| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Biochemistry,
* Department of Biology, Emory University School of Medicine, Atlanta, Georgia, USA
3 Correspondence: Department of Biochemistry, Emory University School of Medicine, 1510 Clifton Rd., Atlanta, GA 30322, USA. E-mail: rkahn{at}emory.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
100 protein sequences analyzed from animals, fungi, plants, and protists clustered into 11 groups, including Arfs, nine Arls, and Sar proteins. To begin functional analyses of the family in a metazoan model organism, we examined roles for all three C. elegans Arfs (Arf-1, Arf-3, and Arf-6) and three Arls (Arl-1, Arl-2, and Arl-3) by use of RNA-mediated interference (RNAi). Injection of double-stranded RNA (dsRNA) encoding Arf-1 or Arf-3 into N2 hermaphrodites produced embryonic lethality in their offspring and, later, sterility in the injected animals themselves. Injection of Arl-2 dsRNA resulted in a disorganized germline and sterility in early offspring, with later offspring exhibiting an early embryonic arrest. Thus, of the six Arf family members examined in C. elegans, at least three are required for embryogenesis. These data represent the first analysis of the role(s) of multiple members of this family in the development of a multicellular organism.—Li, Y., Kelly, W. G., Logsdon, J. M., Jr., Schurko, A. M., Harfe, B. D., Hill-Harfe, K. L., Kahn, R. A. Functional genomic analysis of the ADP-ribosylation factor family of GTPases: phylogeny among diverse eukaryotes and function in C. elegans.
Key Words: Arf family • protein sequence • eukaryotic evolution • RAS superfamily • regulatory protein
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
2
3)
. The ubiquity, essential nature, functional and structural conservation, and increasing numbers of Arfs found through eukaryotic evolution all indicate the importance of this family of regulatory proteins. Using functional criteria and sequence homology, the Arf family can be divided into two groups: the Arf and Arf-like (Arl) proteins. Because Arfs are the most divergent family of 20 kDa GTPases, they are not always included in analyses or descriptions of the Ras superfamily. Similarly, because Sar proteins are the most divergent in sequence within the Arf family, they are not always included in the Arf family. We will use the general term "Arf family" here to include Arfs, Arls, and Sars and the specific term "Arfs" to refer to a subset of proteins within this family that share a number of common functions (see below).
Arfs have been highly conserved throughout eukaryotic evolution, maintaining >60% primary protein sequence identity, in vitro biochemical activities [e.g., cofactor in the ADP-ribosylation of Gs
by cholera or Escherichia coli toxins or activators of phospholipase D (4
5
6
7
8
9)
], and in vivo activity [e.g., suppression of the lethality of arf1– arf2– in Saccharomyces cerevisiae (10
, 11)
]. In contrast, the Arls are more divergent, though still highly related in sequence (40–60% amino acid identity between any two Arls or between any Arl and any Arf; refs 2
, 12
). As indicated, the Sar proteins are the most divergent in primary sequence, having 25–30% identity to Arfs or Arls, but this level of conservation is higher than any other Ras superfamily member.
This highly conserved family is involved in a wide array of cell functions. Arfs have been proposed to function through direct activation of enzyme activities, local alterations in lipid composition, or recruitment of proteins to membranes, particularly vesicle-coat complexes or proteins (13
14
15
16
17
18
19
20)
. Arl1 is the protein most closely related to the Arfs structurally and functionally, including sharing some binding partners (12)
and actions in the secretory pathway (21
22
23
24
25
26)
. Arl1 is an essential gene in Drosophila melanogaster (27)
but not in S. cerevisiae (28)
. Although the most divergent in primary sequence, the Sars may be the most functionally related to Arfs in that they too have been shown to recruit coats (COPII) to ER membranes and are essential in the early secretory pathway (29
30
31
32
33)
. Arl2 has been implicated in yeast [S. cerevisiae (34
, 35)
and S. pombe (36)
], a plant [A. thaliana (37)
], worms [Caenorhabditis elegans (38)
], and mammalian cells (HeLa) (39)
as a regulator of tubulin biosynthesis and microtubule dynamics through its proposed binding to the tubulin cochaperone cofactor D (36
, 37
, 39
, 40)
. Arl4 has been deleted in the mouse and found to be a nonessential gene, with homozygous null animals exhibiting a retardation of germ cell development in males (41)
. Finally, the Arf-related protein Arfrp, or Arfrp1, is a plasma membrane-associated protein that is essential in mice, specifically in early embryonic development during gastrulation (42)
.
We previously analyzed phylogenetic relationships within the Arf family (43)
. In this study, however, we have extended the previous analyses with increased sampling of diverse eukaryotes and the use of more rigorous phylogenetic methods to more accurately describe the evolutionary history of these proteins. The Arf genes/proteins have been partitioned into three classes, termed Arf I, II, and III. With evidence that three different groups of Arfs and at least four different groups of Arls are conserved in metazoans, we made use of RNA interference techniques most highly developed in C. elegans to ask whether the single representative of each of the three Arf classes and three of the Arl classes play a role(s) in the development of this organism. We describe evidence for distinct roles for at least two of the three Arfs [Arf-1 (Arf I) and Arf-3 (Arf II)] and at least one of three Arls tested (Arl-2) in C. elegans development.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
using query sequences from previously characterized proteins (43)
. Sequences were aligned using CLUSTALX 1.81 (45
, 46)
, then edited manually using MACCLADE 4.03 (47)
. Analysis of the entire Arf family (Arfs, Arls, and Sars) was done using a subset of the Arf protein sequences from phylogenetically representative species. Alignments were analyzed using distance and Bayesian maximum likelihood (ML) methods. For the former, TREEPUZZLE 5.0 (48)
was used to calculate pairwise distances with the JTT substitution model and a mixed model of rate heterogeneity (8 gamma-distributed rates and 1 invariable). The ML distance matrix was used to construct a neighbor-joining tree using NEIGHBOR (PHYLIP 3.6a) (49)
. Bootstrap support from 1000 replicates was obtained using PROTDIST (PHYLIP) and the initial gamma shape parameter determined from ML distance analysis. Bayesian analysis was performed using MR BAYES 2.01 (50)
with a JTT substitution model and eight gamma distributed plus one invariable rate heterogeneity categories. Four Markov chains were used (1 heated and 3 cold) and the analysis was run for 1 million generations with trees sampled every 1000 generations. From a plot of likelihood scores vs. generation, the point of inflection was determined and only trees from that point on with the best posterior probabilities were retained for construction of the consensus tree. To compare alternate tree topologies to trees obtained from ML and Bayesian analyses, Kishino-Hasegawa (51)
, Shimodaira-Hasegawa (52)
, and expected likelihood weight tests (48)
were conducted using TREEPUZZLE 5.0.
