Introduction

Acyl-CoA: lysophosphatidylcholine acyltransferases (LPCATs) belong to a large family of enzymes – acyl-CoA: lysophospholipids acyltransferases (LPLATs)- which in forward reaction can transfer acyl group from acyl-CoA to lysophospholipid to synthesize a corresponding phospholipid. According to their preferences for lysophospholipids LPLATs are divided into three groups: LPAATs (preferentially utilizing lysophosphatidic acid), LPEATs (showing the highest activity towards lysophosphatidyletanolamine) and studied in this project LPCATs (with the highest affinity towards lysophosphatidylcholine). All three groups occur commonly in microorganisms, plants and animals1,2,3. In the eukaryotic photosynthetic microalga Phaeodactylum tricornutum so far only one enzyme of LPCAT type has been cloned and characterised. This characterisation, however, was limited only to its forward reaction4.

The diatom P. tricornutum is capable of producing high amounts (up to 30% of total fatty acids) of omega-3 very long-chain polyunsaturated fatty acids (VLC-PUFA) among which eicosapentaenoic acid (EPA; 20:5∆5,8,11,14,17) holds the dominant position5,6,7. The biosynthesis of EPA in P. tricornutum occurs most probably via a combination of ω-6 and ω-3 pathway6. In ω-6 pathway the first step in EPA biosynthesis is the conversion of 18:2∆9,12 to γ-linolenic acid (GLA; 18:3∆6,9,12). Then, GLA is further desaturated by ∆15 desaturase to produce stearidonic acid (SDA; 18:4∆6,9,12,15). All these reactions occur when fatty acids are esterified at the sn-2 position of PC and enzymes participating in such modifications are generally membrane bound6,8,9. The following step in EPA biosynthesis requires, however, a soluble substrate – 18:4-CoA. Thus, stearidonic acid synthesised in PC has to be transferred to cytoplasmic acyl-CoA pool, most probably via the backward reaction catalysed by LPCAT enzyme. The elongated product – 20:4∆8,11,14,17-CoA (ETA-CoA), is in turn esterified to LPC via forward reaction of LPCAT. In created PC, eicosatetraenoic acid (20:4) is further desaturated to 20:5∆5,8,11,14,17 (EPA). The newly formed EPA can be again transferred from PC to acyl-CoA pool via the (most probably) backward reaction of LPCAT. In ω-3 pathway 18:3∆,9,12,15 produced by the desaturation of 18:2∆,9,12 in the sn-2 position of PC via FAD3 action is transferred to acyl-CoA (backward reaction of LPCAT) and then elongated to 20:3∆11,14,17 (ETE, eicosatrienoic acid) The elongated product (20:3-CoA) is first desaturated via ∆8 desaturase action (20:4∆8,11,14,17) and subsequently further esterified to PC were it undergoes next desaturation at the ∆5 position to create EPA6,8,10,11,12.

EPA and other omega-3 VLC-PUFAs are an essential nutrient of human diet. In small amounts they can be synthesized by human body from linolenic acid (18:3∆9,12,15) common in everyday diet, nevertheless marine fishes remain the main source of these fatty acids12,13,14,15,16. Fishes do not produce VLC-PUFA, instead acquire them from consumed microalgae, such as diatoms12,13. Due to overexploitation of marine stocks, a sustainable alternative source of VLC-PUFA has to be found and oilseed crops seem to be natural candidates for production of these important fatty acids. However, naturally existing plants do not produce such fatty acids17. Only transgenic plants with genes of biosynthetic pathway of VLC-PUFA could provide a solution. As a primary producer of omega-3 VLC-PUFA, the marine diatom is one of the potential donors of such genes10,14,15,17.

LPCATs play an important role in the biosynthetic pathways of omega-3 VLC-PUFA. Identifying LPCATs with an appropriate specificity towards the substrates of the biosynthetic pathways of VLC-PUFA seems pivotal to successful production of these fatty acids in transgenic plants. The LPCAT from primary producer of omega-3 VLC-PUFA has probably all of the specificities necessary for those fatty acids biosynthesis. The characterised LPCAT from diatom P. tricornutum possessed all of needed preferences towards acyl-CoAs from biosynthetic pathway of EPA, contrary to e.g. LPCAT from Camelina sativa which is missing specificity towards e.g. 20:4-CoA, a key substrate in EPA production4. As described above, in the biosynthesis of VLC-PUFA not only substrate specificity of LPCAT towards acyl-CoA from the biosynthetic pathway of these fatty acids (in the synthesis of PC with these intermediates) is important, but also the transfer of intermediates (after modification in PC) to acyl-CoA pool for elongation is essential. This transfer could be also attributed to LPCAT activity, namely to its specificity in the backward reaction. So far, however, LPCATs’ specificity in the backward reaction towards PC carrying fatty acids from biosynthetic pathway of EPA has not been characterised. Thus, in the presented research, we have biochemically modified PC of microsomal fraction of transgenic yeast with introduced gene encoding LPCAT of P. tricornutum and characterised remodelling efficiency of these PC species via PtLPCAT1 action. Additionally, we have tested the impact of different environmental factors on remodelling process of PC by PtLPCAT1.

