Identification and characterisation of the cryptic Golgi Apparatus in Naegleria gruberi

Although the Golgi apparatus has a conserved morphology of flattened stacked cisternae in the vast majority of eukaryotes, the organelle has lost the stacked organization in several eukaryotic lineages raising the question of what range of morphologies is possible for the Golgi. In order to understand this range of organellar diversity, it is necessary to characterise the Golgi in many different lineages. Here we identify the Golgi apparatus in Naegleria, the first description of an unstacked Golgi organelle in a non-parasitic eukaryote, other than fungi. We provide a comprehensive list of Golgi-associated membrane trafficking genes encoded in two separate species of Naegleria and transcriptomic support to show that nearly all are expressed in mouse-passaged N. fowleri cells. We then study distribution of the Golgi marker NgCOPB by fluorescence, identifying membranous structures that can be disrupted by Brefeldin A treatment consistent with Golgi localisation. Confocal and immuno-electron microscopy revealed that NgCOPB is localized to membranous structures consistent with tubules. Our data not only identify the Golgi organelle for the first time in this major eukaryotic lineage, but also provide the rare example of a tubular form of the organelle representing an important sampling point for the comparative understanding of Golgi organellar diversity.


Introduction
In mammalian cells and many diverse eukaryotes, the Golgi apparatus appears as a stack of flattened membranes, or cisternae, inside which proteins are modified via glycosylation and transported to the plasma membrane or to endolysosomal organelles. In mammalian cells, disrupting the stacked structure of the Golgi leads to numerous defects in these processes, and Golgi fragmentation is observed in autoimmune diseases, cancer, Huntington's, Parkinson's, and Alzheimer's diseases (Zhang and Wang, 2016 inter alia). However, a few examples of eukaryotes with unstacked Golgi are found across the evolutionary tree, and include the budding yeast Saccharomyces cerevisiae , as well as parasitic taxa such as Plasmodium falciparum, Entamoeba histolytica, and Giardia intestinalis  inter alia).
Many of these organisms were initially thought not to possess Golgi and even to have speciated away from the rest of eukaryotes prior to the origin of the organelle. We now know that this is not the case, based on various lines of evidence. The question therefore becomes, given the conservation of stacked Golgi morphology in the vast majority of eukaryotes, what other structural diversity exists in Golgi organelles? In S. cerevisiae, the Golgi compartments are dispersed in the cytoplasm, appearing as punctae using immunofluorescent staining of Golgi markers (Wooding & Pelham 1998). E. histolytica was originally thought to not have a Golgi, but an ultrastructural study showed that this was an artefact of the fixation process for transmission electron microscopy, and presented micrographic evidence of dispersed cisternae in the cell . Finally, Golgi functions in G. intestinalis are stage specific, and are carried out in encystation-specific vesicles that form dispersed compartments ). Unlike typical Golgi, they are not steady state organelles, but arise in response to cyst wall material formation in the endoplasmic reticulum . 4 Regardless of morphology, organisms with unstacked Golgi encode the membrane trafficking factors necessary for Golgi function ). An extensive set of membrane-trafficking machinery has been characterized as involved in vesicle trafficking events to, from and within the Golgi, with distinct paralogues or complexes acting at these steps. These have been shown to be conserved across eukaryotes ) and have been used as genomic signatures for the presence of Golgi organelles in many of the lineages thought once to lack the organelle. However, further characterization of cryptic Golgi in evolutionarily dispersed linages is necessary to have a better understanding of Golgi organellar evolution and the diversity of form that is possible for this organelle.
Naegleria gruberi is a free-living microbial eukaryote that is evolutionarily distant from animals, yeast, and plants. It is commonly found in both aerobic and microaerobic soil and freshwater environments worldwide . The closely related Naegleria fowleri is an opportunistic neuropathogen of humans and animals, killing ~95% of those it infects within two weeks . In 2010, the N. gruberi genome was published, revealing a remarkably complex repertoire of cytoskeletal, sexual, signalling, and metabolic components (L. K. , as well as a highly complete membrane trafficking system (MTS). Naegleria is a member of the supergroup Excavata, which also contains the trypanosomatids, Trichomonas vaginalis, and Giardia intestinalis, of parasitological importance, and the rodent gut commensal Monocercomonoides sp., which is both anaerobic and amitochondriate. Thus, N. gruberi remains one of the few free-living excavates with a complete and publicly available genome, making it a key sampling point for studying eukaryotic evolution, and potentially a useful model system for studying eukaryotic cell biology outside of the animals, yeast, and plants. One distinctive cellular feature of Naegleria is that it lacks a visibly identifiable Golgi organelle. This is in fact a diagnostic feature of the larger taxonomic group to which Naegleria belongs, the heteroloboseans ) and has been since its inception . 5 Despite some proposals for membranous structures as putative homologues of the Golgi , the only evidence supporting the presence of the organelle in Naegleria has been the bioinformatically predicted Golgi-associated proteins, identified in the genome project (L. K. . We here address the identification and visualisation of the Golgi structure in Naegleria using a multidisciplinary approach and present the first molecular and cellular evidence for the presence of a punctate Golgi in N. gruberi, distinct from other endomembrane organelles.

