Characterization of HAF-4- and HAF-9-localizing organelles as distinct organelles in Caenorhabditis elegans intestinal cells

The intestinal cells of Caenorhabditis elegans are filled with heterogeneous granular organelles that are associated with specific organ functions. The best studied of these organelles are lipid droplets and acidified gut granules associated with GLO-1, a homolog of the small GTPase Rab38. In this study, we characterized a subset of the intestinal granules in which HAF-4 and HAF-9 localize on the membrane. HAF-4 and HAF-9 are ATP-binding cassette (ABC) transporter proteins that are homologous to the mammalian lysosomal peptide transporter TAPL (transporter associated with antigen processing-like, ABCB9). Using transgenic worms expressing fluorescent protein-tagged marker proteins, we demonstrated that the HAF-4- and HAF-9-localizing organelles are not lipid droplets and do not participate in yolk protein transport. They were also ruled out as GLO-1-positive acidified gut granules. Furthermore, we clarified that the late endosomal protein RAB-7 localizes to the HAF-4- and HAF-9-localizing organelles and is required for their biogenesis. Our results indicate that the HAF-4- and HAF-9-localizing organelles are distinct intestinal organelles associated with the endocytic pathway.


Background
The intestine of Caenorhabditis elegans is a multifunctional organ involved in the uptake, metabolism, storage, and distribution of nutrients, as well as in detoxification and immune defense [1]. It consists of a tube comprising only 20 epithelial cells that are not replaced after embryonic development; thus, each intestinal cell is considered to have multiple functions throughout the life of the animal. The intestinal cells are filled with heterogeneous granular organelles that are associated with specific organ functions [2]. Therefore, to understand how these individual cells achieve multiple intestinal functions, these granular organelles should be classified and their physiological roles determined. However, the relationships between different types of granules, their physiological roles, and the regulation of their biogenesis are largely unknown.
The best studied of the intestinal granular organelles are the acidified organelles known as "gut granules", which are easily recognized by the presence of autofluorescent materials [3,4]. Gut granules are regarded as lysosome-related organelles, because GLO-1 localizes to embryonic gut granules, and is required for the biogenesis of the organelles. GLO-1 is a member of the RAB family of small G proteins and a homolog of mammalian Rab38; it is required for the biogenesis of melanosomes (lysosome-related organelles in melanocytes). Several other genetic factors are required for gut granule biogenesis [5][6][7][8][9]. Although some of the physiological roles of gut granules have been identified, such as the storage of zinc [10] and cholesterol [11], and signaling for aging [12], their overall functions remain poorly understood.
Other well-studied intestinal granules are the fat storage organelles known as lipid droplets [13]. The intestine is the major organ responsible for fat storage. Similar to mammalian lipid droplets, those of C. elegans are characterized by a phospholipid monolayer membrane [14]. The short-chain dehydrogenase DHS-3 and the triglyceride lipase ATGL-1 were recently identified as marker proteins that localize on the membrane of lipid droplets [15,16].
