Molecular basis for differential elongation of omega-3 docosapentaenoic acid by the rat Elovl5 and Elovl2.

Functional characterization of the rat elongases, Elovl5 and Elovl2, has identified that Elovl2 is crucial for omega-3 docosahexaenoic acid (DHA) (22:6n-3) synthesis. While the substrate specificities of the rat elongases had some overlap, only Elovl2 can convert the C22 omega-3 PUFA docosapentaenoic acid (DPA) (22:5n-3) to 24:5n-3, which is the penultimate precursor of DHA. In order to better understand the potential for these elongases to be involved in DHA synthesis, we have examined the molecular reasons for the differences between Elovl5 and Elovl2 in their ability to elongate DPA to 24:5n-3. We identified a region of heterogeneity between Elovl5 and Elovl2 spanning transmembrane domains 6 and 7. Using a yeast expression system, we examined a series of Elovl2/Elovl5 chimeras and point mutations to identify Elovl2 residues within this region which are responsible for DPA substrate specificity. The results indicate that the cysteine at position 217 in Elovl2 and a tryptophan at the equivalent position in Elovl5 explain their differing abilities to elongate DPA to 24:5n-3. Further studies confirmed that Elovl2 C217 is a critical residue for elongation of DPA at the level observed in the native protein. Understanding the ability of elongases to synthesize 24:5n-3 may provide a basis for using sequence data to predict their ability to ultimately support DHA synthesis.

previously described ( 5 ). Recombinant yeast expressing chimeric elongase protein was supplemented with 100-200 M of 20:5n-3 (EPA) (Sapphire Bioscience, Waterloo, NSW, Australia) for 24 h. Each chimeric protein was functionally characterized before subsequent constructs were made. Data are expressed as the mean ± SD of incubations from three independent samples.

Fatty acid analysis
Total lipid was extracted from yeast cells and analyzed by gas chromatography as previously described ( 16 ). The amount of each fatty acid was expressed as a percentage of the total amount of all fatty acids. This was done by expressing the peak area for an individual fatty acid as a percentage of the total peak area for all fatty acids.

Statistical analysis
One-way ANOVA with Tukey's post hoc test was performed using Graphpad Prism version 5.03 for Windows (Graphpad Software, San Diego, CA). Statistical signifi cance was set at P < 0.05.

Sequence analysis of rat Elovl5 and Elovl2
The rat Elovl5 and Elovl2 proteins share 56% identity and include the structural features characteristic of microsomal fatty acyl elongases including seven transmembrane domains (TMDs). Twenty-three residues spanning TMD6 and TMD7 in Elovl5 and Elovl2 were identifi ed to have lower identity, 43%, compared with any similar region between TMD1 and TMD5 ( Fig. 1 ).

Native Elovl5 and Elovl2 activity
Although the topic of this investigation is DPA elongation, EPA was used as the substrate because it is common to both enzymes and the DPA that is elongated by Elovl2 is endogenously formed from EPA. EPA accumulation in the yeast cells expressing the rat Elovl5 or Elovl2 was proportional to the concentration of EPA added to the medium (data not shown). Elovl5 and Elovl2 synthesis of DPA increased proportionally with EPA substrate concentration ( Fig. 2A, B ). However, Elovl2 further elongated the newly synthesized DPA to 24:5n-3 ( Fig. 2B ), whereas Elovl5 did not ( Fig. 2A ).

Identifi cation of Elovl2 residues involved in DPA substrate specifi city
Twenty-three residues from T201 to S223 that span Elovl2 TMD6 and TMD7 were replaced with the equivalent residues from Elovl5 to form chimera 1 ( Fig. 3 ). This resulted in a loss of the unique Elovl2 conversion of DPA to 24:5n-3, but retention of the ability to convert EPA to DPA ( Fig. 4A ).
To investigate which combinations of residues were responsible for the DPA to 24:5n-3 function, residues were progressively changed back to the original Elovl2 sequence. Initially, changes were made to leave Elovl5 residues in each TMD with the rationale that these would be adjacent

