural studies showed that the LOSs of M. marinum have a core pentasaccharide in LOS-I of a rhamnosyl diglucosyl-acylated trehalose (Fig. 11C) (Burguière et al. 2005; Ren et al. 2007). The heptasaccharide in LOS-II was derived from LOS-I by adding xylose, accompanied by a novel sugar, “caryophyllose” (see Fig. 11C); repeated addition of caryophyllose gave the octasaccharide LOS-III. LOS-IV had a decasaccharide component, including 4-NH-4,6-dideoxy galactose linked to a range of 3-hydroxy-3-methylated-pyrrolidone cycles (Fig. 11C) (Rombouts et al. 2009, 2010, 2011; Sarkar et al. 2011). LOSs in M. marinum had been signaled earlier by Szulga et al. (1966) who recorded the presence of a lipid similar to that revealed by Minnikin et al. (1989).

Characteristic polar lipid patterns were recorded for M. gordonae (Jenkins et al. 1972) and seven serovars had distinct profiles that were assigned to the LOS family (Brennan et al. 1982). Structural studies showed that the major LOS-I lipids from M. gordonae, strains 989 and 990, are distinct in having methoxylated trehalose bases and branching D-xyloses, with different structures for the two isolates (Fig. 11D) (Besra et al. 1993a). The terminal unit from M. gordonae strain 989 remains to be clarified and the full disparate range of LOSs, identified by Brennan et al. (1982), should be structurally characterized.

The family of LOSs from M. malmoense are shown in Fig. 11E (McNeil et al. 1987a). LOSs I and II have a terminal dimannoside, but LOS III expresses D-galactofuranose that is uncommon outside mycobacterial arabinogalactans. The original “somewhat tenuous” assignment of a major C₂₇ acyl component as 2,4-dimethylpentacosanoate (McNeil et al. 1987a) has been updated by its recognition as 2,4,6-trimethyltetracosanoate (Valero-Guillén et al. 1988; Katila et al. 1991). Again, components that probably correspond to LOSs may have been originally pinpointed in M. malmoense by Jenkins (1985).

Mycobacterium szulgai was proposed as a new species on the basis of distinct polar lipids (Jenkins et al. 1972; Marks et al. 1972; Schaefer et al. 1973). The presence of at least six distinct LOSs in M. szulgai has been documented, but only LOS I has been studied (Fig. 11F) (Hunter et al. 1988).

A range of clinical isolates, considered to be “M. fortuitum complex”, showed a distinct, relatively simple tetraacylated glucosylated trehalose antigen (Fig. 11G: a) (Besra et al. 1992b). However, other studies supported DATs and TATs as the characteristic glycolipid antigens of M. fortuitum sensu stricto (Gautier et al. 1992; Hamid et al. 1993; Sempere et al. 1993) (Sect. 3.1). In a study on the “third biovariant” of the “M. fortuitum complex” (Lanéelle et al. 1996), M. fortuitum produced only the characteristic DATs and TATs, but the third biovariant had the truncated LOS (Fig. 11G: a) (Besra et al. 1992b). The third biovariant was elevated to species status as Mycobacterium houstonense (Schinsky et al. 2004). A related simple LOS with an additional L-rhamnose unit inserted between glucose and the acylated trehalose (Fig. 11G: b) has been characterized from mycobacteria, provisionally labeled “Mycobacterium linda” and associated with Chrohn’s disease (Camphausen et al. 1987). Uncharacterized LOSs have been documented in representatives of “Mycobacterium mucogenicum” (Muñoz et al. 1998). Mycobacterial LOSs have been reviewed by Bai et al. (2015).

Fig. 11B

Lipooligosaccharides (LOSs) of “Mycobacterium canettii”

Fig. 11C

Lipooligosaccharides (LOSs) of Mycobacterium marinum

Fig. 11D

Lipooligosaccharides (LOSs) of Mycobacterium gordonae

Fig. 11E

Lipooligosaccharides (LOSs) of Mycobacterium malmoense

Fig. 11F

Lipooligosaccharides (LOSs) of Mycobacterium szulgai

Fig. 11G

Lipooligosaccharides (LOSs) of Mycobacterium houstonense and “Mycobacterium linda”

M. kansasii strains of smooth colony morphology had cell surface LOSs, whereas rough variants were devoid of such surface antigens (Belisle and Brennan 1989). Previous studies (Collins and Cunningham 1981) had shown that the rough forms of M. kansasii persist longer than smooth variants in experimentally infected mice. Smooth morphology “M. canettii” produces polar LOSs (Fig. 11B) that are not present in rough M. tuberculosis (Soto et al. 2000). This correlates with greatly enhanced hydrophobicity in M. tuberculosis in comparison with relatively hydrophilic “M. canettii” (Minnikin et al. 2015; Jankute et al. 2017). It is probable that rough hydrophobic tubercle bacilli are preferentially suited to aerosol transmission and increased pathogenicity (Jankute et al. 2017). Spontaneous smooth-to-rough “M. canettii” variants, which lacked LOSs (van Soolingen et al. 1997), were found to be mutated in the polyketide-synthase-encoding pks5 locus, a phenotype restored by complementation (Boritsch et al. 2016). These rough variants also showed an altered host–pathogen interaction and increased virulence in cellular- and animal-infection models (Boritsch et al. 2016; Supply and Brosch 2017).

4 Mycobacterial Waxes

Waxes are defined as long-chain fatty acid esters of long-chain alcohols, but glycerol esters will also be included here. Saponification-resistant wax fractions from tubercle bacilli produce a range of (-)-laevorotatory fatty acids, originally termed mycocerosic acids (Ginger and Anderson 1945). These acids were found to be multimethyl-branched, using the alternative name, mycoceranic acids (Marks and Polgar 1955; Polgar and Smith 1963). A “methoxyglycol” from the same waxes was designated as phthiocerol (Stodola and Anderson 1936) and these β-diols were found to be diesterified with mycocerosic acids to produce the phthiocerol dimycocerosate (PDIM) waxes (see Asselineau 1966). Waxes composed of long-chain alcohols esterified to mycolic acids (MEWs) have been included in Sect. 2.3.

4.1 Triacylglycerols (TAGs)

Triacylglycerols (TAGs) ) are found in most mycobacteria, but the full complexity of such mixtures has not been assessed, other than in the saprophyte, M. smegmatis (Purdy et al. 2013). TAGs are likely to be the principal components of internal lipid bodies, expressed under varying culture conditions (Garton et al. 2002). Structurally distinct mycobacterial TAGs have been characterized, with single very long-chain fatty acids analogous to the meromycolate chains of mycolic acids (Kremer et al. 2005; Rafidinarivo et al. 2009) (Fig. 12). Asselineau (1966) named these fatty acids “mycobacteric acids.” The mycobacteric acids include functional groups similar to those of the mycolic acids in the same mycobacterial species (Promé et al. 1966). Detailed profiles of C₃₀–C₅₆ fatty acids, corresponding to mycobacterates, from M. tuberculosis H37Ra have been recorded (Takayama and Qureshi 1978; Qureshi et al. 1980).