Comparative Genomics
For all Golgi-associated MTS genes that were not identified by , the functionally characterized human orthologue was used as a BLASTP or TBLASTN query  to search the N. gruberi NEG-M (Joint Genome Institute, http://genome.jgi.doe.gov/Naegr1/Naegr1.home.html) predicted proteome, EST cluster consensi, and scaffolds. N. gruberi sequences retrieved with an E-value of 0.05 or less were used to reciprocally BLAST the H. sapiens and non-redundant protein databases (NCBI, https://www.ncbi.nlm.nih.gov/). To be considered true orthologues, they must retrieve the initial query or a clear orthologue with an E-value less than the 0.05 cut-off. Comparative genomics searches of the N. fowleri transcriptome were performed using the N. gruberi Golgi-associated MTS sequences as queries, or the human orthologue in cases where the N. gruberi sequence could not be identified, with the same criteria as above.

Phylogenetics
Bayesian and Maximum-Likelihood phylogenetics analyses were performed to assign orthology to sequences in the Qa-and Qb-SNARE families, and the Qc-SNARE subfamily including Syntaxin 6, Syntaxin 8, and Syntaxin 10. Phylogenetically characterized sequences 6 from H. sapiens, Arabidopsis thaliana, and Trypanosoma brucei were aligned with N. gruberi and N. fowleri sequences identified in this study using MUSCLE v.3.8.31 .
Alignments were visualized in Mesquite v.3.2 (Maddison and Maddison 2017), and manually masked and trimmed to remove positions of uncertain homology. ProtTest v3.4 ) was used to determine the best-fit model of sequence evolution, which was LG+G+F for the Qa-and Qb-SNARE alignments, and LG+G for the Syntaxin 6/8/10 alignment. Phylobayes v4.1 ) and MrBAYES v3.2.2 ) programs were run for Bayesian analysis and RAxML v8.1.3 ) was run for Maximum-Likelihood analysis. Phylobayes was run until the largest discrepancy observed across all bipartitions was less than 0.1 and at least 100 sampling points were achieved, MrBAYES was used to search treespace for a minimum of one million MCMC generations, sampling every 1000 generations, until the average standard deviation of the split frequencies of two independent runs (with two chains each) was less than 0.01. Consensus trees were generated using a burn-in value of 25%, well above the likelihood plateau in each case. RAxML was run with 100 pseudoreplicates.