Results

Remodelling of fatty acids of phospholipids naturally present in yeast microsomal fractions by PtLPCAT1 enzyme

In order to study the activity of the PtLPCAT1 enzyme in the process of remodelling of yeast endogenous phospholipids (PC, PE and PA), two types of experiments were conducted: (i) with yeast microsomal fractions isolated from Δale1 yeast cells (with knocked-out encoding gene of ALE1 - the main yeast acyl-CoA: lysophospholipid acyltransferase) transformed with an empty plasmid pYES2/CT (control) and (ii) with microsomal fractions of Δale1 yeast cells transformed with pYES2/CT plasmid carrying the Phatr3_J20460 gene encoding PtLPCAT1. Assays with both types of microsomal fractions were conducted with and without DTNB, an inhibitor of backward reactions catalysed by LPLAT type of acyltransferases; DTNB binds free CoA1. Assays were carried out under conditions promoting the remodelling process with the addition of five different exogenous acyl-CoAs: [14C]16:0-CoA, [14C]18:0-CoA, [14C]18:1∆9-CoA, [14C]18:2∆9,12-CoA or [14C]18:3∆9,12,15-CoA for 30 and 60 min. The differences in activity of remodelling of the tested microsomal phospholipids in assays with microsomes of yeast overexpressing PtLPCAT1 and microsomes of control yeast (both without and with DTNB addition) were considered as remodelling caused by PtLPCAT1 enzyme action. This remodelling process was performed via “backward reaction” and “other reactions” (see last chapter of “Methods”). The intensity of both types of remodelling of microsomal phospholipids via PtLPCAT1 action is presented in Fig. 1.

We understand remodelling of microsomal phospholipids as replacement of fatty acids naturally present in these lipids for fatty acids from added exogenous acyl-CoA. Its intensity/efficiency was measured as the amount of [14C]acyl molecules from acyl-CoA incorporated into remodelled phospholipids in a unit of microsomal fraction containing 1 nmol of microsomal PC per unit of time. Such remodelling was much more efficient in assays with microsomes of yeast overexpressed PtLPCAT1 than in assays with control microsomes, especially in case of PC and PE (Tables S1 and S2). The differences were less pronounced in case of PA, which seems natural, as these microsomes contain yeast endogenous LPAAT - SLC1. Acyl-CoA present in assays had a strong effect on the remodelling intensity. This was especially visible in assays with microsomes of yeast overexpressing PtLPCAT1 (Tables S1 and S2).

Among yeast microsomal phospholipids, PC was the most efficiently remodelled via the action of PtLPCAT1. In case of PE the amount of [14C]acyl molecules from acyl-CoA incorporated into this lipid was 3–4 times smaller and in case of PA less than 10 times smaller (Fig. 1). The differences in percentage of replaced fatty acids in PE and PA were, however, less pronounced due to smaller amount of these lipids in the tested microsomal fractions (PE accounted for about 54% and PA for about 10% of PC content -Table S3). In assays with acyl-CoA containing unsaturated fatty acids it amounted to 39–67% of replacement occurring in PC. In assays with acyl-CoA with saturated fatty acids the percentage of replacement of endogenous fatty acids of PA was even higher than in PC, while in PE it remained much lower than in PC (Table S3).

Remodelling of all these phospholipids occurred both via the “backward reaction” as well as “other reactions” catalysed by PtLPCAT1. Remodelling of PC and PE was several times more efficient in case of acyl-CoA with unsaturated fatty acids (18:1∆9-CoA, 18:2∆9,12-CoA or 18:3∆9,12,15-CoA) present in the reaction mixture, compared to assays with acyl-CoA containing saturated fatty acids (16:0-CoA or 18:0-CoA). The highest efficiency was observed in assays with 18:2-CoA. In assays containing acyl-CoA with unsaturated fatty acids more than 50% of remodelling of PC and PE occurred via the “backward reaction”. The “other reactions” (reactions when PtLPCAT1 creates lysophospholipids via lipase activity or via the transfer of fatty acids from phospholipids to other than CoA acceptors18) were dominating in assays with 16:0-CoA and especially with 18:0-CoA, (Fig, 1 A and B). In case of remodelling of microsomal PA by PtLPCAT1 action, the lowest remodelling efficiency was observed in assays with 18:0-CoA. Assays with all other acyl-CoAs gave similar results. At 60 min incubation time, except for assays with 18:2∆9,12-CoA, remodelling was mainly occurring via the “backward reaction”, however at 30 min incubation time the “backward reaction” was dominating only in case of assays with 18:3∆9,12,15-CoA (Fig. 1C).

Fig. 1
figure 1

Intensity of remodelling of microsomal phospholipids (A – phosphatidylcholine; B – phosphatidylethanolamine; C – phosphatidic acid) of yeast Δale1 mutant transformed with PtLPCAT1 encoding gene (via the backward reaction and other reactions catalysed by this enzyme) in assays with different acyl-CoA. Differences in mean activities of backward reactions and other reactions obtained in assays with microsomes of yeast Δale1 mutant carrying the tested enzymes and the mean control activity (yeast Δale1 mutant transformed with an empty plasmid) are presented. The original activities (mean values and standard deviations obtained in assays without and with addition of DTNB) are presented in supplementary materials: Tables S1 and S2. The calculation formula for remodelling activity via the “backward reaction” and “other reactions” is described in “Methods” and under Tables S1 and S2.