Fig. 12

Monomycobacteroyl diacyl glycerols (MMDAGs) of Mycobacterium kansasii and Mycobacterium brumae

The original wax from M. kansasii was termed a “monomeromycoloyl diacylglycerol” (MMDAG) (Kremer et al. 2005) but precedence dictates that the name should be “monomycobacteroyl diacylglycerol” (Rafidinarivo et al. 2009); the term “meromycolate” should only be used for the designated portion of an intact mycolic acid (Morgan and Polgar 1957). The MMDAGs from M. smegmatis had a C₁₆ straight-chain acid at the glycerol 2-position (Purdy et al. 2013), so this orientation is suggested for the MMDAGs of M. kansasii and Mycobacterium brumae (Rafidinarivo et al. 2009) (Fig. 12). Cellular movement of MMDAGs depends on mmpL11 transporters (Pacheco et al. 2013). Diacylglycerols (DAGs) have been recorded in mycobacteria (Asselineau 1966) but significant structures remain to be determined.

4.2 Phthiocerol Dimycocerosates (PDIMs)

A methoxyglycol, isolated from the chloroform-extracted waxes of M. tuberculosis, was given the name “phthiocerol” by Stodola and Anderson (1936), substantiating the discovery of a “phthioglycol” by Stendal (1934). The C₃₄/C₃₆ methoxydiol character of phthiocerol was established in simultaneous investigations (see Drayson et al. 1958; Demarteau-Ginsburg et al. 1959; Ryhage and Stenhagen 1960). The M. tuberculosis phthiocerol dimycocerosate (PDIM) waxes (Fig. 13A, B) were characterized in early studies (Noll 1957; Asselineau 1966). Detection of mycobacterial PDIMs was achieved by infra-red spectroscopy (Randall and Smith 1964). The basal methoxylated diols, phthiocerol A, and phthiocerol B (Minnikin and Polgar 1965, 1966b; Maskens et al. 1966) are usually accompanied by a ketone, phthiodiolone A (Fig. 13A) (Minnikin and Polgar 1967d). In some cases, phthiotriol A is the main component (Huet et al. 2009) (Fig. 13A). Closely-related PDIMs are characteristic of M. bovis, M. leprae, M. kansasii, M. gastri, and M. haemophilum (Fig. 13A) (Minnikin et al. 2002, 2015; Onwueme et al. 2005). PDIM families from M. marinum and M. ulcerans have alternative absolute stereochemistries for both the acid (Daffé et al. 1984; Daffé and Lanéelle 1988) and diol (Besra et al. 1989, 1990a) components (Fig. 13B).

Nomenclature for the multimethyl-branched fatty acid components of PDIMs is inconsistent and clarification is desirable. The name mycocerosate (Ginger and Anderson 1945) has precedence over mycoceranate, provisionally used for the (-)-laevorotatory multimethyl-branched acids (Marks and Polgar 1955). However, it is inappropriate to assign the label “phthioceranate” to the (+)-dextrorotatory multimethyl-branched acids from the waxes of M. marinum and M. ulcerans (Daffé et al. 1984; Daffé and Lanéelle 1988). Presumably, it was considered that the phthioceranate components of the sulfoglycolipids (SGLs) (Fig. 9) were the closest relatives of the (+)-dextrorotatory multimethyl-branched acids from the M. marinum and M. ulcerans waxes, but the dextrorotatory mycosanoates (Cason et al. 1964) from the M. tuberculosis diacyl trehaloses (DATs) (Fig. 7A) have precedence. Phthioceranates and mycosanoates are readily released on saponification of low-melting M. tuberculosis fats (see Sect. 3.1), but the components of PDIMs require prolonged alkaline hydrolysis (Minnikin et al. 1983). In essence, the M. marinum PDIM-derived C₂₇–C₃₀ multimethyl-branched acids (Fig. 13B) overlap in chain length with the established C₂₇–C₃₄ mycocerosates (Fig. 13A) and are functionally similar. It is proposed, therefore, to recommend the label “S-mycocerosates” for the (+)-dextrorotatory fatty acids from the PDIMs of M. marinum and M. ulcerans (Fig. 13B) and “R-mycocerosates” for the (-)-laevorotatory acids from tubercle bacilli and other mycobacteria. This proposal correlates with a previous rationalization by Onwueme et al. (2005).

The nomenclature of members of the broad phthiocerol family also requires coherence. The main family member was designated phthiocerol A, to distinguish it from a closely-related minor component that was appropriately labelled phthiocerol B (Minnikin and Polgar 1965, 1966b). Diols with terminal ethyl or methyl groups constitute the “A” or “B” series, respectively, and the main ethyl-keto component is phthiodiolone A (Fig. 13A) (Minnikin and Polgar 1967d). This logic is followed in most studies, including intact PDIMs (Minnikin et al. 1985c; Daffé and Lanéelle 1988; Daffé 1991; Onwueme et al. 2005). However, some recent studies (Huet et al. 2009; Rens et al. 2018) have disregarded the existence of phthiocerol B and inappropriately used the abbreviation PDIM B for the wax based on phthiodiolone A!

Overall rationalization of the nomenclature of members of the phthiocerol family has been proposed by Onwueme et al. (2005). The term “DIMs” was limited to labelling the entire “dimycocerosate” families of PDIM waxes and phenolic glycolipids (PGLs) (Sect. 5). The convenient abbreviations PCOL, PDON, and PTOL were suggested for phthiocerol, phthiodiolone, and phthiotriol, respectively (Fig. 14). An A or B could be added to signify terminal ethyl or methyl groups, as exemplified for phthiocerol B (PCOL B) or phthiodiolone A (PDON A). Diol stereochemistry could be shown by use of suffix E or T for erythro- (syn-) and threo- (anti-) forms; E-PCOL A would represent phthiocerol A from M. marinum (Fig. 13B) and T-PDON B would be phthiodiolone B from M. tuberculosis and related taxa (Fig. 14). The label PDIM should be retained for intact waxes based on phthiocerol A (PCOL A) and phthiocerol B (PCOL B). Waxes based on phthiodiolone (PDON) and phthiotriol (PTOL) could be labelled NPDIMs and TPDIMs, respectively. These basic abbreviations are made additionally informative by providing information about the R- or S-mycocerosate (MYCS) fatty acid substituents and the erythro- or threo-diol stereochemistry (Fig. 14). As an example, the main M. tuberculosis PDIM wax is described by the label RT-PDIM A, signifying R-mycocerosates (R- MYCS) esterifying a threo-diol (T-PCOL A) with a terminal ethyl group (Fig. 14). Nomenclatural labels for the so-called phenolic glycolipids (PGLs) (see next Sect. 5) are also included in Fig. 14

Fig. 13A

Phthiocerol dimycocerosate (PDIM) waxes of the Mycobacterium tuberculosis complex, Mycobacterium leprae, Mycobacterium haemophilum, Mycobacterium kansasii and Mycobacterium gastri, based on threo phthiocerols and R-mycocerosates

Fig. 13B

Phthiocerol dimycocerosate (PDIM) waxes of Mycobacterium marinum and Mycobacterium ulcerans, based on erythro phthiocerols and S-mycocerosates

.

