Elovl4 and Fa2h expression during rat spermatogenesis: a link to the very-long-chain PUFAs typical of germ cell sphingolipids[S]

The sphingolipids (SLs) of rodent spermatogenic cells (spermatocytes, spermatids) and spermatozoa contain nonhydroxylated and 2-hydroxylated versions of very-long-chain (C26–C32) PUFAs (n-V and h-V, respectively) not present in Sertoli cells (SCs). Here, we investigated the expression of selected fatty acid elongases [elongation of very-long-chain fatty acid protein (Elovl)], with a focus on Elovl4, and a fatty acid 2-hydroxylase (Fa2h) in rat testes with postnatal development and germ cell differentiation. Along with Elovl5 and Elovl2, Elovl4 was actively transcribed in the adult testis. Elovl4 mRNA levels were high in immature testes and SCs, though the protein was absent. The Elovl4 protein was a germ cell product. All cells under study elongated [3H]arachidonate to tetraenoic and pentaenoic C24 PUFA, but only germ cells produced C26–C32 PUFAs. Spermatocytes displayed the highest Elovl4 protein levels and enzymatic activity. Fa2h mRNA was produced exclusively in germ cells, mostly round spermatids. As a protein, Fa2h was mainly concentrated in late spermatids, in the step of spermiogenesis in which they elongate and their heads change shape. The expression of Elovl4 and Fa2h thus correlate with the abundance of n-Vs and h-Vs in the SLs of rat spermatocytes and spermatids, respectively.

24:4n-6 and 24:5n-3 may be subject to 6 desaturation to produce 24:5n-6 and 24:6n-3, respectively, and these C24 PUFAs may be further elongated to give rise to several n-Vs, 30:5n-6 among others, in the mouse testis (10). Ablation of the Elovl2 gene in mice results in deficiency from the testis of C24 and longer PUFAs, in association with arrested spermatogenesis (10).
In the retina of some vertebrates, PUFAs with chain lengths up to 34:5n-6 and 34:6n-3 also abound, highly concentrated in the phosphatidylcholine present in the rhodopsin-containing disk membranes of rod retinal photoreceptors (11,12). Elovl4, the enzyme responsible for the synthesis of retinal VLCPUFAs, is encoded by the Elovl4 gene. This concurs with the fact that the retina displays a high expression of this gene among primate (13) and mouse (14) tissues. From invertebrates to mammalian species, Elovl4 orthologs share sequence homology with human ELOVL4, suggesting functional conservation during evolution (15). The human ELOVL4 gene and the protein it encodes have received intense research because specific mutations in the human ELOVL4 gene are associated with different forms of retinal macular dystrophies that include Stargardt disease type 3 (15)(16)(17)(18)(19), a group of blindness-causing disorders. Using a gain-of-function approach in cells in culture (20) and photoreceptor-specific conditional knockout mice (21), direct evidence was provided that Elovl4 is required for the biosynthesis of the VLCPUFAs typical of retina (20,21) and for proper photoreceptor function (21). Considering that in addition to sharing the type of VLCPUFAs produced, mouse retina and testes share an important expression of the Elovl4 gene (14), it is not unlikely that ablation or mutations in this gene will affect male fertility.
In testes from immature rats, Cer and SM do not contain PUFAs longer than C24. The proportion of n-Vs increases concomitantly with the appearance of the first round of pachytene spermatocytes (PtSs) and that of h-Vs accompanies that of spermatids (4). The h-V-containing species of Cer and SM may be considered a hallmark of germ cell differentiation, as their concentrations increase from round spermatids (RSs) to elongated spermatids [late spermatids (LSs)] to finally become components of spermatozoa (3,22). The biosynthesis of these fatty acids is expected to require the activity of a fatty acid 2-hydroxylase (Fa2h).
The Fa2h gene is highly expressed in mammalian brain (23,24) and peripheral nerve (25) and also in skin (26,27), where it is thought to encode the enzyme responsible for the 2-hydroxylation of the very-long-chain (saturated) fatty acids that abound in myelin galactosylceramides and skin glucosylceramides. In addition to brain, skin, and intestine, the mouse homolog of Fa2h is markedly expressed in adult testes (23), where it may encode the form of Fa2h that is responsible for the abundance of h-Vs in rodent spermatogenic cell SLs.
In the present study, the seven Elovl family members were investigated for their transcription (mRNA) in adult rat testes. A special emphasis was given to Elovl4, on the assumption that the Elovl4 enzyme participates in the biosynthesis of n-Vs. Elovl5 and Elovl2 were included in this context because the enzymes they code cooperate in the elongation of C18-C24 PUFAs, thus playing important roles as providers of the major C20 and C22 PUFAs of testis on the one hand, and as providers of the C24 PUFAs that will then be further elongated by Elovl4 on the other. The postnatal developmental and cell type-associated transcription levels of the three Elovl genes are compared, and their concurrent activity as enzymes is shown to be maximal in PtSs, decreasing in spermatogenic cells as they mature. Elovl4 and Fa2h are shown to be transcribed and translated differentially during testicular development and to arise at differentiation stages and cell locations that correlate with the abundance of membrane SLs with n-Vs and h-Vs in spermatocytes and spermatids, respectively.