Full size image

Remodelling of phosphatidylcholine with modified fatty acid composition of yeast microsomal fractions by PtLPCAT1 enzyme

To verify the ability of PtLPCAT1 to remodel PC containing fatty acids from biosynthetic pathway of VLC-PUFA, first the PC of yeast microsomal fractions has to be enriched with PC molecules containing such fatty acids. The enrichment of microsomal PC with new PC species was based on very high activity in forward reaction of introduced PtLPCAT. ”New” PC molecules were synthesized de novo from sn-1-18:1-LPC and acyl-CoA added to the assays (with microsomal fractions of yeast overexpressing PtLPCAT1). In preliminary experiments it was established that 5 min long reaction at 30 oC was enough for the synthesis of similar amount of “new” PC from the added substrates to the amount of microsomal PC. Moreover, it was also established that “new” PC species were synthesised only when exogenous sn-1-18:1-LPC was present in the assays; trace amount of such PC was fund in assay with only exogenous acyl-CoA (Fig. S3). Thus, newly synthesised PC species contained in sn-1 position exclusively 18:1 and in sn-2 position fatty acids from added to the assays acyl-CoA. Microsomal membranes with modified PC species were collected by centrifugation at 5 oC (low temperature stops further enzymatic conversions of microsomal lipids) and after removing liquid phase (buffers with all additions favouring the “forward reaction”) further used in assays targeting the activity of PtLPCAT1 towards remodelling microsomal phospholipids (new buffers favouring remodelling process were added to the re-pelleted microsomes and assays were incubated at 30 oC for 30 and 60 min). Part of assays (re-pelleted microsomes) was used to evaluate precisely how much new PC species were de novo synthesised in the forward reaction. This part of the assays (re-pelleted microsomes) was used for TLC/GC analyses of PC. Content of the newly introduced PC and old microsomal PC is presented in Supplementary materials (Table S4).

The performed modifications resulted in five different microsomal fractions enriched with: sn-1-18:1-sn-2-18:3(n-3)-PC, sn-1-18:1-sn-2-18:3(n-6)-PC, sn-1-18:1-sn-2-18:4(n-3)-PC, sn-1-18:1-sn-2-20:4(n-3)-PC and sn-1-18:1-sn-2-20:5(n-3)-PC. The newly introduced PC species in the sn-2 position contained fatty acids of biosynthetic pathway of VLC-PUFA. In assays promoting the activity of PtLPCAT1 towards remodelling of microsomal phospholipids, the efficiency of remodelling of PC of such modified microsomal fractions, via the action of this enzyme, was several times higher than remodelling intensity of PC of unmodified microsomes (Figs. 1 and 2). The PC of microsomal fraction enriched with sn-1-18:1-sn-2-18:3(n-6)-PC was most intensively remodelled. Already after 30 min of incubation, in assays with acyl-CoAs with unsaturated fatty acids, 30 percentages or more of fatty acids present in microsomal PC were replaced with fatty acids from these acyl-CoAs (Fig. 2A). In similar assays with microsomal fractions of the tested yeast containing only natural PC such a replacement did not exceed 5% (Fig. 1A). The next most intensively remodelled PC was PC of microsomal fraction enriched with sn-1-18:1-sn-2-18:4(n-3)-PC followed by PC of microsomal fractions enriched with sn-1-18:1-sn-2-20:5(n-3)-PC, PC of microsomal fraction enriched with sn-1-18:1-sn-2-20:4(n-3)-PC and PC of microsomal fraction enriched with sn-1-18:1-sn-2-18:3(n-3)-PC (Fig. 2). Nevertheless, even the last one was up to 1.5 times more intensively remodelled than unmodified microsomal PC despite the fact that the unit of microsomal fractions (containing 1 nmol PC) possessed only about a half of the amount of PtLPCAT1 present in unmodified microsomes (the amount of microsomal PC was increased by a similar amount of new PC species without increasing the enzyme content).

The remodelling of PC for three out of five of such modified microsomal fractions occurred mainly via the backward reaction of PtLPCAT1. It was the case for PC of microsomal fractions enriched with: sn-1-18:1-sn-2-18:3(n-3)-PC, with sn-1-18:1-sn-2-18:4(n-3)-PC and sn-1-18:1-sn-2-20:5(n-3)-PC (Fig. 2C, D and E). The remodelling of PC of microsomal fractions enriched with sn-1-18:1-sn-2-18:3(n-6)-PC and enriched with sn-1-18:1-sn-2-20:4(n-3)-PC occurred (in assays with most of the tested acyl-CoA) mainly or in significant part as the result of “other reactions” (Fig. 2A and B).

The most intensive remodelling of PC of the tested microsomal fractions enriched with new PC species occurred in assays with acyl-CoA with unsaturated fatty acids. However, it was the type of modification that decided which acyl-CoA was the most effectively utilised. In case of microsomes enriched with sn-1-18:1-sn-2-18:3(n-6)-PC and enriched with sn-1-18:1-sn-2-18:3(n-3)-PC it was 18:2-CoA and in all other cases it was 18:1-CoA. In assays with acyl-CoA with saturated fatty acids remodelling intensity was always higher in the presence of 16:0-CoA than 18:0-CoA (Fig. 2).

The remodelling process concerned most probably sn-2 position of the modified microsomal PC. After treatment of microsomal PC enriched with sn-1-18:1-sn-2-18:4(n-3)-PC (from 30 min assays favouring remodelling process) with phospholipase A2 from bee venom, [14C]18:1 (from present in the assays [14C]18:1-acyl-CoA) was found only in fatty acid fraction; no [14C] was detected in LPC fraction, containing fatty acids esterified to sn-1 position (description of the experiment and obtained results are presented in Fig. S4).

Fig. 2
figure 2

Intensity of remodelling of PC of microsomal fractions from yeast Δale1 mutant overexpressing PtLPCAT1 enriched with new PC “species” (via the backward reaction and other reactions catalysed by this enzyme) in assays with different acyl-CoA. A, B, C, D, E – assays with differently modified microsomal PC; “new” PC species supplementing endogenous microsomal PC are presented above each graph. The original activities (mean values and standard deviations obtained in assays without and with addition of DTNB) are presented in supplementary materials (Table S5 and S5 cont.). The formula for calculation of remodelling activity via the “backward reaction” and “other reactions” is described in “Method” and under Table S5 and S5 cont.

