Review Omega-3 fatty acids in cellular membranes: a unified concept Raymond C. Valentine a , David L. Valentine b, * a Calgene Campus, Monsanto, Inc. and Professor Emeritus, University of California, Davis, CA 95616, USA b Department of Geological Sciences and Marine Science Institute, University of California, Santa Barbara, CA 93106, USA Accepted 27 May 2004 Abstract The Omega-3 fatty acid DHA (docosahexaenoic acid, 22:6) and its sister molecule EPA (eicosapentae- noic acid, 20:5) are highlighted here. These highly unsaturated fatty acids are widespread in nature, espe- cially in the marine environment, and are essential in membranes ranging from deep sea bacteria to human neurons. Studies of DHA/EPA in bacteria have led to a working model on the structural roles of these mol- ecules and are described in this review. The main points are: (a) genomic analysis shows that genes encoding the DHA/EPA pathways are similar, supporting the idea that structural roles in bacteria might be similar, (b) biochemical analysis shows that DHA and EPA are produced in bacteria by a polyketide process dis- tinct from the pathway of plants and animals; this allows DHA and EPA to be produced in anaerobic or oxygen-limited environments, (c) regulatory systems triggered by temperature and pressure have been iden- tified and studied, and add to the understanding of the roles of these molecules, (d) DHA/EPA bacteria are located almost exclusively in the marine environment, raising the prospect of an important linkage between membrane processes and marine conditions, (e) physiological studies of an EPA recombinant of E. coli show that EPA phospholipids contribute essential fluidity to the bilayer and that an EPA-enriched mem- brane supports a respiratory lifestyle dependent on proton bioenergetics; the EPA recombinant displays other physiological properties likely attributed to high levels of EPA in the bilayer, and (f) chemical studies such as chemical dynamic modeling support the idea that DHA and presumably EPA contribute hyperflu- idizing properties to the membrane. We hypothesize that DHA/EPA phospholipids contribute fluidity and other properties to the bilayer which distinguish these highly unsaturated chains from monounsaturates and polyunsaturates such as 18:2 and 18:3. We further hypothesize that the structural properties of 0163-7827/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.plipres.2004.05.004 * Corresponding author. E-mail addresses: [email protected](R.C. Valentine), [email protected](D.L. Valentine). Progress in Lipid Research 43 (2004) 383–402 Progress in Lipid Research www.elsevier.com/locate/plipres
Transcript
Progress in Lipid Research 43 (2004) 383–402
Progress inLipid Research
www.elsevier.com/locate/plipres
Review
Omega-3 fatty acids in cellular membranes: a unified concept
Raymond C. Valentine a, David L. Valentine b,*
a Calgene Campus, Monsanto, Inc. and Professor Emeritus, University of California, Davis, CA 95616, USAb Department of Geological Sciences and Marine Science Institute, University of California,
Santa Barbara, CA 93106, USA
Accepted 27 May 2004
Abstract
The Omega-3 fatty acid DHA (docosahexaenoic acid, 22:6) and its sister molecule EPA (eicosapentae-
noic acid, 20:5) are highlighted here. These highly unsaturated fatty acids are widespread in nature, espe-
cially in the marine environment, and are essential in membranes ranging from deep sea bacteria to humanneurons. Studies of DHA/EPA in bacteria have led to a working model on the structural roles of these mol-
ecules and are described in this review. The main points are: (a) genomic analysis shows that genes encoding
the DHA/EPA pathways are similar, supporting the idea that structural roles in bacteria might be similar,
(b) biochemical analysis shows that DHA and EPA are produced in bacteria by a polyketide process dis-
tinct from the pathway of plants and animals; this allows DHA and EPA to be produced in anaerobic or
oxygen-limited environments, (c) regulatory systems triggered by temperature and pressure have been iden-
tified and studied, and add to the understanding of the roles of these molecules, (d) DHA/EPA bacteria are
located almost exclusively in the marine environment, raising the prospect of an important linkage betweenmembrane processes and marine conditions, (e) physiological studies of an EPA recombinant of E. coli
show that EPA phospholipids contribute essential fluidity to the bilayer and that an EPA-enriched mem-
brane supports a respiratory lifestyle dependent on proton bioenergetics; the EPA recombinant displays
other physiological properties likely attributed to high levels of EPA in the bilayer, and (f) chemical studies
such as chemical dynamic modeling support the idea that DHA and presumably EPA contribute hyperflu-
idizing properties to the membrane. We hypothesize that DHA/EPA phospholipids contribute fluidity and
other properties to the bilayer which distinguish these highly unsaturated chains from monounsaturates
and polyunsaturates such as 18:2 and 18:3. We further hypothesize that the structural properties of
0163-7827/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.
Omega-3 fatty acids such as DHA have become popularized because of their important roles inhuman health and development [1,8,71]. EPA/DHA have important structural roles (e.g. in nerv-ous tissue) and also serve as precursors for hormones such as inflammatory mediators. The studyof eicosanoid hormones is well established [26]. In this review we have chosen to focus on thestructural roles of DHA/EPA in bacteria with applications for higher organisms. This has ledto a new way of thinking about the roles of DHA/EPA in cellular membranes.
Recently, chemists have used 2H-NMR and molecular dynamics simulations to study theshapes of DHA in membranes [6,7,23,25,37,64]. Whereas a detailed summary is beyond the scopeof this article we believe that chemical studies provide some important insights into how thesemolecules impact membrane structure in living cells. For example, there appears to be a generalagreement that DHA is relatively shorter or more compact than more saturated chains. In thisregard Fernandes et al. [25] show that DHA chains have an average length of 8.2 A at 41 �C com-pared to 14.2 A for oleic chains. This is consistent with a DHA conformation with pronounced
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 385
twists of the chain which diminish the separation between both ends. There is also increasing evi-dence that the helical-like structure is neither uniform nor rigid, resulting in backbent shapes inwhich the methyl end does not lie in the luminal region of the bilayer. Instead, the methyl groupwith its added bulk is located in the interior region (see [37]). These authors also show that thepresence of DHA chains in mixed-chain phospholipids leads to a marked increase in the conform-ational disorder of the saturated chain. The compact shapes seen in chemical studies are expectedto result in a larger effective volume which might contribute hyperfluidity to the bilayer. In addi-tion the extra bulk contributed by the methyl ends of the fatty acid chains might influence perme-ability properties of the bilayer [32].
