| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Brain Research Institute, University of Zurich, and Department of Biology, ETH Zurich, 8057 Zurich, Switzerland; and
* Department Biology, University of Konstanz, 78457 Konstanz, Germany
2Correspondence: Brain Research Institute, University of Zurich and ETH Zurich, Winterthurerstr.190, CH-8057 Zurich, Switzerland. E-mail: oertle{at}hifo.unizh.ch
ABSTRACT
Reticulon (RTN) genes code for a family of proteins relatively recently described in higher vertebrates. The four known mammalian paralogues (RTN1, -2, -3, and -4/Nogo) have homologous carboxyl termini with two characteristic large hydrophobic regions. Except for RTN4-A/Nogo-A, thought to be an inhibitor for neurite outgrowth, restricting the regenerative capabilities of the mammalian CNS after injury, the functions of other family members are largely unknown. The overall occurrence of RTNs in different phyla and the evolution of the RTN gene family have hitherto not been analyzed. Here we expound data showing that the RTN family has arisen during early eukaryotic evolution potentially concerted to the establishment of the endomembrane system. Over 250 reticulon-like (RTNL) genes were identified in deeply diverging eukaryotes, fungi, plants, and animals. A systematic nomenclature for all identified family members is introduced. The analysis of exon-intron arrangements and of protein homologies allowed us to isolate key steps in the history of these genes. Our data corroborate the hypothesis that present RTNs evolved from an intron-rich reticulon ancestor mainly by the loss of different introns in diverse phyla. We also present evidence that the exceptionally large RTN4-A-specific exon 3, which harbors a potent neurite growth inhibitory region, may have arisen de novo 
350 MYA during transition to land vertebrates. These data emphasize on the one hand the universal role of reticulons in the eukaryotic system and on the other hand the acquisition of putative new functions through acquirement of novel amino-terminal exons.—Oertle, T., Klinger, M., Stuermer, C. A. O., Schwab, M. E. A reticular rhapsody: phylogenic evolution and nomenclature of the RTN/Nogo gene family
Key Words: reticulon • intron loss • endoplasmic reticulum
THE RTNS are a family of proteins, typically 200-1200 amino acids (aa) in length, described relatively recently in higher vertebrates (1
2
3
4)
. They all share a carboxyl-terminal 150-201 aa reticulon homology domain (RHD; Pfam PF02453) comprising two large hydrophobic regions with an 66 aa loop in between. RTN1 was the first identified member (5)
and is expressed in the central nervous system (CNS) as well as in neuroendocrine cells (1
, 6)
. The functions of RTN1, RTN2, and RTN3 are unknown, yet all (including RTN4/Nogo) are enriched in endoplasmic reticulum membranes (1
, 4
, 7)
.
RTN proteins are ubiquitously expressed in vertebrates and have recently been described in Caenorhabditis elegans (8)
. Their distribution among other metazoan phyla is still entirely undefined.
The fourth vertebrate family member, RTN4/Nogo, gives rise to different isoforms through alternative splicing and alternative promoter usage (9)
. The largest isoform, Nogo-A, has been described as a potent neurite outgrowth inhibitor of CNS myelin (4
, 10
, 11)
. In line with the more oligodendrocyte-restricted expression of Nogo-A (12)
, we have identified a stretch in the Nogo-A-specific region (NiG-
20, aa 544-725) that is strongly inhibitory for neurite outgrowth and cell spreading (13
, 14)
. The loop region between the two carboxyl-terminal hydrophobic domains of the RHD (called Nogo-66) has been shown to induce collapse of neuronal growth cones (10)
through interaction with a GPI-anchored receptor, NgR (15
, 16)
, which simultaneously serves as a receptor for the myelin proteins OMgp (17)
and MAG (18
, 19)
. Since Nogo-66 is present in all known Nogo splice variants, one would anticipate they should all be restrictive for regenerative axons. Nogo-B, for instance, has a rather ubiquitous expression pattern (12
, 20
, 21)
that is not compatible with earlier findings showing that adult CNS myelin is particularly enriched in neurite growth inhibitory factors (22)
. Because the adult CNS of lower vertebrates is able to regenerate after a lesion (23
, 24)
, inhibitory components being largely absent (25
, 26)
, a closer phylogenetic analysis of the nogo gene may elucidate the key evolutionary steps from lower to higher vertebrates required to give birth to its unique growth inhibitory function. In addition to the neurite outgrowth inhibitory character of Nogo, Nogo-B is thought to play a role in apoptosis, particularly in cancer cells (20
, 21)
, although some of the evidence is disputed (27)
.
