A role for very-long-chain fatty acids in furrow ingression during cytokinesis in Drosophila spermatocytes

Summary

Cell shape and membrane remodeling rely on regulated interactions between the lipid bilayer and cytoskeletal arrays at the cell cortex. During cytokinesis, animal cells build an actomyosin ring anchored to the plasma membrane at the equatorial cortex. Ring constriction coupled to plasma membrane ingression separates the two daughter cells [1]. Plasma-membrane lipids influence membrane biophysical properties such as membrane curvature and elasticity and play an active role in cell function [2], and specialized membrane domains are emerging as important factors in regulating assembly and rearrangement of the cytoskeleton [3]. Here we show that mutations in the gene bond [4], which encodes a Drosophila member of the family of Elovl proteins that mediate elongation of very-long-chain fatty acids [5], block or dramatically slow cleavage-furrow ingression during early telophase in dividing spermatocytes. In bond-mutant cells at late stages of division, the contractile ring frequently detaches from the cortex and constricts or collapses to one side of the cell, and the cleavage furrow regresses. Our findings implicate very-long-chain fatty acids or their derivative complex lipids in allowing supple membrane deformation and the stable connection of cortical contractile components to the plasma membrane during cell division.

Results and Discussion

bond Mutants Implicate Very-long-chain Fatty-acids in Cytokinesis

Loss of function of the Drosophila bond gene caused failure of cytokinesis in dividing spermatocytes. Several alleles of bond were identified in a screen for EMS-induced male-sterile mutations that cause defects in cytokinesis during male meiosis [4]. Round spermatids from bond mutant males commonly display two to four nuclei associated with an abnormally large mitochondrial derivative ([4], Figure 1A,B), indicating cytokinesis defects during the meiotic divisions [4]. We mapped the bond locus to a small polytene chromosome region (Experimental Procedures) containing five genes (FlyBase) (Figure 1D). Sequencing of five bond alleles revealed mutations in the annotated gene CG6921 (Figure 1E and F). A 7.5-kb wild-type genomic fragment including CG6921 but not neighboring genes (Figure 1D) rescued the cytokinesis defects of bond mutants (Figure 1C), confirming CG6921 as bond.

Figure 1. Loss of function of bond, a Conserved Elovl Gene, Causes Cytokinesis Defects in Drosophila Spermatocytes.

(A-C) Phase-contrast images of Drosophila round spermatids showing nuclei (arrowheads) and mitochondria derivatives (arrows). (A) Normal mononucleate spermatids of bond^Z3437/TM6 flies. (B) Homozygous mutant bond^Z3437 spermatids with four nuclei and a single enlarged mitochondrial derivative. (C) Mononucleate spermatids of bond^Z3437/bond^Z3437 rescued with a wild-type bond transgene. Scale bar, 20 μm. (D) Map of the bond (CG6921) genomic region with the 7.5-kb rescue fragment. (E) ClustalW -alignment of Bond to yeast Elo2p and Elo3p and human ELOVL4 with the five most strongly predicted transmembrane domains and C-terminal putative ER-retention sequence underlined. (Box) conserved HXXHH motif; (asterisks) sites of mutations in bond alleles. (F) Molecular defects in bond alleles. Amino-acid positions are relative to the predicted start of translation. Allele strengths are based on the frequencies of multinucleate round spermatids in homozygotes. (G) Rescue of a yeast elo2Δelo3Δ strain by bond or the Arabidopsis At2g16280 ORF, but not by the empty expression vector (see Experimental Procedures).

bond encodes a member of the Elovl family of enzymes involved in elongation of very-long-chain fatty acids, which are conserved from yeast to mammals (Figure 1E). Pair-wise Blast analysis (NCBI bl2seq) revealed that the predicted Bond protein has 28% and 29% amino-acid identity, respectively, to the Saccharomyces cerevisiae Elovl proteins Elo2p and Elo3p, and 38% and 47% amino-acid identity, respectively, to human Elovl4 and Elovl7. Bond shares sequence motifs with other Elovl proteins, including five to seven predicted transmembrane domains [5, 6](Kyte-Doolittle analysis), the essential HXXHH motif conserved in the Elovl family [5, 6], and a C-terminal sequence similar to the KXKXX motif of some transmembrane proteins resident in the endoplasmic reticulum. Elovl proteins are enzymes that participate in the elongation of fatty acids with acyl chains longer than 18 carbon atoms [5]. These very-long-chain fatty acids are most commonly found in sphingolipids, and are essential for sphingolipid formation [5] and function [7].