Whereas much of this review deals with DHA/EPA it is clear these molecules share many prop-erties with other polyunsatured fatty acids such as 18:2, 18:3 and 20:4. Indeed it is well knownfrom chemical studies that there is an important linkage between unsaturation and membrane flu-idity, but that increasing the number of cis-double bonds in the chain beyond two (example 18:2)has a lessening effect on gel to liquid crystalline phase transition (Tm) and melting point [14,78].Evidence on the high fluidity of 22:6 phospholipids is reviewed by Salem et al. [65]. At this point itis important to note that the relationship between shape and physiological function of DHA/EPAhas not yet been defined by direct observation in living cells. Hazel [34] has questioned the signif-icance of fluidity in the function of these molecules.
The discoveries of EPA bacteria about 27 years ago [45] followed by DHA bacteria 18 yearsago [19] provided a new research tool for studying the mode of action of these chains in the bilayerof a simple cell. Molecular genetic strategies are emphasized here for understanding the roles ofOmega-3 fatty acids in cellular membranes.
Table 1
Summary of fatty acids discussed in the text
Name Symbola Comments
Saturated
Myristic 14:0 Small amounts found in phospholipids of EPA recombinant
Palmitic 16:0 Common to nearly all organisms; pairs with EPA in recombinant
Stearic 18:0 Common in membranes of chloroplasts, mitochondria and rhodopsin disks
Monounsaturated
Palmitoleic 16:19 Replaced by EPA in recombinant
Oleic 18:19 Constituent of plant oils; common in animals
cis-Vaccenic. 18:111 Widespread in bacteria such as E. coli; often in di-18:111 form
Polyunsaturated
Linoleic 18:29,12 Major constituent of plant oils; important in animal nutrition
Gamma-linolenic a-18:36,9,12 Same as 18:2
Alpha-linolenic c-18:39,12,15 Rare in land plants, but common in marine plants
Highly unsaturated
Arachidonic 20:45,8,11,14 Common in animal membranes; precursor to hormones; very rare in
land plants
Eicosapentaenoic 20:55,8,11,14,17 Major constituent of fish oils; precursor of crucial hormones
Docosahexaenoic 22:64,7,10,13,16,19 Abundant in fish oils; enriched in axons, rhodopsin disks, mitochondria
and chloroplasts of marine algae
a Subscripted numbers denotes locations of cis-double bonds counting from carboxyl group.
386 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
The purpose of this article is three-fold: (a) to summarize recent advances in the moleculargenetics of Omega-3 genes, which set the stage for understanding functions; (b) summarize evi-dence on the unique properties of these chains in bacterial membranes; and (c) apply this knowl-edge toward DHA or EPA-enriched membranes of animals and plants. The different classes offatty acids discussed below are listed in Table 1.
2. Genomic and biochemical analysis of Omega-3 genes
Recently the genomic sequence of an EPA bacterium, Shewanella oneidensis, was reported [35].This is a fresh-water isolate, but is expected to be similar to marine Shewanella discussed below.The EPA genes in this strain were initially observed to be silent [79], but we have found that incu-bation at cold temperatures (e.g., 3 �C) results in expression (unpublished results). A large numberof fatty acid-related genes have been identified in this organism which has evolved three distinctpathways for synthesis of fluidizing chains. The need for cross-talk among these pathways is dis-cussed in Section 3.
2.1. EPA and DHA genes
Initial insight into the molecular genetics of EPA biosynthesis was gained by the cloning,sequencing, and complementation analysis of a 38 kbp genomic fragment from the fish gut isolateShewanella putrefaciens strain SCRC-2738 [Genbank accession number U73935] [87]. Five Shewa-nella genes, designated ORFs 2, 5, 6, 7 and 8, were shown to be necessary for EPA synthesis in arecombinant of E. coli and in the marine cyanobacterium Synechococcus sp. [75,87]. A compari-son of Omega-3 gene clusters from three different marine bacteria is shown in Fig. 1 [Genbankaccession numbers AF409100 (SS9); U73935 (Shewanella); and AB025342 (M. marina)]. It shouldbe noted that the former two strains (i.e. SS9 and Shewanella) produce EPA whereas the latterisolate, M. marina, produces DHA. Fig. 1 and associated nomenclature used here are from thepaper of Allen and Bartlett [3]. Note the strong similarities between EPA versus DHA genes.
2.2. Polyketide (anaerobic) pathway
Cell-free extracts of the EPA recombinant provided a ‘‘low background’’ for biochemical tracerstudies using 13C-labeling. This data, along with the identification of multiple enzyme domainswithin EPA gene products, led Metz et al. [54] to conclude EPA production shares many featureswith polyketide synthesis. The combined activities of these domains include condensation reac-tions (KS domains), acyl CoA:ACP transfer reactions (AT), multiple acyl carrier protein domains(ACP), ketoacyl reduction reactions (KR), chain length factor domains (CL) presumably involvedin decarboxylation reactions, dehydratase/isomerase reactions (DH/I), and enoyl reduction reac-tions (ER) [3]. These domains presumably catalyze the repetitive steps in building the growing acylchain. Molecular oxygen is not involved in any of these steps.
The presence of repetitive ACP domains is unique to EPA and DHA synthases (Fig. 1) withSS9 pfa A possessing five ACP domains, Shewanella sp. six, Moritella five, and Schizochytrium
P. profundum..strain SS9
Shewanella sp.SCRC-2738
Moritella marinastrain MP-1
PfaA PfaB PfaC PfaD
-Ketoacyl-ACP Synthase
Acyl CoA-ACP Transacylase
Acyl Carrier Protein
-Ketoacyl-ACP Reductase
-Hydroxyacyl-ACP DehydraseIsomerase
Enoyl Reductase
Omega-3 Polyunsaturated Fatty Acid Biosynthesis
Involves Large Multiple Domain Proteins
PfaB PfaC PfaD
Fig. 1. Omega-3 gene clusters and enzyme domains of the anaerobic pathway from three marine isolates. The cluster
labeled SS9 and Shewanella are EPA genes, whereas M. marina is a DHA strain. This figure is reproduced from [3], and
is discussed in detail in the text. Note the strong similarities between EPA and DHA gene clusters. The Shewanella
cluster has been transferred to a mutant of E. coli unable to make its own fluidizing fatty acids, resulting in high-level
expression of EPA [54].