Construction of the chimeric elongase protein
Amplifi cation of the 600 bp rat 5 ′ -Elovl2 (position 1-600 bp), 171 bp rat 3 ′ -Elovl2 (position 670-840 bp), and 69 bp (position 643-711 bp) rat Elovl5 fragments was performed using the chimera 1 primers in supplementary Table I, template pYES2-Elovl2 or pYES2-Elovl5, respectively ( 5 ), and Finnzymes Phusion Hot Start High-Fidelity DNA Polymerase (New England BioLabs Inc., Arundel, Qld, Australia). Cycling conditions were as follows: initial denaturation step at 98°C for 30 s, followed by 25 cycles of denaturation at 98°C for 10 s, annealing at 72/70°C (Elovl2/Elovl5) for 20 s, and extension at 72°C for 15 s, followed by a fi nal extension at 72°C for 5 min. The Elovl2 PCR products were gel purifi ed and Dpn I digested to remove template, while the 69 bp Elovl5 PCR product was used directly for subsequent amplifi cations. Chimera 1 was formed using two steps. Amplifi cation of the 5 ′ -Elovl2+Elovl5 and Elovl5+3 ′ -Elovl2 fragments was initially performed with template 5 ′ -Elovl2 and Elovl5 or Elovl5 and 3 ′ -Elovl2, respectively, using the Elovl2 cycling conditions above. Chimera 1 was then amplifi ed using the Elovl2 primers containing restriction enzyme sites fl anking the open reading frame (supplementary Table I, chimera 1) and template 5 ′ -Elovl2+Elovl5 and Elovl5+3 ′ -Elovl2. Cycling conditions were as follows: initial denaturation step at 98°C for 30 s, followed by 10 cycles of denaturation at 98°C for 10 s, annealing at 65-10°C (with each cycle the temperature was reduced by 1°C) for 20 s, and extension at 72°C for 15 s, followed by 20 cycles of denaturation at 98°C for 10 s, annealing at 56°C for 20 s, and extension at 72°C for 15 s, followed by a fi nal extension at 72°C for 5 min. The chimera 1 cDNA and the expression vector pYES2 (Invitrogen Australia Pty. Ltd., Mount Waverley, Vic, Australia) were restriction enzyme treated and ligated using T4 DNA ligase (1.5 Weiss units) (Promega, WI). Transformation of the resulting construct, pYES2-chimera1 into MAX Effi ciency ® DH5a™ Competent Escherichia coli cells (Invitrogen Australia Pty. Ltd.) was performed using heat-shock. Putative transformants were selected using 100 mg ml Ϫ 1 ampicillin and PCR screening. Recombinant plasmids were purifi ed and sequenced at the Institute of Medical and Veterinary Science (Adelaide, Australia).

Site directed mutagenesis of the chimeric elongase, Elovl5, or Elovl2
Site directed mutagenesis (SDM) was used to change Elovl5 amino acids in chimeric protein 1 back to the equivalent Elovl2 amino acids. A series of SDM resulted in the construction of chimeric proteins 2-5. Individual amino acid changes were also made in Elovl5 or Elovl2 using SDM. Complementary primers with a minimum of twelve base pairs on either side of the introduced mutation were designed (supplementary Table I). PCR amplifi cation was performed using the primers and corresponding template outlined in supplementary Table I and Finnzymes Phusion Hot Start High-Fidelity DNA Polymerase (New England BioLabs Inc.). Cycling conditions were as follows: initial denaturation step at 98°C for 30 s, followed by 25 cycles of denaturation at 98°C for 10 s and annealing/extension at 72°C for 4 min, followed by a fi nal extension at 72°C for 5 min. SDM products were cleaned, Dpn I digested, and transformed into E. coli as previously described.