Tissue and cell isolation
Male Wistar rats of different ages were used. The animals, housed in a standard animal facility kept at 22°C, 50-55% humidity, and 12:12 h light:dark cycles, were allowed ad libitum access to food and water. The experimental protocols were approved by the institutional Committee for the Care and Use of Laboratory Animals of the Universidad Nacional del Sur, Argentina. Whole testes were obtained at the postnatal days indicated in the Results, from postnatal day (P)14 to P90 (adult) males. Seminiferous tubules (STs) were isolated from adult testes and, after enzymatic digestion with collagenase, trypsin, and DNase, the resulting germ cell suspensions were subjected to velocity sedimentation separation through BSA density gradients by the STAPUT technique (28). Fractions containing PtSs, RSs, and LSs were obtained and identified as previously described (3,29). As an average, these preparations showed respective purities of 84, 90, and 76%. To ensure absence of germ cells from Sertoli cell (SC) preparations, the latter were obtained from prepubertal testes (rats aged 14-16 days) following standard procedures (30) and cultured during 2 weeks at 37°C under a 5% CO 2 /95% air atmosphere in DMEM:F12 serum-free medium (Gibco, BRL; 1:1, v/v) before use.

Elovls and Fa2h mRNA
RT-PCR of Elovls mRNAs. The RNA from testes, STs, and cell preparations was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Total RNA samples were suspended in RNase-free water, their quality and concentrations were assessed using a PicoDrop spectrophotometer, and they were stored at 80°C until use. From the 260/280 absorbance ratios, the RNA purity ranged from 1.75 to 1.95 (mean 1.85). Total RNA (2 g) was used to synthesize cDNA by reverse transcription in a final volume of 25 l. The reverse transcription reaction vials contained 1 g Random Primers hexamers (Biodynamics), 1× M-MLV RT reaction buffer, 0.5 mM each of dNTP, 25 UI RNase inhibitor (Promega), and 200 UI M-MLV RT (Promega). Of the cDNA solution, 3 l were used for the PCR reaction with Taq DNA polymerase, carried out in a volume of 20 l. The gene-specific primer pairs designed for the amplification of rat Elovls (Elovl1 to Elovl7) are listed in Table 1. Samples were denatured at 94°C for 5 min, followed by amplification rounds consisting of denaturation at 94°C for 30 s, annealing at specific temperature depending of the primer pair for 30 s, and final extension steps at 72°C, first for 30 s for 30 cycles and then for 10 min. The RT-PCR products were stained with ethidium bromide and assessed using 1.5% agarose gel electrophoresis.
Quantitative PCR for Elovl5, Elovl2, Elovl4, and Fa2h. The cDNAs were amplified by real-time quantitative PCR (qPCR) in a final reaction volume of 10 l using KAPA SYBR FAST Master Mix (Kapa Biosystems) and a 0.20 M concentration of each primer. Primer sequences for the specific amplification of rat Elovl2, Elovl5, Elovl4, and Fa2h genes (Table 1) were designed with Primer3 software and purchased from Invitrogen Life Technologies. Gene expression levels and melt-curve analyses were determined using Rotor-Gene 6000 (Corbett Research, Australia). The PCR conditions were as follows: 40 cycles of denaturation at 94°C for 20 s, annealing and extension at 58°C for 30 s, and a final extension step at 72°C for 30 s. The PCR products were confirmed by agarose gel electrophoresis (supplemental Fig. S1).
Data analysis and validation. In each physiological or experimental condition (developmental stage, cell type, or weeks after hyperthermia treatment) a total of 12 measurements for each gene under study were performed: three biological replicates (three independent reverse transcription reactions) and four technical replicates. Using a pool of cDNA from all experimental conditions, standard curves using 1/10 dilution series were prepared and the amplification efficiencies of target and reference genes were calculated, which ranged between 104% and 110% (Table 1).
For normalization of data against validated reference genes, the raw Cq values of the target genes (Elovl5, Elovl2, Elovl4, and Fa2h) and the following five candidates to reference genes: -Actin, alkylglycerone phosphate synthase (Agps), hypoxantineguanine phosphoribosyl transferase (Hprt), TATA box-binding protein (Tbp), and phosphoglycerate kinase (Pgk1) were imported into the qBasePLUS software (31). In qBasePLUS, the ge-Norm module was used to identify the most stable reference gene combination for each experiment (32). In the case of data sets from postnatal development and posthyperthermia treatment samples, the combination of five genes (-Actin, Agps, Hprt, Tbp, and Pgk1) gave a reference gene stability factor (geNorm M) of 0.47 and 0.24 (optimal <0.5) and a variability between sequential normalization factors (geNorm V) of 0.15 and 0.15 (preferable <0.15), respectively. In the case of data sets from isolated cell types (SCs, PtSs, RSs, and LSs), quality analysis (qBasePLUS) indicated that inclusion of -Actin and Tbp reduced reference gene stability when comparing expression levels on the different cell types under study. For this reason, -Actin and Tbp were omitted as reference genes, and the combination of three reference genes (Agps, Hprt, and Pgk1) was found to be optimal to compare the expression of the three elongases and Fa2h, as this combination produced a good M value and the lowest V factor. Using the reliable and validated reference gene combination for each sample, the reported levels of Elovl5, Elovl2, Elovl4, and Fa2h were calculated as calibrated normalized relative quantity (CNRQ) values. These values were then exported from the qBasePLUS software and statistically investigated.
Elovl5, Elovl2, Elovl4, and Fa2h mRNA in the adult testis. The mRNA transcribed from each gene in vivo was evaluated simultaneously in the testis of adult rats. Untreated animals, used as controls, were compared with counterparts experimentally deprived of spermatogenic cells. To generate "germ cell-free" testes in vivo, 3-month-old males were exposed to mild hyperthermia (33). In sedated rats, the caudal segment of the body was immersed into a water bath at 43°C for 15 min, once a day for five consecutive days. Control animals were subjected to the same procedures, except for the bath temperature (22-25°C). All animals were euthanized 2 and 6 weeks after the last of these exposures and their testes were removed. Part of them were fixed in 10% formaldehyde, embedded in paraffin, and cut into 3 m sections for histological observation after hematoxylin-eosin staining, and part of them were subjected to RNA extraction and quantification to determine Elovl5, Elovl2, Elovl4, and Fa2h mRNA transcript levels.