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 387
nine [54,76]. The growing acyl chains are presumably bound covalently to these ACP groups asthioesters with AT domains being required for the loading of the starter and extender units.Clearly, the ability to introduce multiple double bonds into a single acyl chain in the absenceof O2 highlights a major difference with desaturase systems. The ability to produce double bondsanaerobically likely arises from the activities of the DH/I domains present in the microbial synth-ases (bacterial PfaC homologues and Schizochytrium ORF C). Such dehydration/isomerizationreactions might be analogous to those catalyzed by FabA (b-hydroxydecanoyl-ACP dehydratase)in bacterial monounsatured fatty acid synthesis, also an anaerobic process [18].
There are many important biochemical questions yet to be answered such as the nature of thefinal enzyme product, as well as the apparent anchoring of the bacterial complex to the membrane[54]. Harnessing these genes for production of DHA and EPA is an important applied goal [80].The interesting evolutionary question of the possible horizontal transfer of these genes has beendiscussed by Allen and Bartlett [3] as follows:
‘‘The high degree of sequence similarity between the bacterial (Shewanella sp. SCRC-2738,M.marina and SS9) and the Eukaryotic microbe Schizochytrium pfa genes suggests the possibleinvolvement of horizontal gene transfer in the acquisition of the pfa gene clusters in the mar-ine environment. However, among the three bacterial strains whose pfa gene clusters havebeen cloned and sequenced, no sequence conservation flanking the pfa clusters is observedwith the exception of a single undefined ORF located upstream of pfaA in SS9 andMoritella.Furthermore, there is no apparent GC bias among the pfa A–D genes nor is there indicationof flanking genes possessing functions which could facilitate horizontal transfer.’’
388 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
3. Regulation
Progress has been made toward understanding the regulation of bacterial EPA andDHA production, information which provides insight into the roles of these moleculesin the bilayer. This area has recently been summarized by Allen and Bartlett [3]. Indeed,deep-sea isolates subjected simultaneously to high hydrostatic pressures and temperaturesapproaching 0 �C often produce a membrane composition approximating the unsaturationlevels found in axonal membranes [19,65]. The highest enrichment among natural isolatesreported to date is 37% total fatty acid chains with EPA and greater than 20% withDHA [19].
Bacteria are clearly masters at blending fatty acids into their membranes in order to fittheir lifestyles [5,46]. As mentioned above, Omega-3 bacteria have at least three separateroutes for synthesis of fluidizing chains including production of methyl-branched andmono-unsaturated lipids in addition to the EPA or DHA routes; this requires cross-talk withregard to coordination of lipid composition. Different cellular stages of the lifecycle, such asthe dormant state [24], are expected to require different fatty acids [3]. For example, methylbranched chains are fluidizing but lack double bonds, the latter property is expected to sta-bilize the membrane during long periods of exposure to O2 during the dormant stage. Someinitial insight into interplay between pathways has been provided by EPA gene transcriptanalysis of a mutant of P. profundum that overproduces EPA nearly five-fold compared tothe wild-type [3]. In addition, this strain greatly underproduces monounsaturated fatty acids[2]. The requirement for a transcription factor that modulates pfa gene expression is an inter-esting possibility. Transcriptional analyses indicate that the pfa gene cluster is organized intotwo operons, pfa A–C and pfa D of SS9 [3]. The transcriptional start of pfa A has beenmapped to 169 bp upstream of the translational start.
How environmental signals such as changing temperatures and pressures modulate synthe-sis of EPA or DHA in bacteria remains unknown with evidence so far favoring direct bio-chemical control at the level of the EPA or DHA synthase reaction [3]. These authors pointout the possible similarity with the low temperature activation of KASII (b-ketoacyl – ACPsynthase II) catalyzing synthesis of cis-vaccenate (18:111). How ‘‘fluidity’’ acts as a trigger fordesaturase expression and activity in cyanobacteria has recently been reviewed by Murata andcolleagues [55]. These studies provide the clearest picture yet of the mechanism of genetic reg-ulation of desaturases. This mechanism is likely important in modulating synthesis of thebulk of DHA/EPA synthesis in marine phytoplankton. In contrast to phytoplankton, a link-age between fluidity and DHA/EPA synthesis in bacteria has not yet been established but isan interesting area for future research.
4. Distribution and molecular ecology
The list of bacteria with EPA or DHA genes continues to grow and has led to the first prelim-inary classification of these strains [20,44,63]. In this section we consider the distribution and ecol-ogy of these strains.
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 389
4.1. Distribution
So far EPA and DHA genes have been found only in gram-negative bacteria, which synthesizean inner and outer membrane structure similar to E. coli [63]. Indeed, a recombinant of E. coli hasbeen constructed which produces high levels of EPA phospholipids [54] and a variety of bacteriamissing EPA or DHA genes incorporate these molecules when fed exogenously [82]. However,these authors found that E. coli incorporates only traces of DHA fed exogenously and we haveconfirmed this result in finding no growth of unsaturated fatty acid requiring mutants fed EPAor DHA. In contrast, polyunsaturates such as linolenate (e.g. a-18:3) are well known to be effec-tive [17]. This has led to the idea that even when present in the natural environment, the risks ofthese molecules to E. coli might outweigh their benefits. This idea is mentioned here because itprovides a framework for understanding the unique distribution pattern of DHA and EPA inthe biosphere, a point taken up later in this section.