Growth of strains
The wild-type C. elegans strain used in this study was Bristol N2. Worms were cultured at 20°C or 25°C using standard protocols as described by Brenner (53)
.
RNA-mediated interference (RNAi)
cDNAs: Arf-1 and Arf-3 cDNAs from EST clones yk69a11.5 and yk319e2.5, respectively, were obtained from Yuji Kohara (National Institute of Genetics, Japan). Phagemid DNA was excised using standard procedures (Stratagene, San Diego, CA, USA).
The high sequence identity between Arf-1 and Arf-3 prompted concerns about specificity of the RNA-mediated interference of each. Alignment of Arf-1 and Arf-3 nucleotide sequences revealed the longest stretches of identity of 47, 20, and 17 nt in the middle of the open reading frames. Because it is believed that 21-23 nt fragments mediate RNAi effects, this region with the longest stretches of identity could potentially compromise specificity (54)
. Therefore, we generated shorter forms of Arf1 and Arf3 dsRNA in which the regions of high homology were absent. Arf-1 and Arf-3 plasmids were constructed in which 103 bp were deleted corresponding to bp 156-260 of the open reading frame (see list of primers in Table 1
). The PCR products were analyzed by agarose gel electrophoresis to confirm the shorter length before purification and subcloning into pBluescript SK(–), yielding clones pYW135 and pYW136 for short form cDNAs clones of Arf-1 [Arf-1(short)] and Arf-3 [Arf-3(short)], respectively.
|
Arf-6, Arl-1, Arl-2, and Arl-3 cDNAs were generated by amplifying each from an N2 cDNA library (a gift of Steve L’Hernault, Emory University) using nested, gene-specific primers (Table 1)
, then cloned into pCRII-TOPO, yielding clones pYW101 (Arf-6), pYW106 (Arl-1), pYW107 (Arl-2), and pYW108 (Arl-3).
Clone yk40 g9 (obtained from Yuji Kohara (National Institute of Genetics) is an EST that encodes 3 kb of the F16D3.4 cDNA, starting at bp 532 in the open reading frame and ending in the 3' untranslated region. Phagemid DNA excision was performed as for clones yk69a and yk319e.
dsRNA synthesis
Sense and anti-sense RNAs were synthesized using the Ribomax RNA production systems and SP6, T3, or T7 RNA polymerases (Promega, Madison, WI, USA) after linearization by restriction digestion. Sense and antisense RNA reactions were treated with DNase, followed by phenol: chloroform (1:1; v:v) extraction and ethanol precipitation. Annealing of complementary strands at 1 mg/mL resulted in formation of the dsRNA for RNAi injections. Triple mutant Arf-1(RNAi), Arf-3(RNAi), Arf-6(RNAi) -injected animals contained equal amounts of each dsRNA.
dsRNA microinjections
dsRNA was injected into both gonad arms of adult N2 hermaphrodites. Arl2 dsRNA was injected into strain AZ244, carrying a transgene directing expression of tubulin::GFP (55)
. After incubation at 25°C for 6 h, the injected worms were transferred to new plates and allowed to lay eggs for 18 h. Worms were then picked onto a fresh set of plates and allowed to lay eggs for an additional 28 h at 25°C. All progeny were scored 24 h after the injected animal was removed from the plate.
Promoter-specific GFP expression pattern
GFP expression constructs
Plasmids were constructed to allow expression of the GFP reporter protein under control of the Arf or Arl promoters. A list of templates and primers used in these constructs is shown in Table 2
. In the case of Arf-6, a construct was generated that included no protein coding sequence: the promoter was followed directly by the GFP. Two GFP expression constructs were generated for the Arf-1 gene. One contained the GFP fused to the C terminus of the full-length protein and the other replaced the entire open reading frame of Arf-1 with that of GFP. In all other instances, GFP was fused in-frame to an exon of the Arf or Arl coding region, resulting in a C-terminal GFP fusion protein. The promoter regions of Arf-3, Arf-6, Arl-2, and Arl-3 were amplified from N2 genomic DNA. Cosmids B0336 and F54C9 were used as templates for Arf-1 and Arl-1, respectively. Because none of the promoters had been characterized earlier, genomic DNA spanning the region between the relevant gene (Arf-1, Arf-3, Arl-1, Arl-2, or Arl-3) and the nearest upstream open reading frame was used to create the GFP constructs. The Arf-6::gfp construct contained a shorter, 0.8 kb upstream region. The nested, PCR-generated genomic DNA fragments containing the promoters and full-length Arf-1 or Arf-3 coding sequences (without the stop codon) included 2 kb and 0.62 kb upstream sequence for Arf-1 or Arf-3, respectively. PCR with primers for Arl-1, Arl-2, and Arl-3 yielded genomic DNA fragments containing open reading frames and 623 bp, 370 bp, or 742 bp of 5' flanking sequence, respectively. PCR products were verified by agarose gel electrophoresis, followed by purification using Wizard PCR Preps DNA Purification System (Promega). These DNAs were digested with restriction enzymes according to the recognition sites introduced by the PCR primers (Table 2)
and inserted in-frame into the plasmid pPD95.69 (Andrew Fire Vector Kits). The resulting promoter fusion plasmids were named pYW124 (Arf-1::gfp), pYW110 (Arf-1::gfp), pYW109 (Arf-3::gfp), pYW117 (Arf-6::gfp), pYW126 (Arl-1::gfp), pYW113 (Arl-2::gfp), and pYW115 (Arl-3::gfp).
