Báo cáo khoa học: Mutants of Saccharomyces cerevisiae deficient in acyl-CoA synthetases secrete fatty acids due to interrupted fatty acid recycling docx

Results faa1D faa4D mutant cells secrete a large amount of free fatty acids into the medium In a first attempt to explore the fatty acid secretion phenotype for the faa1D mutant described

acyl-CoA synthetases secrete fatty acids due to

interrupted fatty acid recycling

Michael Scharnewski1, Paweena Pongdontri1,*, Gabriel Mora1, Michael Hoppert2and Martin Fulda1

1 Department of Plant Biochemistry, Albrecht-von-Haller Institute, Georg-August University Goettingen, Germany

2 Institute for Microbiology and Genetics, Georg-August University Goettingen, Germany

Fatty acid metabolism pervades large areas of cellular

life In all known ramifications of this metabolism,

fatty acids almost never undergo direct metabolic

conversion as free molecules The fatty acid molecule

usually requires the activation of its carboxyl group

via esterification to either acyl-carrier-protein (ACP) or

to CoA Only as thioesters are fatty acids able to

accom-plish any of their possible metabolic fates In

Saccha-romyces cerevisiae, the fatty acids are synthesized by a

large multienzyme complex designated as fatty acid

synthase (FAS) on the basis of ACP [1] In S

cerevi-siae, the ACP is an integral component of the FAS

complex and, therefore, free acyl-ACP is not available for any other metabolic purposes The synthetic cycle

is terminated by transferring the generated acyl chain directly to CoA [2] Due to this mechanism, the direct product of fatty acid de novo synthesis in S cerevisiae

is acyl-CoA, which consequently is the only activated fatty acid molecule serving all other metabolic path-ways [1] Besides deriving from de novo synthesis, lipid degradation as well as import of exogenous fatty acids may feed into the pool of intracellular fatty acids In both cases, enzymatic activation with CoA is required

to allow further metabolization of these fatty acids In

Keywords

endoplasmic reticulum; FAA; FAT1; fatty

acid accumulation; lipid remodelling

Correspondence

M Fulda, Department of Plant

Biochemistry, Albrecht-von-Haller Institute

for Plant Sciences, Georg-August University

Goettingen, Justus-von-Liebig-Weg 11,

D-37077 Goettingen, Germany

Fax: +49 551 39 5749

Tel: +49 551 39 5750

E-mail: mfulda@gwdg.de

*Present address

Department of Biochemistry, Faculty of

Science, Khon Kaen University, Thailand

(Received 9 January 2008, revised 12 March

2008, accepted 20 March 2008)

doi:10.1111/j.1742-4658.2008.06417.x

In the present study, acyl-CoA synthetase mutants of Saccharomyces cerevisiae were employed to investigate the impact of this activity on cer-tain pools of fatty acids We identified a genotype responsible for the secre-tion of free fatty acids into the culture medium The combined delesecre-tion of Faa1p and Faa4p encoding two out of five acyl-CoA synthetases was necessary and sufficient to establish mutant cells that secreted fatty acids in

a growth-phase dependent manner The mutants accomplished fatty acid export during exponential growth-phase followed by fatty acid re-import into the cells during the stationary phase The data presented suggest that the secretion is driven by an active component The fatty acid re-import resulted in a severely altered ultrastructure of the mutant cells Additional strains deficient of any cellular acyl-CoA synthetase activity revealed an almost identical phenotype, thereby proving transfer of fatty acids across the plasma membrane independent of their activation with CoA Further experiments identified membrane lipids as the origin of the observed free fatty acids Therefore, we propose the recycling of endogenous fatty acids generated in the course of lipid remodelling as a major task of both acyl-CoA synthetases Faa1p and Faa4p

Abbreviations

ACP, acyl-carrier protein; ER, endoplasmic reticulum; FAS, fatty acid synthase; Fat1p, fatty acid transport protein 1; Fat2p, fatty acid transport protein 2; YPR, yeast proteose raffinose medium.