We describe here the identification of >250 RTN and RTNL genes in eukaryotes. We present a systematic nomenclature for all known members of the RTN family. Knowledge of the biological role of a shared protein in one organism can certainly illuminate, and often provide strong evidence of, its role in other organisms. Our analysis is intended not only to elucidate the evolutionary relationships between RTNs, but also to give insight into those residues that play critical roles for RTN structure and function.
MATERIALS AND METHODS
Sequences
Sequences were obtained from GenBank databases (Supplementary Tables). From ESTs, contiguous sequences were established using SeqManTMII (DnaStar) and noncomputational analyses. Exon-intron structures were examined by comparing genomic sequences against cDNA sequences. Where possible, the GT-AG rule of splice donor and acceptor sites was respected.
Nomenclature
The RTN family nomenclature was performed based on the homology of the sequences (see Results). The suggested nomenclature has been approved by the HUGO Gene Nomenclature Committee, the Mouse Genomic Nomenclature Committee, the Zebrafish Nomenclature Committee, the Saccharomyces Genome Database, and the Schizosaccharomyces pombe Genome Project.
Sequence analysis
Hydrophobicity plots were created using DAS (http://www.sbc.su.se/-miklos/DAS), TMHMM (http://www.cbs.dtu.dk/services/TMHMM-2.0/), TMPred (http://www.ch.embnet.org/software/TMPRED_form.html), and TopPred (http://bioweb.pasteur.fr/seqanal/interfaces/toppred.html) using the KD, GVH, and GES scale. A conservation plot (inverted Wu-Kabat variability plot) was calculated manually based on the protein sequences aligned with ClustalW (28)
and plotted using Prism. Sequence logos have been generated using GENIO/logo (http://genio.informatik.uni-stuttgart.de/GENIO/logo/logo.cgi). Bootstrap proportions and Neighbor-Joining Trees of protein sequences have been calculated using ClustalX (v.1.8),with the default parameters and ClustalW. Cladograms and trees have been drawn using TreeView (v.1.6).
RESULTS AND DISCUSSION
In silico and physical cloning of RTN and RTNL proteins
Using the sequences of known RTN proteins, i.e., their carboxyl-terminal RHD, we searched the public databases for homologous proteins, mRNAs, genomic sequences, and ESTs. We physically cloned RTN members from R. norvegicus, B. taurus, G. gallus, X. laevis, C. auratus, and D. rerio cDNA. From a total of 81 mRNA sequences, 48 genomic sequences, 500 genomic traces, and 7500 ESTs, we constructed contiguous strings encoding the entire or partial sequences of 351 mRNA/proteins and isoforms arising from 272 bona fide independent genes from 94 different species (Suppl. Tables). Homologous genes could neither be identified in archaeal or bacterial species nor in the microsporid Enzephalitozoon cuniculi.
These results show that RTN-like genes are present in most eukarya and particularly in plants, where multiple duplication events led to an array of RTN paralogues in the same plant species (see Suppl. Table F).