Fig. 14

Proposed scheme for nomenclature and abbreviations of phthiocerol dimycocerosates (PDIMs) and glycosyl phenolphthiocerol dimycocerosates (phenolic glycolipids, PGLs)

5 Glycosyl Phenolphthiocerol Dimycocerosates (Phenolic Glycolipids, PGLs)

A study (Noll 1957) of the waxes of M. bovis yielded a component with an infra-red spectrum identical to a fraction (G_B) identified by Smith et al. (1954, 1957); this latter investigation also pinpointed a related lipid (G_A) from M. kansasii. These glycolipids were called “mycosides” by Smith et al. (1960b) and relabelled mycosides B and A, respectively. Additional infra-red studies revealed mycoside G from M. marinum (Navalkar et al. 1965). The essential core phenolphthiocerol dimycocerosate units in mycosides B and A, respectively, were elaborated by Demarteau-Ginsburg and Lederer (1963) and Gastambide-Odier et al. (1965). Further sources of this glycolipid type are M. leprae (Hunter and Brennan 1981, 1983; Hunter et al. 1982), M. haemophilum (Besra et al. 1990b, 1991), M. gastri (Gilleron et al. 1990b), M. ulcerans (Daffé et al. 1992), “M. canettii” (Daffé et al. 1987), and the Beijing clades of M. tuberculosis (Huet et al. 2009). The term “phenolic glycolipid” (PGL) is in general use for this category of lipids (Hunter and Brennan 1981; Hunter et al. 1984; Brennan 1984; Puzo 1990; Onwueme et al. 2005) and representative structures are shown in Fig. 15A–D. Abbreviations for the various components of PGLs are included in Fig. 14 and employed in Fig. 15A–D.

Fig. 15A

Glycosyl phenolphthiocerol dimycocerosates (phenolic glycolipids, PGLs) of Mycobacterium kansasii and Mycobacterium gastri, based on threo phenolphthiocerols and R-mycocerosates

Fig. 15B

Glycosyl phenolphthiocerol dimycocerosates (phenolic glycolipids, PGLs) of the Mycobacterium tuberculosis complex, based on threo phenolphthiocerols and R-mycocerosates

Fig. 15C

Glycosyl phenolphthiocerol dimycocerosates (phenolic glycolipids, PGLs) of Mycobacterium leprae and Mycobacterium haemophilum, based on threo phenol-phthiocerols and R-mycocerosates

Fig. 15D

Glycosyl phenolphthiocerol dimycocerosates (phenolic glycolipids, PGLs) of Mycobacterium marinum and Mycobacterium ulcerans, based on erythro phenolphthiocerols and S-mycocerosates

The essential character of M. kansasii mycoside A (Smith et al. 1957, 1960a, b) was elaborated by Gastambide-Odier et al. (1965), Gastambide-Odier and Sarda (1970), and Gastambide-Odier and Villé (1970). Detailed structures were defined by Fournié et al. (1987a, b, c), Rivière et al. (1987), and Gilleron et al. (1990a, b) (Fig. 15A). Similar PGL profiles were produced by Mycobacterium gastri (Vercellone et al. 1988; Gilleron et al. 1990b) (Fig. 15A). Gilleron et al. (1990b) and Watanabe et al. (1997) described up to eight well-defined PGLs in M. kansasii (Fig. 15A).

The major PGL of M. bovis (mycoside B) (Demarteau-Ginsburg and Lederer 1963; Gastambide-Odier and Sarda 1970) was labelled as “M. bovis identifying lipid” (MBIL) (Jarnagin et al. 1983). A single 2-O-methyl-α-L-rhamnopyranose was linked to the phenolic phthiocerol (Daffé et al. 1988b; Chatterjee et al. 1989), along with minor components based on α-L-rhamnopyranose (Fig. 15B). An additional minor disaccharide-containing PGL has been defined in M. bovis BCG (Vercellone and Puzo 1989) (Fig. 15B). M. tuberculosis complex Canetti strains, also produce mycoside B, accompanied by a triglycosyl PGL (PGL-tb) (Daffé et al. 1987) (Fig. 15B). Additional components, based on phenolthiotriol, have been found in unusual M. tuberculosis strains (Watanabe et al. 1994) and Beijing variants (Huet et al. 2009). The methylated PGL aglycone has been isolated from variants of M. tuberculosis and given the name “Attenuation Indicator” (AI) lipid (Fig. 15B) (Goren et al. 1974; Krishnan et al. 2011). Structures of PGL-tb and mycoside B have been confirmed by synthesis (Barroso et al. 2012, 2013).

The specific trisaccharide PGL-I of M. leprae (Hunter et al. 1982) (Fig. 15C) and semisynthetic neo-antigens have found widespread use for the serodiagnosis of leprosy (Cho et al. 1983; Fujiwara et al. 1984; Spencer and Brennan 2011). Demethylation variations in the PGL-I structure are found in PGL-II (Fujiwara et al. 1984) and PGL-III (Hunter and Brennan 1983) (Fig. 15C). A diglycosyl component, lacking the terminal 3,6-di-O-methyl-glucopyranose of PGL-1, was reported by Daffé and Lanéelle (1989) (Fig. 15C). M. haemophilum has a distinct trisaccharide PGL (Besra et al. 1991) that shares mono- and di-methyl rhamnoses with the PGLs from M. leprae and unusual C₃₃ and C₃₄ mycocerosates were common to both (Fig. 15C), suggesting a phylogenetic relationship.

The mycoside G from M. marinum, revealed by Navalkar et al. (1965), was found to be a 3-O-methyl-α-L-rhamnosyl diacyl phenolphthiocerol (Fig. 15D) (Sarda and Gastambide-Odier 1967; Gastambide-Odier 1973; Daffé and Lanéelle 1988). Dobson et al. (1990) have described a series of PGLs from M. marinum, with a predominance of lipids based on phenolphthiodiolone and phenolphthiotriol, in some cases (Fig. 15D). Closely related M. ulcerans produces the monoglycosyl PGL found in M. marinum, with enhanced phenolphthiodiolone and phenolphthiotriol components (Daffé et al. 1984, 1992).