|
Plasmid microinjections
Transgenic lines carrying extrachromosomal arrays that expressed GFP were generated as described (56)
. Plasmids pYW124, pYW109, pYW110, pYW117, pYW126, pYW113, and pYW115 were coinjected with plasmid pRF4 carrying the dominant rol-6 marker into both distal gonad arms of adult N2 hermaphrodites. F2 rollers were picked to establish stable transgenic lines.
Microscopy and staining
F1 RNAi embryos and adults were mounted on 2% agar pads in Tris-buffered saline with 1 mM levamisole and visualized by differential interference contract (DIC) microscopy on a Zeiss Axiophot compound microscope. Images were captured using a DAGE CCD300T-RC digital camera (Dage-MTI Inc., Michigan City, IN, USA).
Animals carrying extrachromosomal arrays for GFP expression were mounted on slides with 1 mM levamisole and visualized with a Leica DMRXA microscope. Pictures were taken with a Cooke Sensicam camera and processed with Image Pro 4.1 software (Media Cybernetics, Inc., Silver Spring, MD, USA).
Whole mount preparation of specimens for staining of DNA was performed as follows. Embryos and/or ovaries were excised from adult hermaphrodites in a drop of M9 buffer (22 mM KH2PO4, 42 mM Na2HPO4, 85 mM NaCl, and 1 mM MgSO4) on polylysine-coated slides. Specimens were fixed by addition of an equal volume of M9 containing 5% paraformaldehyde, a coverslip was added and specimens were incubated at room temperature for 5 min. Animals were frozen on dry ice for 10 min, freeze-cracked by snap removal of the coverslip, incubated in –20°C ethanol for 2 min, and washed in PBS (with 0.2% Tween-20) for 10 min.
Antibody staining was performed by incubating samples in monoclonal -tubulin antibody conjugated to FITC (clone DM1A, Sigma, 1/200 dilution) for at least 2 h and washed three times with PBS plus 0.2% Tween-20. Slides were then blotted dry and Pro-Long (Molecular Probes) antifade containing 1 µg/mL 4', 6-diamidino-2-phenylindole (DAPI) was added for viewing by fluorescence microscopy.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
, 11
, 28
, 57
58
59
60
61
62)
. With the availability of many completed genome sequences, we set about to define the entire Arf family among diverse eukaryotic organisms, including humans. We had focused earlier on finding all predicted members of the Arf family in the completed yeast (S. cerevisiae), fly (D. melanogaster), worm (C. elegans), and human genome databases (43)
; here we also included sequences obtained from protist and plant databases, many of which are not complete or well annotated. A total of 100 sequences was included in the current phylogenetic analyses, as described in Materials and Methods. The resulting tree topologies from the two methods used were not significantly different, and the Bayesian consensus tree is presented (Fig. 1
). Our analyses show that the Arf family proteins cluster into 11 groups: Arfs, Arl1-6, Arl8, Arl10/11, Arfrp, and Sar (Fig. 1B
).
|
Nomenclature can be confusing with large gene families due largely to simultaneous cloning of cDNAs from different labs or from decisions made by curators of different databases. We use the nomenclature adopted by researchers in the field and described in a recent review (see Table 3
; ref 43
). This is especially important in a few cases in which a protein has been given multiple names (e.g., Arl9 previously termed Arf4L or Arl6). We note that human Arl10 and Arl11 are described here for the first time. Although each human protein we described here has been PCR amplified from cDNA libraries or from plasmids obtained from EST clones (R. A. Kahn and S. Kerr, unpublished observation), not all proteins from other organisms have been shown to be expressed. The presence in some (particularly human) databases of additional proteins having sequence similarity to Arf family members but lacking all the conserved consensus GTP binding motifs leaves open the possibility of additional groups of Arls or Arl-related proteins in mammals that appear to be absent in other eukaryotes. The existence of larger proteins that contain regions with high sequence identity to Arfs and do bind guanine nucleotides has been described, e.g., the 64 kDa ARD1 (63)
, but they are not included in our analyses. In this and other cases it appears that the 180 residue-long Arfs may be viewed as a domain when part of a larger protein.
|
The primordial nature of the Arf family among eukaryotes is evident from the finding of six proteins present in the protist Giardia (see Table 3
), an organism thought to represent an early-diverging lineage among eukaryotes (64)
. Although Giardia appears to lack representatives of the Ras or heterotrimeric G-protein families (our analyses and ref 65
), these analyses show it has single representatives in the Arf, Arl2, Arl3, Arl8, Sar, and a sixth protein (GlArl) that does not specifically group with Arf family proteins from other organisms. The presence of orthologs in Giardia (and, in some cases, other protists) points to an early eukaryotic origin for at least six of the Arf family members: Arf, Arl1, Arl2, Arl3, Arl8 (and/or Arfrp) (Fig. 1)
. Thus, much of the evolutionary diversity of the Arf family traces back to the origin of eukaryotes and implies that many of their encoded functions have been retained.