Full size image

Comparison of remodelling of PC of yeast microsomal fractions via PtLPCAT action in competition tests and specificity tests

For this experiment yeast microsomal fractions enriched with sn-1-18:1-sn-2-18:4(n-3)-PC were selected. PC supplementing this microsomal fraction serves as a very important intermediate in VLC-PUFA biosynthesis pathway. Additionally, remodelling of PC of such modified microsomes via the PtLPCAT1 action proceeded relatively fast and occurred almost exclusively via the reverse reaction of the mentioned enzyme (Fig. 2D). In the competition tests all five tested acyl-CoAs were added together in equimolar concentrations (in a given assay only one of them was radioactive). The obtained results were compared with previously obtained results where all acyl-CoAs were tested separately (specificity tests). It turned out that utilisation of the tested acyl-CoAs was similar in both types of assays. Some differences concerned 18:2∆9,12-CoA and 18:3∆9,12,15-CoA. In competition test 18:2∆9,12-CoA was better utilised than 18:3∆9,12,15-CoA contrary to specificity test. In both types of assays 18:1∆9-CoA was utilised preferentially while acyl-CoA with saturated fatty acids relatively poorly (Figs. 2D and 3A and Fig. S1).

Although the utilisation of different fatty acids in remodelling process of PC was similar both in competition tests and specificity tests, the amount of fatty acids from acyl-CoA incorporated to PC was higher in specificity tests (Figs. 2D and 3A). This could have been caused by a five times lower amount of each of fatty acids present in mixtures used in competition tests compared to the amount of acyl-CoA used individually in specificity assays. However, as only part (no more than 16%) of fatty acids from each individual acyl-CoA present in the mixtures were used for PC remodelling during the whole reaction time, it is reasonable to assume that acyl-CoAs present in the mixtures were interfering in different ways with PtLPCAT1 slowing down the remodelling process.

Remodelling of PC of yeast microsomal fractions by three different enzymes of LPCAT type: PtLPCAT1, ATLPCAT1 and ALE1

To compare the ability to remodel PC with fatty acids from VLC-PUFA pathway by other than PtLPCAT1 lysophosphatidylcholine: acyl-CoA acyltransferases, first the microsomal fractions from yeast Δale1 cells overexpressing Arabidopsis AtLPCAT1 (encoded by At1g63050) and overexpressing yeast ALE1 (encoded by ale1) were prepared. These microsomal fractions were enriched with sn-1-18:1-sn-2-18:4(n-3)-PC (as described earlier for microsomes overexpressing PtLPCAT1) and used further in assays favouring remodelling activity of the tested acyltransferases. The competition tests were selected for these experiments as more closely approximating condition present in natural environment of the tested enzymes.

The results have shown that Arabidopsis AtLPCAT1 possesses an even higher ability to remodel such a PC than diatom PtLPCAT1 and that yeast ALE1 can remodel it also, however, with a lower intensity (Fig. 3). The amount of introduced PC in the total microsomal PC could have some impact on the intensity of remodelling of PC of such modified yeast microsomal fractions (enriched with sn-1-18:1-sn-2-18:4(n-3)-PC) by AtLPCAT1 and ALE1. In case of microsomes of yeast overexpressing PtLPCAT1 introduced PC accounted for about 50% of the total PC, in case of microsomes of yeast overexpressing AtLPCAT1 for about 23% and in microsomes of yeast overexpressing ALE1 only for about 15% (Table S4).

To verify that the quality of introduced new PC affected the intensity of remodelling process of microsomal PC by the tested acyltransferases, the above-described microsomal fractions were enriched additionally with sn-1-18:1-sn-2-18:1-PC. In assays with such modifications the remodelling of microsomal PC was much slower than in assay with microsomes enriched with sn-1-18:1-sn-2-18:4(n-3)-PC. In case of microsomes of yeast overexpressing PtLPCAT1 the total amount of fatty acids introduced to microsomal PC from all acyl-CoAs present in assays was about 6.6 times lower, in assay with microsomes of yeast overexpressing AtLPCAT1 about 4 times lower and in assays with microsomes of yeast overexpressing ALE1 about 1.9 times lower (Fig. 4). In the absolute values, the amount of fatty acids from acyl-CoA introduced to microsomal PC in assays with microsomes of yeast overexpressing PtLPCAT1 and ALE1 was similar and more than 2 times higher than in assay with microsomes overexpressing AtLPCAT1.

Each of the tested LPCATs had its own characteristic pattern of fatty acids introduced to PC from acyl-CoA present in the incubation mixture. Generally, unsaturated fatty acids were introduced preferentially by all the tested acyltransferases. Among them 18:1∆9 was introduced with the highest intensity except for assays with microsomes overexpressing AtLPCAT1 enriched with sn-1-18:1-sn-2-18:4(n-3)-PC and assays with microsomes overexpressing ALE1 enriched with sn-1-18:1-sn-2-18:1-PC. In these cases, 18:2∆9,12 was somewhat better utilised than 18:1∆9 for PC remodelling (Figs. 3 and 4).

Fig. 3
figure 3

Utilisation of fatty acids from equimolar mixture of five different acyl-CoA for remodelling of PC of microsomal fraction of yeast Δale1 mutant transformed with genes encoding: PtLPCAT1 (A), AtLPCAT1 (B) and ALE1 (C) enriched with sn-1-18:1-sn-2-18:4(n-3)-PC in assays favouring remodelling process.

Full size image
Fig. 4
figure 4

Utilisation of fatty acids from equimolar mixture of five different acyl-CoA for remodelling of PC of microsomal fraction of yeast Δale1 mutant transformed with genes encoding: PtLPCAT1 (A), AtLPCAT1 (B) and ALE1 (C) enriched with sn-1-18:1-sn-2-18:1-PC in assays favouring remodelling process.