Previously, we reported that HAF-4 and HAF-9, ATPbinding cassette (ABC) transporter proteins that are homologous to the mammalian lysosomal peptide transporter TAPL (transporter associated with antigen processing like, ABCB9), localize to the membrane of a subset of intestinal granular organelles approximately 2 μm in diameter from the late larval to adult stages [17]. HAF-4 and HAF-9 are involved in maintaining the normal formation of their localizing organelles, probably by forming a heterodimer as a functional unit of a transporter [17,18]. The colocalization of HAF-4, HAF-9, and LMP-1, a homolog of mammalian lysosome-associated membrane proteins (LAMPs), suggests their relevance to lysosomes. This interpretation is further supported by the observation that HAF-4 and HAF-9 localize to the enlarged vacuole formed in the mutant of ppk-3, a gene regulating the terminal maturation of lysosomes. Furthermore, HAF-4-and HAF-9-localizing organelles are neither acidified nor autofluorescent granules, and the mutants of haf-4 and haf-9 show the same or a greater number of acidified and autofluorescent granules as the wild-type, suggesting that the HAF-4-and HAF-9-localizing organelles are distinct from gut granules. We also reported that the HAF-4-and HAF-9localizing organelles did not show positive vital staining for the lysochrome dye Nile Red; however, a subsequent study demonstrated that fixed Nile Red staining, but not vital staining, showed specificity to the lipid droplets in the cells [13,14]. Therefore, the possibility that HAF-4and HAF-9-localizing organelles are lipid droplets remained an open question.
Here, we report the classification of the HAF-4-and HAF-9-localizing organelles among intestinal granules using transgenic worms expressing fluorescent proteintagged marker proteins. We confirmed that these organelles are not lipid droplets and do not participate in yolk protein transport. They are also not GLO-1-positive gut granules. Further investigation using the late endosomal protein RAB-7 revealed that the HAF-4-and HAF-9localizing organelles are distinct intestinal organelles associated with the endocytic pathway.
We next determined whether HAF-4 and HAF-9 are involved in the biogenesis of lipid droplets using deletion mutants of haf-4 and haf-9, which exhibit defects in the biogenesis of the HAF-4-and HAF-9localizing organelles [17]. These mutants did not show a decrease in the number of lipid droplets, as shown by Nile Red staining (Fig. 1c) and DHS-3::GFP localization (Fig. 1d), indicating that the organelles are not required for the biogenesis of lipid droplets. This conclusion was also supported by the normal formation of DHS-3::GFP-positive lipid droplets in the lmp-1 mutant, in which HAF-4-and HAF-9localizing organelles show morphological aberrations [17] (Additional file 1: Figure S3).
HAF-4-and HAF-9-localizing organelles are irrelevant to yolk protein transport A previous report indicated that deletion mutants of haf-4 and haf-9 show a reduced brood size [17], which suggests the involvement of the HAF-4-and HAF-9-localizing organelles in the nutrition supply to oocytes. However, these organelles are found in both hermaphrodites and males ( Fig. 2a and Additional file 1: Figure S13a). In C. elegans, VIT-2 (a yp170B homolog) is secreted from the intestinal cells into the pseudocoelomic space, and is taken up by developing oocytes through gonadal sheath cells as a nutritional source [19]. VIT-2 is expressed only in the intestinal cells of the adult hermaphrodite. Although the mechanism by which the oocytes uptake yolk protein via receptor-mediated endocytosis is well characterized [20], precisely how the yolk protein is pooled and trafficked in the intestinal cells has not been elucidated in detail. The yolk protein in the intestinal cells was visualized by VIT-2::GFP, which was distinguishable from the autofluorescent granules and lipid droplets (Additional file 1: Figure S4). However, VIT-2::GFP did not localize to either the HAF-9::mCherrypositive ( Fig. 2b and Additional file 1: Figure S13b) or LMP-1::mRFP-positive granules (Additional file 1: Figure S5). Furthermore, the uptake of VIT-2::GFP in embryos was unaffected in the haf-4 haf-9 double mutant (Fig. 2c). These results suggest that the HAF-4-and HAF-9-localizing organelles are irrelevant to yolk protein transport.