Elovl4 and Fa2h proteins
Protein levels in tissue and cell samples were assessed by Western blot analysis. Immediately after isolation, homogenization was performed using lysis buffer [20 mM Tris-HCl (pH 7.4), 100 mM NaCl, 1 mM EDTA, complete protease inhibitor (Sigma P-8340)] and 1% Triton X-100. Aliquots were taken for protein quantification (DC Protein Assay; BIO-RAD Life Science Group, FL). Protein homogenates were dissolved in denaturing Laemmli buffer, boiled for 5 min (34), and resolved by SDS-PAGE on 10% polyacrylamide gels, loading equivalent amounts of protein (30 g) per lane. After electrophoresis, proteins were electroblotted to PVDF membranes (Millipore, Bedford, MA) and blocked with 5% BSA in TBS-T buffer [20 mM Tris-HCl (pH 7.4), 100 mM NaCl, and 0.1% (w/v) Tween 20] for 2 h at room temperature. The PDVF membranes were incubated overnight at 4°C with the primary rabbit polyclonal antibody, either against Elovl4 (1:1,000 in BSA) or Fa2h (1:500 in BSA). For Elovl4, a rabbit anti-Elovl4 antibody was used, kindly donated by Dr. Robert E. Anderson (University of Oklahoma Health Sciences Center, Oklahoma City, OK), previously validated for Western blot and immunohistochemistry (20). For Fa2h, the antibody, ab111554 anti-Fa2h, from Abcam was employed. After primary antibody contact, samples were exposed for 2 h at room temperature with HRP-conjugated goat anti-rabbit secondary antibody (sc-20049; Santa Cruz Biotechnology) (1:2,000). The membranes were then washed with TBS-T buffer and the immunoreactive bands were detected by ECL (Amersham Biosciences) using standard X-ray films (Kodak X-Omat AR). For loading control, the PVDF membranes were reprobed with a mouse anti- tubulin antibody (1:4,000; DM1-A) (CP06; EMD/Biosciences-Calbiochem) for 3 h and with the corresponding HRP-labeled goat anti-mouse secondary antibody (1:2,000; sc-2005 from Santa Cruz Biotechnology). Protein band intensities were compared using the ImageJ software (National Institutes of Health, Bethesda, MD).

Elovl activity
For detection of long and very-long PUFAs synthesized by the coordinated activity of the Elovls under study, the transformations undergone by [ 3 H]20:4n-6 as precursor were followed. STs, the germ cells isolated therefrom, and SCs prepared as described above, were separately placed in 35 mm diameter dishes suspended in the culture medium. The [ 3 H]20:4n-6 was added and cultures proceeded for 20 h at 32°C in a tissue culture incubator under 5% CO 2 /95% air and 95% relative humidity. The medium was DMEM without glucose (Gibco, BRL, Grand Island, NY), supplemented with 5% (v/v) FBS, 6 mM lactate, penicillin (100 U/ml), and streptomycin (100 U/ml). In a final volume of 2 ml of medium, an amount of STs equivalent to 1.5 mg of protein and an amount of cells adjusted to contain 5 × 10 6 cells per plate were used. Incubations started with the addition of 1 Ci of [ 3 H]20:4n-6 (65.9 Ci/ mmol; Perkin Elmer Life Sciences Inc., Boston, MA), previously diluted with unlabeled 20:4n-6 to reach a 2.5 M final concentration. Tubules and cells were then transferred to glass tubes, separated from the [ 3 H]labeled media by centrifugation (at 800 g for 5 min), and washed twice by suspension in PBS and similar centrifugations. The materials in the sediments were recovered for lipid extraction (35). After partitions and washings of the extracts, the total lipid was dried under N 2 and dissolved in chloroformmethanol. Aliquots from these lipid extracts were taken to check the incorporation of label from [ 3 H]20:4 into lipids and its distribution among lipid classes. After separating the major classes by TLC, most of the label was found to be esterified into lipids, the fraction remaining as free fatty acid varying among PtSs and RSs (supplemental Fig. S2). With no remarkable differences among cells, most of the esterified label was distributed between polar (choline-, ethanolamine-, and inositol-glycerophospholipids) and nonpolar lipids (mainly triacylglycerols).
To explore the transformations undergone by [ 3 H]20:4 into longer and more unsaturated PUFAs after the described 20 h incubations, the [ 3 H]labeled fatty acids from the total lipid extracts of STs and cells were examined. Fractions from these extracts were converted into fatty acid methyl esters (FAMEs), which were then separated according to unsaturation and chain length by a combination of argentation TLC and reverse phase HPLC.
To prepare FAMEs, the classical procedure described by Christie (36) was used. Fractions from the total lipid samples were placed into glass tubes fitted with screw caps and the solvents evaporated under N 2 . Nitrogen-saturated, dehydrated methanol-H 2 SO 4 was then added and, after adding a N 2 gas atmosphere, the tubes were closed and kept at 45°C overnight. The resulting FAMEs were recovered into hexane after adding water and hexane to the methanol solution, mixing thoroughly, and recovering the upper hexane phases. The FAME preparations were subjected to TLC for recovery of (nonhydroxy and 2-hydroxy) FAMEs. Prewashed silica gel G plates and two proportions of hexane:ether were used. The first one (80:20 v/v) was run up to the middle of the plates, and the second one (95:5 v/v) up to near the top (4,37). After location, FAMEs and 2-hydroxy-FAME bands were eluted by thoroughly mixing them with water:methanol:hexane (1:1:1 v/v), recovering the hexane layer, and carrying out two additional hexane extractions. The bulk of the [ 3 H]radioactivity appeared in the nonhydroxy-FAME fraction, whereas the one containing 2-hydroxy-FAME was devoid of detectable label.
To separate the total [ graded series of sucrose concentrations (10 to 20 to 30%) in PBS at 4°C for cryoprotection, embedded in a small amount of OCT compound Tissue Tek (Sakura Finetek, Torrance, CA), and stored at 80°C. Frozen 5-7 m thick sections were prepared using a cryostat (Leica CM 1860), picked up on SuperFrost Plus slides (Thermo Scientific), and refrigerated until stained. Then, tissue sections were incubated in PBS to remove OCT and postfixed for 2 min in 4% paraformaldehyde. The latter had been dissolved in PBS or in acetone:methanol