In accordance with PDIMs (Figs. 13 and 14), the stereochemistry of M. marinum and M. ulcerans PGLs is distinct (Fig. 15D). It would be preferable to label the multimethyl-branched acyl chains as S-mycocerosates rather than the term “phthioceranates” used by Daffé et al. 1984, 1992; Daffé and Lanéelle 1988, and Daffé 1991 (see Figs. 13B and 14). Again, erythro-diols were found in the PGLs of M. marinum (Besra et al. 1989) and M. ulcerans (Besra et al. 1990a) (Fig. 15D), in contrast to the PGL threo-diols in all other taxa (Fig. 15A–C). The stereochemistry at the phenolphthiocerol methyl branch (carbon-4) is considered to be R in M. marinum and M. ulcerans PGLs (Fig. 15D), rather than S in other PGLs (Fig. 15A–C) (Onwueme et al. 2005).

For nomenclature of PGLs, the abbreviations φPCOL, φPDON, and φPTOL (Fig. 14) for the methoxy, keto, and hydroxy phenolic diols, respectively, uses suggestions by Onwueme et al. (2005). Similar to the PDIMs, PGLs based on φPDON and φPTOL could be given the abbreviations NPGL and TPGL, respectively (Fig. 14).

6 Glycopeptidolipids (GPLs) and Peptidolipids (PLs)

Infra-red spectroscopy of extracts from the M. avium complex (MAC) identified a class of lipids (Smith et al. 1957) eventually labelled “mycosides C” (Smith et al. 1960a, b; Smith and Randall 1965). It was also recognized that MAC lipid extracts contained diagnostic glycolipids, as seen by thin-layer chromatography (TLC) (Marks et al. 1971; Jenkins and Marks 1973; Jenkins 1981). Brennan et al. (1978) applied such TLC procedures to the entire Schaefer (MAC) complex and it was demonstrated that the Marks-Jenkins lipids were synonymous with mycosides C and typing antigens reviewed by Schaefer (1965). These lipids were shown to be “polar” glycopeptidolipids (pGPLs) (Lanéelle 1966; Brennan and Goren 1979; Brennan 1981, 1984; Brennan et al. 1981a, b; Tsang et al. 1983, 1992; Denner et al. 1992). Structures of nonspecific “apolar” aGPLs had been established earlier (Jolles et al. 1961; Lanéelle 1966; Lanéelle and Asselineau 1968; Lederer 1967; Jardine et al. 1989).

All GPL structures (Fig. 16A–F), with the exception of those from M. xenopi (Fig. 16G), are based on a tripeptide-amino-alcohol (D-Phe-D-allo-Thr-D-Ala-L-alaninol), N-linked at the D-Phe to a long-chain fatty acyl residue. In the case of the aGPLs, this lipopeptide core is substituted usually by a 6-deoxytalosyl (dTal) unit linked to the allo-Thr residue and by an O-methylated rhamnosyl unit linked to the terminal alaninol. Further glycosylation, usually on the dTal unit, gave antigenic, serospecific pGPLs (Brennan and Goren 1979). Brennan et al. (1981a, b) showed that these oligoglycosyl haptens were liberated on reductive β-elimination with alkaline sodium borohydride, but O-acetyl substituents, important for antigenicity, were lost. Subsequently, the presence of acetates, uronic acids, acetalically linked pyruvic acids, acylaminodideoxyhexoses, and branched-chain sugars was demonstrated (Bozic et al. 1988; Jardine et al. 1989).

Fig. 16A

Glycopeptidolipids (GPLs) of Mycobacterium avium complex (MAC): Group 1, rhamnose/fucose based

Aspinall et al. (1995) classified the oligoglycosyl haptens of 12 of the 31 known serotype/serovar members of the M. avium complex (MAC), based on structural similarities. The presence of three principal groups of MAC GPLs is apparent (Fig. 16A–C) (Chatterjee and Khoo 2001). In Group 1 of MAC GPLs (Fig. 16A), a common L-rhamnose-L-fucose discaccharide, linked to the 6-deoxy-L-talose, is further decorated to produce specific polar (pGPLs). The fucose is lacking in serovar 1, but in serovar 2 it is methylated and acetylated (Fig. 16A); this latter GPL was also characterized from supposed M. paratuberculosis (Camphausen et al. 1985), but it is suspected that the organism examined was M. avium serovar 2 (Chatterjee and Khoo 2001). This correlates with Jenkins (1981) who recorded no specific glycolipid pattern for M. paratuberculosis.

Distinct categories of MAC Group 1 pGPLs can be discerned (Fig. 16A). Firstly, serovars 4 and 20 are limited to an additional methylated rhamnose. Secondly, serovars 3, 9, and 26 add simple disaccharides based on D-glucuronic acid (D-GlcA)-L-fucose. Serovar 25 also included D-GlcA but the disaccharide is completed by acetylated amino-D-fucose. The very individual serovar 14 has rhamnose, but as the D-enantiomer, with the rare 4-N-formyl-L-kansosamine (Fig. 16A). A variant of serovar 4 had D-valine in place of the core phenylalanine (Matsunaga et al. 2012).

The basic core extension in Group 2 MAC pGPLs (Fig. 16B) is a di-L-rhamnose disaccharide, but in all cases, the third and fourth sugars are modified L-rhamnoses and D-glucoses; D-Glc-linked amido substituents are a unifying factor. Serovars 7, 12, and 13 have 2-hydroxy-propanamido and serovars 16 and 17 have 4-methoxy-3-hydroxy-2-methyl-pentanamido and 3-hydroxy-2-methyl-butanamido, respectively. The unusual pGPL member in MAC Group 2 is from serovar 19, with a C-methylated rhamnose and glucuronic acid (Fig. 16B). The core extension in MAC serovars 8 and 21 Group 3 pGPLs (Fig. 16C) is L-rhamnose-D-glucose, with the latter acetylated with pyruvate.

The GPLs of “M. habana” and M. simiae (Fig. 16D) resemble those in MAC Group 2 (Fig. 16B), with a di-L-rhamnose extension (Khoo et al. 1996). The further disaccharide, composed of 6-O-methyl-D-glucose and L-fucose, resembles the D-glucuronic acid-L-fucose units in Serovars 3, 9, and 26 in MAC Group 1 GPLs (Fig. 16A). “M. habana” is closely affiliated with M. simiae, but further detailed investigations revealed that the details of the superficially similar GPLs are able to distinguish the two taxa (Mederos et al. 1998, 2006, 2008).

In the “M. fortuitum-M. chelonae complex,” M. fortuitum sensu stricto lacked GPLs (Tsang et al. 1984) (see Sect. 3.1). M. fortuitum biovar peregrinum expressed GPLs different to those in M. chelonae. The main feature of the simple GPLs from M. chelonae and M. abscessus is an additional L-rhamnose attached to the core L-rhamnose, rather than on the 6-deoxy-L-talose (Fig. 16E).