Nomenclature
The reticulon gene family is based on a cellular component ontology, i.e., proteins are considered to be members of the family if they contain at least part of an identifiable RHD (Pfam PF02453, http://www.sanger.ac.uk/cgi-bin/Pfam/getacc?PF02453) characterized by two large (>30 aa) hydrophobic regions and an 60 aa spacer between the two regions (see Fig. 5A
). The two hydrophobic stretches are responsible for the association of RTN proteins to membranes, in particular those of the endoplasmic reticulum (27)
. The "functional" motif of RTNs and RTN-like proteins is therefore the RHD. A standardization of the RTN nomenclature should help in the identification and understanding of new gene products and facilitate communication between scientists in different disciplines. We suggest that "RTN" serves as the gene symbol for chordate reticulons and "RTNL" for homologous proteins in nonchordate taxa. RTN-like genes in nonchordate metazoans are designated as RTNL, in fungi as RTNLA, in plants as RTNLB, and in protists as RTNLC (Fig. 1
; see Supporting Material on http://www.fasebj.org for further details on the nomenclature guidelines).
|
|
These data are the basis for the construction of a consistent expandable RTN nomenclature.
Phylogenic analysis
On the basis of the alignment of the RHD sequences from 226 distinct genes and 80 different species, a phylogenetic tree of the reticulon family has been inferred. There is an excellent relationship between the clustering pattern of members in the family tree and taxonomic classification (Fig. 1)
. Similar to reticulons, many endomembrane-associated genes have been found in most eukaryotic genomes but are absent in bacteria (29)
.
In the following we will discuss the different subfamilies. First, we will present the RTN genes and their splice isoforms. Second, important evolutionary steps that happened within a subfamily are discussed. Third, we will briefly review known functional data of these RTN genes (e.g., protein interaction partners), with the exception of the vertebrate reticulons, where a large body of publications already covers these facets. Finally, evolutionary features of the entire gene family as well as highly conserved residues will be exposed.
Chordate RTN subfamily
There are four major vertebrate reticulon paralogues: the previously identified rtn1, rtn2, rtn3, and rtn4/nogo genes (Fig. 2A
). All paralogues in fish, amphibians, and mammals have six carboxyl-terminal exons encoding the RHD (Fig. 2C
, see Fig. 6
; M. Klinger et al., unpublished results).
|
|
Mammalian rtn1 is formed by at least 10 exons and can give rise to 3 isoforms (RTN1-A, -B, and -C) by alternative promoter usage (Fig. 2C
; Suppl. Table A; 30). Rtn1 has been duplicated probably in euteleosts (generating the paralogue rtn5).
There are three known RTN2 mRNA variants (2)
derived from 11 different exons (Fig. 2C
). Our analysis has revealed three additional minor splice variants in humans (Suppl. Table A). RTN2 is the most divergent of the mammalian tetralogues, with 36 nucleotide insertions in exon 11.
The nine exons of human rtn3 can give rise to at least five isoforms (Fig. 2C
, Suppl. Table A), from which only RTN3-A1 has so far been described in the literature (3
, 7)
. As previously published (3)
, a RTN3 pseudogene exists in humans on chromosome 4. Since no intron sequences are present, this pseudogene represents a retronuon of the RTN3-A1 transcript (Suppl. Fig. 1
).
Human rtn4 is formed of 14 exons that can give rise to at least 10 splice variants (Fig. 2C
, Suppl. Table A; 9). There is good evidence that rtn4 has been duplicated in X. laevis (M. Klinger et al., unpublished results) while being present as a single gene in S. tropicalis, in conformity with the ancestral diploid karyotype of the genus.
In the urochordate C. intestinalis we were able to ascertain only a single RTN gene, which can give rise to at least six different isoforms originating from alternative splicing and the use of three alternative promoters (Fig. 2C
; Suppl. Table A). The RHD of C. intestinalis is encoded by only four exons.
Vertebrate RTN1-4 are present on four different chromosomal loci, have conserved intron-exon boundaries and intron phasing of the RHD-encoding region (1-2-0-1-0-2), and are more similar to each other than to invertebrate RTN members. The presence of a single bona fide RTN gene in invertebrates and the presence of the vertebrate tetralogues are in accordance with the hypothesis of genome quadruplication in amphioxus-like animals 760-530 MYA before the cyclostome/gnathostome split (31
, 32)
. A reservation remains for the origin of RTN2. Indeed, we were not able to identify any nonmammalian RTN2 sequences. This could mean that 1) RTN2 is a recent duplication during the evolution of mammals followed by rapid sequence divergence or 2) RTN2 is present as a pseudogene (or has been secondarily lost) in lower vertebrates and has been retained or exempted in mammals.
