Emergence and evolution of Zfp36l3

Abstract

In most mammals, the Zfp36 gene family consists of three conserved members, with a fourth member, Zfp36l3, present only in rodents. The ZFP36 proteins regulate post-transcriptional gene expression at the level of mRNA stability in organisms from humans to yeasts, and appear to be expressed in all major groups of eukaryotes. In Mus musculus, Zfp36l3 expression is limited to the placenta and yolk sac, and is important for overall fecundity. We sequenced the Zfp36l3 gene from more than 20 representative species, from members of the Muridae, Cricetidae and Nesomyidae families. Zfp36l3 was not present in Dipodidae, or any families that branched earlier, indicating that this gene is exclusive to the Muroidea superfamily. We provide evidence that Zfp36l3 arose by retrotransposition of an mRNA encoded by a related gene, Zfp36l2 into an ancestral rodent X chromosome. Zfp36l3 has evolved rapidly since its origin, and numerous modifications have developed, including variations in start codon utilization, de novo intron formation by mechanisms including a nested retrotransposition, and the insertion of distinct repetitive regions. One of these repeat regions, a long alanine rich-sequence, is responsible for the full-time cytoplasmic localization of Mus musculus ZFP36L3. In contrast, this repeat sequence is lacking in Peromyscus maniculatus ZFP36L3, and this protein contains a novel nuclear export sequence that controls shuttling between the nucleus and cytosol. Zfp36l3 is an example of a recently acquired, rapidly evolving gene, and its various orthologues illustrate several different mechanisms by which new genes emerge and evolve.

Keywords: Gene evolution; Intronization; New gene formation; Repetitive element expansion; Retrotransposon; Subcellular localization.

FIG. 1. Phylogenetic tree showing the evolutionary history of Zfp36l3

The presence of the gene was mapped using parsimony on a time-calibrated chronogram pruned from the 300-species phylogeny of Schenk et al. (2013). Black branches indicate presence of Zfp36l3, and white branches absence. The arrow points to the branch in the tree on which Zfp36l3 is reconstructed to have evolved, between 23.7 and 37.8 million years ago.

FIG. 2. Acquisition of Rpl37 like sequences and gain of an intron in Mus musculus

(A) Northern blot showing expression of Zfp36l3 mRNA in Mus musculus and Rattus norvegicus placentas. The blots were hybridized with random primed, ³²P dCTP-labeled DNA probes derived from the entire 2178 b of the Mus musculus open reading frame. Arrows indicate the positions of the 18S and 28S rRNAs at approximately 1.8 and 4.7 kb, respectively. (B) Diagram showing elements of the Rpl37-like sequence and the resulting intron in Mus musculus Zfp36l3. The presence of nearly identical flanking direct repeats, and remnants of a poly A tail, indicate that the Rpl37-like sequence was derived by retrotransposition. Note that the poly A tail remnant is shown here as poly T, because the sequence has been inserted into Zfp36l3 in reverse orientation relative to the message sense. The splice donor site is located 18 base pairs after the Zfp36l3 stop codon, and the splice acceptor site is provided by the Rpl37-like sequence. (C) Alignment of the potential splice donor and acceptor sites of the intron found in Mus musculus Zfp36l3, and the genes from the other species that have Rpl37-like sequences. We were unable to obtain the sequence of Arvicanthis niloticus Zfp36l3 in the region containing the splice acceptor site, but based on its position in the phylogeny, it is predicted to possess the Rpl37-like sequences and presumably the intron.