Statistical analysis
One-way ANOVA was used to determine differences among mean values, which were compared using Tukey's multiple comparison test. Data in the figures are presented as mean values ± standard deviation from at least three individual samples, obtained from independent tissue and cell preparations.

Expression of Elovl family genes
In previous work the mRNA of two enzymes involved in PUFA elongation, Elovl5 and Elovl2, were shown to be expressed in rat germ cells at two stages of cell differentiation: PtS and RS (3). Here we extended the study, using RT-PCR analysis, to other cells from the seminiferous epithelium and to other known members of the Elovl family ( Fig. 1). Of the seven mammalian Elovl family members explored, except Elovl3, the transcripts of six isoforms were detected in cells located within STs (Fig. 1). In semiquantitative terms, the most abundant transcript in SCs was that of Elovl6 and, albeit faintly, the Elovl7 mRNA was detected only in these cells. The mRNAs of Elovl1, Elovl2, Elovl5, Elovl6, and Elovl4 were actively expressed in germ cells (PtSs and RSs) as well as in SCs.

Elovl5, Elovl2, and Elovl4 mRNA expression
Given the collaborative activities of Elovl5 and Elovl2 to produce the elongated PUFAs that serve as substrates of Elovl4, and the expected association of the latter with the n-Vs of rat spermatogenic cells, we followed by RT-qPCR their expression patterns through the postnatal period and during germ cell differentiation (Fig. 2). The expression level of Elovl5 was relatively stable from prepubertal ages (P14) to adulthood in testes. Among germ cells, Elovl5 showed far higher levels of expression by the two spermatids than by PtSs. Elovl5 mRNA levels in the latter were similar to those in SCs.
Unlike Elovl5, the Elovl2 mRNA levels were very low at early developmental stages (P14). They augmented significantly at P21 and P25, in coincidence with the appearance of the first generation of spermatocytes (PtSs), and showed an evident upsurge from P25 to P30, concurrently with the timing of appearance in the testis of the first generation of spermatids, to remain high from this age to adulthood (Fig. 2). Among seminiferous epithelium cell types, Elovl2 was strongly expressed in all three spermatogenic cell types studied. Albeit not negligible (Cq value: 24), its expression was lower in SCs than in germ cells.
The Elovl4 mRNA levels in rat testes, against our expectations, were the highest at infantile ages (P14) (Fig. 2). They decreased significantly toward juvenile ages (P25) and, although more slowly, continued to decrease during the postpubertal period (P45). From this moment onwards, the mRNA levels plateaued, remaining in the relatively stable levels typical of adulthood (P90).
The distribution of Elovl4 mRNA among cells of the seminiferous epithelium explained the high expression of this gene observed at P14, taking into account that, at that age, the seminiferous epithelium is mainly composed of SCs (65% of the testicular cells). Thus, Elovl4 transcript levels were remarkably high in these somatic cells (Fig. 2). Because they were obtained from P14 testes and cultivated for as long as 2 weeks with periodic changes of culture medium before the mRNA measurements, and considering that these cells display an inherent phagocytic activity, the probability is low that such mRNA would belong to spermatogonia (SG) or other cells contaminating the samples. The high Elovl4 mRNA levels in the present SC preparations may thus be considered constitutive.
Among adult spermatogenic cells, levels of the Elovl4 mRNA were lower in the meiotic PtSs than in the more differentiated RSs and LSs (Fig. 2).