S cerevisiae, five genes coding for acyl-CoA

syntheta-ses have been identified All these activities are

poten-tially able to mediate between the pool of free fatty

acids and the pool of acyl-CoA molecules Four genes,

termed FAA1 to FAA4 (fatty acid activation), were

characterized in pioneering work by Gordon et al [3–

9] FAA1 and FAA4 encode acyl-CoA synthetases

involved in the activation of imported fatty acids [10]

Although Faa1p represents the major cellular activity,

both enzymes display partial redundancy and a lack of

either one can be compensated by the activity of the

other [6] Faa1p localizes to the endoplasmic reticulum

(ER), the plasma membrane and vesicles, whereas

Faa4p is found at the ER as well as at lipid droplets

[11] FAA2 encodes a peroxisomal activity involved in

the activation of fatty acids scheduled for b-oxidation

whereas the biological role of Faa3p has remained

unclear to date In addition to the four FAA genes,

FAT1was identified as the fifth source of activity

feed-ing into the acyl-CoA pool [12] Initially isolated as a

protein involved in fatty acid transport across the

plasma membrane [13,14], it finally proved to have

acyl-CoA synthetase activity as well, with a preference

for fatty acids with chain length longer than 22

car-bons The protein was localized to lipid droplets and

to the ER [11], but was also reported at the plasma

membrane [15]

Given the diverse functions of fatty acids within the

metabolism, a carefully regulated distribution of fatty

acids within the cell can be anticipated In this respect,

it is of relevance that, besides the pure enzymatic

reac-tion, acyl-CoA synthetase activity is also involved in

fatty acid transport across membranes Cellular uptake

as well as subcellular distribution of fatty acids could

be influenced by this enzymatic activity converting a

hydrophobic substrate into a water soluble CoA-ester

It was first established in a mutant of Escherichia coli

that a defective acyl-CoA synthetase abolished the

uptake of fatty acids from the medium To emphasize

the tight link between fatty acid transport and

acyl-CoA synthetase activity, the mechanism was termed

‘vectorial acylation’ [16] In S cerevisiae, a comparable

model was proposed and evidence was provided for a

participation of Fat1p in transport in concert with the

two acyl-CoA synthetases Faa1p and Faa4p [17] In

this model, Fat1p is involved in the import mechanism

directly, whereas acyl-CoA synthetases Faa1p and

Faa4p is proposed to be responsible for the abstraction

of the delivered fatty acid from the membrane and

concomitantly rendering the fatty acid water soluble

by esterification, thereby trapping the molecule in the

cytoplasm On the other hand, the model is not

with-out controversy and, alternatively, a concept has been

developed describing fatty acid movement across a membrane as a passive process This mechanism of simple diffusion is based on the fast flip-flop of fatty acids through the membrane and has been reviewed comprehensively [18,19]

Reviewing the currently available data, the precise role of acyl-CoA synthetases in regulating certain pools of fatty acids as well as their potential impact on fatty acid transport across membranes both remain elusive In this context, we were intrigued by two reports [20,21] linking the disruption of FAA1 in Candida lipolytica and in S cerevisiae to a fatty acid secretion phenotype In the mutant strains described, the control over certain fatty acid pools appeared to

be restricted and, in addition, fatty acid transport could also be affected Because the mutant strains derived in both cases from random mutagenesis experi-ments, the exact genetic constitutions were not unequivocally established and the molecular basis for the phenotype may have been not fully understood However, if the underlying mechanisms of this pheno-type could be elucidated, such a strain might provide a valuable tool to study the role of distinct acyl-CoA synthetases in regulating certain pools of fatty acids

To assess the role of acyl-CoA synthetases in regu-lating particular pools of fatty acids, we used mutants

of S cerevisiae characterized by various combinations

of deletions in the corresponding genes In mutants defective for Faa1p and Faa4p, we observed a tran-sient fatty acid secretion phenotype We demonstrate fatty acid transport in the absence of any acyl-CoA synthetase activity, and data are presented demonstrat-ing that the direction of transport is dependent on the growth phase of the mutants In addition, we provide evidence for the hypothesis that Faa1p and Faa4p are involved in the recycling of fatty acids deriving from lipid remodelling processes

Results

faa1D faa4D mutant cells secrete a large amount

of free fatty acids into the medium

In a first attempt to explore the fatty acid secretion phenotype for the faa1D mutant described earlier by Michinaka et al [21], we employed the faa1D mutant strain YB497 instead (similar to the other YB strains generously provided by J I Gordon) and tested its ability to secrete fatty acids By contrast to the data presented by Michinaka et al [21], we were unable to detect even traces of free fatty acids in the culture medium by using this different strain harbouring the same mutation Because the strains investigated by