HAF-4-and HAF-9-localizing organelles are not acidified gut granules
Although the HAF-4-and HAF-9-localizing organelles are not acidified, LMP-1, a homolog of mammalian LAMPs, also localizes to these organelles, suggesting a potential relationship with the lysosomes [17]. HAF-4 and HAF-9 do not localize to organelles with autofluorescent materials that are characteristic of gut granules, and haf-4 and haf-9 were found to be unnecessary for the formation of gut granules. To investigate the relationship between these organelles and gut granules in more detail, we compared the localization of HAF-4 and HAF-9 with that of GLO-1, a small GTPase required for the biogenesis of gut granules. GLO-1::GFP localized to the surface of autofluorescent granules but not to the Nile Red-positive lipid droplets (Additional file 1: Figure  S6). In adult worms expressing both GLO-1::GFP and HAF-9::mCherry, the mCherry-edged intestinal granules (arrows in Fig. 3a) and the mCherry-internalized intestinal granules (arrowheads in Fig. 3a) were observed, the latter of which are thought to be lysosomes or lysosomerelated organelles accumulating mCherry in their lumen. Although GLO-1::GFP localized to the surface of the mCherry-internalized granules, it did not localize to the mCherry-edged intestinal granules where HAF-9::mCherry was localized on the membrane ( Fig. 3a and Additional file 1: Figure S13c). Similarly, GLO-1::GFP did not localize to the LMP-1::mRFP-edged intestinal granules (Additional file 1: Figure S7). We also investigated the localization of HAF-4 and HAF-9 in the glo-1 mutant. The localization of HAF-4::GFP and HAF-9::mCherry to the surface of intestinal granules was observed even in the glo-1(zu391) mutants ( Fig. 3b and Additional file 1: Figure S13d Figure S8). These results indicate that the HAF-4-and HAF-9-localizing organelles are not acidified gut granules.
HAF-4-and HAF-9-localizing organelles are associated with the endocytic pathway Although the HAF-4-and HAF-9-localizing organelles are prominent intestinal granules in late larval and young adult worms, they are distinct from acidified gut granules and lipid droplets. Furthermore, HAF-4::GFP and HAF-9::mCherry were not localized to either the peroxisome marker-positive organelles or the mitochondrion markerpositive organelles (Additional file 1: Figure S9). Therefore, what are these organelles?

Discussion
We investigated the relationships among the intestinal granular organelles of C. elegans using two approaches. The first was through comparison of the localization of fluorescent protein-tagged proteins and fluorescently visualizable materials inside the organelles, facilitating the identification of organelles to clarify the interrelationships in their formation processes. The other approach involved investigating how a defect in a subset of intestinal organelles affected other subsets, which could help to determine whether or not these organelles require each other for their biogenesis and function. These investigations clearly showed that the HAF-4-and HAF-9-localizing organelles are not lipid droplets, based on the lack of Nile Red stainability and DHS-3::GFP localization, and that they are not required for the biogenesis and lipid storage of lipid droplets. It was also confirmed that the organelles are not yolk granules and are not necessary for the transport of yolk protein to developing oocytes. The involvement of these organelles in the storage or distribution of nutrients has been suggested, because haf-4 and haf-9 mutants, which are defective in organelle biogenesis, produced phenotypes with slow growth and reduced brood size. However, the organelles seem to be irrelevant for the storage and distribution of lipids and yolk protein. The storage of polypeptides and/or carbohydrates is a possible candidate for their function. In particular, the storage of peptides seems to be a plausible function, because HAF-4 and HAF-9 are believed to transport peptides from the cytosol to the lumen of the organelles, based on their homology with the mammalian peptide transporter TAPL [17].
Furthermore, HAF-4-and HAF-9-positive intestinal granules did not overlap with GLO-1-positive granules, and the HAF-4-and HAF-9-localizing organelles and acidified gut granules did not require each other for their biogenesis, suggesting that they are distinct organelles that are formed by different pathways. Experiments using GFP::RAB-7 and the rab-7 mutants demonstrated that RAB-7 associates with HAF-4-and HAF-9-localizing organelles and participates in their biogenesis. The small GTPase RAB-7 is a key regulator of the endocytic pathway, and replacement of RAB-5 with RAB-7 occurs when early endosomes mature into late endosomes [22]. Therefore, our results strongly suggest that the HAF-4-and HAF-9-localizing organelles are formed through the maturation or fusion of RAB-7-positive late endosomes. Transport substrates of HAF-4 and HAF-9 may possibly be necessary for the maturation or fusion process. The multivesicular structure observed in the lmp-1 mutant under a transmission electron microscope also suggested a defect in the endocytic pathway [17]. More interestingly, the biogenesis of acidified gut granules requires the Rab38 homologue GLO-1 but not RAB-7 [9], providing further evidence that HAF-4-and HAF-9-localizing organelles and acidified gut granules are irrelevant. Mammalian Rab38 associates with melanosomes and platelet dense granules and is necessary for their biogenesis [23,24]. In contrast to the acidified gut granules, which are lysosomerelated organelles, HAF-4-and HAF-9-localizing organelles are thought to be more related to the canonical lysosomes in terms of their biogenesis, even though they are not acidified and are much larger than the average lysosome (approximately 2 μm versus 0.05-0.5 μm in diameter). The possibility that they are residual bodies is also unlikely because autofluorescent materials known as age pigments, which are characteristic of residual bodies [25], are attributed to acidified gut granules but not to HAF-4and HAF-9-localizing organelles.