Between the four vertebrate paralogues, no homology of the amino-terminal exons could be identified. The divergence of the amino-terminal exons within the same orthologs is also higher compared with the RHD exons, demonstrating nonhomogeneous evolution of the genes (Fig. 2B
). Homologous ESTs to the amino-terminal exons of rtn1 could be identified down to D. rerio, whereas ESTs homologous to the amino-terminal exon 1 of nogo were found in S. tropicalis and X. laevis (Fig. 2B
). Homologous exons to the huge 2400 bp long, Nogo-A-specific exon 3 that encodes the neurite outgrowth and cell spreading inhibitory region of this protein could be found in X. laevis, but not in fish (Fig. 2B
). Thus, some of the amino-terminal exons seem to have arisen 360 MYA during the transition to land vertebrates or have not yet been identified in the fish genomes. However, there are also more recent gains of exons, e.g., exon 1F of the rtn4/nogo gene has been newly acquired in primates after their divergence from rodents (100 MYA) (9)
.
These results demonstrate that all vertebrate RTN paralogues originated by gene duplication from an ancestor with the same RHD gene structure (see below).
Insect RTN-like subfamily
Two RTN-like genes can be found in the D. melanogaster genome. The rtnl1 gene is formed of 12 exons that can give rise to at least 11 splice variants (Fig. 3A
, Suppl. Table B). The RHD is formed by four carboxyl-terminal exons that are conserved in the Anopheles gambiae ortholog (Fig. 3A
).
|
A rtnl2 gene (formerly tropomyosin-like gene) has not been identified in the Anopheles genome and no ESTs originating from rtnl2 could be found in public databases. Since the gene is in a single exon (thus lacking the typical reticulon exon-intron structure of insects) and its homology is relatively low to the other insect homologues, we speculate that this is a retronuon with pseudogene character.
(DROME)Rtnl1 was trapped using a GFP-carried P1 element (33) (
it has not been recognized as a reticulon protein by the authors and was called G9). The given insertion point of the transposon would be exon 1D, according to our data, and could potentially be spliced into the -A, -B, -C isoforms of the gene. Based on these data these transcripts are expressed ubiquitously in early embryonic development and become CNS specific late in embryogenesis (34)
. The fusion protein is localized to the ER (33)
and during cell division is in a ring-like structure that surrounds the spindle rotating with it during neuroblasts division (34)
.
Nematode RTN-like subfamily
In the C. elegans genome, a single RTN-like gene could be identified with 11 exons giving rise to at least 5 isoforms originating from 3 alternative promoter sites (Fig. 3B
). Orthologous RTNL sequences were identified in a number of other nematodes (Suppl. Fig. 2b
; Suppl. Table C). The RHD is formed by three carboxyl-terminal exons (Fig. 3B
).
(CAEEL)RTNL-C (formerly nRTN-C) has been shown to be associated with ER membranes (8)
. The protein is detectable almost exclusively during the late gastrulation stage of the embryo whereas mRNA is present in developing gonads and oocytes of larvae (8)
. The amino-terminal region specific for RTNL-C interacts with RME-1 (8)
, a protein localized to the endocytic recycling compartment (ERC) and involved in the exit of proteins from ERC to the plasma membrane or to the trans-Golgi network (35)
.
Platyhelminth RTN-like subfamily
We were able to identify RTNL ESTs in S. mansoni, a parasitic digenetic trematode belonging to the family of blood flukes (Suppl. Table D).
Although some vaccines against schistosomiasis are currently in human trials, none promises high efficacy so that the need for the identification of effective antigens for vaccination is strong (36)
. Since (LEIIN)RTNLC has been successfully used as a vaccine against visceral leishmaniasis in hamster models (see below), it would be of interest to envision vaccination with (SCHMA)RTNL against schistosomiasis in order to lower the TH2 response of chronic infection.