To determine if Bond is a functional Elovl protein, we tested whether the Drosophila gene can substitute for Elovl genes. In S. cerevisiae, simultaneous loss of ELO2 and ELO3 causes lethality [8]. Overexpression of mammalian or Arabidopsis very-long-chain-fatty-acid elongases had previously been shown to restore viability of an elo2Δelo3Δ double mutant and to generate very-long-chain fatty acids in yeast cells [9, 10]. Overexpression of Bond in yeast also restored the viability of an elo2Δelo3Δ double mutant (Figure 1G), indicating that Bond indeed has elongase activity and can function as an Elovl enzyme in yeast.

Expression of bond Is Cell-Type Specific

Bond is one of 20 predicted Elovl proteins in the Drosophila genome, based on tBlastn analysis (Figure 2A). Three of the corresponding genes were previously described as having restricted expression patterns [11-13], suggesting tissue-specific expression for different elovl genes. Analysis of bond expression by in situ hybridization using two different probes that distinguish bond from other elovl homologs revealed bond transcripts in spermatocytes, but not in spermatogonia (Figure 2B). This expression pattern is consistent with the cytokinesis defect observed in bond mutants: multinucleate bond spermatids contained a maximum of four nuclei, indicating failure of cytokinesis in the two meiotic divisions but not in the preceding mitotic divisions. Two other elovl homologs, CG17821 and CG31141, are expressed in the adult testis (FlyAtlas,[14]). We confirmed testis expression of CG31141 and CG17821 by RT-PCR, and identified CG17821 expression in spermatocytes by in-situ hybridization; this gene, however, failed to show Elovl activity in yeast (not shown).

Figure 2. bond Is Expressed in Spermatocytes.

(A) Phylogenetic tree [31] of the Elovl proteins from S. cerevisiae, human, and D. melanogaster. (B-D) In situ hybridization. (B) Expression of bond mRNA in spermatocytes (arrow) but not in spermatogonia at the apical region of the testis (arrowhead). (C) Embryonic expression in oenocytes (arrowhead) and in a segment of the intestine (arrow). (D) Expression in the ovary in stage 9-10 egg chambers. Scale bar, 100 μm.

bond transcripts were also detected in specific cell types in embryos and ovaries. In embryos, bond transcripts localized to groups of cells that participate in the formation of steroid-derived molecules, such as oenocytes (Figure 2C) and corpus alatum (not shown; Flybase/BDGP http://www.fruitfly.org/cgi-bin/ex/insitu.pl). In ovaries, bond transcripts were detected in stage 9-10 egg chambers (Figure 2D). Mutants harboring loss-of-function bond alleles (Figure 1F) were viable and developed to adulthood with no overt morphological defects, but were male- and female-sterile. It is possible that other elovl family members can provide enzymatic function in tissues other than spermatocytes.

Bond Is Required for Successful Cleavage-furrow Ingression

Analysis of living spermatocytes revealed that loss of Bond function dramatically affects cleavage-furrow ingression. In wild-type spermatocytes, a GFP-tagged myosin-regulatory light chain (Sqh-GFP, [15]) assembled into a ring at the equator of dividing cells (n=10) in early anaphase (Figure 3A, 0 min; movie 1), then immediately initiated constriction accompanied by membrane deformation into a furrow (Figure 3A, 5-25 min; movie 1). The myosin regulatory light chain, a subunit of myosin, can localize to the cleavage furrow independently of F-actin [16].

Figure 3. Cleavage Furrows Fail to Ingress and Contractile Rings Eventually Detach from Plasma Membranes in bond Spermatocytes.

(A-C) Frames from time-lapse sequences. Phase-contrast (upper panels) and Sqh-GFP (lower panels) images of spermatocytes undergoing cytokinesis in bond^Z3437/TM6 (A), bond^Z3437/bond^Z5274 (B), and bond^Z3437/bond^Z3437 (C) flies. Time 0 is the time of earliest Sqh-GFP detection at the cell equator. Arrows, detached Sqh-GFP rings; arrowheads, plasma membranes (outlined in white). Scale bars, 10 μm. (D) Dynamics of plasma-membrane ingression and contractile-ring contraction during cytokinesis in wild-type and bond spermatocytes, with diameters of Sqh-GFP rings and cleavage furrows (measured from the phase-contrast images) plotted over time. (Solid circles) Sqh-GFP ring and furrow diameters in a typical wild-type spermatocyte; ring and furrow diameters were identical throughout cytokinesis in these cells. (Open circles) Sqh-GFP rings; (dots) furrows in bond cells. (Left) cell in B; (right) cell in C.

