Interplay of TRIM2 E3 Ubiquitin Ligase and ALIX/ESCRT Complex: Control of Developmental Plasticity During Early Neurogenesis

Tripartite motif 2 (TRIM2) drives neurite outgrowth and polarization, is involved in axon specification, and confers neuroprotective functions during rapid ischemia. The mechanisms controlling neuronal cell fate determination and differentiation are fundamental for neural development. Here, we show that in Xenopus, trim2 knockdown affects primary neurogenesis and neural progenitor cell survival. Embryos also suffer from severe craniofacial malformation, a reduction in brain volume, and the loss of motor sensory function. Using a high-throughput LC-MS/MS approach with GST-Trim2 as bait, we pulled down ALG-2 interacting protein X (Alix) from Xenopus embryonic lysates. We demonstrate that the expression of trim2/TRIM2 and alix/ALIX overlap during larval development and on a cellular level in cell culture. Interestingly, trim2 morphants showed a clustering and apoptosis of neural progenitors, which are phenotypic hallmarks that are also observed in Alix KO mice. Therefore, we propose that the interaction of Alix and Trim2 plays a key role in the determination and differentiation of neural progenitors via the modulation of cell proliferation/apoptosis during neurogenesis.


Introduction
Tripartite motif (TRIM) proteins represent a family with more than 80 members in humans. TRIM proteins are characterized by a similar domain structure consisting of a RING-finger domain, one or two zinc finger domains, so-called B boxes, and a coiled-coil domain. Most TRIM family members have been proposed to function as E3 ubiquitin ligases due the presence of their RING-finger domain. The RING-type E3 ligases operate as scaffolds to recruit both ubiquitin coupled E2 and substrate, thereby facilitating ubiquitination [1,2]. Despite their overall similar structural organization, TRIM proteins regulate a wide variety of cellular processes.

Immunostaining of Xenopus Embryos
To analyze the protein expression, either embryos injected with morpholino or mRNA into one cell of a two-cell stage embryo or non-injected embryos were used. Embryos grown until the desired Nieuwkoop-Faber (NF) stage were fixed and left overnight in Dent's solution at -20 • C for permeabilization. Next, embryos were washed in a series of decreasing methanol concentration steps, blocked for approximately 6 h in 20% horse serum/PBS and incubated with appropriate antibody (see Immunostaining and Western blotting) at 4 • C overnight. Embryos were washed thoroughly in PBSTB twice, 2 h each and in PBSTBN for 2 h. Then, they were washed thoroughly 3 times in a 5 min wash and a wash step in PBSTB at 4 • C overnight. Embryos were transferred to respective secondary antibody conjugated with either alkaline phosphatase or fluorophores and incubated at 4 • C overnight. This was followed by wash steps as above for 30 min and overnight. The staining of embryos was performed as described for Wmish. In case of fluorescence staining, embryos were washed twice for 5 min and left in PBS buffer for further analysis.

Motility-Escape Assay in Xenopus Embryos
A modified assay based on Sztal et al., 2016 was used in Xenopus embryos to perform motility-chase assay [30]. Xenopus late tailbud stage control and morphant embryos were put in a Petri dish with 2% agarose cut out in the center into a 1.5 cm × 1.5 cm square well containing 0.1× MBS. Using a micropipette tip, the sensorimotor response of embryos was observed by poking/touching embryo tails. The response to external stimuli was captured by Zeiss microscope and Motic camera and software.

Cell Culture and Transfections
HEK 293T cells were used for the Western blot analysis. For pull-down assay, approximately 1 × 10 6 cells were seeded in a 6-well plate and maintained in DMEM supplemented with 10% FCS and 100U/mL penicillin G and streptomycin. All media and sera were purchased from PAA laboratories. For transient transfection, 60% confluent cells were transfected with indicated plasmid DNA polyethyleneimine (3 µg DNA in 1 µL of 5mg PEI/mL) (Sigma-Aldrich, Darmstadt, Germany). After approximately 6 h of incubation with the PEI/DNA complex, medium was replaced by fresh serum containing medium. Cells were harvested 24 h post transfection. The lysates were cooked in loading buffer and loaded on 10% SDS gels for further analysis.

GST-Pulldown and LC-MS/MS Mass Spectrometry
Xenopus trim2 was cloned into pGEX4-T1 vector. Fifty micrograms of bacterially expressed and purified GST or GST-Trim2 was added to 1 mL (1 mg protein) lysate and after 1 h at 4 • C under continuous rotation, 50 µL GSH-Sepharose beads were added to the mixture. After 60 min of incubation, the beads were washed three times with IP lysis buffer and once with 0.1 mM Tris, pH7.4. Proteins were eluted with 2× Laemmli buffer, separated by SDS/PAGE, and analyzed by Western blot. GST-Trim2 as a bait was applied on Xenopus embryonic lysates (NF stages 30-36) followed by affinity enriched LC-MS/MS on an LTQ Orbitrap instrument [31]. Three independent biological replicates were analyzed. Proteins detected were evaluated based on the intensities of the MS signals and spectral counts to assess the enrichment of protein interactions with GST-Trim2 relative to GST alone. First, proteins are required to be reproducibly present, which is detected by at least five spectral counts. Second, proteins showing less than 100-fold spectral count enrichment over proteins co-isolated with GST alone are deemed nonspecific and are excluded. Third, the base value for Student t-test with a significance of 10 −5 is set as standard (Table 1).