Fungi RTN-like subfamily (RTNLA)
Despite other reports (8)
, RTN-like proteins could be identified in the genome of Saccharomyces cerevisiae as well as of S. pombe (Fig. 1
, Suppl. Fig. 2C
). In both cases single, intron-less genes encode the proteins (see Fig. 6
). S. cerevisiae is the only fungus where we identified two separate genes, suggestive of a separate gene duplication event consistent with previous findings (37)
. Orthologs were identified in different fungi, but the divergence between them is very high (Suppl. Fig. 2C
, Suppl. Table E). The homology between (YEAST)RTN1 and (SCHPO)rtn is only 25%, which is comparable to the homology between (YEAST)RTN1 and vertebrate reticulons. This is not uncommon (38)
; there is evidence that yeasts are generally fast-evolving organisms (39)
and that S. cerevisiae and S. pombe had already diverged 420-330 MYA (38)
.
Overexpression of (SCHPO)rtn (formerly cwl1+) has been shown to reduce the cell wall content (20%), inhibit cell separation after division in the absence of thiamine, and cause cell lysis in the absence of sorbitol (40)
.
Fission and budding yeasts are viable and have no obvious phenotype after gene disruption (40
, 41)
.
(YEAST)RTN2 is induced by the G-protein regulator Sst2 probably via a 5' stress response element in its promoter region (42)
and is generally up-regulated in response to stressors (http://www.transcriptome.ens.fr/ymgv/).
It has been shown that (YEAST)RTN1 interacts with YOR285w and YDL089w, two genes of unknown function (43)
. Furthermore, RTN1 has been shown to interact with the H3/H4 histone acetyltransferase HAT2 (YEL056w), a yeast homologue of the human retinoblastoma binding protein 4 (44)
. (YEAST)RTN2 interacts with the inositol-1,4,5-trisposphate 5-phosphatase INP54, with the osmosensor SHO1, with karyopherin KAP104, with SEC27 involved in the ER-to-Golgi protein trafficking, and with YMR110c, an aldehyde dehydrogenase type III homologue (43
, 44)
.
In conclusion, our data present not only new members of the RTN family but also new interaction partners pointing to possible functions.
Plant RTN-like subfamily (RTNLB)
In plants we were able to identify >180 members of the RTNLB subfamily (Suppl. Fig. 3
; Suppl. Table F). A. thaliana has at least 15 RTNLB genes with conserved exon-intron structure and intron phasing (Fig. 4
). The RHD domain for Arabidopsis and Oryza sativa RTNLB genes is usually formed by four exons (generally exon 2-5) and the conserved intron phasing is 1-2-0-1. We have, however, identified some exceptions. In (ARATH)rtnlb5 intron IV is absent whereas in (ARATH)rtnlb12 and (ARATH)rtnlb13 intron III is absent and the two exons are fused. In (ARATH)rtnlb6 exons 3 and 4 are nonhomologous to the other paralogues, i.e., not encoding a RHD domain. The phasing of intron III in this gene is indeed different and therefore might represent an example of exon shuffling. (ARATH)rtnlb14 lacks the last exon and (ARATH)rtnl15 lacks the first two exons of the RHD (Fig. 4)
. At least in two of the A. thaliana RTNLB genes (10 and 11) there is an additional, noncoding exon located 3' to the four RHD encoding ones. The phylogenetically most ancient group is probably represented by the rtnlb10-12/18 orthologs, since they are the closest homologues of the fern C. richardii (rtnlb34-35) and the moss P. patens (rtnlb36-40) reticulons (Fig. 1
; Suppl. Fig. 3
).
|
The majority of the RTNLB genes have a single amino-terminal exon with relatively low homology between the members.
RTN-like proteins in early branching eukaryotes (RTNLC)
Besides the three eukaryotic kingdoms of metazoans, fungi, and plants, we have identified RTNL genes in euglenozoa (Leishmania infantum and major), pelobiontida (Mastigamoeba balamuthi), and mycetozoa (Dictyostelium discoideum) (Fig. 1
; Suppl. Table G). Of these three subfamilies, (DICDI)RTNLC and (MASBA)RTNLC have the closest relationship in concordance with the literature (45)
. It is noteworthy that Mastigamoeba lack recognizable Golgi bodies. (LEIMA)RTNLC is encoded by a single, intron-less exon whereas (DICDI)RTNLC is encoded by three exons (see Fig. 5
).