An interesting contrast to Giardia is the fungus S. cerevisiae, which has just seven Arf family members, with three Arfs (two ArfI/II proteins and Arf3), three Arls (Arl1, Arl2/Cin4), Arl3 (an ortholog of Arfrp), and Sar. Animals, however, have representatives in all 11 groups. Three of these groups (Arl4, Arl6, and Arl10/11) are composed of only animal proteins indicating their recent origins, with Arl10/11 (present only among vertebrates) being quite recent. Apparently the number of Arf family proteins has increased in the metazoan lineage, suggesting an animal-specific function(s) to some of these newly evolved proteins. Given the phylogenetic distribution of Arf family members among eukaryotes, some specific losses of genes are predicted. For example, Arl3 is present in animals and protists (Fig. 1)
. The absence of Arl3 in plants and fungi is most easily interpreted as a loss from those organisms.
Arl2 and Arl3 have been found to share some unique structural features that influence their handling of guanine nucleotides (66
, 67)
and must have diverged very early, as they are present in Giardia. Although Arl2 has turned up in numerous genetic screens and is functionally linked to microtubule assembly, Arl3 never has. Given the role of Arl2 in the cytoskeleton, it is tempting to speculate that Arl3 shares a related function, perhaps specifically involved in motility. The apparent loss of Arl3 from fungi and plants (each of which possess cell walls) may have been made possible in these organisms due to their nonmotile life style. If so, we predict that zygomycete fungi (which remain flagellated) will retain Arl3.
A total of 51 Arf proteins from animals, fungi, plants, and protists were mined from existing databases for an in-depth phylogenetic analysis of this subfamily (Fig. 2
). We found that each metazoan has a minimum of three Arfs, with as many as six in mammals. The six mammalian Arfs have been described and shown to fall into groups: class I (mammalian Arf1-3), class II (Arf4-5), and class III (Arf6) (60
, 68)
. This grouping is based on conservation of primary sequence and of intron boundaries within Arf classes. Our phylogenetic analysis provides specific support for these groups (Fig. 2)
. This set of six Arfs likely represents a "final" number of Arfs in mammals. Fly and worm genomes were found to encode single representatives from each of these three Arf classes, further supporting the class structure of mammalian Arfs.
|
In contrast to animals, specific representatives of class I and II Arfs are not found in fungi and plants (and protists; see below). The lack of phylogenetic support for a grouping of fungal and plant Arfs with ArfI or ArfII (Fig. 2)
suggests the most parsimonious interpretation that (animal) ArfI and Arf II duplicated and diverged early in animal evolution after their divergence from fungi and plants. Similarly, we found that all of the protists analyzed contain only one Arf as opposed to at least two Arfs (class I/II and III) present in all animal, fungal, and plant genomes. Thus, our analyses suggest the simple possibility of a single founding Arf in protists that was later duplicated after the divergence of plants: first, into I/II + III (in plants, fungi and animals), then even later into I + II (in animals). The topology shown in Fig. 2B
does not directly indicate this hypothesis; however, a few unsupported branches can be rearranged to produce a tree consistent with this interpretation (see below). Table 4
summarizes several branch moves and their statistical support values. Move #5 (ArfIIIs ancestral to ArfI+ArfI/II clade, illustrated in inset of Fig. 2B
) represents the evolutionary scenario described above and is not statistically significantly different from the topology shown in Fig. 2B
. Alternatively, the ArfI/II + III split could have predated protist divergences, followed by multiple subsequent losses in protists. Additional taxon sampling among more diverse protist lineages will further distinguish between these hypotheses.
|
The essential S. cerevisiae pair Arf1 and Arf2 (97% identical) falls into a larger class I/II group (see Fig. 2
) whereas Arf3 groups with class III. The latter observation is surprising given the fact that mammalian Arf6 can rescue lethality resulting from deletion of S. cerevisiae Arf1 and Arf2 (11)
, as can the single Arf representative from the protist Giardia lamblia (11)
, whereas apparently the S. cerevisiae ArfIII, ScArf3, cannot (69)
. We note that ScArf3 has the longest branch in this group and thus may have diverged in function more so than other class III Arfs. This explanation is consistent with the fact that mammalian Arf6s localize and function in distinct ways from Arf1-5 (70)
.
The trees shown in Fig. 1B
, Fig. 2B
are consensus Bayesian likelihood trees. Some details of these phylogenetic results led us to question whether slightly alternative topologies would allow more straightforward interpretations (e.g., inset in Fig. 2B
). We hypothesized that rearranging some (mostly unsupported) branches would simplify the apparent phylogenetic relationships and the inferred evolutionary scenarios implied from them (Table 4)
. For example, we asked whether it was appropriate to move GlArl into the Arl6 or Arl8/Arfrp clades (Fig. 1B
), whether animal Arf I could be moved to form a clade with animal Arf IIs (Fig. 2B
), or whether protist Arfs could be moved ancestral to ArfI/II/III so as to root the Arf-specific subtree. In each case we applied three different tests for statistical analysis of significance (Table 4
; data not shown for expected likelihood weight tests). In each case, the topology changes were not statistically significantly different (except for negative controls, in which rearrangements are expected to fail the tests), suggesting that these topologies are as good as those in the trees shown (Fig. 1B
, Fig. 2B
). Therefore, given the lack of resolution for certain clades in the consensus trees, alternative tree topologies may in fact represent more straightforward evolutionary histories of Arf and Arl proteins.