FIG. 3. Alignments of key protein domains in rodent ZFP36L2 and ZFP36L3

Shown are portions of several rodent ZFP36L2 proteins, and the analogous regions of the ZFP36L3 protein sequences discussed in this paper. We excluded ZFP36L3 sequences from Peromyscus polionotus and various members of the Rattus family, because of very high similarities to Peromyscus maniculatus and Rattus norvegicus, respectively. The original alignment was made using the entire proteins, with Clustal Omega and its default settings, and the three regions of the proteins shown are from the original alignment of the full-length proteins shown in Fig. S2. Species names in blue type are members of Muridae family; red are Cricetidae; the single green species is from Nesomyidae, and in black are two representative rodent species that do not have Zfp36l3. L2 refers to ZFP36L2; all other sequences are from ZFP36L3 from various species. The asterisks indicate amino acid identity at that site; the colons indicate amino acid similarity; and the periods indicate lesser similarity, according to the Clustal conventions. The location of the single intron in mouse and rat Zfp36l2 is indicated by the arrowhead in A. The locations of the “diagnostic” P and D residues in ZFP36L3 are indicated in B and C, respectively, and the probable NOT1 binding site near the C-terminus in is indicated in C (NOT1 BD). In C, the shaded residues represent probable phosphorylation sites in Mus musculus ZFP36L3, except that, using the numbering system in NP_001009549.1, only two of the four serines highlighted at positions 694–697 are phosphorylated, and probably only one of the two residues at 718 and 721 is phosphorylated. See section 3.52 for further details.

FIG. 3. Alignments of key protein domains in rodent ZFP36L2 and ZFP36L3

Shown are portions of several rodent ZFP36L2 proteins, and the analogous regions of the ZFP36L3 protein sequences discussed in this paper. We excluded ZFP36L3 sequences from Peromyscus polionotus and various members of the Rattus family, because of very high similarities to Peromyscus maniculatus and Rattus norvegicus, respectively. The original alignment was made using the entire proteins, with Clustal Omega and its default settings, and the three regions of the proteins shown are from the original alignment of the full-length proteins shown in Fig. S2. Species names in blue type are members of Muridae family; red are Cricetidae; the single green species is from Nesomyidae, and in black are two representative rodent species that do not have Zfp36l3. L2 refers to ZFP36L2; all other sequences are from ZFP36L3 from various species. The asterisks indicate amino acid identity at that site; the colons indicate amino acid similarity; and the periods indicate lesser similarity, according to the Clustal conventions. The location of the single intron in mouse and rat Zfp36l2 is indicated by the arrowhead in A. The locations of the “diagnostic” P and D residues in ZFP36L3 are indicated in B and C, respectively, and the probable NOT1 binding site near the C-terminus in is indicated in C (NOT1 BD). In C, the shaded residues represent probable phosphorylation sites in Mus musculus ZFP36L3, except that, using the numbering system in NP_001009549.1, only two of the four serines highlighted at positions 694–697 are phosphorylated, and probably only one of the two residues at 718 and 721 is phosphorylated. See section 3.52 for further details.

FIG. 3. Alignments of key protein domains in rodent ZFP36L2 and ZFP36L3

Shown are portions of several rodent ZFP36L2 proteins, and the analogous regions of the ZFP36L3 protein sequences discussed in this paper. We excluded ZFP36L3 sequences from Peromyscus polionotus and various members of the Rattus family, because of very high similarities to Peromyscus maniculatus and Rattus norvegicus, respectively. The original alignment was made using the entire proteins, with Clustal Omega and its default settings, and the three regions of the proteins shown are from the original alignment of the full-length proteins shown in Fig. S2. Species names in blue type are members of Muridae family; red are Cricetidae; the single green species is from Nesomyidae, and in black are two representative rodent species that do not have Zfp36l3. L2 refers to ZFP36L2; all other sequences are from ZFP36L3 from various species. The asterisks indicate amino acid identity at that site; the colons indicate amino acid similarity; and the periods indicate lesser similarity, according to the Clustal conventions. The location of the single intron in mouse and rat Zfp36l2 is indicated by the arrowhead in A. The locations of the “diagnostic” P and D residues in ZFP36L3 are indicated in B and C, respectively, and the probable NOT1 binding site near the C-terminus in is indicated in C (NOT1 BD). In C, the shaded residues represent probable phosphorylation sites in Mus musculus ZFP36L3, except that, using the numbering system in NP_001009549.1, only two of the four serines highlighted at positions 694–697 are phosphorylated, and probably only one of the two residues at 718 and 721 is phosphorylated. See section 3.52 for further details.