Michinaka et al [21] resulted from a repeatedly

employed mutagenesis with ethyl-methane sulfonate,

the precise genotype of their mutant was not well

established Nevertheless, the data presented appeared

to indicate a role of Faa1p in the fatty acid secretion

phenotype Assuming that one of the additional four

acyl-CoA synthetases in S cerevisiae could mask the

phenotype by overlapping functions, we assumed that

a combined elimination of two or more of those

activi-ties might induce fatty acid secretion Therefore, we

tested strains deficient in combinations of FAA1 and

one or more additional FAA The experiments were

complicated by the fact that strains carrying deletions

in both FAA1 and FAA4 were flocculent in liquid

cul-ture, as described previously [10] This characteristic

trait resulted in large cellular lumps swimming in an

essentially clear medium, making it impossible to

determine the status of the culture by measuring cell

density By testing alternative medium compositions, a

YP medium containing raffinose (YPR) instead of

dex-trose was identified to allow proper growth as

homoge-nous cultures for all strains tested Because the use of

raffinose was essential for the accurate measurement of

cell density in different stages of the cultures, this

sugar was used as carbon source for all following

experiments unless otherwise noted The different

strains were grown for 48 h in YPR before the cells

were removed by centrifugation and the supernatant

was extracted and tested for free fatty acids For

strains carrying a combined deletion of FAA1 and

FAA2or of FAA1 and FAA3, we did not observe any

fatty acid secretion phenotype However, the combined

deletion of FAA1 and FAA4 resulted in the

accumula-tion of significant amounts of free fatty acids in the

culture medium Whereas no fatty acids were detected

in the media of wild-type cells of 48-h-old cultures,

approximately 220 lmolÆL)1 free fatty acids were

found in the media of the mutant strain YB525 (see

supplementary Fig S1) To investigate whether the

deletion of additional Faap would further increase the

amount of secreted fatty acids, we also tested the

qua-druple mutant YB526 deficient in four Faap activities

The detected fatty acid concentrations in the media

proved to be similar for the double and for the

qua-druple mutant, indicating that Faa2p and Faa3p are

not significantly involved in the observed phenotype

Surprisingly, the detected amounts of free fatty acids

in the media were high enough to result in whitish

flakes on the surface of the culture The composition

of the secreted fatty acids reflects the profile of fatty

acids found in membrane lipids of yeast and consists

of myristic acid (14:0), palmitic acid (16:0), palmitoleic

acid (16:1), stearic acid (18:0) and oleic acid (18:1) (see

supplementary Fig S1) We conclude from this experi-ment that the deletion of FAA1 and FAA4 is necessary and sufficient to produce the fatty acid secretion phenotype No other combination of FAA deletions keeping either FAA1 or FAA4 intact resulted in a com-parable phenotype

To characterize the observed phenotype in more detail, we first determined the progress of fatty acid secretion in a growing culture Therefore, we measured the concentration of free fatty acids in the media at several time points during the exponential and station-ary growth phases In these experiments, a tight corre-lation between stage of growth and fatty acid secretion was observed, revealing several striking characteristics (Fig 1A) First, the total amount of fatty acids in the media increased continuously during early exponential phase Second, the fatty acid concentration in the media did not reach a stable plateau Instead, the total amount of fatty acids in the media started to decline during the late exponential phase and continued to decrease during the stationary phase Third, the point

in time of maximum concentration in the media was different for specific fatty acids Whereas 16:0 reached its maximum concentration at 48 h and subsequently declined, the level of the unsaturated fatty acids 16:1 and 18:1 kept their maximum concentration until approximately 70 h before they also began to decline Taken together these results clearly identified the growth stage of the culture as being an important parameter influencing the fatty acid secretion pheno-type

In the next step, the underlying mechanisms were analysed Whereas export of free fatty acids from the cells during early exponential growth phase most prob-ably explains the accumulation of fatty acids in the media, several alternative models may explain the subsequent decline in fatty acid concentration One possible interpretation would be a re-import of the ini-tially secreted fatty acids back into the cells If this assumption is correct, this transport should result in

an increase of free fatty acids within the cells because the mutant cells are unable to utilize these fatty acids due to the lack of acyl-CoA synthetase activity To evaluate this possibility, we measured the concentra-tion of intracellular free fatty acids at the same time points during the exponential and stationary growth phases that had been used for the estimation of the extracellular fatty acids In addition, we determined the amount of lipid-bound fatty acids to cover all major pools of fatty acids (Fig 1B,D)

The amount of cellular free fatty acids could be easily affected by lipid hydrolysis during the extraction procedure To rule out the possibility that major

proportions of the detected free fatty acids are result-ing from such effect, we analysed extractions with and without boiling and the results obtained were not sig-nificantly different Therefore, boiling was omitted in subsequent experiments To differentiate between free fatty acids and esterified fatty acids, the cellular lipid extracts were subjected to two different protocols to achieve either methylation of free fatty acids or trans-methylation of esterified fatty acids [22,23]