The major point from classification studies is that genes for anaerobic production of EPA orDHA are virtually always found in the marine environment [19,63] raising the obvious possibilitythat these genes have selective advantage in and are tailor-made for this environment. However,as mentioned above, the marine-only rule seems to be broken with the finding of EPA genes in afreshwater isolate called Shewanella oneidensis, although this is the only exception to date [35].Also, one of the biggest surprises is the presence of the anaerobic pathway in certain marine fungi[54].
The classification of Omega-3 bacteria has shed light on another interesting aspect regardingthe vertical distribution pattern of EPA versus DHA bacteria in seawater. The point is thatDHA bacteria were first thought to be found only in the deep sea, compared to EPA strains whichare found in both deep and shallow seas [19,44,63]. However, later studies showed that DHAstrains are present in high levels in the guts of marine invertebrates living in cold, shallow seasor on sea ice [44]. These authors also discuss the possibility that Omega-3 bacteria living in thegut might be involved in symbiotic associations with their hosts, although direct evidence on thispoint is lacking.
4.2. Ecology
Studies on the distribution pattern of Omega-3 genes in the biosphere have provided otherimportant clues about the role of these fatty acids in cellular adaptation to the marine world. Sev-eral key ecological parameters are summarized as follows:
� O2 – EPA and DHA are the most unsaturated and most oxidatively unstable fatty acids foundin nature [30] and consequently require specialized protective strategies [28]. Bacteria appear touse, perhaps, the most primitive mechanism – avoiding oxidation as much as possible, seekingout environments low in molecular oxygen, or where kinetics of oxidation are slow. The anaer-obic nature of this pathway is consistent with this concept. It is also likely that Omega-3 bac-teria must avoid direct sunlight and warm temperatures which are known to dramaticallyincrease oxidation rates [30]. As discussed above, marine bacteria may also need to ‘‘time’’Omega-3 synthesis in order to avoid oxidation of their dormant-stage bilayers where cellsare often exposed to O2 for long periods of time while adrift in sea water [24].
390 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
� Temperature – The authors are not aware of DHA bacteria which grow above 20 �C [19]. Thisis consistent with the extreme instability of these membranes at warmer temperatures, inessence requiring that DHA bacteria maintain a strictly cold-dependent lifestyle. For example,DeLong and Yayanos [19] reported that cells of Vibrio (now Moritella) marinus MP-1 grown at20 �C contained about 17% of the levels of DHA compared to cells grown at 5 �C. Clearly,DHA synthesis is sharply down-regulated as temperatures approach 20 �C. We believe that thismechanism helps protect the cell against oxidative damage.
� Hydrostatic pressure – This important parameter has been reviewed by Allen and Bartlett [3] interms of regulation of EPA synthesis and by DeLong and Yayanos [19] concerning growth ofdeep-sea isolates. For example, DeLong and Yayanos [19] point out that a majority of isolatesof deep-sea bacteria synthesize DHA or EPA. Also these workers showed that synthesis ofDHA/EPA increases as a function of pressure. Therefore, it seems likely that DHA/EPA playan important but not exclusive role in adaptation to high pressure.
� Salinity – The high surface to volume ratio characteristic of marine bacteria exposes these cellsto toxic levels of Na+ which must be constantly pumped from the cell. Haines [32] has pointedout that the energy cost to drive Na+-pumps is often staggering and apparently consumes amajority of total cellular energy. Hulbert [38] also emphasizes this point. We believe thatEPA/DHA production is one of many mechanisms of adaptation to high salt but that themechanism is indirect and involves energy production as discussed in detail in Section 5. Thefact that most marine bacterial isolates lack EPA/DHA but are obviously adapted to high saltshows that this is not a universal mechanism for salt tolerance. One of the most interesting eco-logical questions in this field concerns the widespread occurrence of EPA/DHA in membranesof marine plants (i.e. phytoplankton) in contrast to the complete absence in land plants. Howmarine plants protect their EPA/DHA enriched membranes against photo damage is especiallyinteresting.
In summary, Omega-3 bacteria stand out in living in an extremely narrow ecological zone al-most always characterized by relatively high Na+ levels, which is toxic and costs considerable en-ergy to pump out of the cell. In many cases various combinations of cool temperatures, highhydrostatic pressures and darkness prevail. This means that in natural environments, Omega-3cells are likely subjected to cycles of ‘‘energy stress’’ in contrast to the idealized conditions inwhich these organisms are generally cultured in the laboratory. It is interesting to speculate thatenergy stress is the selective force behind the evolution of Omega-3 genes.
5. Functions of EPA and DHA in cellular membranes
EPA and DHA are found in membranes throughout the biosphere. In this section potentialroles of EPA and DHA in bacterial as well as Eukaryotic bilayers are discussed.
5.1. Bacterial membranes
The roles of EPA or DHA in bacterial membranes is of great interest and has implications forunderstanding the biochemical functions of these unique fatty acids in plants, animals and hu-
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 391
mans. A classic approach toward understanding an unknown pathway or product is to block thegene(s) using mutational strategies. Several years ago, we succeeded in isolating a class of EPAminus mutants of Shewanella sp. [22] using a mutational approach involving chemical mutagen-esis followed by screening of large numbers of cold-sensitive mutants (Fig. 2(a)). The parent is amarine Shewanella putrefaciens isolate which produces up to 10–15% EPA [88] in contrast to mu-tants such as cs-22 and cs-52 which lack EPA. Growth of the mutants at 4 �C was found to bedependent on EPA spread as a thin layer on the surface of a petri dish (the plate on the right con-tains 50 lg EPA/plate). This bioassay is adapted from studies of the unsaturated fatty acidrequirement of E. coli [16]. EPA was omitted from the control on the left. As shown in Fig.2(a), growth at 4 �C requires EPA which reaches levels of 5% in the membrane phospholipidsof the mutant cells. However, other polyunsaturated fatty acids common in marine oils, includingc-18:3 and 20:4, are also effective (Fig. 2(b)). Linoleic (18:2) shows only traces of activity whereas16:1, 18:19 and 18:111 are not effective in this assay. Preliminary experiments showed that DHAsupports growth, but in later studies it was found that this effect is due to preferential uptakeof EPA present in trace amounts in the DHA preparation. There is also a possibility thatDHA is taken up and then converted to EPA in the cell [82]. The lack of activity of a-18:3 isnot understood, but might be accounted for by the lack of passage through fatty acid porins.It should be noted that the specificity of the fatty acid requirements of Shewanella (4 �C assay)are markedly different compared to E. coli (42 �C assay). As discussed below, unsaturated fattyacid auxotrophs of E. coli accept fatty acid chains with 1–3 double bonds, but not more. As shownin Fig. 2(c) the levels of EPA needed to satisfy the maximum growth requirement are comparableto the values reported for E. coli (i.e., about 10–100 lg fatty acid/plate surface). However we haveobserved in a number of bioassays that EPA levels above 100 lg/plate surface leads to growthinhibition of Shewanella. The basis of this effect is unknown.