FIG. 4. Subcellular localization of ZFP36L3 from Mus musculus and Peromyscus maniculatus

Plasmids expressing ZFP36L3 proteins were transfected into HEK 293 cells and localized using GFP tags, or by immunostaining, with confocal microscopy. Where indicated, cells were treated with leptomycin B (LMB), an inhibitor of the nuclear export factor CRM1, or with vehicle (MeOH), for 3h. (A) Cells were transfected with plasmids expressing Mus musculus or Peromyscus maniculatus ZFP36L3, with GFP fused to either the N- or C-terminal ends. Mus musculus ZFP36L3 and Peromyscus maniculatus ZFP36L3 are abbreviated to MusL3 and PeroL3, respectively. The end of ZFP36L3 to which GFP was fused is indicated by the order in which they are written. In the top row of (A) are shown cells treated with vehicle only, with GFP expression mainly in the cytosol in all cases, except in the case of GFP alone, which was expressed in both the nucleus and cytosol. In the bottom row of (A), after treatment with LMB, Mus musculus ZFP36L3 remained primarily cytosolic, regardless of which end was attached to GFP, whereas Peromyscus maniculatus ZFP36L3 was largely nuclear, indicating that the latter is a shuttling protein. Similar findings were observed with both N- and C-terminal tagging of both proteins. (B) Untagged Peromyscus maniculatus ZFP36L3 was transfected and localized by immunostaining, using an antibody (#2050) raised against a fragment of this protein. The Peromyscus maniculatus ZFP36L3 protein was again retained in the nucleus in response to LMB. The specificity of the antibody is shown using pre-immune serum under identical conditions (right panel in Fig. 4B). (C) Mutations that removed the N- and C-terminal predicted NESs were made to the Peromyscus maniculatus ZFP36L3 - GFP fusion constructs. The diagram above each image indicates the end to which GFP was fused and which predicted NES was deleted. Images 2 and 4 indicate that deletion of the C-terminal NES alone results in accumulation of the protein in the nucleus with both of the GFP tagged proteins, while image 1 shows that deletion of the predicted N-terminal NES has no effect on localization. Proteins containing both putative NES deletions were also retained in the nucleus (image 3).

FIG. 5. Core residues of the novel Peromyscus maniculatus ZFP36L3 NES are conserved in other ZFP36L3 proteins, but not in ZFP36L2

Shown are sections of the ZFP36L2 and ZFP36L3 protein alignment from Fig. S2 that correspond to the newly identified NES from P. maniculatus. The alignment was performed with Clustal Omega, and is formatted according to Clustal conventions, as described in the legend to Fig. 3. A general consensus for a leucine-rich NES has been proposed by la Cour et al. (la Cour, et al. 2004): [LIVFM]X_{(2 or 3)}[LIVFM]X_{(2 or 3)}[LIVFM]X[LIVFM], where letters within the square brackets are critical hydrophobic residues, any one of which may be present at that position, X indicates any amino acid, and subscript values indicate the number of amino acids. The NES that we identified in P. maniculatus follows the pattern LX₃LX₂LX₁G, which matches the consensus (grey highlight), with the exception of the last residue, which is a G (cyan highlight). This suggests that the G residues found in this position, in several of the other species (cyan highlight), might also support a functional NES. Alternatively, the immediately preceding residue in all species is either an F or L, both hydrophobic amino acids known to contribute to NESs at other sites. Virtually all of the remaining core residues in the ZFP36L3 proteins from the other species match the NES consensus (grey highlight). In the only exceptions, there was an S, or an N present at the fourth position. We did not examine what effect this might have on the activity of those putative NESs. None of the ZFP36L2 proteins contain sequences resembling the NES consensus at this position, although the sequence DSL, found in the middle of the known P. maniculatus NES, is completely conserved in all ZFP36L2 and ZFP36L3 proteins sequenced to date. The position of this short sequence is indicated by the asterisks below the alignment. Underlined residues in the P. maniculatus sequence indicate the seven amino acids that were deleted in ZFP36L3-GFP fusion constructs that destroyed the function of the novel NES. Species names in blue type are members of Muridae family; red are Cricetidae; the single green species is in Nesomyidae; and in black are two representative rodent species that do not have Zfp36l3.