Full size image

Effect of temperature and pH value on remodelling intensity of microsomal PC via PtLPCAT1 action

For these assays microsomal fractions of yeast Δale1 mutant transformed with PtLPCAT1 encoding gene enriched with sn-1-18:1-sn-2-18:4(n-3)-PC were used. Assays favouring remodelling process of microsomal PC by the tested enzyme were carried out with [14C]18:1∆9-CoA as fatty acid donor.

We have established that the optimal temperature for remodelling process of PC via the PtLPCAT1 action was 40 °C. A decrease of temperature to 30 °C only slightly lowered the remodelling efficiency – by about 13%. However, further decrease of the temperature to 20 °C diminished this process by about 87%. At 10 °C the remodelling efficiency was at very low level of about 3.5% of the peak remodelling efficiency at 40 °C. The increase of the temperature to 50 °C and further to 60 °C reduced the remodelling process of PC of the tested microsomal fraction to about 10% and about 1% respectively of the remodelling efficiency observed at 40 °C (Fig. 5A).

The effect of pH on the activity of PtLPCAT1 in remodelling of PC of the tested microsomal fractions was examined with four buffers with different pH ranges: 0.1 M phosphate buffer (pH 5.0–8.0); 0.1 M Tris-HCl (pH 8.0–10.0); 0.1 M NaHCO3-NaOH (pH 10.0–11.0) and 0.1 M Na2HPO4-NaOH (pH 11.0–12.0). The highest remodelling efficiency occurred at pH 8.0 in assays with phosphate buffer. Lowering pH value of this buffer to 5.5 caused almost linear decrease of remodelling process of microsomal PC to about 54% of the remodelling efficiency observed at pH 8.0. Further decrease of pH to 5.0 caused a slight increase of remodelling efficiency (Fig. 5B). Replacing the phosphate buffer with Tris-HCl buffer of pH 8.0 reduce remodelling efficiency by 50%. With increasing alkalinisation of Tris-HCl buffer, a non-linear increase in remodelling efficiency was noted. At pH 10 it accounted for 70% of the highest activity recorded at pH 8.0 in assays with phosphate buffer. Changing the Tris-HCl buffer at pH 10.0 to the NaHCO3-NaOH buffer of the same pH did not alter the efficiency of remodelling process. Alkalinisation of NaHCO3-NaOH buffer to pH 11.0 decreased significantly remodelling activity (to about 18% of the highest recorded efficiency at pH 8.0). However, after changing this buffer to a Na2HPO4-NaOH buffer of the same pH, the reaction intensity increased to a high level of approximately 90% of the activity observed at pH 8.0 (phosphate buffer). The increase of this buffer alkalinity to pH 12.0 caused a sharp reduction in remodelling of PC of the tested microsomal fractions to about 6% of the maximum activity recorded at pH 8.0 (Fig. 5B).

The effect of temperature and pH on remodelling intensity of PC of microsomal fractions of yeast Δale1 mutant transformed with empty plasmid pYES2/CT was not investigated in this research. This was due to the very low efficiency of remodelling of microsomal PC in assays with such microsomes. During 30 min incubations at 30 oC only about 1.3 pmol of [14C]FA was introduced to 1 nmol of microsomal PC (Table S1), while in the same conditions in the assays with microsomal fractions of yeast Δale1 mutant overexpressing PtLPCAT1 enriched with sn-1-18:1-sn-2-18:4(n-3) (used in assays testing the effect of temperature and pH) it was about 400 pmol of [14C]FA per nmol microsomal PC (Fig. 5A). Thus, the effect of background reactions on remodelling activity observed in performed assays was negligible.

Fig. 5
figure 5

Effects of temperature (A) and pH (B) of incubation buffer on remodelling intensity of PC of microsomal fraction of yeast Δale1 mutant transformed with PtLPCAT1 encoding gene enriched with sn-1-18:1-sn-2-18:4(n-3)-PC in assays (favouring remodelling process) with [14C]18:1-CoA.

Full size image

Discussion

The involvement of enzymes of LPLAT type (acyl-CoA: lysophospholipid acyltransferases) in remodelling process of phospholipids is relatively poorly characterised. The first hypothesis concerning this process was presented by Lands19. In his model of phospholipid remodelling two types of enzymes have to be involved: lipases and lysophospholipid acyltransferases (LPLATs). The former produce lysophospholipids (LPLs) and the latter reacylate these LPLs with new fatty acids from cytosolic acyl-CoA pool (Lands’ cycle). Later on, it has been suggested that remodelling of phospholipids may occur also via LPLAT actions in reverse (named also “backward”) reaction followed by the forward reaction 20,21. After cloning the genes encoding enzymes of LPLAT type, the remodelling of phospholipids (most research concerned PC) via the backward and subsequent forward reaction has been experimentally proven1,21. Based on these results and own studies Klińska et al.22 proposed involvement of two cycles termed: “Lands’ cycle” and “LPLAT cycle” in phospholipid remodelling. Recently it has been shown, that LPLAT can be involved in first step of phospholipid remodelling (i.e. creation of LPL) also in another way than via the catalysis of the “backward reaction”18. The above mentioned studies presented evidence that LPLAT could produce LPL via their lipase activity or via the transfer of fatty acids from phospholipid to other than CoA acceptor. Created LPL could be further reacylated by LPLAT to produce appropriate phospholipids with a new set of fatty acids. It was also shown that the acyl-CoAs present in LPLAT environment can strongly affect LPLAT activity in remodelling process of phospholipids18. The results obtained in the presented study show that acyl-CoAs are also affecting the remodelling activity of PtLPCAT1. Similarly to the results concerning remodelling of phospholipids via the action of Arabidopsis AtLPCAT218 the diatom PtLPCAT1 showed much stronger remodelling activity in the presence of acyl-CoA with unsaturated fatty acids than with saturated ones which suggests that it could be a more general feature of plant LPCATs. Moreover, the obtained results added to the characteristic of remodelling of PC via the LPCATs action a new parameter affecting this process: fatty acid compositions of PC subject to remodelling.