Functional analyses of the Arf family members have focused on the role of Arfs in membrane traffic, and genetic analysis has been mainly limited to S. cerevisiae as a model genetic organism. With a much clearer picture of the diversity and evolutionary relationships of the Arf family, we sought a metazoan model that would allow us to extend the functional studies in yeast and specifically address the question of functional importance of redundancy among the Arfs themselves. For these reasons, we used C. elegans and the technique of RNA interference (RNAi) to provide insight into potential roles for these regulatory GTPases in development. A summary of the worm Arf/Arl genes/proteins and proposed names is shown in Table 3
.
RNAi in C. elegans
Arf(RNAi): Arf family members in C. elegans were first identified by Murtagh et al. (Arf-1 and Arf-3; J. Murtagh, J. Moss, and M. Vaughan, personal communication); with genomic sequence data, others were noted (T. Barnes and S. Hekimi, personal communication). A third gene, Arf-6, is described here for the first time and is named by virtue of its closest ortholog in mammals (see Fig. 2
and Table 3
). Arf-6 was found in genome sequence on a sparsely populated region of chromosome IV; its representation in cDNA libraries used for EST databases appears to be rare. We amplified the predicted open reading frame from a C. elegans cDNA library and documented a specific pattern of expression from the Arf-6::GFP transgene (see below), consistent with it being an actively transcribed gene whose expression is turned on early in embryogenesis. This is consistent with data reported for the Arf-6 ORF in genomic microarray analysis of C. elegans development (71)
.
Presentation of dsRNA (RNAi) corresponding to an endogenous mRNA results in the specific degradation in C. elegans of the mRNA (reviewed in ref 72
). Both the maternal and zygotic RNA components of most genes can be efficiently abolished by this method (73)
. We targeted six members of the Arf family and triple mutant animals that lacked all three Arfs. Defects in embryogenesis were clearly evident in worms made deficient in two of the three Arf proteins. Embryonic lethality occurred in 71% (see Table 5
) of Arf-1(RNAi) progeny. No elongation or morphogenesis was evident (Fig. 3
A, B), although these embryos contained gut granules as determined by polarized light microscopy (not shown). A few of these embryos were moving, indicating that although some muscle differentiation had occurred, it was poorly organized.
|
|
Different results were obtained from injections of dsRNA encoding Arf-3. About 80% of Arf-3(RNAi) embryos were arrested during embryogenesis (see Table 5
). However, these embryos arrested at the 2-fold stage of elongation (Fig. 3C, D
), which was clearly later than that seen in Arf-1(RNAi) animals. Animals that hatched exhibited a "Lumpy-Dumpy" morphology, similar to embryos containing mutations in genes involved in muscle development (74)
. In the arrested embryos, gut granules and well-defined pharyngeal structures were visible but gut cells exhibited an abnormal, disorganized arrangement (not shown). Malformed embryos were observed to roll and twitch within the eggshell, indicating the presence of differentiated muscle cells. The pharynx in these embryos also appeared misshapen (Fig. 3)
, possibly as a result of a defect in elongation.
In contrast, Arf-6(RNAi) animals had no obvious phenotypes. They exhibited wild-type movement, had no detectable developmental defects, and were fertile. Arf6 is unique among the mammalian Arfs in being stably associated with membranes, especially the plasma membrane, functioning in the regulation of endocytosis (75
76
77)
and the cytoskeleton (78
79
80
81
82)
. It was therefore surprising that of the three worm Arfs only Arf-6(RNAi) lacked an observed phenotype, even though similar amounts of dsRNA were injected for each target. Neuronal tissue in C. elegans is known to be inherently refractory to RNAi effects by exogenous dsRNA (83)
. The neuronal expression of the arf-6::gfp promoter construct (Table 6
) could be consistent with a neuron-restricted essential function that cannot be targeted by the method we used. However, the expression observed in other tissues (e.g., muscle) that are not refractory to RNAi suggests that its function in most tissues is not essential.
|
Injection of dsRNA encoding either Arf-1 or Arf-3 caused a progressive sterility in the injected animals. In both cases, all of the injected animals became sterile within 24 h postinjection. Sterility of injected animals correlated with changes in the morphology of the gonadal arm and apparent defects in cellularization, as visualized by light microscopy (Fig. 4
). This phenotype was not observed in control-injected animals or those injected with Arf-6 dsRNA. Sterility caused by Arf-1 and Arf-3 dsRNA in the injected P0 animals indicates that these genes are needed not only during embryogenesis but also in adult germline function. Thus, results described here provide the first evidence for multiple roles in metazoan development for the two most closely related Arfs, Arf-1 and Arf-3. Despite the high degree of structural similarities between C. elegans Arf-1 and Arf-3, the distinct phenotypes of embryonic arrest resulting from RNAi, timing of their expression in the developing embryo (see below), and their overlapping but distinct expression patterns in adults support the conclusion that despite similarities in biochemical activities, their roles in the developing organism are separate. The description of developmental roles for Arf-1 and Arf-3 provides an interesting contrast to results from single cell studies, and may provide a broader framework within which these regulatory proteins can be better understood.
|
Specificity of Arf(RNAi) effects
The high similarities between Arf-1 and Arf-3 in sequence (85% protein and 76% nucleotide identity) and in phenotypes resulting from RNAi, raised concerns about the specificity with which these dsRNAs were acting. Two additional controls were performed to ensure the loss of the targeted message and retention of the homologous message. Because RNAi is thought to be mediated by 21-23 nt fragments that can anneal to identical sequences in mRNAs, we compared the nucleotide sequences of the Arf-1 and Arf-3 messages. The presence of two regions of identity of 21 nt or greater were found in the middle of the open reading frames and were both contained in one 100 nt fragment. Plasmids were constructed in which these long regions of identity were deleted, as described in Materials and Methods, and dsRNAs prepared as before. With the shorter constructs, the longest stretch of identity in nucleotide sequences was 14 nt. The phenotypes of Arf-1(short)(RNAi) and Arf-3(short)(RNAi) animals were indistinguishable from those resulting from injection of the full-length dsRNAs, as described above (data not shown).