We analyzed cell division in 15 spermatocytes from flies trans-heterozygous for the two strong alleles, bond^Z5274 and bond^Z3437, and 33 spermatocytes from bond^Z3437 homozygotes. The homozygotes and trans-heterozygotes displayed similar phenotypes, indicating that the defects observed were not due to secondary mutations. Seemingly normal Sqh-GFP rings assembled during early anaphase in the bond mutants. However, constriction of the Sqh-GFP rings was minimal or substantially slowed in 10 of the 15 bond^Z5274/bond^Z3437 and in 21 of the 33 bond^Z3437 homozygous spermatocytes (Figures 3B, C; movies 2 and 3), and these cells ultimately failed to complete cytokinesis during the period of observation. Measurements of Sqh-GFP ring diameter over time indicated that the rate of ring constriction in the bond cells that went on to fail cytokinesis was slower than in wild type (Figure 3D and Figure S1). Sqh-GFP rings remained poorly constricted in these cells for 15-27 min after the rings first appeared (Figures 3B and C; Figure S1). By this time, wild-type rings had constricted almost to the size of the stable ring canals that remain to connect sister germ cells following cytokinesis (Figure 3A, 20 min; Figure S1).

Bond function also appeared to be required for long-term stable attachment of the contractile machinery to the cell cortex. Late in cytokinesis, after furrow ingression had stalled for some time, the plasma membrane detached from the contractile ring in bond mutants, and the furrow regressed rapidly (Figures 3B and C, arrowheads; Figure S1). At exactly the same time, the poorly constricted Sqh-GFP rings constricted or collapsed rapidly and directionally from the site of membrane detachment towards the opposite side of the cell, where the ring remained attached to the cell cortex (Figure 3B and C, arrows; Figure 3D; Figure S1). The rapid constriction of the Sqh-GFP ring after detachment from the membrane indicates that at least in some cells, the actomyosin ring retained the capacity to constrict in bond mutants. The slow rate of furrow ingression observed in bond spermatocytes could be due to defects in the ability of the membrane to deform and bend inwards under the contractile force of the actomyosin ring. Very-long-chain fatty acids present in phosphoinositides have been implicated in the stability of high membrane curvature [17], and sphingolipid- and cholesterol-rich domains at the plasma membrane appear to modulate membrane deformability [18]. Thus, it is possible that very-long-chain fatty acids that depend on Bond can facilitate membrane curvature during cleavage furrow ingression.

Previous analyses failed to detect F-actin rings in most bond mutant spermatocytes, which also displayed poorly constricted anillin rings [4]. However, examination of fixed mutant spermatocytes with an improved F-actin staining technique revealed the presence of F-actin rings in 39 of 69 bond spermatocytes in telophase. These actin rings were substantially thinner than the constricted F-actin rings of wild-type cells and were extended or only partially constricted (Figures 4A and B), consistent with the defect in cleavage-furrow ingression observed in live bond mutant cells.

Figure 4. Actomyosin Rings and Central Spindles Are Abnormal in bond Mutants.

bond^Z3437/TM6 (A) and bond^Z6275/bond^Z3734 (B) telophase spermatocytes stained for actin (red), tubulin (green), and DNA (blue). Arrowheads, actin rings. (C-D) Living bond^Z6275/TM6 (C) and bond^Z6275/bond^Z6275 (D) spermatocytes expressing EGFP-tagged β-tubulin. Arrows, bundled astral microtubules that contact the equatorial membrane. Scale bars, 10 μm.

Mutations in bond Affect Organization of the Central Spindle

Previous studies showed that telophase bond spermatocytes exhibit severely defective central spindles [4]. However, it was not clear when and how the defects in microtubule organization arose in bond spermatocytes. We analyzed the central spindles in living spermatocytes from wild type and bond^Z6275 mutants expressing EGFP-tagged β-tubulin. In agreement with previous observations [19], we found that in wild type, a population of microtubules extended from the poles to contact the cell cortex at the future site of cleavage-furrow ingression in early anaphase (Figure 4C, 0-15 min, arrows; movie 4; n=8). As these microtubules began to bundle, the cell elongated (Figure 4C, 10-20 min) and the microtubule array of the central spindle became visible in the cell midzone (Figure 4C, 15-25 min). With ingression of the furrow, the pole-to-equatorial-cortex microtubule-bundles compacted together with the internal microtubule array to form a tight telophase central spindle (Figure 4C 25-35 min).