The main goal of the presented study was to validate the suggestion that PtLPCAT1 can efficiently remodel PC molecules containing fatty acids from biosynthetic pathway of VLC-PUFA. This would fulfil the second precondition for PtLPCAT1 involvement in this pathway, i.e. delivering the intermediates of VLC-PUFA (modified in PC) for elongation process. Previously we have shown that PtLPCAT1 can successfully acylate all intermediates of VLC-PUFA to LPC creating a PC with these intermediates which could be modified by different desaturases4. From a theoretical point of view, the most favourable transfer of such intermediates for elongation process would be via the backward reaction performed by PtLPCAT1; intermediates transferred from PC as acyl-CoAs - direct substrates for elongation. However, even if intermediates are removed from PC as non-esterified fatty acids, they can be attached to CoA via the action of acyl-CoA synthesis23. We started the research by demonstrating that PtLPCAT1 can remodel PC of yeast microsomal fractions containing only fatty acids naturally present in yeast cells. Obtained results show that this remodelling occurred predominantly via the backward reaction of PtLPCAT1 and its highest intensity was in assays with acyl-CoA containing unsaturated fatty acids. Building on the observation that PtLPCAT1 in the forward reaction very efficiently acylates fatty acids from biosynthetic pathway of VLC-PUFA to LPC4 we tried to modify biochemically the PC of the tested yeast microsomal fractions. In the assay favouring forward reaction we used microsomal fractions of Δale1 mutant overexpressing PtLPCAT1, exogenous 18:1-LPC and different acyl-CoAs with fatty acids from VLC-PUFA biosynthetic pathway. Our efforts were successful and we obtained microsomal fractions enriched with five new PC species: sn-1-18:1-sn-2-18:3(n-3)-PC, sn-1-18:1-sn-2-18:3(n-6)-PC, sn-1-18:1-sn-2-18:4(n-3)-PC, sn-1-18:1-sn-2-20:4(n-3)-PC and sn-1-18:1-sn-2-20:5(n-3)-PC. The new PC species accounted for about 50% of all PC molecules present in the modified microsomes. The remodelling of PC of such modified microsomal fractions was much more effective than remodelling of PC naturally occurring in yeast. Bering in mind that remodelling process concerning mainly sn-2 position of PC, in the most efficient case fatty acids from this position were replaced by that from acyl-CoA in about 95% of PC molecules. These results suggest that enrichment of microsomal fraction with PC molecules with polyunsaturated fatty acids in the sn-2 positions, not only affect the remodelling process of such PC species but also has an impact on LPCAT activity towards other PC molecules present in the membranes. In any case, the obtained results provide strong evidence that PtLPCAT1 can deliver the intermediates of VLC-PUFA present/modified in PC for further metabolism. The very good activity of PtLPCAT1 towards all the tested PC with VLC-PUFA intermediates indicates also, that PtLPCAT1 specificity does not eliminate any of the branches of VLC-PUFA biosynthesis pathways, and consequently that the flow of intermediates would depend on the activity of other enzymes from the biosynthetic pathway of these fatty acids.

To verify whether this very good activity in remodelling of PC containing “intermediates” of VLC-PUFA is specific only/mainly to PtLPCAT1 or is a broader feature of LPCAT enzymes, the microsomal fractions of yeast Δale1 mutant overexpressing AtLPCAT2 and microsomal fractions of yeast Δale1 mutant overexpressing ALE1 were enriched with sn-1-18:1-sn-2-18:4(n-3)-PC. The enrichment process was conducted biochemically in the same way as in case of microsomes of yeast Δale1 mutant overexpressing PtLPCAT1. However, after enrichment process the microsomal fractions of yeast overexpressing AtLPCAT2 contained only about 23% of new PC and microsomal fractions of yeast overexpressing ALE1 only about 15% of this PC in total PC molecules. It was probably due to lower specificity in the forward reaction of these two enzymes to 18:4(n-3)-CoA compared to PtLPCAT1. In the same experiments we enriched also all three discussed microsomal fractions with sn-1-18:1-sn-2-18:1-PC (the same method as for enrichment with sn-1-18:1-sn-2-18:4(n-3)-PC). The remodelling process of PC of those modified microsomes depended heavily on the type of modification carried out. In all three types of microsomes enriched with sn-1-18:1-sn-2-18:4(n-3)-PC remodelling process of microsomal PC was much faster than in case of enrichment of the membrane with sn-1-18:1-sn-2-18:1-PC. In membranes of yeast overexpressing ALE1 remodelling was about 2 times more intensive (the amount of introduced new PC was the lowest) and in membranes of the two other microsomes (yeast overexpressing PtLPCAT1 and AtLPCAT2) about 4 times more intensive. The absolute remodelling intensity of PC of yeast microsomal fractions enriched with sn-1-18:1-sn-2-18:4(n-3)-PC was similar or even faster in membranes from yeast overexpressing AtLPCAT2 compared to membranes of yeast overexpressing PtLPCAT1. This suggests that diatom’s PtLPCAT1 probably does not differ much in its ability to remodel PC with “intermediates” of VLC-PUFA from the higher plants’ LPCATs. The results from these two types of enrichment of microsomal membranes (with sn-1-18:1-sn-2-18:4(n-3)-PC and with sn-1-18:1-sn-2-18:1-PC) indicate clearly that for remodelling efficiency of microsomal PC via LPCAT enzyme, the kind of PC molecules building these membranes is important and not the relation of PC to other microsomal lipids. Only enrichment with PC containing polyunsaturated fatty acids positively affected the remodelling process.