The concentration of bound fatty acids at different time points resulted in a typical curve corresponding

to the growth of the culture and is independent of the observed changes in the pools of free fatty acids (Fig 1C,D) For the free fatty acids, the results clearly demonstrated that the decline of fatty acid concentra-tion in the media was indeed paralleled by a constant increase of the concentration of intracellular free fatty acids (Fig 1D) This could indicate an import of fatty acids into the cells On the other hand, the increase of intracellular fatty acids in absolute amounts was approximately three-fold greater than the concomitant decrease of extracellular fatty acids Therefore, besides the import of exogenous fatty acids, additional release

of internal fatty acids had to contribute to the increase

of intracellular fatty acids However, analysis of the previously mentioned fatty acid specificity of the trans-port appeared to provide additional indications for ongoing fatty acid import As described earlier, the amount of 16:0 specifically decreased strongly in the media between time points 48 h and 65 h ()30.0 lmolÆL)1), whereas the concentration of the other fatty acids in the media changed only moderately during this period ()4.6 to +5.6 lmolÆL)1) Strikingly, it was also 16:0 that increased within the cells during this time interval in a much stronger fashion than any other fatty acid (+68.8 lmolÆL)1 versus +3.6 to +30.3 lmolÆL)1), suggesting re-import of 16:0 into the cell during this growth stage In summary, the results demonstrated an export of free fatty acids out of the

A

B

C

D

Fig 1 Relationship between fatty acid concentration and stage of the culture (A) Extracellular free fatty acids in the culture medium

of YB526 were determined at the time points indicated The bars representing individual fatty acids are indicated as 14:0 (black), 16:0 (dark gray), 16:1 (spotted), 18:0 (hatched), 18:1 (light gray) (B) Time-course of intracellular free fatty acid accumulation Intracellu-lar free fatty acids accumulated by cells of YB526 were determined

at the time points indicated (C) Growth curve of the culture [YB332 (triangle), YB526 (diamonds) and MS51 (square)] (D) Con-centration of fatty acids in the pools of free fatty acids in the media (diamonds), of free fatty acids in the cells (square) and of esterified fatty acids (triangle) The error bars represent the SEM from three independent experiments.

cells during the early exponential growth phase,

whereas the strong increase of intracellular free fatty

acids at the late exponential phase is compatible with

the hypothesis that the reduction of fatty acid

concen-tration in the media is due to a significant re-import of

free fatty acids into the cells

Fatty acid transport across the plasma membrane

is functional in absence of all known acyl-CoA

synthetases

The translocation of fatty acids across the plasma

membrane in the absence of Faap activity appeared to

be in contrast to the model suggesting vectorial

acyla-tion as a basis for fatty acid transport [17] On the

other hand, the strains used so far might contain

resid-ual acyl-CoA synthetase activity due to the presence of

Fat1p, which was shown to possess this enzymatic

activity also To address the question of whether fatty

acid transport protein 1 (Fat1p), or even fatty acid

transport protein 2 (Fat2p) for which no enzymatic

activity has been demonstrated to date, is responsible

for the fatty acid transport observed in the previous

experiment, we generated knockout deletions of FAT1

and FAT2 in the background of the well established

FAA quadruple mutant YB526 [6] The obtained

strains with five (MS51) and six gene deletions

(MS612), respectively, are devoid of any detectable

acyl-CoA synthetase activity (data not shown) Both

strains were subjected to the same measurements of

free and bound fatty acids as described for the

quadru-ple mutant The results indicated that the additional

gene deletions did not change the capacity of the cells

to transport fatty acids As shown in Fig 2, the loss of

Fat1p resulted in even higher levels of intracellular free

fatty acids during the stationary phase More

impor-tantly, the same two phases of early fatty acid export

followed by re-import at later stages were observed in

exactly the same chronological order as in the

quadru-ple mutant Therefore, it was concluded that cells of

S cerevisiae lacking any cellular acyl-CoA synthetase

activity are nevertheless able to transport fatty acids

efficiently across the plasma membrane

Fat1p is responsible for remaining acyl-CoA

synthetase activity of strain YB526

As shown above, the deletion of Fat1p in the

back-ground of the FAA quadruple mutant further

increased the concentration of intracellular

accumu-lated free fatty acids This suggested a capacity for

Fat1p to access the pool of free fatty acids, most likely

via its proven acyl-CoA synthetase activity [12]

Despite in vitro assays for the acyl-CoA synthetase activity of Fat1p showing a strong specificity for very long chain fatty acids [12], our data also appeared to indicate activity against C16 and C18 fatty acids To test this possibility in vivo, we incubated the different strains with radiolabelled oleic acid Successful feeding

of the exogenous fatty acid into the cellular acyl-CoA pool should result in incorporation of the label in the various lipid classes Cells of wild-type, YB526, MS51, MS52 and MS612 were grown in YPR to the early sta-tionary phase before radiolabelled oleic acid was added Following incubation for 24 h, the cellular lipids were extracted and subjected to TLC (Fig 3)