Transcriptomics
N. fowleri (Ax) V212 were grown axenically at 37°C in Oxoid medium in T75 culture flasks . A mouse-passaged strain of V212 was obtained by intranasal inoculation of the amoebae in B 6 C 3 F 1 mice. Mice were sacrificed when symptoms of infection were evident. The amoebae were harvested from brain tissue and then continuously passaged through mice two times (MP2), four times (MP4) and six times (MP6). Amoebae were subsequently cultured as above for 7-10 days to remove residual brain tissue, and then RNA was extracted and converted to cDNA with the Affymetrix/USB M-MLV (cat 78306) kit using standard protocols. Illumina libraries were constructed using the Nextera Workflow and sequenced in an Illumina MiSeq at the TAGC facility (UAlberta). 7 Between 2.8-4.0 million paired-end 300bp reads remained after pre-processing with Trimmomatic v0.36 ) using the arguments SLIDINGWINDOW:50:30 TRAILING:20. Reads were aligned to an unpublished genome of N. fowleri strain V212 produced as part of an on-going project (unpublished) using the program Tophat v2.0.10 ). Transcripts were assembled for each condition using Cufflinks v2.1.1 and then merged with Cuffmerge ). Cuffdiff v2.1.1 ) was then used to map the reads to the merged transcripts and determine the relative expression. For some Golgi-associated MTS genes, the merged transcript appeared to be partial or incorrectly fused with another gene product, based on comparison with the N. gruberi homologues. To more accurately assess the expression of these transcripts, the transcript boundaries were manually modified to correct for this before mapping the reads. All newly generated N. fowleri gene sequences have been deposited in Genbank as Accession (XXXXX-YYYYY).

Naegleria cell culturing
Naegleria gruberi strain NEG-M (kindly provided by Lillian Fritz-Laylin) was grown axenically at 28 °C in M7 medium . Cells were passaged every 3 -5 days depending on their density.

RNA extraction
Total RNA extraction of Naegleria gruberi's trophozoites was performed using RNAeasy Midi Kit (Qiagen) according to the manufacturer's protocol. cDNA was amplified according to the manufacturer's guidelines using the SuperScript III RT Reaction (Invitrogen).

Generation of antisera to N. gruberi proteins
The entire N. gruberi NEG-M strain ORFs of COPB (XP_002673194) and Sec31 (XM_002669379) were PCR amplified from the cDNA using the primers shown in Table S1 and 8 subsequently cloned into pET16 or pET30b (Novagen), sequenced and transformed into Escherichia coli BL21(DE3) PLyS cells. The expressed proteins were purified using a Ni-NTA column under native conditions. The proteins were further purified by gel electrophoresis and 4 mg of each purified proteins were used to make chicken and rat polyclonal antisera, respectively (two animals per protein; Davids Biotechnologie GmbH; Germany).

Cell Fractionation of Naegleria
Naegleria gruberi cellular fractions were achieved using differential centrifugation of the cell homogenate (passed five times through a gauge 33 hypodermic needle). All steps were carried out at 4 °C and in the presence of the protease inhibitors (Complete Mini EDTA-free cocktail tablets, Roche). To separate cellular fractions, the cells were centrifuged at 1,000 × g for 10 min, and washed and resuspended in the buffer (250 mM sucrose and 20 mM MOPS, pH 7.4). The homogenate was centrifuged at 1,000 × g for 10 min to remove unbroken cells. The supernatant was then centrifuged at 3,000 × g for 15 mins to collect the pellet that contained the nuclei. The supernatant was carefully collected and centrifuged at 7,000 × g for 30 minutes to discard the mitochondrial fraction ). The membrane fraction was centrifuged at 20,000 × g for 3 hours and the supernatant was used as the cytosolic fraction. The separated fractions were quantified using a Bradford assay (Bio-Rad) and then were analyzed via western blot analysis.

Western blotting
Western blots of total protein extracts from N. gruberi trophozoite cells were incubated with the chicken anti-Cop1 (1:200) and rat anti-Sec-31 (1:200) antisera, followed by secondary anti-chicken and anti-rat antibodies respectively conjugated to peroxidase (Sigma). The blots 9 were developed using the ECL protocol (Amersham) and visualized using the Syngene G:BOX XT4 machine on the GeneSys software.