In bond mutant spermatocytes (n=8), microtubules from the pole also contacted the equatorial cortex and bundled there in early anaphase (Figure 4D, 5-10 min, arrows; movie 5). However, the internal microtubules failed to form an organized central-spindle array (Figure 4D, 10-20 min). The cleavage furrow ingressed only minimally and stalled (Figure 4D, 20-25 min). Eventually, the microtubule bundles that contacted the equatorial cortex became disorganized and the cleavage furrow regressed, as seen also with Sqh-GFP (see above). Examination of living cells expressing Pavarotti-YFP, a marker of microtubule-plus ends, indicated that internal microtubule arrays initially formed in early anaphase of bond mutants, but were destabilized and lost during the failure of furrow ingression (see Supplemental Material).

Observation of living cells indicated that mutant spermatocytes did not elongate properly during anaphase and telophase. Whereas wild-type cells increased in pole-to-pole length by∼51% during this period, bond mutant cells elongated by only∼24% (Figure S3). This difference became apparent in early anaphase B, when wild-type cells initiated furrow ingression accompanied by rapid elongation. The maximum elongation of bond mutant cells was comparable to that achieved by wild-type cells at the initiation of furrow ingression.

The behavior of β-tubulin and Pavarotti in bond spermatocytes indicates that Bond function is not required for the formation and behavior of the pole-to-equatorial-cortex microtubules, which are thought to determine the site of contractile-ring assembly and cleavage-furrow ingression (reviewed in [20]). However, Bond function is required for formation of a robust central spindle, suggesting that its activity is necessary for the stabilization of midzone microtubules during anaphase and telophase. There are several non-mutually-exclusive explanations for the abnormal central-spindle assembly in bond mutants. It is possible that changes in membrane fatty-acid composition due to loss of Bond function affect elements of the contractile apparatus that normally stabilize midzone microtubules to ensure the formation of a robust and well organized central spindle [21, 22]. It is also possible that side-to-side compression of midzone microtubule bundles by ingression of the cleavage furrow in wild-type cells may facilitate the formation of cross-bridges that stabilize the dense array of the telophase central spindle. In bond mutants, the slow cleavage furrow ingression may delay formation of those cross-bridges, allowing disassembly of microtubules from their minus ends. Finally, proper organization of the central spindle may depend on the capacity of the cell to elongate along the spindle axis, which appears compromised in bond cells (Figures 4D and S3).

Role of Bond in Spermatocyte Cytokinesis

The phenotype of bond spermatocytes differed from those of spermatocytes mutant in genes thought to affect membrane-trafficking events, such as loss-of-function of Arf6 [23], Rab11 [24], or the phosphatidylinositol-transfer protein Vibrator/Giotto [25, 26]. In arf6 and rab11 mutant spermatocytes, initial rates of cleavage-furrow ingression were comparable to wild type for more than 15 min, and telophase central spindles formed transiently. When cytokinesis failed in arf6 mutants, the furrow regressed slowly, and the ring of cortical Sqh-GFP expanded with it, returning to a large diameter [23]. In contrast, the rate of cleavage-furrow ingression was slow almost from the outset in bond spermatocytes, telophase central spindles failed to organize, and in most cells the Sqh-GFP ring eventually detached from the cortex and rapidly collapsed, rather than accompanying the regressing cleavage furrow. The striking difference between the bond and arf6 phenotypes indicates that bond may contribute to cytokinesis thorough a role different from the support of membrane expansion mediated by membrane trafficking [27].

Our observations indicate that very-long-chain fatty acids dependent on Bond function are required for successful cleavage-furrow ingression during cell division in spermatocytes. Specialized very-long-chain fatty acids dependent on Bond may be required to allow the plasma membrane to deform for supple invagination during cleavage-furrow ingression. The biophysical properties of very long acyl chains, proposed to stabilize highly curved membranes [17], are consistent with a role in membrane deformation during furrow ingression. Membrane composition that depends on bond may also be required to mediate proper membrane-contractile ring interactions. Growing evidence from studies of T-cell activation suggests that specialized membrane domains rich in sphingolipids and cholesterol regulate cytoskeletal-membrane interactions [3]. Sphingolipids have also been implicated in the activation of the Rho-Dia pathway that controls the tubulin and actin cytoskeletons [28]. Thus, it is possible that bond-dependent lipids may participate in specialized membrane domains that promote actin-ring formation and stability at the cortex. Although we were unable to detect concentration of sphingolipids or cholesterol at the equator of dividing Drosophila spermatocytes or cultured S2 cells (data not shown), similar lipids have been found to localize at the cytokinesis furrows of fission yeast and sea urchin eggs [29, 30]. Consistent with these observations, our findings indicate that very-long-chain fatty acids that depend on bond, and their derived lipids, may coordinate membrane deformation with the contractile activity of the actomyosin ring, which leads to successful furrow ingression during cytokinesis in Drosophila spermatocytes.