The core L-rhamnose in M. peregrinum GPLs is also the point of modification and the D-allo-threonine carries 3-O-methyl-L-rhamnose in place of the usual 6-deoxy-L-talose (Fig. 16F); a sulfated variant is notable (López Marín et al. 1991, 1992a; Lanéelle et al. 1996). Individual GPLs support the rejuvenation of the species Mycobacterium peregrinum (Kusunoki and Ezaki 1992). GPLs are similar in M. peregrinum, M. senegalense, and M. porcinum (Fig. 16F) (López Marín et al. 1993; Besra et al. 1994a).

The GPLs of M. xenopi (Fig. 16G) have different lipopeptide cores, L-serine providing an anchor point for 3-O-methyl-6-deoxy-L-talose (Rivière and Puzo 1991). Up to four decorated L-rhamnose units are linked to the terminal D-allo-threonine methyl ester (Rivière et al. 1993; Besra et al. 1993b). Notably, these GPLs have two relatively short acyl chains, one of which is attached to the rhamnosyl oligosaccharides.

Apolar aGPLs probably fulfil a structural role in the mycobacterial outer membrane (MOM), with the antigenic polar pGPLs having specific cell surface activity. Rough M. avium variants lack GPLs (Barrow and Brennan 1982), as confirmed by genomic studies (Belisle et al. 1993a, b). GPLs facilitate sliding motility, biofilm formation, and cell wall integrity (see Deshayes et al. 2005; Ripoll et al. 2007; Schorey and Sweet 2008) and possibly mycobacteriophage attachment (Goren et al. 1972). Rough variants of M. abscessus, lacking GPLs, are more invasive (Howard et al. 2006; Catherinot et al. 2007; Julián et al. 2010; Mukherjee and Chatterji 2012; Pang et al. 2013; Bernut et al. 2014; Brambilla et al. 2016; Gutiérrez et al. 2018). Rough GPL-free M. abscessus strains are relatively hydrophobic and possibly more likely to be spread in aerosols (Minnikin et al. 2015; Jankute et al. 2017; Viljoen et al. 2018). Fregnan et al. (1962) noted that smooth to rough colony changes correlated with GPL loss in a scotochromogenic Mycobacterium, “the culture aspect changing from hydrophilic to hydrophobic.” Hydrophobicity was a key factor in facilitating aerosol transmission of related MAC organisms (Parker et al. 1983; Falkinham 2003).

Fig. 16B

Glycopeptidolipids (GPLs) of Mycobacterium avium complex (MAC): Group 2, rhamnose/rhamnose based

Fig. 16C

Glycopeptidolipids (GPLs) of Mycobacterium avium complex (MAC): Group 3, rhamnose/glucose based

Fig. 16D

Glycopeptidolipids (GPLs) of “Mycobacterium habana” and Mycobacterium simiae

Fig. 16E

Glycopeptidolipids (GPLs) of Mycobacterium abscessus and Mycobacterium chelonae

Fig. 16F

Glycopeptidolipids (GPLs) of Mycobacterium peregrinum, Mycobacterium senegalense and Mycobacterium porcinum

Fig. 16G

Glycopeptidolipids (GPLs) of Mycobacterium xenopi

Some rough mutants of M. avium serovar 2 express peptidolipids (PLs) that represent the core of the usual GPLs (Fig. 16A) (Belisle et al. 1993a). Initial detection of distinct peptidolipids in M. paratuberculosis (Lanéelle and Asselineau 1962) was confirmed by subsequent studies (Rivière et al. 1996; Eckstein et al. 2006; Biet et al. 2008). PLs based on a pentapeptide (L5P) and a tripeptide (L3P) were characterized (Fig. 17) (Bannantine et al. 2017), the former in cattle M. paratuberculosis strains and the latter in sheep strains. A peptidolipid was isolated from cattle-associated “Mycobacterium minetti,” a synonym of M. fortuitum, and given the name “fortuitine” (Vilkas et al. 1963; Asselineau 1966). Introducing mass spectrometry for peptide sequencing (Barber et al. 1965), a draft structure was assigned for fortuitine (Fig. 17).

Fig. 17

Peptidolipids (PLs) of Mycobacterium paratuberculosis and Fortuitine

7 Isoprenoid Lipids

Isoprenoid quinones are essential in respiratory processes and pigments are characteristic of a number of mycobacterial pathogens. Isoprenoid hydrocarbon chains are favored for a range of carrier molecules and specific halimane diterpenoids, such as 1-tuberculosinyladenosine, are encountered.

7.1 Isoprenoid Quinones

The most common mycobacterial isoprenoid quinone is a dihydrogenated menaquinone with nine isoprene units, abbreviated MK-9(H₂) (Fig. 18) (Minnikin 1982; Collins et al. 1985). Initial isolation from Mycobacterium phlei (Gale et al. 1963) was followed by structural characterization (Azerad et al. 1967; Azerad and Cyrot-Pelletier 1973). A related sulfated menaquinone from M. tuberculosis has been labelled sulfomenaquinone (Fig. 18) (Holsclaw et al. 2008; Sogi et al. 2016). An unusual non-isoprenoid quinone from M. avium, mavioquinone (Fig. 18) (Scherrer et al. 1976), requires substantiation.

Fig. 18

Mycobacterial isoprenoid quinones

7.2 Isoprenoid Pigments

Carotenoid pigments in mycobacteria were established by Chargaff and Lederer (1935). Opportunist pathogens, such as M. kansasii, M. marinum, M. gordonae, M. scrofulaceum, and M. szulgae, express carotenoids (Tsukamura and Mizuno 1969; Goodwin 1972; Liaanen-Jensen and Andrewes 1972; Ratledge 1976; Goren and Brennan 1979; Minnikin 1982; Ichiyama et al. 1988; Robledo et al. 2011). Photochromogenic mycobacteria, such as M. kansasii and M. avium, are pigmented in light but not in the dark, but scotochromogenic mycobacteria, like M. gordonae and M. scrofulaceum, are also pigmented in the dark (Goren and Brennan 1979). Ubiquitus β-carotene is often the principal mycobacterial pigment, accompanied by related carotenoids, as illustrated for M. kansasii by David (1974), where eight additional compounds were characterized (Fig. 19). Distinct carotenoid groups have been proposed (Ichiyama et al. 1988), including incompletely characterized oxygenated pigments, classed as xanthins or xanthophylls (see Goodwin 1972; Minnikin 1982). A xanthophyll example is the glucoside “phlei-xanthophyll,” produced by the saprophyte Mycobacterium phlei (Fig. 19) (Hertzberg and Liaaen-Jensen 1967). Unpigmented colonial variants of M. avium are less hydrophobic (Stormer and Falkinham 3rd. 1989).