Conclusions
Although the physiological function of the HAF-4-and HAF-9-localizing organelles remains elusive despite their abundance in intestinal cells, our data demonstrate that they are distinct intestinal organelles associated with the endocytic pathway. The storage of polypeptides and/or carbohydrates is a possible candidate for their function. Uncovering the specific function of the organelles and their relationship with other intracellular compartments would provide insight into the organelle diversity to support the multifunctionality of C. elegans intestinal cells.
To generate the GFP::rab-7 construct in which enhanced green fluorescent protein (EGFP) is fused inframe at the N-terminus of RAB-7, three DNA fragments were polymerase chain reaction (PCR)-amplified and fused as follows. A 1.6-kb DNA fragment (corresponding to the promoter region of rab-7) and a 1.9-kb DNA fragment (corresponding to the coding sequence and 3′ untranslated region of rab-7) were PCR-amplified with the primer sets rab-7_A#F3/rab-7_A#R3 and rab-7_A#F1/ rab-7_A#R1, respectively. A 0.7-kb sequence of the EGFP coding region was PCR-amplified with the primer set rab-7_A#F2/ rab-7_A#R2. These DNA fragments were fused by PCR sewing with the primer set rab-7_A#F3/ rab-7_A#R1, and subsequently cloned between the NotI and BamHI sites of the pBlueScript vector.
The primer sequences used for establishment of the constructs were as follows: Microinjection of DNA into the C. elegans germ line was performed as described by Mello et al. [27] using pRF4 [rol-6(su1006)] as a selection marker. Integrant formation was performed according to Mitani's method [28]. Our study does not require any animal ethics approval.

Nile Red staining
The worms were washed with S-basal (50 mM potassium phosphate buffer [pH 6.0] containing 0.1 M NaCl) and fixed with 0.5 % paraformaldehyde in phosphate-buffered saline for 1 h at room temperature. Fixed worms were rinsed with M9 buffer (42 mM Na 2 HPO 4 , 22 mM KH 2 PO 4 , 86 mM NaCl, 1 mM MgSO 4 , 0.02 % gelatin) and stored at 4°C until required. They were stained with 1 μg/ mL Nile Red in M9 buffer prepared from a stock solution (1 mg/mL Nile Red in acetone) for 15-30 min at room temperature.

Optical microscopic observation
The differential interference contrast (DIC) and autofluorescence images were obtained using an epifluorescence microscope (BX51; Olympus Corp., Tokyo, Japan) with DIC optics. Fluorescence images of GFP, mCherry, mRFP, and Nile Red were obtained using a confocal microscope (FV1000; Olympus Corp., Tokyo) with 473-nm or 559-nm laser excitation. Spectral scanning and unmixing were performed using the spectral deconvolution program in the FV1000 software FV10-ASW. For the unmixing from autofluorescence, autofluorescence was enhanced by preexposure to ultraviolet rays. All images were obtained using hermaphrodites on adult day 1 after anesthetization with 50 mM NaN 3 in M9 buffer, unless otherwise indicated. The glo-1::GFP and GFP::rab-7 transgenic worms subjected to the microscopic investigation were reared at 16°C for clearer identification. All images presented are of the middle part of the intestine (int3 to int7) unless otherwise noted, oriented with the anterior to the left. The presented images are the representatives of at least three animals.