As expected, the lipid extract of wild-type cells showed the label spread to phospholipids and neutral lipids In comparison, the level of incorporation in YB526 cells was drastically reduced, but the label was still clearly detectable in phospholipids as well as an additional spot tentatively identified by co-migration with lipid standards as fatty acid ethyl ester (Fig 3) By contrast, there was no longer any incorporation of labelled fatty acids to either phospholipids or TAG if, in addition to the FAA genes, FAT1 was deleted, as shown by the lipid extracts of MS51 and MS612 The combined deletion of all FAA genes and FAT2 in strain MS52,

on the other hand, resulted in the same level of incor-poration as in YB526, indicating that Fat2p is not responsible for the remaining capacity to channel exog-enous fatty acids into lipids Control strains carrying deletions in only FAT1 or FAT2, or in both FAT1 and FAT2, incorporated labelled fatty acids comparable to the wild-type (see supplementary Fig S2) Surprisingly,

Fig 2 Total amount of intracellular and extracellular fatty acids of YB526 and MS51 The cells were grown in YPR medium and har-vested at different time points during the exponential and stationary phases Medium and cells were extracted and the total fatty acid concentrations of the different pools were determined The fatty acid concentrations are given for YB526 intracellular (square), YB526 extracellular (diamonds), MS51 intracellular (circle), and MS51 extracellular (triangle) The error bars represent the SEM from three independent experiments.

even in MS51 and MS612, the radioactive label

showed up in the spot identified as fatty acid ethyl

ester, demonstrating a potential to incorporate free

fatty acids into fatty acid ethyl esters in absence of any

acyl-CoA synthetase activity This observation

sup-ports a recent description of enzymatic activity found

in microsomal preparations from plants and yeast that

is able to acylate aliphatic alcohols without prior

acti-vation to acyl-thioesters [24] Irrespective of this side

reaction, the labelling experiments identified Fat1p as

the remaining acyl-CoA synthetase activity in the

strain YB526 that is able to activate imported fatty

acids prior to their incorporation into phospholipids

The accumulated free fatty acids derive from lipid

remodelling processes

The strong accumulation of free fatty acids in the

media and within the mutant cells raised questions

regarding the metabolic origin of these molecules Obviously, acyl-CoA synthetase activity in wild-type cells is masking a permanent internal generation of sig-nificant amounts of free fatty acids To test whether these fatty acids were derived from lipid turn-over pro-cesses, we aimed to achieve a fatty acid modification that is restricted to lipid bound fatty acids For this purpose, we introduced a D12 specific lipid desaturase from sunflower into strain MS51 By contrast to the endogenous stearoyl-CoA desaturase, the heterologous D12-desaturase converts exclusively lipid-bound 18:1 to 18:2 [25,26] Due to the nature of this desaturase, the generated 18:2 was considered as marker for the pool

of bound fatty acids Therefore, the occurrence of 18:2

in the pool of free fatty acids would argue for lipid turnover as the source of fatty acid accumulation The expression of the D12-desaturase resulted in diminished growth of the yeast cells, causing reduced levels of fatty acids in all pools measured However, analysis of lipid extracts of the transformed cells demonstrated 18:2 to be a significant constituent of the pool of ester-ified fatty acids, indicating successful expression of the D12-desaturase (Fig 4A) More interestingly, the pool

of free fatty acids contained 18:2 as well; indicating a release of formerly lipid bound fatty acids into this pool (Fig 4B,C) Moreover, the ratio of the sum of all natural fatty acids of S cerevisiae (14:0 to 18:1) to the artificially produced 18:2 was almost identical in the fraction of bound fatty acids compared to the fraction

of free fatty acids (11.19 and 11.03, respectively; see supplementary Table S1) This constant ratio would be

in line with lipid remodelling processes as a source for the accumulated fatty acids in the mutant cells, provided that the release of the fatty acids is rather unspecific

The direction of fatty acid transport is dependent

on the metabolic state of the cells

To gain further insight into the correlation of growth phase of the culture and the direction of fatty acid transport, we investigated the possibility of manipulat-ing the transport by changmanipulat-ing the medium conditions

of the growing culture Therefore, we grew cells as described above in raffinose containing medium Approximately 3 h after the cultures reached the sta-tionary phase and the cells already started to re-import fatty acids from the media, we again fed raffinose to the cultures (Fig 5A–C) In controls, water was added

to the cells As expected, the culture fed with raffinose started to grow again indicated by increasing attenu-ance (data not shown) We then analyzed the amount