Immunofluorescence microscopy
Naegleria gruberi cells were seeded at 30,000 cells/ml on NUNC LabTek Chamber slides prior to the experiment and grown for 24 hours. Cells were incubated for 20 min with the ER-tracker blue-white DPX marker (Molecular Probes) and then fixed using 2% formaldehyde followed by permeabilization with 0.1% Triton-X in 1X PBS. After blocking for 1 hour in 3% BSA -1X PBS, the cells were probed with the chicken anti-COPI (1:250) and rat anti-Sec

Fixation of cells for electron microscopy and immuno-gold labelling
Aspirated cultures of N. gruberi were fixed for 1 hour in freshly prepared PBS solution containing either 2.5% Glutaraldehyde or 4% Formaldehyde, for Contrast Transmission Electron Microscopy (CTEM) or Immuno-Electron Microscopy (IEM) respectively. Both sample preparations were then washed several times with PBS. CTEM samples were stained and postfixed with 1% Osmium. Both sample preparations were then dehydrated through an ethanol series (30%, 50%, 70%, 90% and three times in 100%). The ethanol was then aspirated and replaced with the appropriate resin mixture. CTEM samples were suspended in Propylene Oxide (PO, Agar Scientific) and mixed thoroughly by rotor, the PO was then replaced by a 50:50 mix 10 of PO and Agar Scientific Low Viscosity (LV) Resin and mixed again. The PO:LV resin mix was then replaced by 100% LV resin. IEM samples were similarly suspended in LR white Resin (Agar Scientific). Resin permeation was aided by placing the samples in a vacuum for 2 minutes.
The resin was then aspirated and replaced with fresh resin and the samples transferred into Beem

Brefeldin A experiments
Prior to each experiment (immunofluorescence microscopy or protein extraction/cell fractionation), Naegleria cells were incubated at 28 °C in M7 medium for 3 hours with 10 nM, 100 nM and 1 µM of Brefeldin A (S7046-SEL, Stratech Scientific Ltd).

Results
Naegleria encode and express Golgi-associated membrane trafficking machinery Naegleria, along with the rest of the heteroloboseans, have been diagnostically described as lacking a visible stacked Golgi ). However, sequencing 11 and annotation of the N. gruberi genome suggested that it encodes many of the necessary  (Table S1). We identified several additional homologues not originally reported in the genome paper, including three members of the COG tethering complex, a nearly complete EARP/GARP complex, and a single Syntaxin 6 orthologue. As the Qa-, Qb-, and Qc-SNAREs are highly paralogous gene families, phylogenetic analyses were performed in order to classify orthologues (Supplementary Figure   1a-c). Therefore, the set of Golgi-associated MTS machinery in N. gruberi is even more complete than previously thought (Figure 1, Table S1).
The presence of these genes in one Naegleria species suggests functional relevance.
Nonetheless, the possibility remains that the genes may not be translated. In order to determine whether these genes are present in other Naegleria species, and more importantly expressed, we generated transcriptomic data from N. fowleri. mRNA was extracted from N. fowleri grown axenically and mouse-passaged. Homology searching was then performed to identify Golgirelated MTS genes in the resulting transcripts, and the relative expression of these genes under each condition was calculated. We identified expressed transcripts for all 66 Golgi-associated MTS genes in N. fowleri, (Table S1). All N. gruberi sequences were shown to have an orthologue in N. fowleri, with the exception of three N. gruberi-specific paralogues (Vps53A, Ykt6B, and Vps45B). Furthermore, N. fowleri encodes and expresses two additional members of the COG complex. These results suggest that Naegleria not only encodes, but also expresses Golgi trafficking and structural proteins.