The effects of environmental factors on remodelling activity of LPCAT enzymes have not been tested so far. In the presented studies we examined the impact of two environmental factors: temperature and pH. In both cases, the effect of these factors on enzyme activity towards remodelling of microsomal PC differed from the effect on PC synthesis (forward reaction). The optimal temperature in remodelling process was 40 oC while in case of forward reaction it was 30 oC. Changes in reaction temperature exerted a stronger effect on remodelling process than on PC synthesis (presented results and4). Contrary to the effect of pH on the forward reaction catalysed by PtLPCAT14, the remodelling process occurred also in pH 5–6 with relatively high efficiency. The remodelling of PC of the tested microsomal fractions (microsomes of yeast Δale1 mutant overexpressing PtLPCAT1 enriched with sn-1-18:1-sn-2-18:4(n-3)-PC) in assays at pH 7.2 and incubation temperature of 30 oC proceeded mostly via the backward reaction. Thus, we can conclude that the changes in efficiency of forward and backward reaction of PtLPCAT1 caused by environmental factors are not similar. However, the participation of “backward reaction” and “other reaction” in remodelling of PC at different pH values has not been tested. In the previous studies it was shown that the efficiency of utilisation of a given acyl-CoA in the forward and backward reactions differed substantially2.

To summarize, it is possible to conclude that from biotechnological point of view P. tricornutum LPCAT1 fulfils all the requirements necessary to deal with substrate dichotomy during biosynthesis of VLC-PUFA. In the forward reaction it can acylate all fatty acids from the biosynthetic pathway of VLC-PUFA to LPC4 and in the backward reaction or in other reactions involved in remodelling process of PC it provides all the necessary “intermediates” present/modified in PC to the elongation process. From the point of general knowledge, this research brings new evidence on mode of action of LPCAT enzymes. Especially significant is the observation that the structure of PC building membranes affects the activity of LPCAT in remodelling of overall microsomal PC.

Methods

Chemicals

Most of non-radiolabelled and 14C labelled acyl-CoA were synthesised in our laboratory according to modified method described by Sanchez et al.24. 14C-labelled fatty acids were purchased from PerkinElmer Life Science (Waltham, MA, USA). Non labelled fatty acids were purchased from Sigma Aldrich. All other chemicals were purchased from Merck, Sigma Aldrich or MP Biomedicals.

Yeast strains

Yeast Δale1 cells (Saccharomyces cerevisiae mutant Y02431 disrupted in main endogenous LPCAT enzyme activity), with introduced plasmids pYES2/CT or plasmids pYES2/CT harbouring P. tricornutum LPCAT1 gene (Phatr3_J20460) obtained as described in our previous paper4, were used in the experiments.

Yeast cultivation and isolation of microsomal fractions

Yeast cells were cultured for 24 h with shaking (220 rpm) at 30 °C in synthetic uracil dropout medium (Yeast Nitrogen Base, Sigma-Aldrich; CSM-URA, MP Biomedicals) containing 2% glucose. After that time galactose was added (final concentration of galactose in the medium was 2%) and the cells were grown for additional 24 h. After that time yeast (OD reached 3–4) was harvested by centrifugation at 1,500x g for 10 min. The obtained pellets were washed twice with 50 mL distilled water. After second centrifugation washed pellets were suspended in “glass bead disruption buffer” (20 mM Tris-HCL, pH 7.9, 10 mM MgCl2, 1mM EDTA, 5% glycerol, 0.3 M ammonium sulphate) supplemented with protease inhibitors (Roche). Yeast suspensions were transferred to two 2 mL plastic tubes with screws, filled to ¾ volumes with glass beads (0.45–0.5 mm in diameter). The tubes were shaken 10 times for 30 s in Mini Bead Beater (BioSpec Products, Bartesville, Ok, USA). Crushed yeast cells with glass beads were transferred to 50 mL plastic tubes and centrifuged in cold room for 10 min at 1,500x g. The resulting supernatants were filtered through Miracloth into centrifuge tubes, which were subsequently centrifuged for two hours at 100,000x g in Beckman L-70 ultracentrifuge (Beckman). The pellets formed after centrifugation, containing microsomal fractions, were washed briefly with phosphate buffer (0.1 M; pH 7.2) and then suspended in a small amount of this buffer (modified method according to Dahlqvist at al.25. Obtained microsomal fractions were stored at -80 °C for further analysis. Aliquots of obtained microsomal fractions were used for phosphatidylcholine content determination. PC concentration in microsomes was determined by performing lipid extraction using modified Bligh and Dyer method26, separation of lipids of chloroform fractions by thin-layer chromatography and measuring the fatty acid content of the isolated PC by gas chromatography analysis (according to the procedure described in details in “Lipid separation and analysis” section).