Interestingly, the EPA minus cells have lost their natural tolerance to low levels of severalantibiotics including streptomycin, kanamycin, and neomycin. This effect is studied using plateassays at 14 �C. EPA was found to restore tolerance in this assay, but the specificity pattern isless stringent than for the cold temperature assay. For example, monounsaturates showed someactivity. Some background information is necessary to explain this experiment. It is wellknown that deenergization of the cytoplasmic membrane results in the apparent permeationof certain antibiotics [58]. However, it is now clear that it is not permeability itself but ratherenergy-requiring efflux systems that are often responsible for this effect. For example, uncou-plers and presumably any form of severe energy stress would simply inhibit the active effluxprocesses, thereby increasing the cellular concentration of these toxic substances. Interestingly,any toxic substance from Na+ to classic antibiotics, capable of being actively effluxed from thecell, might display this effect. This is because the efflux of toxic molecules against a concentra-tion gradient requires energy in the form of ATP or proton-motive force. In a general senseany interruption in energy supply would be expected to increase sensitivity or decrease toler-ance. EPA bacteria such as Shewanella oneidensis [70] are known to possess efflux systemsof this kind. Initially, we thought that the EPA requirement for antibiotic resistance mightbe explained by some specific effects of fluidity on the efflux enzymes themselves. However,it is difficult to explain the cold sensitive phenotype by this mechanism. An alternative expla-nation is that the EPA mutants are energy stressed because of under-fluidization of their res-piratory membranes.
392 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
The next experiment involves a novel EPA recombinant of E. coli used here as a ‘‘reporter’’ forEPA function. The fatty acid profile of the EPA recombinant shows that EPA is the only signif-icant fluidizing fatty acid available to the cells (Fig. 3). This confirms earlier studies with yeast,which show that DHA/EPA are able to fluidize the membranes of a Eukaryotic cell [83,81]. Notethat about 40% of the total fatty acids in the membrane of the EPA recombinant are EPA andthat other potential fluidizing species are largely missing. The large peak on the left side of thefatty acid profile corresponds to the major saturated fatty acid, 16:0 which pairs with EPA[87]. Some 14:0 is also made (5–9%) and might contribute some fluidity. This is the highest levelof EPA yet reported in bacteria and also represents the most highly unsaturated membranes seenin E. coli. As mentioned above, we have found that an unsaturated fatty acid-requiring mutant ofE. coli fails to incorporate significant levels of EPA fed exogenously. However, it is clear that EPA
Fig. 3. EPA supplies essential fluidity to E. coli. Cells for fatty acid analysis [54] were grown at 18 �C for four days on
the surface of L-broth plates containing a total of 0.3 M NaCl. Note the major peak for 20:5 (right) and its saturated
partner 16:0 (left). The high 20:5 content (�40%) makes this one of the purest Omega-3 bilayers yet available for study.
Also, note the virtual absence of other fluidizing chains; monounsaturated chains normally comprise >50% of total
fatty acids under these conditions. Mass spectrometric analysis shows that EPA chains are not a substrate for
cyclopropanylation, leaving their cis-double bonds exposed.
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 393
generated internally enters the membrane and contributes properties essential for growth of E. colieven on strictly respiratory substrates such as proline or succinate. The point is that an EPA-en-riched membrane clearly supports proton bioenergetics in E. coli.
Fig. 2. Unsaturated fatty acid requirement of a cold-sensitive mutant of Shewanella putrefaciens (2738). The strain and
culture medium are described by Metz et al. [54]. Chemical mutagenesis was carried out by growing a culture of S.
putrefaciens at 30 �C for 24 h in marine broth containing 5 lg ml�1 of nitrosoguanidine. About 12,000 colonies were
plated on marine agar and incubated at 30 �C for 48 h. Colonies from each plate were transferred by replica plating
methods to two fresh plates, one incubated at 4 �C (non permissive temperature) for four days and compared with sister
plates incubated at 30 �C (permissive temperature). About 50 cold-sensitive mutants showing poor or no growth at 4 �C,while displaying near-normal growth at 30 �C, were selected in this first screen. In the second screening the 50 or so
mutants were individually transferred to fresh plates using a grid design, with one plate serving as a control and the other
containing EPA spread evenly over the plate surface. Plates were incubated at 4 �C for several days until control patches
showed growth. The purpose of a second screen was to detect EPA-requiring mutants. Five mutants showed a strong
growth response to EPA with the mutant cs-52 chosen for further study. (a) A mutant strategy for studying the functions
of EPA in bacterial membranes. To obtain this picture, the cold-sensitive mutant, cs-52was spread along with 50 lg EPAon the surface of a marine agar plate (right). The left plate was a control spread with water-ethanol solvent used to dilute
the fatty acid sample. Incubation was at 4 �C for one week. Note the complete requirement for EPA at this cool
temperature. (b) Fatty acid specificity. Plate assay as for A, except 50 lg/plate surface of various fatty acids were used.