Visualization of the Golgi in N. gruberi
A commonly used marker for Golgi in eukaryotic cells is the COPI complex, which forms vesicle coats for trafficking from the cis-Golgi to the ER ) and acts at cis and intermediate Golgi compartments . In order to visualize the N. gruberi Golgi using immunofluorescence microscopy, we generated antibodies specific to the beta subunit of the COPI complex (NgCOPB) and to the Sec31 protein of the COPII complex (NgSec31), which is specific to the ER as a comparison point for an endomembrane organelle of the early secretory system. The generated antisera, from chicken and rat respectively, showed high specificity for NgCOPB and NgSec31 in western blots, as they recognized a protein with expected size of 114.5 KDa and 145.7 kDa in the N. gruberi's extracted cell lysates (Supplementary Figure 2). Immunofluorescence microscopy of N. gruberi cells showed distinct patterns for the two antibodies (Figure 2). Consistent with standard ER morphology, the Sec31 antisera showed a network-like localisation around the nucleus of the organism. By contrast, the COPB antisera showed some cytosolic staining, consistent with COPB being a cytosolic/peripheral membrane protein, but strikingly, we see punctuated and tubular localisation around the cell. Some apparent overlap was observed as expected for organelles that span the breadth of the cells. However, clear areas of non-overlap were seen consistent with these structures being discrete organelles.

Brefeldin A treatment disrupts COPB localisation
To further assess whether the COPB antisera was marking Golgi organelles, we tested whether the observed discrete localisation is inhibited by Brefeldin A (BFA), a fungal metabolite that rapidly and reversibly inhibits transport of secretory proteins resulting in relocation of Golgi resident proteins to the ER and COPB to the cytoplasm . Sub-cellular fractionation of membrane constituents showed that treatment with BFA at 10 nM, 100 nM and 1 µm for 3 hours shifted the COPB intensity from the internal membrane to the cytosolic fraction 13 (Figure 3a). Consistent with the previous results, immunofluorescence microscopy demonstrated that the punctuated/tubular localisation disappears as concentrations of BFA increase ( Figure   3b).

The Naegleria Golgi appears as a tubular organelle
We performed localisation experiments captured with confocal microscopy, in order to better assess the ultrastructure of the Naegleria Golgi. Using the same antibody concentrations as before, we captured the localisation of COPB in various cells. The 3D rendering of several sections per cell, demonstrate a discrete tubular organisation (Figure 4a- Both localisation patterns were unlike the one seen for Sec31 (Supplementary Figure 5).
Finally, to investigate the subcellular localisation of NgCOPB even further, we utilized immune-gold electron microscopy ( Figure 5, Supplementary Figure 6), which showed with high confidence localisation of this protein in distinct membrane organelles (1 -4 µm in length) as opposed to the cytosol, nucleus, larger membrane organelles and membrane vesicles.