Alignment, Phylogeny, Synteny
Fasta sequences for the protein families analyzed were obtained by the Blast tool (http://blast.ncbi. nlm.nih.gov) and aligned using the T-Coffee and Box shade tool (http://tcoffee.vital-it.ch/apps/tcoffee/ index.html). A phylogenetic tree of the proteins was generated through maximum-likelihood using one-click mode (http://phylogeny.lirmm.fr/phylo_cgi/index.cgi). The synteny analysis is based on data derived with the help of metazome v. 3.2 (https://metazome.jgi.doe.gov/pz/portal.html). The individual gene sequences and the corresponding information regarding Xenopus laevis gene loci were obtained from Xenbase (http://gbrowse.xenbase.org) and depicted accordingly.

Statistical Analysis
A Chi-square test of homogeneity examines the extent to which the occupied variables for a four-field table deviate from an expected, homogeneous distribution. Unpaired Chi-square distribution of independence was used to test the null hypothesis. p-values were deduced from the abundance values of the replicates of independent biological samples. The significance of the p value was set at p < 0.001 and degrees of freedom (df) n-1. Null hypothesis is rejected, if the chi square is greater than the value in the chi-square table with p ≤ 0.001 and 1 degree of freedom.

During Xenopus Development, trim2 Is Expressed in Neural and Non-Neural Ectoderm
Trim2 expression was first described in the adult mouse brain [5]. However, its expression during development has not been reported. At first, we examined the temporal expression, performed a sequence analysis, and elucidated the evolutionary relationship of Xenopus trim2 (Figure 1). Temporal expression analysis showed a gradual decrease of maternal transcripts. First zygotic expression was detected at low levels at the onset of gastrulation and was maintained throughout early development with an increase at the onset of neurulation. A comparison of amino acid sequences revealed that the known domains of TRIM2/Trim2 are fully conserved between mammals and Xenopus. Analysis of synteny and phylogeny showed a high conservation of Trim2/trim2 particularly in tetrapods ( Figure S1). To analyze the spatial expression of trim2 during early neural development, we performed  (d-f) NF stage 21, lateral, anterior and dorsal views. trim2 transcripts were detected in eye vesicle (ev), neural tube (nt), olfactory (ol), and otic placode (op), as well as cement gland anlage (cg). (g-i) At NF stage 25, trim2 expression became more distinct in eye vesicle (eb), forebrain (fb), midbrain (mb), olfactory placode (ol), otic placode (op), and neural tube (nt). (j-l) NF stage 33, trim2 in fore-(fb), mid-(mb), and hindbrain (hb), eye (ey), olfactory pit (olp), otic vesicle (ov), spinal cord (sc), and trigeminal ganglion (tg). Scales as indicated. (strd-mo n = 20 embryos; trim2-mo n = 16 embryos (E); p = 1.97E-09, unpaired χ 2 test, χ 2 = 36, df =1, phenotype occurrence (po) =16, p < 0.001).
To investigate whether the timing and appearance of mRNA and protein expression correlate, we compared the whole-mount in situ detection of trim2 mRNA and immunohistochemical detection of Trim2 protein in late tailbud stage embryos (Figure 2A embryos revealed that the temporal and spatial expression of trim2 transcripts and the Trim2 protein overlap. We observed an elevated expression of trim2/Trim2 in the region of prospective motor neurons, crossing fibers of the floor plate, and cells of the medial and lateral ganglionic eminences ( Figure 2(Ad-Ag,Bd-Bh)). Lower levels of expression were detected throughout the forebrain (Figure 2(Ad,Ai,Bd)). At the level of diencephalon, Trim2 expression was also observed in the ventral midline region, where radial glial cells are present [32] ( Figure 2B). Furthermore, trim2/ Trim2 was found in the hypothalamus, prethalamus, as well as in ganglion cells of the retina and in the lens epithelium ( Figure 2(Ae,Bf)). Double whole-mount in situ with tubb2b revealed that trim2 was expressed at particularly high levels in secondary neurons of the hindbrain, the inner otic vesicle, dorsal root entry zone, and in dorso-lateral sensory neurons of the spinal cord ( Figure S2). Generally, trim2/Trim2 were highly expressed in committed and differentiated neuronal cells, whereas the expression level in neural progenitors was almost absent. Outside the forming neural system, we observed weak trim2 expression only in somites, notochord, and the ventral blood island and in a spotted pattern within the skin (Figure 2(Ab,Aj,Ak,Bc,Bh)). The localization of trim2 transcripts and protein suggest a functional role of Trim2 during the formation of the central nervous system. To investigate whether the timing and appearance of mRNA and protein expression correlate, we compared the whole-mount in situ detection of trim2 mRNA and immunohistochemical detection of Trim2 protein in late tailbud stage embryos (Figure 2A,B). Histological analysis of sectioned embryos revealed that the temporal and spatial expression of trim2 transcripts and the Trim2 protein overlap. We observed an elevated expression of trim2/Trim2 in the region of prospective motor neurons, crossing fibers of the floor plate, and cells of the medial and lateral ganglionic eminences (Figure 2A(d-g); 2B(d-h)). Lower levels of expression were detected throughout the forebrain ( Figure  2A(d,i); 2B(d)). At the level of diencephalon, Trim2 expression was also observed in the ventral midline region, where radial glial cells are present [32] ( Figure 2B€). Furthermore, trim2/Trim2 was found in the hypothalamus, prethalamus, as well as in ganglion cells of the retina and in the lens epithelium ( Figure 2A(e); 2B(f)). Double whole-mount in situ with tubb2b revealed that trim2 was expressed at particularly high levels in secondary neurons of the hindbrain, the inner otic vesicle, dorsal root entry zone, and in dorso-lateral sensory neurons of the spinal cord ( Figure S2). Generally, trim2/Trim2 were highly expressed in committed and differentiated neuronal cells, whereas the expression level in neural progenitors was almost absent. Outside the forming neural system, we observed weak trim2 expression only in somites, notochord, and the ventral blood island and in a spotted pattern within the skin (Figure 2A   views, trim2 transcripts were detected in the forming brain (br), olfactory pit (olp), spinal cord (sc), and the ventral blood island (vbi). (d-h) Twenty-micrometer gelatine/albumin sections at the level of fore-, mid-, hindbrain, and spinal cord. trim2 is expressed in the prosencephalon (pc), olfactory pit, ganglion cell layer (gc), lens epithelium (ls), hypothalamus (hy), motor neurons (mn), otic vesicle (ov), and spinal cord (sc). Faint expression was observed in the ventral blood island. (i-k) Sagittal sections reveal trim2 expression in the ganglion cell layer, lens epithelium, brain, spinal cord, notochord (nc), somites (so), and ventral blood island. (B) Whole-mount immunochemistry. (a-c) At NF stage 35 (anterior, dorsal, and lateral views), Trim2 protein expression overlaps strongly with trim2 mRNA. (d-i) Twenty-micrometer gelatine/albumin transverse sections along the fore-, mid-, hindbrain, and spinal cord region. Trim2 protein was detected in the developing CNS, cement gland, dorsal root entry zone (drez), ganglion cells, hindbrain, hypothalamus, ion-secreting cells (isc), motor neurons, olfactory pit, otic vesicle, prosencephalon (pc), prethalamus (pth), presumptive radial glial cells (rg), somites (so), and ventral blood island.