Fa2h mRNA expression
The transcription of the Fa2h gene clearly differed from that of Elovl4 during postnatal testicular development, as shown by RT-qPCR (Fig. 3). Thus, Fa2h mRNA levels were undetectable in testes up to P25, augmented rapidly from P25 to P30, in coincidence with the appearance of RSs, and showed an even steeper increase from P30 to P35, in coincidence with the timing of appearance of the first elongating spermatids. After this peak, the mRNA levels underwent a temporary, but significant, decrease from P35 to P45, and a less steep, but steady, increase thereafter up to adulthood (P90).
Unlike the high levels of Elovl4 mRNA, the Fa2h mRNA was absent altogether from SCs (Fig. 3). Among spermatogenic cells, the mRNA levels of Fa2h were lower in PtSs than in the two spermatids, being the highest in RSs.

Elongases and Fa2h mRNA levels and germ cell depletion
Short exposures of the testis of adult rats to mild hyperthermia, repeated once a day for 5 days, results in a significant loss of spermatogenic cells with permanence of SCs (33). Albeit affected in some of their functions, the latter cells remain viable (38) and, 12 weeks thereafter, the spermatogenic cells are mostly recovered. As an indirect means of exploring whether the high Elovl4 mRNA levels observed in immature SCs were also a feature of those from the adult testes, we followed the changes in gene expression of Elovl4 in comparison with those of Elovl5, Elovl2, and Fa2h in testes after exposures to hyperthermia, in comparison with untreated controls (Fig. 4). Two weeks after such exposures, and concomitantly with the depletion of spermatogenic cells with permanence of the hyperthermia-resistant somatic SCs and SG within tubules, Elovl2 mRNA levels decreased significantly and the Fa2h mRNA expression was virtually lost. Interestingly, Elovl5 transcript levels markedly increased at the same time point. Those of Elovl4 showed a similar trend of increase, although not significantly.
Six weeks after the heat exposures, the seminiferous epithelium had started to be repopulated with spermatogenic cells. Consistently, this was accompanied by the return to the control levels of Elovl5 and Elovl4 mRNA levels and a significant partial recovery of those of Elovl2 and Fa2h. Resembling the tendency during postnatal development, as both situations encompass a change from a "mostly Sertoli" to a "mostly spermatogenic" cell-rich tissue, the relative Elovl2 mRNA levels tended to increase, while those of Elovl4 tended to decrease, from week 2 to week 6 posthyperthermia.

Elovl4 protein expression
In contrast to the high mRNA levels from Elovl4, the Elovl4 protein was absent from testes at early stages of testicular development (P14), when the STs are populated by proliferating SG and SCs (Fig. 5). Elovl4 was detected in Western blots at P21, reached a high expression at P25 (PtSs already present), and persisted at similar levels from P37 onwards (Fig. 5A). The distribution of Elovl4 among isolated cell populations complemented these observations by showing that the protein was absent altogether from SCs (Fig. 5A), while its levels were high in PtSs, RSs, and LSs. The time of appearance in the testis and the abundance among germ cells was confirmed in testicular tissue sections (Fig. 5B). Thus, the protein was immunodetected intensely at P21, in coincidence with the start of the pachytene phase of meiosis undergone by the first round of primary spermatocytes (PSs). These were virtually the only members of the spermatogenic cell line that were labeled at this developmental time point. SG and SCs, detected by the single row of close nuclei forming a layer adjacent to the basal membrane of the tubules, were unstained. The Elovl4 protein continued to be detected in differentiating germ cells from this age to adulthood. In different sections of the adult seminiferous epithelium the Elovl4 staining was concentrated in adluminal PSs, it was paler in the population of spermatids located above them (mainly RSs), and became bright-red again in elongating (early and late) spermatids close to the tubular lumen (Fig. 6), strongly condensed in small spot-like structures located in the areas close to tubular lumen. Interestingly, the staining of Elovl4 varied with the stage of the spermatogenic cycle (Fig. 6A). Thus, the staining in PtSs was the most intense among tubule cells at stages II-IV and VII-VIII, but became paler to the point of virtually fading in PtSs at stages V-VI, IX-XI, and XII-XIV (a scheme illustrating the stages accompanies Fig. 6B). This suggests that the enzyme concentration per cell decreases in PtSs in the last period of the long meiotic phase.
The Elovl4 localization was cytoplasmic in PtSs, and restricted to the perinuclear area, as expected for an ER-located enzyme (Fig. 6C). In PtSs and also in RSs, a higher concentration of the protein was observed in a distinctly perinuclear and more condensed structure. In elongating spermatids, Elovl4 was even more highly concentrated in a form of densely packed structure with an unspecified cytoplasmic localization that could correspond to the cytoplasmic lobe of spermatids.