The effect of Elovl2 point mutations on EPA activity
The substitution of cysteine for tryptophan in the Elovl5 W231C mutant showed the importance of Elovl2 cysteine at position 217 for elongation of DPA to 24:5n-3. When Elovl2 C217 was substituted into the equivalent position in Elovl5, there was a restoration of 24:5n-3 synthesis ( Fig. 5A ). In the reverse mutant, where Elovl5 W231 was substituted into the equivalent position in Elovl2, the ability to convert DPA to 24:5n-3 was lost but EPA to DPA synthetic capability was retained ( Fig. 6A ). To further examine the role of tryptophan in the loss of Elovl2 DPA elongation, Elovl2 point mutations were made with either the less spacefi lling residue alanine or another bulky residue phenylalanine. The Elovl2 C217A ( Fig. 6B ) and Elovl2 C217F ( Fig. 6C ) mutants retained DPA to 24:5n-3 activity, although at reduced levels with 24:5n-3 reaching 2.6 and 1.8%, respectively, after 200 M EPA supplementation compared with 3.8% with native Elovl2 ( Fig. 2B ).

The effect of Elovl5 point mutations on EPA activity
Results with chimeric proteins 3, 4, and 5 demonstrate that the residues important for native Elovl2 DPA activity include L204-A206 in TMD6 and C217-I219 in TMD7. These residues correspond with Elovl5 S218-G220 in TMD6 and W231-Y233 in TMD7. Within these six residues, only Elovl2 L218/Elovl5 L232 is conserved. The effect of individual substitutions of the other fi ve Elovl5 residues Fig. 1. A deduced amino acid sequence alignment of Elovl5 and Elovl2 from rat. Elovl5 amino acid numbering is shown above the alignment and Elovl2 numbering shown below the alignment. An asterisk indicates every ten amino acids. Identity/similarity shading was based on the Gonnet series matrix produced by ClustalX where primary black shading indicates identical residues and secondary and tertiary gray shading indicates similar residues with an 80% and 60% cut off, respectively. Seven predicted transmembrane spanning domains were predicted using the HMMTOP transmembrane topology prediction server version 2.0 and are shown with a solid lined box and a region of lower identity between transmembrane domains 6 and 7 is shown with a dashed lined box. only one position where a residue is conserved across Elovl5 from all species which is different to the conserved residue across Elovl2 from all species, and this is the tryptophan at position 231 in rat Elovl5 which is equivalent to the cysteine at position 217 in rat Elovl2 ( Fig. 7 ).

Comparing functionally characterized fi sh and mammalian Elovl5 and Elovl2
The region of 23 residues examined in chimera 1 was compared with the deduced Elovl5 and Elovl2 protein sequences across other mammals and fi sh ( Fig. 7 ). There is  targeted for involvement in DPA substrate specifi city due to its lower sequence identity compared with the overall sequence.
Confi rmation that the TMD6 and TMD7 region was important for the differing substrate specifi cities between Elovl5 and Elovl2 was provided by chimera 1 in which 23 residues that span Elovl2 TMD6 and TMD7 were replaced with the equivalent residues from Elovl5. This resulted in a loss of the unique Elovl2 conversion of DPA to 24:5n-3, but with retention of the ability to convert EPA to DPA. This success provided the platform for the sequential changes in the Elovl5 insert to determine which residues were important in restoring the Elovl2 functionality of converting EPA to DPA and then to 24:5n-3. Chimeras 4 and 5 demonstrated that fi ve residues were potentially important, three in TMD6 and two in TMD7. Of these fi ve