trim2 Knockdown Leads to Smaller Head and Affects Cranial Axis Positioning
In order to understand the function of Trim2, we utilized antisense morpholino oligonucleotides to block the translation of trim2, while synthetic egfpor ß-gal-RNA was co-injected as a tracer into one cell of two-cell stage Xenopus embryos. The specificity of the morpholino was confirmed by immunohistochemistry and rescue experiments ( Figure 3, Figure S3A) and by injecting a morpholino reporter made of a GFP-tagged complimentary sequence as reported earlier [33]. The immunohistochemistry of trim2-morphants at NF stage 17 confirmed the suppression of trim2 translation within the neural plate territory (Figure 3a). Tadpoles at NF stage 35 showed a strong reduction of Trim2 expression in the eye, brain, and spinal cord upon trim2-morpholino injection (Figure 3(b-b",b1-b3)) along with a reduction in pigmentation on the injected side ( Figure S3B). Moreover, a structural disorganization of presumptive brain tissue toward the non-injected side was also observed (Figure 3b1). Cells were dissociated, misaligned, and clustered all over the presumptive brain and in the forming eye.

trim2 Knockdown Leads to Smaller Head and Affects Cranial Axis Positioning
In order to understand the function of Trim2, we utilized antisense morpholino oligonucleotides to block the translation of trim2, while synthetic egfp-or ß-gal-RNA was co-injected as a tracer into one cell of two-cell stage Xenopus embryos. The specificity of the morpholino was confirmed by immunohistochemistry and rescue experiments (Figure 3, Figure S3A) and by injecting a morpholino reporter made of a GFP-tagged complimentary sequence as reported earlier [33]. The immunohistochemistry of trim2-morphants at NF stage 17 confirmed the suppression of trim2 translation within the neural plate territory (Figure 3a). Tadpoles at NF stage 35 showed a strong reduction of Trim2 expression in the eye, brain, and spinal cord upon trim2-morpholino injection (Figure 3b-b'', b1-b3) along with a reduction in pigmentation on the injected side ( Figure S3B). Moreover, a structural disorganization of presumptive brain tissue toward the non-injected side was also observed (Figure 3b1). Cells were dissociated, misaligned, and clustered all over the presumptive brain and in the forming eye.