Cells from about 60 colonies were quantitatively washed from the plate surface and diluted using fresh medium for
measurement of cellular density at 600 nm using a spectrophotometer. (c) Saturation curve. Assay conditions as for (a)
and (b), except EPA levels were from 0–100 lg per plate were used. Note that results here were obtained by the plate
assay, with cultures aerated by shaking retaining their EPAminus phenotype but showing elevated levels of 16:1 or 18:111which appear to compensate for the absence of EPA. Standard errors are not included. We found that while the relative
activities of the fatty acids remained constant in different experiments that the growth yield, and thus the absolute
activities of the fatty acids (and the absolute error) varied.
b
394 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
However, the EPA recombinant does not behave like a typical E. coli cell. Some highlights ofmicrobiological experiments are summarized as follows:
� The temperature range of the recombinant is restricted to about 12–22 �C and more closelyresembles a psychrotolerant marine isolate than E. coli.
� Cellular growth near the maximal range of temperature is dependent on osmotic strengthapproaching sea-water and a break-point for surface growth occurs around 0.2 M NaCl. Aroughly linear relationship between NaCl levels, growth and EPA levels occurs in the range0.2–0.35 M NaCl.
� A variety of experiments such as the requirement for micro-aerophilic conditions for maintain-ing cultures in the laboratory highlight a novel relationship with O2 levels; for example, Facci-otti [57] showed that recombinant cells are markedly more sensitive to photo-oxidation than acontrol culture. This confirms earlier results on the oxidative instability of a DHA-enrichedEukaryotic bilayer [28].
The main point in describing these various experiments is to show that EPA and DHA providekey properties to bacterial membranes. Perhaps the most important generality to be drawn frommicrobiological experiments is that EPA is a ‘‘double-edged sword’’ for E. coli offering apparentlymajor benefits as well as risks. This might help explain the virtual absence of these fatty acids inmembranes of land plants [86] and terrestrial bacteria [44] where risks might outweigh benefits.There is also the idea that benefits must be substantial to compensate for the risks.
5.2. Mitochondria and chloroplasts
The membranes of mitochondria and chloroplasts, which are the major energy transducingorganelles in Eukaryotic cells, are sometimes enriched in DHA or EPA [10,15,33,38,49,61,66,67,72,73]. In this section, evidence supporting the role of DHA and EPA in enhancing energyproduction in these organelles is summarized. Mitochondria are discussed first.
The literature on the role of DHA in energy metabolism of whole animals has recently beenreviewed by Hulbert [38]. These studies have led to an interesting theory in which DHA-enrichedmembranes are correlated with high metabolic rates of tissues such as heart and skeletal muscles[39,40]. This concept is called ‘‘membrane pacemaker theory’’ and seems to account for a greatdeal of data on the correlation of Omega-3 levels with metabolic rates [38]. It is interesting to notethat the ‘‘double-edged sword concept,’’ mentioned above for EPA or DHA bacteria, seems toapply to DHA mitochondria. In this case, proton barrier properties of DHA-enriched mitochon-dria are negatively affected [38], while the net energy production is increased. Although these aresome of the most interesting studies published to date on Omega-3 functions in animal mitochon-dria, the molecular roles of DHA phospholipids are not yet understood.
The pathways of electron transport and energy production in mitochondria are known in greatdetail [69]; we emphasize the many similarities with chloroplasts.
The current picture of photosynthetic electron transport in EPA or DHA enriched chloroplastsis as follows: Two photosystems (i.e., PSI-PSII) cooperate to transfer electrons from H2O toNADP+ and to produce ATP. According to this model, mechanisms for long distance electrontransfer must exist because the two systems are spatially separated by long distances across the
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 395
bilayer (see reviews by [51,62,84]). In other words, ‘‘long distance electron transport’’ is definedhere as the movement or shuttling of electron carriers, coupling donor and acceptor enzyme com-plexes, which are separated by relatively long distances along the chloroplast membrane. It isimportant to note that this definition does not apply to water soluble carriers such as plastocyaninwhich move through the aqueous phase. Only lipid-soluble, plastoquinone carriers, working theirway through the interior portion of the bilayer, would be expected to be influenced by fatty acidcomposition. What roles might DHA and EPA play in long distance electron transport?
There is increasing biochemical evidence supporting the concept that the plastoquinol(PQ)-cyt-ochrome b6f reaction along the electron transport chain represents the rate-limiting step in pho-tosynthesis. This step involves collisions between plastoquinone-cytochrome b6f [9,43,48,85]. Asalready mentioned, plastoquinone works as a mobile substrate within the membrane linkingthe photo-system II complex and the cytochrome b6f complex. Blackwell [9] show that this mech-anism often involves long distance diffusion of PQ within the lipid bilayer phase of the thylakoidmembrane. They summarize the evidence that the quinol oxidation rate is determined by the PQ-cytochrome b6f encounter frequency.
We propose highly fluidized phospholipids might increase the PQ-cytochrome b6f encounterfrequency, thus enhancing energy production. For example, quinone diffusion rates measured inpure lipid vesicles were found to be 1–2 orders of magnitude higher than those measured in theinner mitochondrial membrane, an effect attributed to obstruction by respiratory chain proteins[21,31]. In other words, in addition to long distances, it is now believed that quinone moleculestraveling through the lipid phase must journey or percolate their way through an obstaclecourse consisting of densely packed proteins or enzymes. Many years ago, McElhaney summa-rized physical evidence that growth of bacteria is not slowed until greater than 50% of theirphospholipids are hardened or in the gel-phase [52]. This means that quinone carriers encoun-tering these ‘‘frozen’’ patches or islands [68] must detour or otherwise move about 2 orders ofmagnitude slower through these islands [27]. In short, a bilayer packed with obstructing pro-teins as well as islands of phospholipids which must be circumvented is expected to slow elec-tron transport. These islands tend to disappear at warmer temperatures, but it seems thatnature has developed a second mechanism to avoid or disperse these patches [89]. This involvesthe use of phospholipids, most commonly alpha-linolenic acid, which apparently are able tofluidize the membrane even at cool temperatures. We propose that a hierarchy of fatty acidsor phospholipid structures have evolved for this purpose with 22:6 being at the high end ofthe scale.