of fatty acids in the medium and inside the cells

Fig 3 TLC of lipid extracts of yeast strains fed with radiolabelled

oleic acid Cells of wild-type and the mutants indicated were grown

for 24 h in the presence of radiolabelled oleic acid The cells were

harvested by centrifugation, lipids were extracted, and these

extracts were separated by solvent A [acetic acid methyl ester ⁄

iso-propanol ⁄ chloroform-methanol ⁄ 0.75% KCl (25 : 25 : 28 : 10 : 7,

v ⁄ v)] followed by solvent B (chloroform ⁄ acetone (8 : 2 v ⁄ v) + 1%

NH3) on silica plates Comparison between YB526 and MS51

clearly indicate that Fat1p in YB526 is responsible for the

incorpora-tion of exogenous oleate into phospholipids The band labelled as

EE was tentatively identified as fatty acid ethyl esters produced by

all strains even in absence of any acyl-CoA synthetase activity.

Lipid classes were determined by co-migration of lipid standards

and staining with copper sulfate TAG, triacylglycerol; EE, fatty acid

ethyl ester; OA, oleic acid; PC, phosphatidylcholine; PI,

phosphati-dyinositol; PS, phosphatidylserine The figure shows one

represen-tative result out of three independent experiments.

Surprisingly, the accumulation of free fatty acids inside

the cells stopped immediately after adding raffinose

(Fig 5A) Simultaneously, large quantities of free fatty

acids were secreted by the cells (Fig 5B) At the same time, the concentration of the total esterified fatty acids remained constant (Fig 5C) To allow for simple comparison, the amount of fatty acids in the medium prior to the addition of either water or of raffinose was set as 100% In control experiments, we observed

A

B

C

Fig 4 Profile of free and esterified fatty acids upon expression of

desaturase in MS51 Cells of strain MS51 expressing a

D12-desaturase of sunflower were grown for 70 h in minimal media

before the cells were harvested and lipids were extracted The

dif-ferent pools of fatty acids were analyzed and presented as: (A)

intracellular esterified fatty acids; (B) intracellular free fatty acids;

(C) extracellular free fatty acids As a control, cells transformed

with empty vector were extracted The bars representing individual

fatty acids are depicted as 14:0 (dark gray), 16:0 (hatched),

16:1 (gray), 18:0 (spotted), 18:1 (light gray), 18:2 (black) The error

bars represent the SEM from three independent experiments.

A

B

C

Fig 5 Feeding of raffinose to MS51 cells in stationary phase Total fatty acid concentration in different pools was measured in cultures fed with raffinose (diamonds) at time point 65 h (arrows) during the stationary phase (A–C) As a control, the same amount of water was added to separate cultures (square) (A) Free fatty acids inside the cells (B) Free fatty acids in the medium (C) Esterified fatty acids The error bars represent the SEM from three independent experiments.

a decrease of free fatty acids in the medium by 25% within 30 h This decrease is accompanied by an increase of cellular free fatty acids partly due to re-import, as described earlier By contrast, the amount of free fatty acids in the medium of the cultures fed with raffinose increased by 100% within

30 h after adding the raffinose The supply of raffinose allowed the cells, on the other hand, to maintain the concentration of free fatty acids inside the cells at a constant level When the supply of raffinose was finally exhausted, after approximately 95 h of the experiment, the cells again started to re-import fatty acids from the medium and to accumulate these fatty acids intra-cellularly in even higher amounts than in the control culture From these results, it was concluded that the direction of transport was reversible and that it was regulated by the metabolic state of the cells

Cytological features of MS51 During the stationary phase of the culture, the concen-tration of free fatty acids found in the mutant cells was up to 50-fold greater than that observed in wild-type cells High concentrations of free fatty acids are believed to be rather unfavourable to cells due to their detergent character Nevertheless, the mutant cells were not only viable, but also showed only minor dif-ferences in their growth behaviour compared to wild-type To investigate whether the level of free fatty acids might have observable consequences for the cell,

we inspected the subcellular morphology by electron microscopy Whereas most organelles, the cytoplasm and the plasma membrane showed no obvious anom-aly, we observed a strikingly enlarged ER in the mutant cells (Fig 6) The mutant cells were pervaded

by strands of ER with conspicuously dilated lumen This is especially apparent in the cortical ER, which is closely apposed to the cytoplasmic membrane (Fig 6C,D) The lumen appears darker than the cyto-plasm (Fig 6B–D) and is filled with dark stained, lam-inated material This striking phenotype was observed

in all mutant cells inspected, whereas it was never found in wild-type cells To a moderate extent, the dilated ER-phenotype was visible already in cells har-vested during the exponential phase but it became