Elovl5, Elovl2, and Elovl4 enzymatic activity
As a complement to their expression, the activity of elongases as enzymes was explored. Previous work showed that Elovl5 displays elongase activity with C18 and C20 PUFAs, but not C22 PUFAs, and that Elovl2 has lower activity with C18 PUFAs than with C20-C22 PUFAs (39), albeit C20 PUFAs (20:5n-3, 20:4n-6) are similarly good substrates for the two elongases (40). When [ 3 H]20:4n-6 was used as a substrate, ST segments and cells isolated therefrom clearly elongated it to 22:4 and further to 24:4, an anticipated function of Elovl5 and Elovl2 (Fig. 7). This activity was shared by SCs and germ cells. In consistency with their lack of Elovl4 protein, SCs did not produce quantifiable amounts of PUFAs longer than C24. Only in germ cells, significant proportions of the label appeared in longer tetraenes up to 28:4, as expected for a functional Elovl4 enzyme using the available [ 3 H]24:4 as its substrate (Fig. 7). Among the cells surveyed, and on the basis of a fixed amount of protein, PtSs showed the highest ability to produce C22-C24 PUFAs (22:4n-6 and 24:4n-6). As judged from the production of labeled tetraenoic n-Vs with C26-C32 chains, these cells also exhibited the highest Elovl4 activity.
An important proportion of the label from [ 3 H]20:4n-6 appeared in longer pentaenoic fatty acids, mostly [ 3 H]22:5n-6 and [ 3 H]24:5n-6 (Fig. 7). The formation of these two pentaenes manifests the presence of highly active desaturases in all seminiferous epithelium cells under study, including SCs. However, only in spermatogenic cells, the expected substrates of Elovl4, [ Fig. 7, where comparable amounts of protein from isolated spermatogenic cells were used, it was clear that the PUFA elongation activity of Elovl4 during spermatogenesis tended to decline with the progression of cell differentiation in the direction PtS → RS → LS (Fig. 7). The fact that Elovl4 protein also decreased its levels in the order PtS → RS, but then reappeared concentrated in LSs (Fig. 6), supports our interpretation that, although present, most of the enzyme existent in LSs was partially inactive.

Fa2h protein expression
In parallel to the reported age-related increase in the proportions of h-Vs in the SM and Cer of rat testis (4), Fa2h protein levels increased in this organ as its maturation proceeded (Fig. 8). With the antibody used, Fa2h was detected    6. A: Distribution of Elovl4 in the adult testis at the indicated stages of the seminiferous epithelium cycle. The protein was strongly expressed in PSs, mainly PtSs, in a stage-dependent way. Their staining was the most intense among tubule cells at stages III-IV and VII-VIII, but was faint at stages V-VI, IX-XI, and XII-XIV. The protein level was even fainter in RSs and reappeared concentrated in small spot-like structures in elongating (early and late) spermatids (LS). B: Representative scheme of the rat spermatogenic cell differentiation cycle and the associated tubular stages, redrawn from material published by Russell et al. in 1983 (54). SGs (B, type B; In, intermediate; m, meiotic SGs); PlP, preleptotene spermatocytes; DiS, diplotene spermatocytes; LpS, leptotene spermatocytes; ZgS, zygotene spermatocytes; 2°S, secondary meiotic spermatocytes; RB, residual body. C: Confocal images (bar = 50 m). Elovl4 localization was mostly perinuclear in PSs and RS (arrow heads). The protein, in an indeterminate cytoplasmic localization, appeared densely packed in LS (arrows). Basal lamina (BL, yellow dotted lines) and tubular lumen (TL) are indicated. faintly in Western blots before P25, when it showed a marked increase (Fig. 8A). This timing coincided with the start of the second meiotic division and emergence of the first haploid spermatids. The Fa2h protein levels reached their apparent maximum at P37, as they were similar from this age to adulthood.
Spermatogenic cells were the only cells in STs in which the Fa2h protein was expressed (Fig. 8B). Such expression was lower in spermatocytes than in spermatids, in agreement with the enrichment in h-V-containing SLs associated with the PtS → RS progression (3).
In spermatids, Fa2h levels and intracellular location varied with the seminiferous epithelium stages. The protein reached its maximum level in LSs at stages III-IV, V-VI, IX-XI, and XII-XIV of the spermatogenic cycle. In spermatids at IX-XI stages, the protein was located in the cell cytoplasm, whereas in the other three stages it was concentrated in the posterior region of the heads, partially covering the cell chromatin-condensing nuclei. These three spermatogenic stages contain spermatids undergoing the most advanced steps of spermiogenesis, up to steps 17 and 18. Over the curved heads of these spermatids, Fa2h was clearly observed to surround the nuclear envelope partially: its fluorescence was intensely focused over the posterior aspect (i.e., that pointing to the lumen) of the spermatid heads. The thin and slightly curved spermatid head apices (pointing backward into the epithelium) were uncovered, as indicated by the Hoechst blue staining of the spermatid nuclear chromatin protruding out from the fluorescent structure. Confocal imaging confirmed that, in these LSs, Fa2h was concentrated over the nuclear envelope (Fig. 8C, D), in a location that suggested coincidence with the structure known as "manchette," a temporary cuplike formation that surrounds the spermatids head during their elongation. The high Fa2h concentration in this LS area was also transient. At spermatogenic stages VII-VIII, which contain spermatids at the step just preceding spermiation (step 19), Fa2h had totally disappeared (Fig. 8B),   Fig. 7. Conversion of arachidonic acid into PUFAs with longer chains by cell populations from the adult rat seminiferous epithelium cells. Total STs, and the indicated cells isolated therefrom, were independently maintained for 20 h in culture with [ 3 H]20:4n-6 as a precursor. In each case the total lipid was extracted, its fatty acids were converted into methyl esters, the latter were separated into tetraenoic (T) and pentaenoic (P) fatty acid fractions, and their components were resolved by HPLC. To comparatively summarize the activity of Elovl2 and Elovl5 on the one hand, and that of Elovl4 on the other, the sum of label in (T + P) C22-C24 and C26-C32 fatty acids, respectively, is shown in the right panels. ND, not detected.
for it was also absent from the free spermatozoa located in the tubule lumen.