Suppression of trim2 Function Affects Neurogenesis, Proliferation, and Cell Survival
The suppression of trim2 function resulted in smaller heads, affecting the developing brain and eyes. To investigate whether early forebrain development was impaired in trim2 morphants, we investigated the expression of early forebrain markers in those embryos. At stage 15, otx2 expression in the fore-midbrain region was expanded (Figure 4(Aa)). Similarly, the expression of pax6 in the prospective eye field (Figure 4(Ab)) and that of fb1 in the presumptive forebrain were expanded. Altered cell proliferation and/or survival might have contributed to the impaired neuronal phenotype upon the suppression of trim2. Thus, we monitored mitotic and apoptotic cells accordingly (Figure 4(Ad-Ah)). TUNEL assay revealed a profound increase in the number of apoptotic cells at early neural (NF stage 15) as well as early organogenesis stages (NF stage 27) within the brain, eye, and cement gland (Figure 4(Ad-Ae"), red arrow heads), and all other territories of trim2 gene expression. As observed in the cranial transverse section of an NF stage 27 embryo, we often observed an aggregation of apoptotic cells in the presumptive brain and eye on the injected side (Figure 4(Ae")). Cell proliferation assay identified only half the number of mitotic cells on the injected side compared to the non-injected side.

Suppression of trim2 Function Affects Neurogenesis, Proliferation, and Cell Survival
The suppression of trim2 function resulted in smaller heads, affecting the developing brain and eyes. To investigate whether early forebrain development was impaired in trim2 morphants, we investigated the expression of early forebrain markers in those embryos. At stage 15, otx2 expression in the fore-midbrain region was expanded ( Figure 4A(a). Similarly, the expression of pax6 in the prospective eye field ( Figure 4A(b)) and that of fb1 in the presumptive forebrain were expanded. Altered cell proliferation and/or survival might have contributed to the impaired neuronal phenotype upon the suppression of trim2. Thus, we monitored mitotic and apoptotic cells accordingly ( Figure 4A(d-h)). TUNEL assay revealed a profound increase in the number of apoptotic cells at early neural (NF stage 15) as well as early organogenesis stages (NF stage 27) within the brain, eye, and cement gland ( Figure 4A, d-e'', red arrow heads), and all other territories of trim2 gene expression. As observed in the cranial transverse section of an NF stage 27 embryo, we often observed an aggregation of apoptotic cells in the presumptive brain and eye on the injected side ( Figure  4A(e'')). Cell proliferation assay identified only half the number of mitotic cells on the injected side compared to the non-injected side. and fb1) appeared apparently expanded (red arrow heads). However, this is accompanied by an increase in apoptotic cells (d, TUNEL), while proliferation (f, pH3) was not affected at this early stage. In young tadpoles (NF stage 27), the rate of apoptosis was highly increased in the brain region including the forming eye (e',e") and the number of mitotic cells was strongly reduced (g-g") (strd-mo n = 12E; trim2-mo n = 20E; p = 2.1E-06, unpaired χ 2 test, χ 2 =22.5, df =1, po = 18, p < 0.001). A similar effect was observed for fb1 and otx2 genes (Figure 4 To further characterize trim2 morphants during early neural differentiation, we monitored the expression of three key markers ( Figure 4B). Neural precursor cells express the pan-neural marker sox3 throughout the neural plate at NF stage 15. In trim2 morphants, sox3 expression appeared expanded on the injected side, similarly to pax6 (Figure 4(Ba)). This result suggested a regular, though expanded induction of primary neural precursor cells in trim2-deficient cells. However, neural specification and differentiation markers such as myt1 and neural-beta-tubulin (tubb2b), respectively, which are expressed in medial, intermediate, and longitudinal stripes and in the trigeminal placode within the neural plate, were clearly reduced upon the suppression of trim2 function (Figure 4(Bd,Bg)) and at the tailbud stage, the expression of myt1, and tubb2b was significantly reduced (Figure 4(Be-Bi)). Correspondingly, as observed in transverse sections of trim2 morphants, secondary neural precursor cells marked by sox3 expression were almost absent in the dorsal regions of the forebrain and midbrain (Fiure 4(Bb)). In placodal-derived structures such as olfactory, otic, and trigeminal placodes, the expression of myt1 and tubb2b was also reduced. Taken together, our results suggest that the suppression of trim2 function did not impair primary neural induction; rather, it affected neural differentiation and impaired cell proliferation as well as cell survival, maintaining neural fate during early neurogenesis.

trim2 Morphants Show Loss of Motor Sensory Function
Recently, a human patient with a heterozygous missense mutation in TRIM2 was reported to exhibit a loss of gross motor neuron development that resulted in an early onset of axonal neuropathy, muscle hypotonia, and sarcopenia [9]. Trim2-depleted mice displayed tremors with episodes of spontaneous seizures [5]. At late tailbud stages, spinal neurons differentiate and connect to create a system that performs sensorimotor function, which is crucial for survival. This sensorimotor function also includes a response to external stimuli such as touch or poking by escaping with a rapid beating movement. Therefore, we performed motility-escape assays on late tailbud stage embryos to investigate whether trim2 knockdown affects motor-sensory function. The analysis revealed loss of motor and sensory function in trim2 morphants injected into one cell at the 2-cell stage (Movie 1). After several attempts of poking/stimulating the embryos on the injected side, embryos moved in circles with one-sided beating against stimulation on the non-injected side, which is indicative of a non-spontaneous response to sensory stimulation. A complete loss of motor sensory function accompanied by severe developmental defects and high mortality was observed upon injection of trim2-morpholino into both blastomeres of a two-cell stage embryo (Movie 2). We were able to rescue this phenotype as well as a marker gene expression of primary neurogenesis by injecting synthetic mutated Xenopus trim2 RNA (∆trim2) resistant to the trim2-morpholino (Movie 3).