DISCUSSION
The initial reason for examining the TMD6 and TMD7 region of the rat Elovl5 and Elovl2 arose from a report that the TMD6 and TMD7 region of the yeast elongase, Sur4p, was responsible for the elongation of C 18 substrates to C 26 and the determination of the chain length ( 13 ). If this region is important for elongation activity in the rat enzymes, including the different substrate specifi cities between Elovl5 and Elovl2, it is expected that within the two sequences there must be regions of homology which enable both proteins to elongate EPA and other regions of heterogeneity which enable Elovl5 and Elovl2 to elongate SDA or DPA, respectively. An alignment of the rat Elovl5 and Elovl2 proteins highlighted a 23 residue region of heterogeneity spanning TMD6 and TMD7. This region was residue such as phenylalanine, which contains a benzyl side chain similar to tryptophan or the structurally simple alanine, retained DPA to 24:5n-3 activity, although at reduced levels. The ability of Elovl2 to elongate DPA may be due to the effect of C217 in TMD7 on the structure of Elovl2.
An alignment of the deduced Elovl5 and Elovl2 protein sequences from other functionally characterized mammals and fi sh supports the essentiality of cysteine at the equivalent position across all Elovl2 proteins ( Fig. 7 ). Likewise a tryptophan is found at the equivalent position across all 16 of the Elovl5 sequences used in the alignment ( Fig. 7 ).
Although chimera 1 and Elovl2 C231W resulted in a loss of DPA activity making the enzymatic activity of these proteins more Elovl5-like, these proteins did not gain signifi cant Elovl5-like SDA activity (data not shown). Similarly, the gain of DPA activity by Elovl5 W231C did not result in a loss of Elovl5-like SDA activity (data not shown). These fi ndings suggest that the residues responsible for SDA substrate specificity are not within TMD6 and TMD7, but instead in a separate region of Elovl5.
The opposite chimeric construct was made by replacing the 23 residues from I215 to G237 that span Elovl5 TMD6 and TMD7 with the equivalent from Elovl2. Interestingly, this chimeric protein was inactive when expressed in yeast and no longer able to convert SDA or EPA (data not shown). A similar fi nding was reported in the fungi elongases when chimeric proteins of PirELO and PinELO were made. The inclusion of a region of PirELO residues in PinELO resulted in a gain of EPA substrate specifi city, whereas the reciprocal chimera resulted in an inactive PirELO which was no longer able to convert GLA or EPA ( 15 ).
We have reported that the chicken Elovl5 has some ability to elongate DPA ( 9 ). This is unlike the Elovl5 enzymes of rat ( 5,17 ), human ( 18 ), and most, but not all, fi sh ( 6,8,16,19,20 ). The current study does not identify sequence differences between the chicken and rat Elovl5, which may explain their different abilities to elongate DPA. Also, it does not identify sequence variability which could explain the higher activity of the chicken Elovl5 which converts 20% DPA to 24:5n-3 compared with Elovl5 DPA conversion activities of 5-9% in sea bream, zebrafi sh, cobia, and Atlantic bluefi n tuna ( 8,10,11,21 ). The sites within these Elovl5 enzymes that confer DPA elongation ability may not be within the transmembrane regions examined in this study with rat enzymes.
The results of this study provide a starting point for further examination of the differing abilities of Elovl5 and Elovl2 to elongate DPA and the differing abilities of Elovl5 enzymes in different species to elongate DPA. A comprehensive understanding of the molecular differences responsible for these differing activities could allow sequence data to be used to assess the ability of a species or different breeds of a domestic species to elongate DPA to 24:5n-3, a critical and perhaps rate-limiting reaction for DHA synthesis. residues, the point mutations revealed that it was the C217 residue of Elovl2 that was critical for elongation of DPA at the level observed in the native protein. A further fi nding was that the loss of DPA to 24:5n-3 activity in Elovl2 C217W appeared to be caused at least in part by the inclusion of a tryptophan residue, which is at the equivalent position in Elovl5, and not simply due to the loss of the cysteine residue at this position. Cysteine is a less spacefi lling residue than tryptophan, which may facilitate the entry of DPA further into the transmembrane channel. However, the same substitution with another hydrophobic   7. A deduced amino acid sequence alignment of mammalian and fi sh Elovl5 and Elovl2. Identity/ similarity shading was based on the Gonnet series matrix produced by ClustalX where primary black shading indicates identical residues and secondary and tertiary gray shading indicates similar residues with an 80% and 60% cut off, respectively. The conserved tryptophan in all Elovl5 sequences and cysteine in all Elovl2 sequences is indicated with an arrow.