A

B

C

D

Fig 6 Electron microscopic analysis of cells of wild-type and MS51 in stationary phase Overview of wild-type (A) and mutant cells (B,C) In wild-type cells, the ER lumen is inconspicuous, ER lumen in mutant cells appears to be dilated; the ER is marked by arrows The ER lumen appears darker than the cytoplasm (C,D) and

is filled with dark stained, laminated material (D) A detail of (C) is shown at higher magnification.

drastically enhanced in cells harvested from stationary

phase (see supplementary Fig S3) Therefore, the

severity of the symptoms appeared to correlate directly

with the level of accumulated free fatty acids This

observation might suggest that the mutant cells are

able to deposit excess of free fatty acids specifically in

the ER, thereby excluding an excess of free fatty acids

from other cellular compartments

Discussion

The present study was designed to improve our

under-standing of acyl-CoA synthetase activity on different

pools of fatty acids in S cerevisiae The experiments

were initiated by data showing a fatty acid secretion

phenotype for different yeast cells deficient of Faa1p

[20,21] Despite not verifying the fatty acid secretion

phenotype for strains of S cerevisiae deficient of

Faa1p alone [21], we did observe fatty acid secretion

in strains deficient not only in Faa1p, but also in

Faa4p Because the faa1D mutant strain employed in

the previous study [21] was isolated by a screen

involv-ing two independent rounds of mutagenesis with

ethyl-methane sulfonate, one possible explanation for the

observed phenotype would be an unidentified

addi-tional mutation in FAA4 resulting in a faa1D faa4D

genotype for the strain finally described The fatty acid

secretion phenotype observed in the present study

indi-cated that the presence of either Faa1p or Faa4p is

necessary during exponential growth to keep

endoge-nous free fatty acids inside the cell During the

station-ary phase, we then observed a re-import of those fatty

acids previously secreted during the exponential

growth phase This re-import finally resulted in a ratio

of approximately 10 : 1 of free fatty acids to esterified

fatty acids in the mutant cells, whereas this ratio was

approximately 1 : 20 in wild-type cells The fatty acid

secretion and re-import was observed not only in the

faa1D faa4D double mutant, but also in the strains

YB526, MS51 and MS612, indicating fatty acid uptake

even in the absence of any cellular acyl-CoA synthetase

activity Surprisingly, the direction of transport proved

to be reversible Upon addition of raffinose to cells in

the stationary growth phase, the import of fatty acids

stopped immediately and export was again initiated

The fast response resulting in an instantaneous

rever-sion of the direction of net transport strongly argues for

an active mechanism to achieve fatty acid export

It is important to note that the additional deletion

of FAT1 did not change the biphasic mode of fatty

acid transport but it did further increase the amount

of accumulated free fatty acids Whereas convincing

data had been presented showing that Fat1p is

impor-tant for fatty acid import in S cerevisiae [27], the transport processes of fatty acid export and re-import observed in the present study were essentially indepen-dent of this previously described activity Instead, our results for Fat1p suggest partly overlapping functions with Faa1p and Faa4p with respect to the capacity to channel released fatty acids back into lipids This was demonstrated not only by the increased amounts of accumulated free fatty acids in the fivefold mutant MS51, but also by feeding experiments with radiola-belled fatty acids On the other hand, the activity of Fat1p alone was not strong enough to prevent the secretion of fatty acids upon combined deletion of FAA1and FAA4

To allow for speculation about the biological role of the release of fatty acids, it was essential to gain infor-mation about their precise metabolic origin By expressing a lipid specific D12-desaturase from sun-flower in the five-fold mutant MS51, we obtained data supporting the hypothesis that the majority of the secreted fatty acids were released from phospholipids The reasons for this release are not clear yet, but it appears to be legitimate to assume that lipid remodel-ling processes are involved that ensure continuous adaptation of membrane parameters to cellular needs Initial evidence for prominent lipid remodelling was obtained from analyses of the fatty acid composition

of different lipid classes described as molecular species Despite the fact that the biosynthesis of each different phospholipid class involves common precursors, strong differences in the molecular species were detected [28]