Trim2 Interacts with Alix Physically
Previous studies have shown that NEFL (Neurofilament Light) is a substrate for TRIM2-catalyzed ubiquitination and accumulated in the cerebellum of Trim2 mutant mice [5]. However, using a GST-Trim2 protein as bait, we were unable to pull down NEFL from Xenopus embryonic lysates or from HEK293T cells. To identify proteins that interact with Xenopus Trim2, we applied GST pulldown along with high-throughput LC-MS/MS screening ( Figure 5A). We identified the Alix protein in very high abundance as a prominent candidate for a Trim2-interacting protein (Table 1). We confirmed the interaction of Trim2 with Alix by a GST pulldown assay using GST-Trim2 and lysates of flag-Alix transfected HEK293T cells ( Figure 5B, Figure S4).
To address the intrinsic relationship between Trim2 and Alix, we performed a ubiquitination assay (data not shown). Trim2 and Alix were overexpressed in HEK293 cells in the presence of ubiquitin, but we were unable to see any mono-or polyubiquitylation of Alix (data not shown). Thus, Alix is not a substrate of Trim2.  (Lanes 1 and 2). GST alone serves as a control (Lane 3). Eluted flag-tagged Alix after binding GST-Trim2 (Lane 4, upper blot). Note: images of the blots are cropped accordingly to the area of exposed protein bands. Uncropped blots in Figure S4. Two different gels/bolts were used to show the eluted flag-tagged Alix and precipitated GST-Trim2 of the same sample.

Expression of Alix Overlaps with trim2 in Xenopus Embryos
Direct interactions between Trim2 and Alix suggest that both genes are either expressed in overlapping or at least in adjacent domains of the embryo. Indeed, we observed an overlapping pattern of expression for alix and trim2. During early neurulation, alix transcripts were detected in the neural tube, eye vesicle, and the cement gland (Figure 6). At late tailbud stages, high level of alix transcripts appeared within the forming CNS and cranial placodes. A high level of expression was further detected in the pronephros, which did not show trim2 expression (Figure 2A(g), 2B(h)), indicating Trim2 independent functions of Alix as well. In addition, we tested by immunofluorescence using commercial antibodies whether endogenous TRIM2 and ALIX do show overlapping expression in a human neuroblastoma cell line SH-SY5Y. Signals for endogenous TRIM2 and ALIX overlapped strongly and showed a filamentous pattern including some vesicular-like or granular-like structures within the cytoplasm ( Figure 6B). Note: images of the blots are cropped accordingly to the area of exposed protein bands. Uncropped blots in Figure S4. Two different gels/bolts were used to show the eluted flag-tagged Alix and precipitated GST-Trim2 of the same sample.

Expression of Alix Overlaps with trim2 in Xenopus Embryos
Direct interactions between Trim2 and Alix suggest that both genes are either expressed in overlapping or at least in adjacent domains of the embryo. Indeed, we observed an overlapping pattern of expression for alix and trim2. During early neurulation, alix transcripts were detected in the neural tube, eye vesicle, and the cement gland (Figure 6). At late tailbud stages, high level of alix transcripts appeared within the forming CNS and cranial placodes. A high level of expression was further detected in the pronephros, which did not show trim2 expression (Figure 2(Ag,Bh)), indicating Trim2 independent functions of Alix as well. In addition, we tested by immunofluorescence using commercial antibodies whether endogenous TRIM2 and ALIX do show overlapping expression in a human neuroblastoma cell line SH-SY5Y. Signals for endogenous TRIM2 and ALIX overlapped strongly and showed a filamentous pattern including some vesicular-like or granular-like structures within the cytoplasm ( Figure 6B).

Functional Interactions of Trim2 and Alix
To provide first insight into the functional relationship of Trim2 and Alix, we performed single and co-suppression experiments of trim2 and alix and compared the expression of early neural marker genes. The injection of both morpholinos at a combined dose equal to the final dose of a single trim2-morpholino injection resulted in similar alterations of marker gene expression compared to trim2 morphants (Figure 7). At NF stage 15, sox3 expression within the neural plate appeared widened on the injected side. The expression of myt1 and tubb2b in intermediate and longitudinal stripes and the trigeminal placode were again reduced. Moreover, at the early tail bud stage (NF stage 32), alix morphants showed a similar loss of motor-sensory function as observed for trim2 morphants (Movie 4). Since Trim2 and Alix physically interacted and combined suppression resulted in an almost identical phenotype, we asked whether an epistatic relationship exists or both act interdependently. To this end, we performed rescue experiments for co-injected morphants with synthetic trim2-RNA or human ALIX-RNA. However, we were unable to rescue the double morphants with either trim2 or ALIX-mRNA. Further, we tested a dose of trim2-mo or alix-mo, which alone resulted in only a very mild phenotype. Since the injection of both morpholinos in combination resulted in an additive effect and the trim2-and alix-morphants could not be rescued by injecting the other RNA, we concluded that Trim2 and Alix interact not linear but in parallel in the same pathway ( Figure 7B, Figure S5).