Documentation of the degree of loss of a specific protein as a result of RNA interference is difficult in the absence of a specific antibody for the target protein. To monitor protein levels in injected animals and their progeny, we constructed lines expressing Arf-1 or Arf-3 as C-terminal GFP fusion proteins driven by their own promoters from multi-copy, extrachromosomal DNA arrays. Transgenic animals were injected with Arf-1 or Arf-3 dsRNA and GFP expression was monitored by fluorescence microscopy. A nearly complete loss of GFP signal was observed when transgenic animals were injected with dsRNAs corresponding to the GFP fusions (data not shown), thus confirming loss of the transgenic message and protein. In addition to knockdown of the reporter signal, the previously identified embryonic arrest phenotypes were observed for each dsRNA. The depletion of the multi-copy transgene transcript, coupled with the arrest observed in RNAi embryos, indicates that the dsRNA treatment was at or above saturation for the endogenous targets. In contrast, strong fluorescence was still evident when the noncorresponding RNA was injected (data not shown). These data demonstrate a high level of specificity in the loss of protein and message as a result of injection of dsRNAs encoding members of the Arf family. As this holds true for the most homologous messages, we infer it also holds true for less related messages that have no stretches of identity of 19 nt or longer.
GFP fusion protein expression
If, as suggested by the results above, the Arfs have separable roles in development, they might be expected to show differences in the developmental timing of their expression and/or in their tissue specificity. To test this, transgenic lines carrying C-terminal GFP fusions of either the full-length Arf proteins (Arf-1::GFP and Arf-3::GFP) or replacement of the coding region with that of GFP (Arf-1::gfp and Arf-6::gfp) were generated and analyzed. The results are summarized in Table 6
.
Arf-1::GFP or Arf1::gfp expression was first detectable in very young embryos with as few as 48-64 cells. Arf-3::GFP was only seen at later stage embryos (>200 cells). These two results are consistent with those from RNAi in which animals depleted of Arf-1 arrested at an earlier stage than those depleted for Arf-3. The Arf-6::gfp promoter construct resulted in very early expression, similar to that seen with Arf-1::GFP (data not shown).
The Arf::GFP lines exhibited broad and overlapping spectra of postembryonic tissue expressions, with all three Arfs expressed in the ventral nerve cord, vulva, intestine, and body wall and pharyngeal muscle tissues (Table 6)
. Expression of each Arf was more evident in some tissues than other Arfs; for example, Arf-1::GFP was clearly seen in the uterus, spermatheca, and gonadal sheath, Arf-3::GFP was in seam cells, and Arf-6::GFP in the excretory cell. In contrast to the Arf genes, the Arl-1::gfp and Arl-3::gfp constructs each exhibited a more narrow pattern of expression. Whereas Arl-1::gfp was expressed in head neurons, the ventral nerve cord, and intestine, Arl-3::gfp was limited to a subset of neurons in the head and tail regions. Transgenes in these animals were nonintegrated and exhibited mosaicism in patterns of expression between transgenic siblings. The patterns summarized in Table 6
represent consistent patterns observed within animal populations for each transgene.
Arl RNAi: The C. elegans genome encodes 12 proteins in the Arf family, including the three Arfs, and orthologs of human Arl1, Arl2, Arl3, Arl5, Arl6, Arl8, Arfrp, and Sar proteins. In this study we focused on Arl1-3 and propose to name the genes Arl-1 (F54C9.10), Arl-2/evl-20 (F22B5.1), and Arl-3 (F19H8.3). Arl-1(RNAi) and Arl-3(RNAi) did not result in any detectable phenotypes in the injected P0 animals or in any of the injected animals’ progeny (data not shown). Injection of dsRNA encoding Arl-2/evl-20 caused 25% of the F1 offspring produced 6–24 h after injection to arrest during embryogenesis; this lethality increased to 100% in embryos laid after 24 h postinjection. The Arl-2/evl-20(RNAi) F1 offspring from the earlier part of the brood that survived grew up sterile. These animals exhibited a variable disorganized germline phenotype, including unreflexed gonadal arms (Fig. 5
). A protruding vulva defect was observed in a fraction of these survivors (not shown). The germ cells often appeared to have defects in cellularization, forming regions with numerous nuclei sharing a complete cytoplasm (as opposed to the partial syncytium normally observed). After 24 h, all offspring of Arl-2 dsRNA-injected animals arrested as embryos. These dead embryos were found to arrest at a very early stage with little evidence of differentiation observed (see Fig. 5C, D
).
|
The apparent cellularization defect in the germline of surviving Arl2/evl-20(RNAi) animals was further investigated. Ovaries were dissected from offspring of mock-injected and Arl-2/evl-20(RNAi) animals before fixing and staining DNA with DAPI for viewing by fluorescence microscopy. We observed far fewer developing germ cells in ovaries from Arl-2/evl-20(RNAi) animals compared with mock-injected controls (Fig. 6
A vs. D). When chromosomes in diakinesis stage oocytes from each ovary were viewed at higher magnification (Fig. 6B, E
), it was evident that oocyte nuclei in Arl-2(RNAi) animals often contained many more chromosome figures than the six bivalents observed in control oocytes. The loss of physically joined bivalents that normally occurs as a result of meiotic recombination and formation of chiasmata indicates additional defects in meiosis-specific processes that may not be related to the mitotic defects. Seventeen or more chromosome figures were often observed in the defective oocytes (Fig. 6E
, arrows). Polyploid nuclei were observed in nuclei of earlier meiotic prophase I germ cells, indicating that the ploidy defect likely arose from poor segregation during mitotic division, which precedes entry into meiosis (not shown).