DISCUSSION
In the present study, six of the seven members of the Elovl family are shown to be expressed in the adult rat testis. All of their mRNAs are produced by cells located within the STs, varying in levels among spermatogenic cells and most being also expressed in somatic SCs. Our analysis showed that Elovl5, Elovl2, and Elovl4 are differentially expressed during postnatal development, until relatively stable expression levels of each are reached in the adult testis, where the three transcripts are mostly contributed by spermatogenic cells.

Elovl5 and Elovl2
Elongation of diet-derived 18:2n-6 to C20-C22 PUFAs is mostly a function of Elovl5, while elongation of the latter to C24 PUFAs is mostly associated with Elovl2. These two elongases, but especially Elovl2, are regarded as rate-limiting in the production of the C24 PUFA acyl-CoAs (10). The , as revealed by immunofluorescence. The Fa2h protein was exclusively expressed in spermatids. It reached its maximum in LSs that were at the III-IV, V-VI, IX-XI, and XII-XIV stages of the spermatogenic cycle. Of these stages, the protein location was cytoplasmic (#) at IX-XI (steps 9-11), and in the rest of stages, which contain more advanced spermatids (up to steps 13-18 of spermiogenesis), it was concentrated around the highly condensed chromatin of the blue-stained cell nuclei (arrows). In later spermatids (stages VII-VIII, step 19 of spermiogenesis), Fa2h was no longer stained (*), indicating that its expression was temporary. C: Confocal microscopy to show the location of nuclei (TO-PRO, red) in comparison to Fa2h (green) at stages XII-XIV, using immunofluorescence (bar = 10 m). present quantitative comparisons showed that Elovl5 mRNA is already expressed in rat testes at developmental ages at which Elovl2 (and C20-C22 PUFA) levels are still very low, whereas Elovl2 is strongly upregulated in parallel with the emergence of the C20-rich spermatocytes and C22-rich spermatids (3,22,41). Interestingly, in liver from Elovl5 /mice, Elovl2 is upregulated, and the synthesis of C20 and C22 PUFAs is not interrupted (42). This reflects the partial overlapping between the activities of these two enzymes. In comparison, in testes of Elovl2 / mice, the deficiency of C22, and especially that of C24 PUFAs and longer polyenes like 30:5n-6, is not compensated (10), demonstrating the importance of Elovl2 in providing the PUFAs that serve as substrates to Elovl4.
The high mRNA levels of Elovl2 in spermatogenic cells, and the fact that PSs and early spermatids are the cells most susceptible to temperature in the adult testis (43) explains our present finding that Elovl2 mRNA levels were the most reduced in the adult rat testis 2 weeks after the massive germ cell depletion induced by exposures to hyperthermia. In contrast, Elovl5 mRNA levels had increased more than twice in the same 2 week period, indicating upregulation of this gene in the reemerging germ cells. These are the newly formed PtSs requiring the synthesis of the C20 PUFAs normally abundant in their membrane glycerophopholipids. A similar rising trend seems to have affected the Elovl4 expression, which, by the same token, could respond to PtSs requiring the n-Vs that normally abound in their Cer and SM (3). The finding that, after their significant drops at week 2, Elovl2 (and also Fa2h) mRNA levels tended to increase at week 6 posthyperthermia, coincides with the timing of reappearance in the testis of spermatids, concomitantly with their 22:5-rich glycerophospholipids and h-V-rich SLs.