Enzyme assay

The assays measuring remodelling of microsomal phospholipids (naturally present in yeast microsomes) contained 1 mg BSA, 0.2 µmol free CoA, 10 nmol of acyl-CoA: palmitoyl-CoA ([14C]16:0-CoA), stearoyl-CoA ([14C]18:0-CoA), oleoyl-CoA ([14C]18:1Δ9-CoA), linoleoyl-CoA ([14C]18:2Δ9,12-CoA) or α-linoleoyl-CoA ([14C]18:3Δ9,12,15-CoA), 40 mM potassium buffer (pH 7.2) and aliquots of microsomal fractions containing 5 nmols of microsomal PC in final assays volume of 100 µL. Reactions were carried out without and with addition of 0.5 µmol DTNB (5,5’-dithiobis-2-nitrobenzoic acid). The reaction mixtures were incubated in a Thermomixer Compact (Eppendorf) at 30 °C with shaking (1,250 rpm) up to 60 min. The reactions were terminated by addition of 375 µL chloroform/methanol (1:2; v/v), 125 µL 0.15 M acetic acid and 125 µL of chloroform. After vigorous mixing, the reaction tubes were centrifuged for 2 min at 2,000x g and the chloroform fractions (containing lipids) were transferred to new tubes. After separation of obtained chloroform fractions by TLC (see “Lipid separation and analysis” section) products of the reactions were visualised and identified on the basis of 14C-labelled standards and quantified using an autoradiograph (Instant Imager; Packard Instruments Co.)

In order to determine the activity of PtLPCAT1 in remodelling of phosphatidylcholine containing fatty acids other than those naturally occurring in yeast PC, the above-described procedure was modified. In the first stage of assays, forward reactions were performed to enrich endogenous phosphatidylcholine with de novo synthesised PC molecules containing “new” fatty acids.

Individual reactions contained 10 nmol of respective acyl-CoAs (18:1Δ9-CoA, 18:3Δ9,12,15-CoA, 18:3Δ6,9,12-CoA, 18:4Δ6,9,12,15-CoA, 20:4Δ8,11,14,17-CoA or 20:5Δ5,8,11,14,17-CoA) and 5 nmol of sn-1-18:1-LPC (Larodan). The assays were carried out with 0.1 M potassium phosphate buffer (pH 7.2; 100 µL). Microsomal fractions (source of the tested enzyme) were added in amount equivalent to 1 nmol of microsomal PC. Assays were incubated at 30 °C in Eppendorf Thermomixer Compact with continuous shaking (1,250 rpm) for 5 min. After that time, the reaction mixtures (usually from three repetitions) were combined and subject to centrifugation at 16,000x g for 20 min (cold room with temperature of about 4 °C). The supernatants were discarded and the resulting pellets (containing microsomal fractions enriched with newly synthesized PC) were suspended in the reaction mixtures and assays determining the efficiency of remodelling of PC of such modified microsomal fractions were performed according to procedure described above for non-modified yeast microsomal fractions.

Some of the pellets, obtained by centrifugation after modification procedure, were used for lipid extraction and analyses (method described below). The phosphatidylcholine content in the obtained extracts and the composition of PC fatty acids were determined on GC after separation of the extracts’ lipids on TLC.

Lipid separation and analysis

For analysis of individual lipid classes, obtained chloroform extracts were evaporated to dryness under a stream of nitrogen, dissolved in 50 µL of chloroform, applied to silica-coated plates (Silica Gel 60, Merck) and developed in a glass chamber with chloroform: methanol: acetic acid: water (85:15:10:2.5; v: v:v: v). After drying, the plates were developed by brief exposure to iodine vapor. Identification of individual lipid classes was performed by comparison of their localisation with appropriate standards. After evaporation of the iodine, the parts of gel containing appropriate lipid classes were scraped off, dried and incubated with 2 mL of the methylation mixture (2% sulfuric acid in dry methanol) for 40 min at 90 °C. After the end of transmethylation, the obtained methyl esters of fatty acids present in the analysed lipid classes were extracted to hexane and analysed on gas chromatograph (Shimadzu GC-2010) with a flame ionization detector equipped with a 60 m × 0.25 mm CP-WAX 58 CB column (Perlan Technologies).

Definition and calculation of the intensity of LPLAT “backward reaction” and “other reactions” involved in remodelling process of phospholipids.

The term remodelling process of phospholipids refers to a process of replacement of fatty acids esterified to the glycerol backbone of these phospholipids by other fatty acids from cytosolic acyl-CoA pool. In this process first lysophospholipids (LPLs) are created and subsequently these LPLs are esterified with fatty acids from acyl-CoA pool via acyl-CoA: lysophospholipid acyltransferases (LPLATs) action (forward reaction). The term remodelling via “backward reaction” refers to a part of remodelling process where LPLs are created by LPLATs in the backward reaction. In turn, the term remodelling via “other reactions” concerns a part of remodelling process where LPLs are created in reactions different than the backward reaction performed by LPLATs (LPLs can be produced e.g. by lipases, phospholipid: diacylglycerol acyltransferases, phospholipid: sterol acyltransferases or by LPLATs in reactions other than the “backward reactions” – description of these reactions in 18), (Fig. 6). In assays without addition of DTNB, LPLs are created both via the backward reaction of LPLATs and via other reactions. When DTNB is added to the assays the backward reactions performed by LPLATs is blocked (DTNB is bounded to -SH group of CoA making it unavailable to LPLATs;1). Thus, in assays with DTNB occur only remodelling via “other reactions”. By subtracting the amount of remodelled phospholipid (PL) via “other reactions” (i.e. in assays with DTNB) from the amount of remodelled PL produced in assays without DTNB we obtain the amount of remodelled LP produced solely via the “backward reaction” performed by LPLATs. Figures 1 and 2 present intensity of “backward reaction” and “other reactions” calculated according to the above formula; the data used in calculations are presented in Tables S1 and S2 and in Table S5 and S5 cont. respectively.

Fig. 6
figure 6

Schema of PC remodelling in assays without DTNB (A) and with DTNB (B) in the presence of [14C]acyl-CoA. Subtracting the amount of remodelled PC in assays with DTNB from the amount of remodelled PC in assays without DTNB results in the amount of remodelled PC obtained via the backward reaction catalysed by LPCAT.

Full size image