To establish distinct molecular species, a system involving sequential deacylation by phospholipases and reacylation by acyltransferases was proposed [29] Lipid labelling experiments strongly supported this model for S cerevisiae [28] and the results obtained from studies with rat hepatocytes provided information suggesting, for phosphatidylcholine and phosphatidyl-ethanolamine, a similar rate of de novo synthesis and

of remodelling activities [30] Taken together, these data indicate lipid remodelling as a quantitatively important process, which probably is responsible for the establishment of distinct molecular species of cer-tain lipid classes In this respect, the free fatty acids observed in the present study are most likely the result

of interrupted lipid remodelling processes Conse-quently, we propose the activation of endogenous free fatty acids produced within this recycling mechanism

as a most important role for Faa1p and Faa4p Although being metabolically inaccessible to the mutant cells, the re-import of the released free fatty acids during stationary phase finally resulted in the accumulation of large amounts of free fatty acids

inside the cells By electron microscopy, the effect of

this accumulation on cellular morphology was

inspected and diagnosed as a severe phenotype of the

ER To our knowledge, similar structures of the ER

have not been described previously Because the

strength of the phenotype corresponded to the level of

accumulated free fatty acids, it is fair to assume that

the free fatty acids themselves account for the dark

stained dilated lumen of the ER How the free fatty

acids are organized in these structures is currently

unknown The accumulation of free fatty acids

result-ing in electron-dense material has recently been

described in lipid bodies of pex5D cells that are unable

to degrade fatty acids by b-oxidation [31] The

observed structures were termed gnarls and it was

speculated that they could represent self assembled

fatty acid structures Different to the conspicuous ER

phenotype of the faaD mutant cells, the gnarls were of

a delicate nature and visible only upon permanganate

staining [31] This might reflect the significantly lower

amount of free fatty acids in cells containing gnarls

compared to cells described in the present study Given

the significant concentrations of free fatty acids, it

was, nevertheless, remarkable to note that the

morpho-logical changes were limited to the ER, whereas the

plasma membrane or membranes of other organelles

were essentially unaffected This selectivity appeared to

rule out passive dissolving of fatty acids into the

lipo-philic phase of membranes in general, but rather

indi-cates specific channelling to the ER Such targeted

intracellular transport of free fatty acids strongly

sug-gests the existence of an active and most likely protein

mediated process

Taken together, the data obtained so far clearly

demonstrate fatty acid uptake into the cell in the

absence of any cellular acyl-CoA synthetase activity

Therefore, the observed fatty acid transport and

acti-vation are not coupled but rather are separate tasks

that are independent of each other These results

appear to be in contradiction to the model of vectorial

acylation postulating acyl-CoA synthetase activity as a

prerequisite of fatty acid import In this model, it was

suggested that fatty acids might be removed from the

inner leaflet of the plasma membrane by acyl-CoA

synthetase activities releasing acyl-CoA into the

cyto-plasm, thereby giving space for newly incoming fatty

acids The model was based on a complete set of

experiments showing severely diminished capacity of

the strains faa1Dfaa4D and fat1D to metabolize

exo-genous fatty acids [10,27] The results obtained were

interpreted as consequence of impeded fatty acid

trans-port The crucial question is why these experiments

failed to detect the transport that we observed in our

investigations Initially, the cells in the previous studies were taken from cultures in the exponential growth phase During this stage, we showed, at least for mutants deficient of Faa1p and Faa4p, a phenotype of active fatty acid secretion obviously causing conflict with simultaneous fatty acid import By contrast, the strong capacity to import fatty acids in our experi-ments was observed for mutant cells in the stationary phase In addition, a comparison of the results obtained by uptake assays with C1-BODIPY-C12 [10,27] to those of the present study revealed the time scale as being another important difference The fatty acid uptake described in the present study is a rather slow process, observed over hours during the station-ary phase, whereas the uptake involving acyl-CoA synthetase was measured with a fluorescent fatty acid analogue in the range of 60 s Nevertheless, the two different velocities in transport might fit into a common model describing the removal of fatty acids from the inner leaflet of the membrane by two differ-ent active compondiffer-ents On the one hand, acyl-CoA synthetase activities were shown to make a contribu-tion On the other hand, a targeted channelling of free fatty acids from the inner leaflet of the plasma mem-brane towards the endoplasmic reticulum could provide an alternative mechanism for removing free fatty acids from the membrane This alternative removal system might be less efficient and result in a slower net transfer across the membrane On the other hand, the proposed model is able to balance the results obtained in the present study with the model of vecto-rial acylation and would also reconcile previous results from other studies showing fatty acid transport in the absence of acyl-CoA synthetase activity [8,32]

Experimental procedures

Yeast strains and media The yeast strains used are shown in the supplementary (Table S2) YPR consisted of 1% yeast extract, 2% peptone and 2% raffinose Yeast proteose dextrose medium con-sisted of 1% yeast extract, 2% peptone and 2% dextrose Yeast supplemented minimal media contained 0.67% yeast nitrogen base, 2% raffinose, amino acid content according

to Brent Supplement Mixture Dropout Powder (MP Biomedicals, LCC, Illkirch, France) (1.16 gÆL)1) 2% agar was added for solid media

Cell growth Yeast cultures were grown for 18 h in YPR and diluted until D600of 0.02 was reached in triplicate in flasks