In summarizing this section, the discovery of EPA bacteria [45] occurred almost simultaneouslywith the finding by Gudbjarnason et al. [29] that DHA levels in heart muscle are higher in smallmammals with fast heart rates than in larger animals with slow heart rates. Hulbert and his col-leagues have confirmed and greatly expanded this finding leading to a new unified concept of therole of DHA in animal physiology [38], with likely applications in plants and marine bacteria. Inanimals, one of the most interesting findings concerns the strong correlation between DHA levelsin the membrane and certain kinds of muscle tissues. These include fast heart muscle, as men-tioned above, hummingbird wing muscle, and rapid rattle muscle of rattlesnake [39,42]. As re-viewed by [38]; the mitochondria from these muscles are also enriched in DHA. This fits withthe concept proposed here that DHA works to boost or maximize electron transport in mitochon-dria leading to increased energy output. The increased leakiness of DHA-enriched membranes
396 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
[10,60,73] might represent the price that this class of muscle must pay for increased or sustainedenergy output (double-edged sword again).
5.3. Rhodopsin disks
The outer segment of rod cells of the eye contains an array of some 1000 or so membrane disksor wafers stacked one on top of the other. These disks, which are structurally separated from theplasma membrane, contain visual rhodopsin and serve as antenna for detecting low light in vision.These disks are the most studied, highly unsaturated membranes in nature [28]. Giusto et al. [28]have compiled 497 original references emphasizing membrane biochemistry of rhodopsin disks.The overall picture coming from this overview is a highly unsaturated membrane composed ofabout 50% DHA [4], often paired with a saturated chain such as 18:0 in the phospholipid. How-ever, a significant number of phospholipid molecules contain DHA in both positions of the glyc-erol backbone (i.e., di-DHA; [56]). These reviewers point out that pioneering researchers studyingrhodopsin disks were the first to recognize the rapid turnover and re-synthesis of these highlyunstable structures. Indeed, the need for a sophisticated repair and turnover system has beenclearly demonstrated for these disks and evidence that the oxidative instability of DHA itself isat least partially responsible is convincing [28]. All available evidence points toward an importantfunction for DHA phospholipids in these crucial membranes, but their physiological or biochem-ical roles have remained elusive.
To get at the possible biochemical contribution of DHA phospholipids in rhodopsin disks, it isnecessary to go back to classic experiments on membrane motion carried out 30 years ago [50,59].These authors were able to calibrate the lateral motion of rhodopsin protein as it moves across themembrane surface and found that this molecule travels at amazing speed. In essence, they cali-brated a property of the lipid portion of the membrane since it is now known that rhodopsin pro-tein structure itself is not designed for rapid motion [27]. A simple example might help clarify thispoint. Imagine the difficulty in moving a child�s top, frozen in ice, sideways, compared to the sameexperiment in warm oil. The journey of rhodopsin is believed to be largely unobstructed with re-spect to other proteins since electron transport complexes are absent from these bilayers.
Since these classic experiments on membrane motion were conducted, biochemists specializingin the mechanism of the visual cascade have discovered why rhodopsin must move so fast [12,74].Rapid-fire collisions between rhodopsin and transducin are necessary to trigger the visual cascade.However, details of how DHA phospholipids take part in this reaction have remained largely dor-mant for some years, although the linkage to high fluidity is not a new idea (see [28,65]). A theoryon how high fluidity might work in vision is described in Section 6. Lancet et al. [47] have pointedout the similarity in design between trigger reactions in vision and other sensory perception sys-tems where DHA-enriched membranes are likely involved.
5.4. Axons
Based on the high and invariant levels of DHA in mammalian brain tissue [41,65], it is believedthat DHA phospholipids likely play a crucial role in brain biochemistry. Interestingly, a giant lob-ster neuron was found to be composed of a significant proportion of EPA [13], in contrast to
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 397
mammalian axons where only traces of EPA are found [65]. Also, severe starvation of rats foressential fatty acids has been found to depress DHA levels which are replaced by more-saturatedchains [11]. It can be argued from this data that DHA in the rat brain is part of some sort of func-tional hierarchy. This is consistent with a mechanism based on fluidity, a topic that has been re-viewed by Salem et al. [65].
Compared to rhodopsin disks, less is known about axon bilayers. However, it is clear that thisDHA enriched membrane is involved in maintaining the precise differentials in Na+ (outside) andK+ (inside) necessary for transmission of the electrochemical signal. In recent years, the permea-bility contributions of the DHA-portion of the axonal bilayer have been overshadowed by thepowerful cation gates and pumps operating in these membranes. The axonal membrane networkwhich altogether stretches over untold miles, is thought to be devoid of respiratory complexes andmight offer a relatively unobstructed surface with respect to proteins. However, there is no evi-dence that we are aware of demonstrating the need for collisions, such as the case with rhodop-sin-transducin. This is not the case with synaptic membranes which receive chemical signals [47].The importance of high fluidity in axons is considered in the following section.
6. Towards a unifying theory
The purpose of this section is to synthesize information presented above in order to explain thestructural roles of DHA and EPA in membranes spanning deep-sea bacteria to human neurons.Picture a membrane of a deep-sea bacterium carrying out respiration at temperatures near 0 �Cand under tremendous hydrostatic pressure. Earlier pioneering studies show that bacteria con-tinue to grow and replicate, albeit slowly, at temperatures when only �10–15% of their total lipidremains in the fluid state [17,52]. McElhaney [52] has pointed out that growth of A. laidlawii at agiven temperature is not significantly altered until about half of the membrane lipid is convertedfrom the liquid crystalline to the gel phase. However, growth ceases when about 90% of the mem-brane is in the gel phase. Studies with E. coli show clearly that fluidity is essential for growth [16]and these authors point out that E. coli can grow despite much of the membrane lipid being in theordered or gel phase [17]. Conversion of a high proportion of the total membrane lipids of deep-sea bacteria into the gel phase would be expected to cause two major problems, disrupting perme-ability and severely restricting movement of electron carriers necessary for energy production.Also, key respiratory complexes buried in the interior of islands would make their redox sites lessaccessible to quinone electron carriers which must continue to work in this extreme environment.How do deep-sea bacteria solve this problem?