Elovl4
As previously shown in the adult mouse (14), our work shows that Elovl4 mRNA is actively produced in the adult rat testis. It goes on to demonstrate that such mRNA is differentially expressed among spermatogenic cells. In addition, of the three elongases compared here, Elovl4 was the one whose mRNA levels were higher in SCs than in germ cells, an intriguing finding considering that the former somatic cells failed to express Elovl4 protein and to produce C26-C32 PUFA. Conversely, in the n-V-rich PtSs (3), where the protein was highly concentrated and most active as an elongase, the mRNA levels were the lowest. Although the reasons for these features remain elusive at present, the high mRNA levels in SCs provided a reasonable explanation to our present in vivo observations in two situations in which SCs outnumber germ cells: the infantile gonads (not yet producing the latter) and the adult testes undergoing the consequences of hyperthermia (having lost most of them). In each of these cases, Elovl4 mRNA transcripts were present because they were mainly contributed by SCs.
The transcriptionally expressed, but translationally silent, Elovl4 mRNA in SCs suggests that it may be subject to negative regulation or silencing in these cells. If this were the case, however, the next question would be why SCs should need to produce Elovl4 mRNA. Extracellular microvesicles are part of the cell secretome, and carry proteins and functional RNA species (including mRNA, noncoding RNA, and miRNA) from one cell to another, where mRNAs may be translated into appropriate proteins (44). This horizontal transfer of RNA could potentially occur from SCs to spermatogenic cells, especially to PtSs. Small membrane-bound vesicles of endosomal origin, known as exosomes, are released by many cells, mediating intercellular transport of proteins, lipids, and RNAs (45,46).
Although not yet characterized in rat SCs, the presence of exosomes has been described in turtle testes (47).
Also intriguingly, we also detected significant Elovl4 mRNA at embryonic ages in the primordial testes of the rat (unpublished observations), i.e., far from the moment in life in which membrane lipids that contain n-Vs are expected to arise. Small amounts of translated Elovl4 enzyme should suffice to give rise to n-Vs that could be modified by other enzymes to synthesize products involved in regulatory functions. In this regard, a recent discovery in the mouse retina showed that two n-Vs from the retina, 32:6n-3 and 34:6n-3, in free form, can be converted by oxygenases into a novel series of compounds termed "elovanoids" (48). These molecules, formed in cells of the retinal pigment epithelium and released to the medium in nanomolar concentrations, were shown to enhance the expression of pro-survival genes in photoreceptors (48).
As a testicular protein, Elovl4 was not detected in extratubular locations and it showed a differential spatiotemporal pattern of expression among germ cells. Meiotic spermatocytes intensely expressed Elovl4 at specific spermatogenic cell stages and in a cell location that was compatible with its concentration in Golgi structures and in the ER. The protein staining decreased with differentiation between PtSs and RSs, as also did the ability to produce C26-C32 PUFAs from PtSs to RSs, suggesting that Elovl4 was gradually less necessary as differentiation proceeded. This is reasonable, as an important part of the lipid materials synthesized in the large-sized PtSs, including the polar lipids of plasma membrane and intracellular organelles, are "inherited" by its two postmeiotic daughter RSs.
In spermatids at more advanced stages of differentiation, namely those subject to the steps of spermiogenesis, the Elovl4 protein appeared concentrated in a small rounded structure within the cytoplasmic lobe. This formation may correspond to the "radial body" previously described in rat elongated spermatids, an aggregate of collapsed and radially arranged remnants of ER cisternae (49). The high Elovl4 protein concentration detected by the antibody, together with the weakest elongation activity observed in LSs, suggest that the protein, active in the ER at earlier stages, eventually loses part of its functionality at the last stages of spermiogenesis. Its high concentration in the mentioned small round structures probably reflects the final destination of former ER membranes, together with other disposable organelles and extranuclear components, to the particles known as "residual bodies". These small membrane-bound structures, destined to be phagocytized by SCs, are known to concentrate, in a highly compact form, organelles and other intracellular materials no longer needed in the nascent gametes. This was apparently the fate of Elovl4 and Fa2h, as both proteins were absent from testicular spermatozoa.
CerS3 is the member of the Cer synthase gene family that is specifically expressed in spermatogenic cells (6), where it is responsible for the synthesis of Cer species with VLCPUFAs (7,50), thereby leading to SM and other VLCPUFA-containing SLs. The pattern of expression of Elovl4 observed here among cells of rat STs coincided with that of the CerS3 protein described in mice by Rabionet et al. (7) in that both protein levels were highest in PtSs, declining in RSs, and concentrated again in small round structures present in LSs. Taken together, these results suggest that the expression of Elovl4 and CerS3, at the mRNA and protein level, tend to precede that of Fa2h during spermatogenesis, and that these enzymes concertedly collaborate in the biosynthesis of the SLs with VLCPUFAs that are destined to spermatozoa.

Fa2h
Unlike Elovl4 mRNA, the Fa2h mRNA expression was undetectable in SCs. Consistently, it was apparently absent from immature rat testis and it vanished from the adult rat testis when the loss of germ cells with permanence of SCs was experimentally forced as a consequence of repeated exposures to hyperthermia. Not surprisingly, the Fa2h expression, at the mRNA and protein levels, was highest in spermatids, where its products, h-Vs, tend to concentrate in Cer and SM species (3).
The most interesting finding about Fa2h protein was its spermatogenic stage-specific localization. Meager in PtSs, its staining appeared abundant and disseminated in the cytoplasm of RSs, as expected from the ER localization of Fa2h as an integral membrane protein (23,51) and from the accretion of its h-V products in these cells (3). At late steps of spermiogenesis, the protein had evidently undergone a change in its position within cells, indicative of intracellular membrane transport. In the spermatids situated most distally from the limiting membrane of the tubules, the fluorescence associated to Fa2h was found to surround the nuclear envelope partially, intensely focused over the posterior aspect of the head of the elongating spermatids, in coincidence with the formation of the temporary structure known as manchette. The changes in shape that the heads of rat spermatids undergo during spermiogenesis, from roughly spherical to sickle-shaped, critically depend on this complex microtubule-based structure (52). Actin filaments, microtubules, and GM1-rich raft microdomains participate in allowing the correct assembly and anchoring of the manchette complex (53). The function of the enzymes, Elovl4 and Fa2h, is evidently to work in partnership to produce germ cell membrane SLs with VLCPUFAs, which, unlike those with saturated fatty acids, occur in membrane regions distinctly apart from "raft-like" domains (22). The present observations suggest that the relationship between Fa2h and manchette functions may be simple coexistence and cooperation, the former facilitating, the latter operating the changes in head shape that take place in LSs. The n-Vs, and especially the rodent-specific h-Vcontaining SLs, may provide the necessary flexibility to the membrane of spermatids as their heads increase their curvature, enabling the formation of the hook-shaped head that is typical of mouse and rat spermatozoa.