Functional Interactions of Trim2 and Alix
To provide first insight into the functional relationship of Trim2 and Alix, we performed single and co-suppression experiments of trim2 and alix and compared the expression of early neural marker genes. The injection of both morpholinos at a combined dose equal to the final dose of a single trim2-morpholino injection resulted in similar alterations of marker gene expression compared to trim2 morphants (Figure 7). At NF stage 15, sox3 expression within the neural plate appeared widened on the injected side. The expression of myt1 and tubb2b in intermediate and longitudinal stripes and the trigeminal placode were again reduced. Moreover, at the early tail bud stage (NF stage 32), alix morphants showed a similar loss of motor-sensory function as observed for trim2 morphants (Movie 4). Since Trim2 and Alix physically interacted and combined suppression resulted in an almost identical phenotype, we asked whether an epistatic relationship exists or both act interdependently. To this end, we performed rescue experiments for co-injected morphants with synthetic trim2-RNA or human ALIX-RNA. However, we were unable to rescue the double morphants with either trim2 or ALIX-mRNA. Further, we tested a dose of trim2-mo or alix-mo, which alone resulted in only a very mild phenotype. Since the injection of both morpholinos in combination resulted in an additive effect and the trim2and alix-morphants could not be rescued by injecting the other RNA, we concluded that Trim2 and Alix interact not linear but in parallel in the same pathway ( Figure 7B, Figure S5).  Fig. 7a;b: strdmo, n = 10E; co-inj. trim2/alix-mo, n = 20E; p = 9.47E-06, unpaired χ 2 test, χ 2 =19.6, df =1, po = 17, p < 0.001; fig. 7c-d: myt1: strd-mo n = 10E; trim2/alix-mo n = 12E; p = 1.85E-05, χ 2 =18, df =1, po = 11; fig. 7cd: tubb2b: strd-mo, n = 10E; trim2/alix-mo, n = 13E; p = 1.12E-05, unpaired χ 2 test, χ 2 =19, df = 1, po = 12, p < 0.001). (B) Effect of combined injection of low doses of trim2 and alix mo. To determine the dose of morpholino injection that caused only a very mild phenotype compared to standard morpholinoinjected embryos, decreasing amounts of trim2-mo or alix-mo were injected into one cell of a two-cell stage embryo. The width of the sox3 expression domain was measured using ImageJ, and the ratio between the non-injected and the injected side was calculated. Combined injection of 0.4 pmol of trim2 and alix-mo each led to an additive effect compared to individual injections (0.4 pmol trim2-mo + 0.4 pmol alix-mo). The trim2 morpholino phenotype could not be rescued by co-injection with either halix-RNA (1ng) n = 36E, and the alix-morpholino phenotype could not be rescued by Δtrim-2 RNA (1ng) n = 29E.