It is proposed that deep-sea bacteria have evolved EPA and DHA phospholipids for this pur-pose but that this is only one among several different classes of phospholipids serving a similarrole [2,19]. We hypothesize EPA or DHA phospholipids work like miniature icebreakers to blockor disrupt the formation of islands of gel-phase lipids [68,89]. The mechanism is unknown butmight be explained on the basis of the hyperfluid properties of DHA/EPA predicted by 2H-NMR and molecular dynamic models discussed above. What we are proposing is the net effectmakes more energy available to the cell via both energy conservation [77] and enhanced energyproduction [38].
398 R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402
It is important to point out that membrane heterogeneity caused by islands made up of gel-phaselipids is not restricted to cold temperatures, but rather appears to be a general property seen overthe entire range of growth temperatures [52,68]. It is hypothesized that DHA phospholipids in mit-ochondrial membranes, even in endothermic animals, work by a similar mechanism as describedabove for bacteria – contributing significantly to fluidity. However, the purpose in this case mightbe to maximize or sustain energy production [38]. In chloroplasts the current working model is thathigh fluidity plays an important role in long distance electron transport. This has major implica-tions for understanding the role of DHA and EPA in oceanic productivity. One of the most inter-esting questions for future research involves the levels of energy gained by this mechanism. Giventhat energetic loss in DHA mitochondria is around 20% [38], it seems logical to expect gains wellabove this value. There is also the question of energy efficiency. For example, it may turn out thatDHA-enriched mitochondria in animal muscle might offer high and even continuously high levelsof energy production to match the needs of their specific muscle tissue or organ. However, to meetthese demands, there seems to be a price to pay in terms of energy efficiency. This is not a majorproblem for an organism capable of acquiring an excess of respiratory substrates in its diet (e.g.a deep-sea bacterium living in the gut of its host), but may be for other organisms.
The role of high fluidity in maintaining a membrane suitable for proton bioenergetics seemsreadily applied to specialized membranes such as rhodopsin disks and axons. As mentionedabove, rhodopsin disks are highly enriched in DHA (�50%) and contain relatively high levelsof di-DHA phospholipids [4,56]. It seems likely this membrane will turn out to set the standardfor high fluidity and would sit at the top of a biological fluidity scale. High fluidity in this caseappears to work solely for maintaining the membrane lipids in a homogeneous, liquid crystallinestate for the purpose of maximizing collisions between rhodopsin-transducin.
Understanding the structural role of DHA in the brain is an important area for future re-search. Like rhodopsin disks this membrane is not populated by electron transport complexes.DHA phospholipids obviously contribute to permeability properties of axonal membranes, butwould seem to be a poor choice if fidelity of cation permeability were the only consideration [53].Indeed, rhodopsin disks have been found to be leaky for Na+ [36]. We hypothesize that DHA-phospholipids work in axons by a mechanism similar to that described for mitochondria. It iseasy to imagine that should excessive gel-phase islands be formed, that the resulting membraneislands might cause short-circuiting or drastically change the electrophysiology of the axon.Once again, DHA is proposed to be the best design for this purpose due to its biofluidizingproperties.
Finally, pioneering experiments on membrane fluidity in bacteria conducted several decadesago have been revisited and integrated with new work on Omega-3 bacteria. This has led to a uni-fied concept of the structural roles of Omega-3 fatty acids in cellular membranes based on the flu-idity provided by these chains. The concept of high fluidity seems to be an important missing linkin understanding membrane structure-function. It is envisioned that cells are able to choose fromamong a hierarchy of chains or phospholipid structures suitable for maintaining membrane flui-dities in different environments or for their special biochemical needs. It seems likely that DHAand EPA phospholipids represent just the ‘‘tip of the iceberg’’ regarding highly fluidizing struc-tures. Integrated biochemical, chemical and genetic studies are needed to define the molecularroles of DHA/EPA and the contributions of these molecules to membrane fluidity and otherimportant physiological properties of the bilayer.
R.C. Valentine, D.L. Valentine / Progress in Lipid Research 43 (2004) 383–402 399
Acknowledgement
We are grateful to Calgene, Inc. (later Monsanto, Inc.) for sponsoring the research on DHAand EPA bacteria forming the basis of this article. Special thanks go to Nordine Cheikh, VicKnauf, Christine Shewmaker, Daniel Facciotti, Bill Hiatt and Michael Saxton for encouragementand advice. We are grateful to Jim Metz for a gift of the EPA recombinant, Tim Hickman andTom Hayes for analytical, Marilyn York for media preparation and stock cultures, and CindyAnders for manuscript preparation. The work of JL Harwood and three anonymous reviewersgreatly improved the manuscript. DLV was funded by the National Science Foundation (EAR0311894 and OCE 0085607). We thank Doug Bartlett for providing space, advice and editorialcomments.
References
[1] Albert CM, Hennekens CH, O�Donnell CJ, Ajani UA, Carey VJ, et al. Fish consumption and risk of sudden
cardiac death. JAMA 1998;279:23–8.
[2] Allen EE, Facciotti D, Bartlett DH. Monounsaturated but not polyunsaturated fatty acids are required for growth
of the deep-sea bacterium Photobacterium profundum SS9 at high pressure and low temperature. Appl Environ
Microbiol 1999;65:1710–20.
[3] Allen EE, Bartlett DH. Structure and regulation of the Omega-3 polyunsaturated fatty acid synthase genes from
the deep-sea bacterium Photobacterium profundum strain SS9. Microbiology 2002;148:1903–13.
[4] Anderson R. Lipids of ocular tissues IV. A comparison of the phospholipids from the retina of six mammalian
species. Exptl Eye Res 1970;10:339–44.
[5] Annous BA, Becker LA, Bayles DO, Labeda DP, Wilkinson BJ. Critical role of anteiso-C15:0 fatty acid in the
growth of Listeria monocytogenes at low temperatures. Appl Environ Microbiol 1997;63:3887–94.
[6] Applegate AR, Glomset JA. Computer-based modeling of the conformation and packing properties of
docosahexaenoic acid. J Lipid Res 1986;27:658–79.
[7] Barry JA, Trouard TP, Salmon A, Brown MF. Low-temperature 2H NMR spectroscopy of phospholipid bilayers