Fig. 19

Mycobacterial isoprenoid pigments

7.3 Isoprenoid Lipid Carriers

A partially saturated C₃₅ octahydroheptaprenol and a C₅₀ decaprenol (“Dec”) (Fig. 20A–J) are the predominant isoprenoid lipids involved in “carrying” carbohydrates to their biosynthetic destinations in mycobacteria. Phosphorylated forms of these two acted as glycosyl acceptors from GDP-mannose (Takayama and Goldman 1970; Takayama et al. 1973; Yokoyama and Ballou 1989) and UDP-glucose (Schultz and Elbein 1974). C₃₅-P-Man and C₅₀-P-Man (Fig. 20a, c) are donors for the synthesis of the higher-order phosphatidylinositol mannoides (PIMs, see Sect. 8) (PIM₄–PIM₆) and also lipomannan (LM) and lipoarabinomannan (LAM) (Scherman et al. 2009); the simpler PIM₁–PIM₃ derive their Manp units directly from GDP-mannose. The structure of the mycobacterial decaprenol (Fig. 20c–i) is unusual, with mono-trans, octa-cis olefins (Wolucka et al. 1994; Wolucka and de Hoffmann 1998).

Fig. 20

Mycobacterial isoprenoid lipid carriers

Mikušová et al. (1996) established that C₅₀-P-P-GlcNAc (“GL-1”) (Fig. 20i) and C₅₀-P-P-GlcNAc-Rha (“GL-2”) (Fig. 20j) are acceptors for the subsequent glycosylation steps leading to the synthesis of the C₅₀-P-P-GlcNAc-Rha-(Galf)x- (Araf)y precursor of mycobacterial cell wall core formation. The precursor of the galactosamine (GalNH₂) monomers, attached to the branching Ara units of the mycobacterial cell wall core, is C₅₀-decaprenol-P-GalNAc (Fig. 20j), along with a minor C₃₅-heptaprenol analogue (Škovierová et al. 2010). The 5-methylthioxylose units, substituting the mannoside caps of the LAM from M. tuberculosis, originate from a decaprenol-P-linked precursor (Fig. 20d) (Angala et al. 2017), whose oxidized form (Fig. 20e) was also characterized.

A β-arabinofuranosyl-1-monophosphoryldecaprenol (C₅₀-P-Araf) (Fig. 20f) is the precursor of Araf units in mycobacterial cell walls (Wolucka et al. 1994). Mycobacterial β-D-ribosyl-1-monophosphodecaprenol (Fig. 20g) is the probable precursor of C₅₀-P-Araf (Fig. 20f) (Wolucka and de Hoffmann 1995; Scherman et al. 1995, 1996; Mikušová et al. 2005) and a β-arabinofuranosyl-1-monophosphorylheptaprenol (C35-P-Araf) (Fig. 20b) was identified but its role is undefined.

“Myc-PL”, is an “apparent carrier in mycolic acid synthesis” (Besra et al. 1994b). This unusual lipid, from M. smegmatis, was a mycoloyl-mannosylphosphopolyprenol (Fig. 20k), the C₃₅ heptaprenol having similarities to the mannose and arabinose carriers described above (Fig. 20a, b). However, despite indications of this lipid in M. tuberculosis (Besra et al. 1994b), the presence of Myc-PL in pathogenic mycobacteria remains to be established.

7.4 Halimane Diterpenoids

In a search for the role of the terpenoid cyclase-associated M. tuberculosis gene Rv3377c, a diterpene was characterized and named tuberculosinol (Fig. 21) (Nakano et al. 2005); it is a member of the halimane diterpenoids (Roncero et al. 2018). A distinct bioactive diterpene was recognized in M. tuberculosis (Mann et al. 2009), but this material, designated as isotuberculosinols, was shown to be a mixture of enantiomers of halimane diterpenoids (Fig. 21), identical to nosyberkols from sponges (Spangler et al. 2010; Maugel et al. 2010; Hoshino et al. 2011; Mann et al. 2012; Mann and Peters 2012). Scrutiny of the M. tuberculosis genome revealed that Rv3378c was responsible for the production of 1-tuberculosinyladenosine (Fig. 21) (Layre et al. 2011b, 2014). Biosynthetic studies pinpointed a distinct isomer, N⁶-tuberculosinyladenosine (Fig. 21) (Young et al. 2015; Oldfield 2015); both isomers were confirmed by chemical synthesis (Buter et al. 2016). This family of lipids correspond to uncharacterized M. tuberculosis components, labelled X and possibly Z1, Z2 in Fig. 8 of Dobson et al. (1985).

Fig. 21

Halimane diterpenoids of Mycobacterium tuberculosis

8 Polar Lipids and Related Lipoglycans

The mycobacterial inner membrane (MIM) (Minnikin et al. 2015) includes universal polar lipids, such a diphosphatidylglycerol (DPG), phosphatidylethanolamine (PE), and phosphatidylinositol (PI). PI is extended to give the glycophospholipid phosphatidylinositol mannosides (PIMs) that are further developed into the lipomannan (LM) and lipoarabinomannan (LAM) water soluble lipoglycans.

8.1 Polar Phospholipids, Glycolipids, and Ornithine Lipids

The principal mycobacterial phospholipids are cardiolipin (diphosphatidylglycerol DPG), phosphatidylethanolamine (PE), and phosphatidylinositol (PI) (Fig. 22) (Goren and Brennan 1979; Goren 1984). A polar lipid acyl component is “tuberculostearic acid” (TSA) (Fig. 22), first recognized by Anderson and Chargaff (1929), and confirmed as 10-D(R)-methyloctadecanoic acid (Spielman 1934; Prout et al. 1948). TSA has been variously synthesized, as reviewed by ter Horst et al. (2010b) who also synthesized regioisomers of mycobacterial PE and proved that TSA is located on the sn-1 position of glycerol as shown in Fig. 22. By analogy, TSA is also assumed to be on the sn-1 positions of glycerol in mycobacterial DPG and PI (Fig. 22).

Fig. 22

Mycobacterial polar lipids: diphosphatidylglycerol (DPG), phosphatidyl-ethanolamine (PE), phosphatidylinositol (PI), diglucosyl diacyglycerol (DGDAG) and ornithine lipid (OL)

An ornithine-based polar lipid (OL) (Fig. 22) has been detected in M. tuberculosis, M. bovis, and possibly M. marinum (Promé et al. 1969; Lanéelle et al. 1990). Ornithine lipids replace PE in reduced phosphate cultures in other bacteria (Minnikin and Abdolrahimzadeh 1974). PEs from M. tuberculosis have minor 3-OH-C_{16, 18, 20} straight-chain fatty acids (Alugupalli et al. 1995), similar to that in the ornithine lipids (OL) (Fig. 22). A diglycosyl diacylglycerol was isolated from M. tuberculosis (Hunter et al. 1986a) (Fig. 22).