(LEIIN)RTNLC (formerly papLe22) has recently been described as a protein that potentially aggravates visceral leishmaniasis, stimulates IL-10 release from patient-derived peripheral blood mononuclear cells in vitro, and is found in nuclear fractions of the promastigote form (46)
. The protein itself as well as anti-RTNLC antibodies have been detected in the serum of patients and specific IgG titers correlated with the clinical status of the patients (46)
. Intramuscular vaccination in hamster models with RTNLC cDNA down-regulated the TH2 response after Leishmania infection and reduced episodes of Leishmania circulation by 50% (47)
.
Evolution of the intron-exon structure
Intron loss or gain represents critical steps in the divergence of expression patterns of individual gene family members, since introns are known to host regulatory elements. Intron loss can occur through an mRNA intermediate mechanism as processed retrotranscripts (48)
or can be due to direct genomic deletion (49)
. Figure 5
shows the conservation of the intron-exon structure and intron phasing in different phyla. The most parsimonious scenario would propose an intron-rich reticulon ancestor ("Proto-Reticulon") in early eukaryotes and multiple independent intron losses in distantly related lineages. The ancient exon-intron structure of the proto-reticulon would still be conserved in most plant genes and has been reacquired by urochordates after intron losses (Fig. 5)
. Our proposal would best fit with the intermediate view postulating that introns arose at the initiation of multicellularity. In different phyla, extensive intron loss therefore has to be assumed. However, two episodes of intron gain also have to be claimed in our model. A first phase 2 intron insertion occurred around the time of the fungi/metazoan split. It is probable that the second insertion of the phase 0 intron antedates the tetraploidization event before the fish/tetrapode split (500 MYA; ref 50
) but took place after the urochordata-craniata split (Fig. 5)
. The absence of introns in yeast reticulons would fit with the theory of runaway reverse transcriptase leading to the loss of introns as a result of retroinsertions of spliced messengers (48)
. Similarly, this could be the case for euglenozoa. The overrepresentation of phase 1 and 2 introns with respect to the mean percentage, as well as the overrepresentation of asymmetrical exons, suggests that the proto-reticulon did not arise by classical exon shuffling mechanisms (51
52
53)
. In contrast, the amino-terminal exons probably are newly acquired and are most often symmetrical 1-1 exons. The high rate of divergence of these amino-terminal exons between both paralogues and orthologs suggests a low evolutionary pressure on these regions.
Variability and patterns of sequence conservation
The systematic approach of our study allows us to analyze the conservation of single amino acid residues within the RHD (Fig. 6
). The functional importance of sites is generally believed to be inversely related to the evolutionary rate of amino acid replacement, i.e., sites of greatest functional significance are under the strongest selective constraints. Amino acids conserved in all orthologous sequences would point to the importance of such sites for a putative common function of these genes, whereas amino acids conserved in paralogues but different in orthologs would point to specific relevance of these sites for newly acquired functions. Figure 5B
represents the graph for hyperconserved regions of the chordate RHD. Hyperconserved residues are concentrated in the calculated hydrophobic regions. The second putative transmembrane domain, shown to be de rigueur for ER membrane association of the reticulon proteins (27)
, has highly conserved amino acids (Fig. 6A, B
). Nine specific residues as well as five amino acid types within the canonical pattern of RHD are outstandingly conserved between all species of the various taxa (Fig. 6B
). Two are prolines that might be critical structural residues and five of them are aromatic amino acids that could represent essential functional residues. Proline is one of the most strongly conserved amino acids and occurs often in protein structures where the polypeptide backbone is required to adopt an unusual conformation, like maintenance of short loops (54)
. Pro68RHD could thus be crucial for the correct folding of the 66 aa loop. Future studies mutating these hyperconserved amino acids could provide important insights into structure and function of the reticulons. Since Nogo-66 has been described as an inducer of growth cone collapse (10)
via binding to NgR with residues 1-40 of the loop-region (NEP1-40; 16), and thus to be one of the responsible components for the lack of regeneration in the CNS of higher vertebrates, we wondered whether this unique feature is reflected by particular amino acid divergence with respect to lower vertebrates Nogo-66 and to the other RTN members, respectively. Although significant differences can be noted between Nogo-66 and the other RTN-66 regions, no major changes (i.e., amino acid substitution with different chemicophysical properties) from fish to higher vertebrates Nogo-66 can be prime facie observed that would allow us to spot key mutations responsible for the putative gain of inhibitory activity (Fig. 6C
).