|
The DNA content and distribution were examined in later brood cohort embryos from control and Arl-2(RNAi) animals (Fig. 6C, F
). Controls revealed a normal DAPI staining pattern and distribution in all cells (Fig. 6C
). In contrast, embryos from Arl-2(RNAi) animals often contained clumps of large amounts of DNA, as well as cells that lacked DAPI staining as determined by DIC and fluorescence microscopy (Fig. 6F
and data not shown). These results indicate that cytokinesis was defective in the germline and embryos as a result of the loss of Arl2 expression.
Previous data from C. elegans and other model genetic systems have linked Arl2 function to that of microtubules, so two additional experiments were performed to investigate defects in this cytoskeletal protein as a possible explanation for the Arl-2(RNAi) phenotype. Transgenic lines expressing GFP fused to the C terminus of ß-tubulin were injected with Arl-2 dsRNA. Much less GFP signal was evident in embryos from Arl-2(RNAi) animals and very few mitotic spindles in addition to the accumulation of polyploid nuclei in only a few locations (Fig. 7
, compare panel A to B). We asked whether there was a corollary decrease in endogenous tubulin similar to that seen for the tubulin-GFP expressed from the transgene by staining fixed embryos with tubulin antibodies. A clear decrease in staining was likewise observed using tubulin antibodies in the Arl2 depleted embryos (Fig. 7
, compare panels C and D). These data are consistent with evidence from multiple model organisms that suggest a required role for Arl2 proteins in the assembly or regulation of microtubules, particularly during mitosis and cytokinesis.
|
Arl2 function has been linked to that of cofactor D, which functions as a cochaperone in the folding of tubulin and formation of the mature heterodimeric tubulin. Genetic screens in both S. cerevisae (34
, 35)
and S. pombe (36)
identified both the Arl2 and cofactor D orthologs and mutations in each yielded similar phenotypes in these yeasts. For these reasons, we asked if the phenotypes described above for arl-2 (RNAi) may be linked to defects in tubulin biosynthesis and may therefore be phenocopied by knockdown in expression of the worm otholog of cofactor D. A PBLAST search of the worm protein database with the full-length human cofactor D sequence yielded a single hit, with a probability value of 6.7e-83. The predicted protein, F16D3.4, shares 28.4% identity overall with human cofactor D over 1024 residues (the human and worm proteins are 1192 and 1232 residues in length, respectively). By contrast, human and bovine cofactor D share 80% identity. dsRNA encoding F16D3.4 was prepared and injected into N2 animals, as described in Methods. Progeny produced within 6–24 h after injection with dsRNA grew to sterile adults (90%; 124/137) or were dead embros (10%; 13/137). Sterile adults had clearly observible morphological defects in the germline and produced no progeny. They also all exhibited everted vulva. Embryos laid 24–48 h post-injection were all dead (99%; 278/281); arrested during embryogenesis (see Fig. 7H
). Embryos depleted of F16D3.4 were stained for tubulin and DNA (Fig. 7E
) and found to exhibit large clumps of DNA and reduced tubulin levels, very comparable to those seen in arl-2(RNAi) embryos. These results are consistent with ARL-2 and F16D3.4 having very similar functions in animals, having to do with tubulin and microtubules.
Arl2 from S. cerevisiae (Cin4) and S. pombe (Alp41) have been shown to be capable of affecting microtubule-mediated processes and likely perform highly related or identical functions, yet Arl2 (Cin4) is not an essential gene in S. cerevisiae (34)
whereas it (ALP41) is in S. pombe (36)
. This apparent role for Arl2 orthologs in tubulin folding is likely conserved in mammals (39)
and should therefore affect every cell in those organisms. The phenotypes of Arl-2(RNAi) in C. elegans may be quite similar to those described (37)
for mutations in TITAN5, the Arabidopsis ortholog of Arl2. These include blockage of endosperm cellularization, decreased numbers of germ cells, and defects in embryonic development, and could be explained by changes in tubulin folding and resulting changes in microtubule dynamics. The germ cell defects and embryonic arrest phenotype observed in Arl-2(RNAi) animals are consistent with problems with mitotic spindle regulation.
Depletion of Arl-2 message and protein in C. elegans by RNAi resulted in sterility in the earliest affected offspring (embryos laid 6–24 h post-dsRNA injection), with early embryonic arrest in the later portion of the brood (embryos laid >24 h postinjection). The 6–24 h brood probably received a sufficient load of maternal RNA or protein to allow completion of embryogenesis, and perhaps grew up sterile because the germline is simply the most sensitive tissue to depletion of maternal Arl-2 message, as has been observed with numerous essential genes (W. Kelly, unpublished observation). The early embryonic arrest in the earliest affected animals may be more indicative of the null phenotype in which maternal and zygotic Arl-2 components have been depleted. The cytokinesis defects observed and the marked depletion of tubulin in these embryos strongly suggest a conservation of the role of Arl-2 in tubulin dynamics in C. elegans.