Discussion
Our present work provides insight into the differential expression and function of Trim2 during vertebrate embryogenesis. Trim2 is a conserved member of the TRIM-NHL family of E3 ligases [34]. Evolutionarily, Trim2 is highly conserved across vertebrates ( Figure S1). Analysis of trim2/Trim2 mRNA and protein expression revealed that trim2/Trim2 is predominantly expressed in the developing neural system of Xenopus laevis embryos. By the suppression of trim2 function, we demonstrated that trim2 affects the proliferation, determination, and differentiation of neural progenitors during embryonic neurogenesis, while primary neural induction was not impaired. Interestingly, swimming tadpoles of trim2 morphants showed affected motor sensory function at later stages. In addition, we provide evidence that Trim2 physically and functionally interacts with Alix using pulldown techniques, mass spectroscopic analysis, and functional analysis.
In adult mice, Trim2 expression was reported in regions of the cerebellum, hippocampus, retina, and the spinal cord [5]. In Xenopus embryos, we found trim2 expression in the presumptive ventral brain and spinal cord, including the floor plate, otic vesicles, and olfactory pits. Both Trim2 protein and trim2 mRNA levels were high in committed and differentiated neuronal cell layers of the future central nervous system. Thus, Trim2/trim2 expression appeared highly similar in the CNS of Xenopus tadpoles and adult mice. In non-neural tissues, only weak expression in head mesenchyme, notochord, somites, and in the ventral blood island was observed.
Recent insights into the mechanisms of neuronal polarization during axon specification revealed TRIM2-dependent post-translational regulation of NEFL [5,6,35]. First reported in the molluscan Lymnaea stagnalis, the orthologue L-Trim is up-regulated during the neurite outgrowth of central neurons and in the postnatal brain during neuronal regeneration. A strong inhibition of neurite outgrowth was observed upon the knockdown of l-trim mRNA as well [35]. Similarly, cultured hippocampal neurons and motor neurons from mice showed a hyperpolarization-like phenotype upon TRIM2 overexpression and a hypopolarization-like phenotype after trim2 knockdown. Likewise, proteasomal degradation of NEFL by TRIM2 is necessary for neural polarization of the hippocampal neurons. All data thus far indicate a functional role for TRIM2 in secondary neurogenesis [4,35]. Our findings complement this previous work, providing further evidence of trim2 function in embryonic neurogenesis. We showed that in trim2 morphants, the determination and differentiation of proneural genes is impaired at the open neural plate stage, while the precursor cell pool was unaffected or appeared even enlarged. This was also the case as observed for early forebrain markers. Presumably, the suppression of trim2 function inhibited the cells from exiting the cell cycle properly, thereby resulting in an accumulation of precursor cells with a higher frequency of apoptotic cells. In later stages, we also observed severe gross morphological malformations and a loss of structural organization within the craniofacial region, in contrast to findings in Trim2 GT mice. However, this may be due to the integration of the gene trap vector inside the Trim2 locus between exons 6 and 7, which may result in a partially active protein, since the catalytically active RING domain of Trim2 is located on exon 2. In addition, the authors reported a residual expression of nearly 5% of wild-type mRNA due to an alternative splicing effect [5]. On the other hand, the suppression of trim2 function in Xenopus embryos resulted in defective motor neurons as seen by an ataxia-like phenotype of swimming tadpoles, similar to that reported in Trim2 GT mice.
Little is known about the direct interaction partners of TRIM2/Trim2. We identified Alix (originally described as a pro-apoptotic protein) as a candidate Trim2-interacting protein. First isolated from Xenopus oocytes, the alix homologue Xp95 is phosphorylated by src kinases [36]. In mice, ALIX has been described as a calcium-dependent interaction partner of apoptosis linked gene-2 (ALG-2) [37,38] and recently of two ABC-type cholesterol transporters of mainly neuronal cellular identity [10]. The overexpression of ALIX in cerebellar neurons of mice revealed activation of the caspase cascade leading to neuronal cell death, independent of the JNK or p38 MAPK signaling pathways. However, a C-terminal fragment of ALIX protected neurons from potassium withdrawal-induced cell death [20]. Our studies suggest a direct, functional interaction of Alix and Trim2 during embryonic neurogenesis and development. However, Alix is not a substrate of Trim2. In general, the expression of alix correlated strongly with trim2 during Xenopus development. In tadpoles, both transcripts were found in an overlapping pattern in the developing central nervous system and particularly in the retina and lens, brain, spinal cord, and the olfactory and otic placodes. Moreover, the loss of neural differentiation and the microcephalic-like phenotype observed in trim2 morphants mimics that of Alix-deficient mice [25].
To date, there is no evidence for a functional role of TRIM2 in the regulation of apoptosis or cell survival. However, there is clear evidence for ALIX in this respect [22,37,38]. Thus, Alix may possess a more general function as an adapter protein, which is important to convey substrate specificity and selection between an E3-ubiquitin ligase such as NEDD4-1 or Trim2 for mono-or polyubiquitination as reported [10,39]. We could show a significant decrease in cellular proliferation, which is most probably due to a high rate of apoptosis within the developing CNS of trim2 morphants. Along similar lines, a loss of neural progenitors was observed during the transient phase of apoptosis in Alix KO mice [25]. ALIX is involved in the regulation of the endolysosomal system through binding to endophilins and to endosomal sorting complexes required for the transport (ESCRT) proteins TSG101 and CHMP4b. Since TRIM2 appears to localize to vesicular-like structures of so-far unknown identity, this may provide a subcellular platform for functional interactions between these referred proteins [11,[13][14][15][16]40].