8.2 Phosphatidylinositol Mannosides (PIMs)

A “phosphatide” fraction from tubercle bacilli yielded glycerophosphoric acid, mannose, and the hexahydric alcohol “inositide” (Anderson and Roberts 1930; Anderson 1939, 1941, 1943). Saponification of the phosphatides gave a “phosphorous- containing glycoside,” which on dephosphorylation produced a “mannoinositose.” The essential structure of phosphatidyl myo-inositol dimannoside (PIM₂) (Fig. 23) from M. tuberculosis was eventually established (Vilkas and Lederer 1960; Ballou et al. 1963; Lee and Ballou 1964a). Tri-, tetra-, penta-, and hexa-mannosides were recognized and the previously identified pentamannoside (PIM₅) (Nojima 1959; Ballou et al. 1963) was defined by Lee and Ballou (1965) (Fig. 23). PIM₁ was isolated in small amounts, and it was proven that the mannose was attached to the 3-hydroxyl group of the myo-inositol ring (Ballou and Lee 1964).

Fig. 23

Mycobacterial polar lipid phosphatidylinositol inositol mannosides (PIMs): monoacyl and diacyl PIMs and point of linkage (X) for formation of lipomannan (LM) and lipoarabinomannan (LAM) lipoglycans

Pangborn and McKinney (1966) isolated from M. tuberculosis a series of PIM₂ lipids containing 2, 3, and 4 acyl residues. Brennan and Ballou (1967, 1968) also indicated the presence of di- and mono-acyl PIM₂ and PIM₅. Subsequent publications (Gilleron et al. 2001, 2003, 2006, 2008; Hsu et al. 2007a, b) have described the full extent of acylation in the PIM family in most mycobacteria, the principal components being mono- and di-acyl-PIM₂ and PIM₆ lipids (Fig. 23); M. leprae lacks the di-acylated components (Minnikin et al. 1985d). The biosynthesis of PIMs has been reviewed (Guerin et al. 2010; Angala et al. 2014; Sancho-Vaello et al. 2017). The major PIM₂ and PIM₆ lipid classes locate tuberculostearic acid on the sn-1 position of glycerol (Fig. 23) (Gilleron et al. 2003, 2006; Dyer et al. 2007; ter Horst et al. 2010b).

The mycobacterial nonlipid, water-soluble, lipomannans (LMs) and lipoarabinomannans (LAMs) incorporate a lipid anchor that is essentially PIM₂ (Fig. 23) (Hunter et al.1986b; Hunter and Brennan 1990). These highly important antigenic lipoglycans have been extensively reviewed (Chatterjee et al. 1992; Chatterjee and Khoo 1998; Nigou et al. 2003; Gilleron et al. 2008; Angala et al. 2014, 2018).

9 Other Lipophilic Molecules

9.1 Mycobactins

Mycobactins are lipophilic siderophores, involved in mycobacterial iron uptake and metabolism (Fig. 24) (Snow 1954a, b, 1965a, b, 1970; Ratledge 1982, 1999, 2004, 2013; Ratledge and Dover 2000; Quadri and Ratledge 2005; Quadri 2008; Horwitz and Horwitz 2014; Patel et al. 2018). Families of closely related mycobactins are found in most mycobacteria, excepting Mycobacterium paratuberculosis that requires mycobactin supplementation or added ferric ions (Snow 1970; Ratledge 2004). The original intracellular mycobactins are exemplified by the structure of mycobactin T from M. tuberculosis (Fig. 24) (Snow 1965b, 1970). More recently, a more hydrophilic variety of lipophilic iron chelaters has been recognized and termed “carboxymycobactins,” as shown for the example from M. avium (Fig. 24) (Lane et al. 1995; Ratledge and Ewing 1996); methyl esters of essentially the same compounds were isolated independently by Gobin et al. (1995). These amphipathic mycobacterial mycobactins and carboxymycobactins operate with water-soluble, peptide-based “exochelins” to assimilate essential iron (Ratledge and Dover 2000; Ratledge 2004; Quadri 2008).

Fig. 24

Mycobacterial mycobactins and carboxymycobactins: lipophilic siderphores for iron assimilation

9.2 Mycolactones

The causative agent of African Buruli ulcer, M. ulcerans, was found to produce a lipid toxin (George et al. 1998) that was characterized as a polyketide and given the name mycolactone (Fig. 26) (George et al. 1999; Gunawardana et al. 1999). Subsequent studies elaborated a family of related mycolactones (Fig. 25) (Stinear and Small 2008; Kishi 2011; Chany et al. 2013; Gehringer and Altmann 2017). The definitive lipids of M. ulcerans are interconvertible E/Z geometric isomers, at the 4′, 5′ double bond, labelled mycolactones A/B (Fig. 26) (Song et al. 2002). Investigations of reduced virulence Australian and Asian strains (Mve-Obiang et al. 2003) produced mycolactone C (Hong et al. 2003; Judd et al. 2004) and mycolactone D (Hong et al. 2005a), respectively (Fig. 25). The Japanese strain, M. ulcerans subsp. shinshuense had two keto variants, mycolactones S1 and S2 (Fig. 25) (Hande et al. 2012). Mycolactone E and a keto-analogue (E keto) were provided by “Mycobacterium liflandii,” a frog pathogen (Fig. 25) (Mve-Obiang et al. 2005; Hong et al. 2005b; Aubry et al. 2008; Spangenberg et al. 2010). M. marinum, infecting fresh-water fish, gave mycolactone F (Ranger et al. 2006; Kim and Kishi 2008), but M. marinum strains infecting salt-water fish had diastereomeric mycolactone dia-F (Fig. 25) (Kim et al. 2009).

Fig. 25

Mycolactone lipophilic toxins of Mycobacterium ulcerans and related taxa

Fig. 26

Mycoketides of Mycobacterium tuberculosis and Mycobacterium avium

9.3 Mycoketides

A CD1c-restricted, mycobacteria-specific T-cell line recognized previously unknown mannosyl lipids, of apparent isoprenoid nature, from M. tuberculosis and M. avium (Moody et al. 2000). Detailed studies revealed that these glycolipids were of polyketide origin and they were designated as mycoketides (Fig. 26) (Matsunaga et al. 2004; Quadri 2014). The structures of these mannosyl phosphomycoketides (MPMs) were confirmed by synthesis (van Summeren et al. 2006; Scharf et al. 2010; Buter et al. 2013).