Evolution of Nogo/RTN4 in vertebrates
Concerning the evolution of RTN4/Nogo as a relevant inhibitor for neurite outgrowth, it is of interest to consider whether this function agrees with the data gathered in the present study. Nogo-66, which is part of the RHD, and NgR (M. Klinger et al., unpublished results) are both present down to teleosts, where regeneration does occur. The Nogo-66 region is also present in all known Nogo splice variants and therefore has an almost ubiquitous tissue distribution that strictu sensu does not fit with the postulated constraint of inhibitors to the CNS. Sequence comparison did not permit us to identify key evolutionary mutations that could have allowed Nogo-66 to become a potent inhibitory site in higher vertebrates, although the overall variability of course does not exclude an evolutionary gain of function (Fig. 6)
. Future studies will show whether the orthologous fish Nogo-66 displays binding properties to and inhibitory activity via NgR and its fish ortholog or not. In contrast, the Nogo-A-specific 2400 bp long exon 3 seems to appear during the transition to land vertebrates. Both NiR-2 and NiG-20, shown to be inhibitory for neurite outgrowth and cell spreading (14), so far have no homologous counterpart in fish and are present with weak similarity in amphibians. This would be in concordance with the observation that the lack of regeneration during phylogeny was acquired during amphibian evolution and ontogenetically after metamorphosis (55)
. However, based on ESTs, the amphibian Nogo-A orthologs are expressed in embryos and tadpoles and not only after metamorphosis, similar to the situation in mammals, where Nogo-A is present in the CNS before myelination and thus before the CNS-specific lack of axonal regeneration. Developmental regulation of the membrane topology and surface expression of Nogo-A might explain some of the discrepancies, but more in-depth studies are needed on this issue. If this exceptionally long exon originated de novo in amphibians or was acquired by horizontal gene transfer from a prokaryotic or viral entity cannot be decided at this time.
FINAL REMARKS
This study demonstrates that the evolution of the RTN genes, probably arising during early eukaryote origin from an intron-rich ancestor, has been inhomogeneous. There is no evidence for evolutionary pressure on the highly variable amino termini of all subfamilies, in contrast to the conserved RHD. This is suggestive of a basal function of the common RHD, a function that might be shared by most of the identified members in the distinct taxa. Since the collective feature of RTNs is the association with ER membranes and because the RHD seems to be restricted to the eukaryotic world, we might speculate that the evolution of the RTNs was concomitant with the development of the endomembrane system 1.7 billion years ago and its function is related to it. Based on this hypothesis, the in-depth study of simple model organisms might provide important insight into the role of reticulons in other eukarya, including plants and vertebrates, where the functions of most members are still elusive.
ACKNOWLEDGMENTS
Thanks to Dr. A. Buss for reading the manuscript and special thanks to Dr. H. Wain, HUGO Gene Nomenclature Committee, for excellent support, as well as Dr. L. Maltais (MGNC), Dr. K. Frazer (ZFIN), Dr. V. Wood (S. Pombe Genome Project), and Dr. S. Dwight (SGD) for collaboration on the gene nomenclature. T.O. and M.E.S were supported by the Swiss National Science Foundation (Grant No. 31-63633) and by the NCCR on Neural Plasticity and Repair, C.A.O.S. by the Deutsche Forschungsgemeinschaft (DFG, SPP), and M.K. by the "Stiftung der Deutschen Wirtschaft für Qualifizierung und Kooperation e.V."