Our results provide evidence for an important function of Trim2-Alix interaction during early neural development. In concert, they may link developmental cues to cellular proliferation/cell survival and thus may take an important function during neuronal determination and differentiation.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4409/9/7/1734/s1, Figure S1. (A) Xenopus Trim2 protein consists of 748 amino acids and is organized in domains containing a single RING zinc finger, two B-box zinc fingers, a single coiled-coiled, Filamin/ABP 280 domain, and six NHL repeats. (B) Semi-quantitative RT-PCR analysis on consecutive developmental stages. Maternal expression of trim2 is observed followed by gradual decrease in expression by gastrulation (NF stage 10) and increasing again. Strong expression at neurulation (NF stage 18) until the latest stages analyzed (NF stage 42) in a gradually increasing manner. Expression of histone H4 was monitored to control total RNA input. (C) Semi-quantitative RT-PCR analysis of trim2 expression in adult tissues. The brain, eye, and spinal cord showed strong expression, while the skin, heart, kidney, muscle, testis, intestine, and pancreas showed weak expression of trim2. (D) Phylogenetic analysis: An evolutionary rooted tree is constructed based on maximum likelihood and bootstrap analysis. The branch length is proportional to the number of substitutions per site. The numbers in red next to nodes represent bootstrap support values. The red arrow indicates the root of the tree, the blue arrow indicates the out-group, and the green arrow indicates the common ancestral node between mammals and amphibians. The bar at the bottom of the phylogram indicates the evolutionary distance to which the branch lengths are scaled based on the estimated divergence. There is a relatively very little divergence observed over time. (E) Synteny organization of Trim2/trim2: The chromosomal localization of the gene encoding trim2 is conserved in mammals. In vertebrates, the locus appeared grossly conserved. Each arrow stands for a single gene. The arrowhead indicates the direction of the ORF. Orthologues are marked with identical colors. trim2 (black arrow) is present in all species analyzed. Upstream trim2 is flanked by the same gene with minor exceptions. Downstream trim2 is flanked by the same gene, except for mammalian. Arrow colors; black: trim2, red: meiotic nuclear divisions 1/mnd1, yellow: transmembrane 131 like/tmem131l, turquoise: FH2 domain containing 1/fhdc1, orange: ADP ribosylation factor interacting protein 1/arfip1, white 1-10: yet uncharacterized proteins. (F) Comparison of amino acid sequences of TRIM2/Trim2. H. sapiens (GenBank Accession no. NP_001123539), B. taurus (GenBank Accession no. NP_001077204), M. musculus (GenBank Accession no. NP_109631), R. norvegicus (GenBank Accession no. NP_001102022), G. gallus (GenBank Accession no. NP_001244243), D. rerio (GenBank Accession no. NP_001014393), X. tropicalis (GenBank Accession no. NP_001005680), X. laevis (GenBank Accession no. NP_001085492). Conserved domains are indicated by colored boxes according to A. Figure S2. Whole-mount double in situ hybridization of trim2 and tubb2b at NF stage 35/36. (a) Lateral view. (b-d) Transverse section of midbrain, hindbrain, and the spinal cord region. trim2 is mainly expressed in differentiated neurons (e.g., motor neurons, dark blue) such as tubb2b (red). Figure S3. (A) Phenotypic rescue of trim2 morpholino-injected embryos. Synthetic Xenopus trim2 RNA (∆trim-2), which did not contain a complementary sequence of the morpholino, was injected along with trim2 morpholino into one cell of a two-cell stage embryo. (a) The expanded expression of sox3, (c, e) and the medial, intermediate, and longitudinal stripes of myt1 and tubb2b-expressing cells were rescued on the injected side. Red arrows indicate the rescued side. (b, d, f) The percentage of embryos rescued (sox3 strd-mo n = 20 embryos, trim2-mo n = 24 embryos, p = *** 7.21E-09; myt1 strd-mo n = 18 embryos; trim2-mo n = 23 embryos; p = 6.46E-09; tubb2b strd-mo n = 20 embryos; trim2-mo n = 19 embryos; p = 1.78E-08, unpaired χ 2 test). Scale as indicated. (B) Phenotype of trim2 morphants. (A) Upper panel, whole embryo, lateral views, non-injected side (nis) and injected side (is). Bottom panel shows enlarged cranial lateral view. Red arrowhead indicates a suppression of proper head development (eye, otic vesicle, and branchial arches) and pigmentation. The bar plot represents the survival percentage of trim2 morphants compared to non-injected control and standard morpholino-injected embryos. Survival rate was 24% for trim2 morphants. Total injected (ctrl n = 186 embryos, strd-mo n = 288 embryos, trim2-mo n = 254). Survived (ctrl n = 177 embryos, strd-mo n = 253 embryos, trim2-mo n = 62). Scales as indicated. Figure S4. (A) Uncropped blots of the GST pulldown assay ( Figure 5B) at different exposure times. Blot left of red line probed against Alix and the right side against Trim2. The rectangles with dashed lines indicate the presented detail in Figure 5B. The GST-Trim2 lane was loaded as a positive control for the ab reaction only. Figure S5. (A) alix and trim2 morphants show an almost identical alteration of marker gene expression. (a-b) Widened expression for pax6 at the prospective forebrain and eye field region on the injected side in both trim2 and alix morphants was observed. (c-d) forked head box G1 (fb1) also displayed a relatively expanded band in trim2 and alix morphants. Yellow dotted line indicates the midline. (B) Phenotypic rescue of trim2 or alix morpholino-injected embryos. Synthetic Xenopus trim2 RNA (∆trim-2) or Human alix RNA, which did not contain a complementary sequence of the morpholino, were injected along with trim2 or alix morpholino respectively into one cell of a two-cell stage embryo. (a, b) The width of the sox3 expression domain on the morpholino-injected sides was expanded but (a', b') rescued by co-injection of synthetic RNA. The percentage of embryos with no phenotype, phenotype, or rescued phenotype was calculated (trim2-mo n = 42 embryos, trim2-mo plus ∆trim-2 RNA n = 24 embryos, p = 3,46E-04; alix mo n = 52, alix-mo plus halix-RNA n = 14 embryos, p = 2,96E-4).