9.4 Polymethylated Polysaccharides (PMPS)

During an investigation of the PIMs of M. tuberculosis and M. smegmatis, Lee and Ballou (1964b) originally reported on the presence of a 6-O-methyl-D-glucose-containing “polysaccharide-like material,” later characterized as an extraordinary mycobacterial lipopolysaccharide (Lee 1966). An early structure for these polymethylated polysaccharides (PMPSs) (Saier and Ballou 1968a, b, c) was revised (Forsberg et al. 1982), the key components comprising glucose, 3- and 6-O-methylglucoses, and glyceric acid (Fig. 27). Various acylation combinations of PMPSs involve acetate, propionate, isobutyrate, octanoate, and succinate (Keller and Ballou 1968; Gray and Ballou 1972; Smith and Ballou 1973; Narumi et al. 1973; Jackson and Brennan 2009). Acyl function heterogeneity of the PMPS of M. tuberculosis was confirmed (De et al. 2018), so only a representative structure is shown in Fig. 27. Succinate content can vary from zero to three, resulting in four main components. PMPSs have been described in a variety of mycobacteria and Nocardia (Lee 1966; Smith and Ballou 1973; Tuffal et al. 1995, 1998a, b; Jackson and Brennan 2009).

Fig. 27

Polymethylated polysaccharides (PMPS) of Mycobacterium tuberculosis

These important minor lipophilic PMPS molecules (0.01% of biomass; De et al. 2018) are possibly involved in the regulation of fatty and mycolic acid synthesis. The PMPSs and the related 3-O-methyl-mannose-containing polysaccharides from M. smegmatis (MMPs; Gray and Ballou 1971; Maitra and Ballou 1977) are likely fatty-acyl carriers, facilitating processing of long, insoluble fatty-acyl CoAs (Yabusaki and Ballou 1979; Yabusaki et al. 1979).

10 Conclusions

The object of this chapter is to provide, for the first time, a structural database for the lipids of clinically significant mycobacteria, appropriately referenced; however, it is informative to include a summary of the likely cellular location of such lipids. The wide array of lipid types, detailed in Sections 2 to 9, are principally associated with the two distinct mycobacterial cell envelope membrane bilayers (Fig. 28) (Minnikin et al. 2015). The cytoplasmic mycobacterial inner membrane (MIM) includes conventional polar lipids (Fig. 22); however, phosphatidylinositol dimannosides (PIM₂; Fig. 23) are the main components of the inner leaflet with high proportions of phosphatidylinositol hexamannosides (PIM₆; Fig. 23) in the outer leaflet (Bansal-Mutalik and Nikaido 2014; Minnikin et al. 2015). Non-lipid lipoarabinomannans (LM) and lipoarabinomannans (LAM) are anchored to the MIM outer leaflet by an analogue of PIM₂ (Fig. 23). It is considered that isoprenoid quinones (Fig. 18), carotenoid pigments (Fig. 19), halimane diterpenoids (Fig. 21), and mycolactones (Fig. 25) are MIM-associated. The isoprenoid lipid carriers (Fig. 20) and mycoketides (Fig. 26) probably originate in the MIM domain. Iron sequestering mycobactins (Fig. 24) associate with the MIM outer leaflet; polymethylated polysaccharides (PMPS, Fig. 27) are cytoplasmic but close to the MIM inner leaflet

Fig. 28

Essential anatomy and role of mycobacterial lipids in the inner and outer leaflets of mycobacterial cell envelope inner (MIM) and outer (MOM) membranes. Non-lipid lipomannan (LM) and lipoarabinomannan (LAM) lipoglycans are included, as is the arabinogalactan-peptidoglycan (ARA-GAL-PG) macromolecule to which the mycolic acids (MAs) are attached. See text and Figures 1–27 for other abbreviations

(Fig. 28).

The mycobacterial outer membrane (MOM) is the main location for the majority of the other types of mycobacterial lipids (Fig. 28). The inner leaflet of the MOM is considered to be a monolayer of long-chain mycolic acids (MAs, Fig. 1A–G), covalently bound to terminal arabinose units presented by an arabinogalactan-peptidoglycan macromolecule (Minnikin 1982; Brennan and Nikaido 1995; Minnikin et al. 2002, 2015). MAs can adopt different conformations to ensure appropriate integrity of the MOM inner leaflet (Villeneuve et al. 2005, 2007, 2010, 2013; Minnikin et al. 2015) (Fig. 28). Essentially, three classes of MOM outer leaflet free lipids, interacting with the bound mycolates to complete the MOM bilayer, can be discerned (Fig. 28). In the first class, phthiocerol dimycocerosates (PDIMs; Fig. 13A, B), pentaacyl trehaloses (PATs; Fig. 7B), trehalose polyphleates (TPPs; Fig. 10), and tetraacylated sulphoglycolipids (Ac₄SGLs; Fig. 9) are relatively inert and probably enhance cell surface hydrophobicity (Minnikin et al. 2015; Jankute et al. 2017). The intermediate second class comprises antigenic di- and triacyl trehaloses (DATs and TATs; Figs. 7A, B and 8), diacylated sulphoglycolipids (Ac₂SGLs; Fig. 9), apolar glycopeptidolipids (aGPLs; Fig. 16A–G), peptidolipids (PLs; Fig. 17), and apolar phenolic glycolipids (PGLs; Fig. 15B, D). Strongly antigenic hydrophilic lipooligosaccharides (LOSs; Fig. 11A–G), polar glycopeptidolipids (GPLs; Fig. 16A–G), and polar phenolic glycolipids (PGLs; Fig. 15A–D) are the third class. Information is required regarding the precise location of mono-mycoloyl glycerols (MMGs; Fig. 4), glucose monomycolates (GMMs; Fig. 3), monomycobacteroyl diacylglycerols (MMDAGs; Fig. 12), and mycolate ester waxes (MEWs; Fig. 6), though MOM association is suspected. Similarly, mono- and dimycoloyl arabinoglycerols (MMAG and DMAGs; Fig. 5), analogues of the terminal unit of the mycoloyl arabinogalactan-peptidoglycan macromolecule, are probably MOM associated. Trehalose monomycolate (TMM) (Fig. 2) is involved with incorporation of MAs into the mycoloyl arabinogalactan-peptidoglycan and trehalose dimycolate (TDM) (Fig. 2) is likely to be associated with the MOM inner leaflet.

11 Research Needs

Specific research needs have been highlighted in the majority of the structural figures, so only general points are listed below.

• Complete the structural details of all lipids from known pathogenic mycobacteria.

• Fully investigate the lipid composition of all related pathogenic mycobacteria.

• Define the physical and conformational behavior of all mycolic acids and various conjugates in order to comprehend the organization of the covalently bound mycolic acid inner leaflet of the mycobacterial outer membrane (MOM).

• Study the interaction of representative mycolic acid monolayers with individual and combinations of MOM outer leaflet free lipids to explore the essential composition and behavior of the lipid domains of the mycobacterial outer membrane.

• Perform a thorough investigation of the characteristics of the special mycobacterial inner membrane (MIM), particularly the role of the phosphatidylinositol mannosides (PIMs).

• In short, assemble a reliable database comprised of information about the structures, cellular location, and function of mycobacterial lipids to inform parallel studies on lipid genomics, biosynthesis, biological activity, and drug targets.