FOOTNOTES
1 The nucleotide sequences reported in this paper have been submitted to the GenBankTM/EBI Data Bank with accession numbers AY164656-AY164662, AY164697-AY164996, and AY262091-AY262105. 
Received for publication January 2, 2003. Accepted for publication April 4, 2003.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  W. He, Q. Shi, X. Hu, and R. Yan The Membrane Topology of RTN3 and Its Effect on Binding of RTN3 to BACE1 J. Biol. Chem., October 5, 2007; 282(40): 29144 - 29151. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Lauren, F. Hu, J. Chin, J. Liao, M. S. Airaksinen, and S. M. Strittmatter Characterization of Myelin Ligand Complexes with Neuronal Nogo-66 Receptor Family Members J. Biol. Chem., February 23, 2007; 282(8): 5715 - 5725. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-O. De Craene, J. Coleman, P. Estrada de Martin, M. Pypaert, S. Anderson, J. R. Yates III, S. Ferro-Novick, and P. Novick Rtn1p Is Involved in Structuring the Cortical Endoplasmic Reticulum Mol. Biol. Cell, July 1, 2006; 17(7): 3009 - 3020. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. T. Wang, G. Kodama, P. Baldi, and R. K. Moyzis Global landscape of recent inferred Darwinian selection for Homo sapiens PNAS, January 3, 2006; 103(1): 135 - 140. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Diekmann, M. Klinger, T. Oertle, D. Heinz, H.-M. Pogoda, M. E. Schwab, and C. A. O. Stuermer Analysis of the Reticulon Gene Family Demonstrates the Absence of the Neurite Growth Inhibitor Nogo-A in Fish Mol. Biol. Evol., August 1, 2005; 22(8): 1635 - 1648. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Geng, M. E. Shin, P. M. Gilbert, R. N. Collins, and C. G. Burd Saccharomyces cerevisiae Rab-GDI Displacement Factor Ortholog Yip3p Forms Distinct Complexes with the Ypt1 Rab GTPase and the Reticulon Rtn1p Eukaryot. Cell, July 1, 2005; 4(7): 1166 - 1174. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. A. Dodd, B. Niederoest, S. Bloechlinger, L. Dupuis, J.-P. Loeffler, and M. E. Schwab Nogo-A, -B, and -C Are Found on the Cell Surface and Interact Together in Many Different Cell Types J. Biol. Chem., April 1, 2005; 280(13): 12494 - 12502. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. A. Eckardt Host Proteins Guide Agrobacterium-Mediated Plant Transformation PLANT CELL, November 1, 2004; 16(11): 2837 - 2839. [Full Text] [PDF]
|
|
|
|
|
|
H.-H. Hwang and S. B. Gelvin Plant Proteins That Interact with VirB2, the Agrobacterium tumefaciens Pilin Protein, Mediate Plant Transformation PLANT CELL, November 1, 2004; 16(11): 3148 - 3167. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Klinger, J. S. Taylor, T. Oertle, M. E. Schwab, C. A. O. Stuermer, and H. Diekmann Identification of Nogo-66 Receptor (NgR) and Homologous Genes in Fish Mol. Biol. Evol., January 1, 2004; 21(1): 76 - 85. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
) and intron losses (
) are itemized.
=hydrophobic residue; 
=aromatic residue). C) Sequence logos of vertebrate 66-loop region. NEP1-40 is the peptide shown to bind to NgR, particularly via residues 1-10 and 30-33, as indicated by dotted lines. Peptide4/NEP31-55 has been shown to be a relatively weak inducer of growth cone collapse despite its lack of binding affinity to NgR. This region is thought to be mainly responsible for the observed inhibitory properties of Nogo-66. Logos for hv-Nogo-66 (10 sequences), lv-Nogo-66 (3 sequences), Nogo-66 (13 sequences), RTN1-66 (9 sequences), RTN2-66 (4 sequences), and RTN3-66 (10 sequences) are shown (hv=higher vertebrates, i.e., amphibian to mammalian sequences; lv=lower vertebrates, i.e., fish sequences).