Discussion
In this study, we have employed a comprehensive set of tools to firstly identify the cryptic Golgi organelle in Naegleria and then investigate its ultrastructure. Prior to this study, the existence of a Golgi apparatus in Naegleria had only been indirectly inferred through its phylogenetic relationship with stacked Golgi-possessing lineages and, slightly more directly shown, by the presence of a moderate list of genes from a genome of single species, which in 14 canonical systems are localised at the Golgi. We therefore, sought to provide an increasingly direct and convincing array of evidence. Additional homology searching within N. gruberi allowed the identification of several previously unreported proteins that are characteristically associated with the Golgi apparatus, while transcriptomic analysis of a second species demonstrated that their presence in another Naegleria representative (N. fowleri) and more importantly that these genes are indeed expressed. The complement of Naegleria Golgi genes suggests both cis-Golgi and trans-Golgi Network (TGN) functions and retrograde pathways from the cis-Golgi to ER, as well as multiple routes from the TGN outward. This is consistent with complex Golgi function in Naegleria fits well with other reports of vesicle coat complexity As COPI is a well-established marker for the Golgi, we generated a homologous polyclonal antibody against the entire NgCOPB protein and we used it for subsequent localisation studies. COPB is a peripheral membrane protein that translocates to membranes from the cytosol during vesicle formation for the transport of material from cis-Golgi to the ER . In the case of N. gruberi, our immunofluorescence microscopy studies have demonstrated discrete tubular localisation that occupies 17% of the total cellular volume (Supplementary Figure 7) and does not show the same localisation patterns as DAPI, Sec31 or ER-tracker. Both confocal and immuno-electron microscopy has provided evidence of the association of this protein with membranous structures.
The identity of this membrane structures as Golgi homologues was further confirmed with BFA experiments. Brefeldin A acts on the Arf-GEF, GBF1, blocking COPI vesicle formation . Exercising both biochemical and microscopical means, we have demonstrated that BFA observations are consistent with the behaviour of Golgi markers in other organisms with unstacked Golgi (e.g. Plasmodium, Entamoeba and Giardia) upon BFA treatment . 15 Based upon the sum of our data, we conclude that the Golgi in Naegleria takes the form of discrete tubular compartments, which did not exceed 1 µm in diameter and 4 µm in length.
These tubules did not appear concentrated in any specific region of the cell but were dispersed throughout.
There are several intriguing avenues for future investigation of Naegleria Golgi. Our data addresses only a cis-Golgi marker, leaving open the question of TGN organization. We observed the Golgi in the most common life-stage of Naegleria, the trophozoite, but the behaviour of the organelle in the flagellate and cyst forms would be worthy of enquiry. Finally, organellar dynamics would be fruitfully investigated using live-cell imaging and higher-resolution microscopy, such as light-sheet technologies. In Giardia and yeast, there is evidence of transient communication intermediates , either tubular or vesicular between the compartments. Understanding the mechanism of material transfer between Naegleria Golgi will be an exciting challenge. All of these aspects would be greatly facilitated by additional molecular cell biological tools in Naegleria that are currently being developed (protocols.io). Given the complexity of the metabolic and cell biological complement encoded in its genome (L. K. Fritz-Laylin et al. 2010), we suggest that Naegleria is a promising model organism for comparative eukaryotic cell biology.
In theory, there is an array of forms that a stacked Golgi could take upon an evolutionary reorganization in a lineage. These include a single large (but unstacked) organelle, a vesicular tubular network, a tubular network, or discrete dispersed smaller compartments. While a single perinuclear organelle is reported in Plasmodium , to our knowledge its ultrastructure has not been examined at the EM level. The organelles of Giardia, Entamoeba, and Saccharomyces are best described as discrete dispersed small compartments. By contrast, the Golgi in microsporidian lineages is reported as either an array of lamellar membranes in the cysts or as a tubular (but not vesicular) network ). This latter organization is most similar to that which we observe in Naegleria. From this diversity of form, it appears that there are no obvious constraints excluding any of the potential organellar morphologies that Golgi may take upon leaving the otherwise nearly ubiquitous stacked organization. It will be exciting to investigate in additional taxa, and using sophisticated cell biological techniques, to further elucidate the ultrastructure of these compartments. This also raises the fundamental question of why stacking is so pervasive amongst eukaryotes, given the demonstration that other morphologies can readily exist. yeast cells. Molecular biology of the cell, 9(9), pp.2667-80. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=25539&tool=pmcentrez&render type=abstract. Zhang, X. & Wang, Y., 2016. GRASPs in Golgi Structure and Function. Frontiers in Cell and Developmental Biology, 3(January), pp.1-8. Rat anti-N.gruberi Sec31 antiserum (1:250; red) shows a discrete localisation in the cells, while DAPI stains the Naegleria nucleus and mitochondrial DNA. DIC shows the differential interference contrast image of the cells used for immunofluorescence. Scale bar: 10 µm. c.
Cellular localisation of COPI and Sec31 in N. gruberi cells. Chicken anti-N.gruberi COPB antiserum (green) shows a discrete localisation in the cells that is not co-localised with the rat anti-N.gruberi Sec31 antiserum (red), while DAPI stains the Naegleria nucleus and mitochondrial DNA. DIC shows the differential interference contrast image of the cells used for immunofluorescence. Scale bar: 10 µm.  Immuno-gold localisation of COPI in Naegleria gruberi cell by transmission electron microscopy shows localisation associated with membrane bound organelles. The inset shows a higher-magnification image of the organelles. Four additional images can be found in Supplementary Figure 6. The graph demonstrates the densities of labelling in the different compartments of N. gruberi cells suggesting that COPI is mainly localized in the membrane bound organelles of the cell.