These results are similar to those reported for the evl-20 mutation in C. elegans, which was shown to be a mutation in the arl-2 locus (38)
. In those studies, however, no germline defects in the offspring of dsRNA-injected animals were found; but analysis was limited to offspring >24 h postinjection, at which point the germline defects were not evident in our experiments. They also concluded, based on germline mosaic studies, that evl-20 (Arl-2) was not required for any process in the germline. Our results directly conflict with this conclusion. One possibility is that the mosaic analysis as performed cannot rule out that maternal contribution (perhaps perdurance of maternally synthesized mRNA) allows the mosaic animals to grow up fertile, but their offspring, which do not inherit maternal products, die. The RNAi-induced defects we observe in very early cleavage embryos are consistent with obliteration of maternal products, since zygotic transcription is not required for cellular proliferation until approximately the 100 cell stage (84)
. Because RNAi targets maternal and zygotic mRNAs, the early brood sterility we observe is strong evidence that the evl-20 (Arl-2) gene product has an essential function in germline and soma.
We describe expression patterns for five of the six GTPases studied based on the insertion of the GFP reporter either as a C-terminal fusion or replacing the GTPase (Arf-1::GFP and Arf-6::GFP). The results are summarized in Table 6
and indicate those tissues expressing the highest levels of each protein. Because each of these reporter constructs is retained as extrachromosomal arrays, expression levels from these transgenes likely differ from those of the endogenous loci. Because each promoter has not been rigorously characterized, it is possible that regulatory elements could be missing. Nevertheless, these lines give an indication of the differences in the expression patterns of the different family members that agrees well with expression studies from other organisms: high levels of expression of each GTPase were seen in neuronal cells (either ventral nerve cord or head and tail neurons), the same cells expressing the highest levels of each Arf and at least some Arls in mammals. Because all three classes of Arfs and Arls have been found to be expressed in every tissue and cell line examined previously in other organisms and were found to be widely expressed in the studies reported here, it is likely that every C. elegans cell expresses all six of the Arf family members and the patterns observed represent those with the highest levels of expression.
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
and Arl2 (see above; ref 38
) have distinct requirements in development suggests that conservation of biochemical function in vitro does not fully reflect the biological role of classes of proteins such as the Arfs and Arls. The identification of mutants in the Arf family members in C. elegans will add considerably to our understanding of the role(s) of these regulatory proteins in development. We have tested many likely candidate mutants mapping to the Arf-1 and Arf-3 regions, but mutations have not been found in either gene. This indicates that the path ahead involves genetic screens, which is one of the strengths of C. elegans as a model organism. This work was supported by grants from the National Institutes of Health (Y.L. and R.A.K., GM67226 and GM68029. We thank Steve L’Hernault for his support and the use of his equipment and reagents during the course of this work and Sue Jinks-Robertson for her support of this work.
|
|
FOOTNOTES |
|---|
2 Current address: Department of Molecular Genetics and Microbiology, University of Florida College of Medicine, Gainesville, FL 32610-0266, USA.
Received for publication May 19, 2004. Accepted for publication August 5, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. M. Clark, L. D. Goldstein, M. Tevlin, S. Tavare, S. Shaham, and E. A. Miska The microRNA miR-124 controls gene expression in the sensory nervous system of Caenorhabditis elegans Nucleic Acids Res., February 21, 2010; (2010) gkq083v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. A. Cunningham and R. A. Kahn Cofactor D Functions as a Centrosomal Protein and Is Required for the Recruitment of the {gamma}-Tubulin Ring Complex at Centrosomes and Organization of the Mitotic Spindle J. Biol. Chem., March 14, 2008; 283(11): 7155 - 7165. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Bowzard, D. Cheng, J. Peng, and R. A. Kahn ELMOD2 Is an Arl2 GTPase-activating Protein That Also Acts on Arfs J. Biol. Chem., June 15, 2007; 282(24): 17568 - 17580. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Zhou, L. Cunningham, A. I. Marcus, Y. Li, and R. A. Kahn Arl2 and Arl3 Regulate Different Microtubule-dependent Processes Mol. Biol. Cell, May 1, 2006; 17(5): 2476 - 2487. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Hofmann and S. Munro An N-terminally acetylated Arf-like GTPase is localised to lysosomes and affects their motility J. Cell Sci., April 15, 2006; 119(8): 1494 - 1503. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. A. Kahn, J. Cherfils, M. Elias, R. C. Lovering, S. Munro, and A. Schurmann Nomenclature for the human Arf family of GTP-binding proteins: ARF, ARL, and SAR proteins J. Cell Biol., February 27, 2006; 172(5): 645 - 650. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Lay, B. L. Grosshans, H. Heid, K. Gorgas, and W. W. Just Binding and Functions of ADP-ribosylation Factor on Mammalian and Yeast Peroxisomes J. Biol. Chem., October 14, 2005; 280(41): 34489 - 34499. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. A. Volpicelli-Daley, Y. Li, C.-J. Zhang, and R. A. Kahn Isoform-selective Effects of the Depletion of ADP-Ribosylation Factors 1-5 on Membrane Traffic Mol. Biol. Cell, October 1, 2005; 16(10): 4495 - 4508. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
24 h postinjection exhibited an early arrest and failed to hatch. A–D) Ovaries dissected from 6–24 h adult hermaphrodite offspring, as described in Materials and Methods. C, D) Higher magnifications of diakinetic chromosomes in individual oocytes in control and Arl-2 RNAi offspring, respectively. D) Arrows point to oocyte nuclei with unusually high numbers of chromosome figures. A DAPI-stained embryo laid 24 h after control (E) or Arl-2 (F) dsRNA injections. F) Arrows point to large polyploid nuclei and asterisks indicate large cellularized areas that appear to lack nuclear DNA.
