An ankyrin-binding motif regulates nuclear levels of L1-type neuroglian and expression of the oncogene Myc in Drosophila neurons

L1 cell adhesion molecule (L1CAM) is well-known for its importance in nervous system development and cancer progression. In addition to its role as a plasma membrane protein in cytoskeletal organization, recent in vitro studies have revealed that both transmembrane and cytosolic fragments of proteolytically cleaved vertebrate L1CAM translocate to the nucleus. In vitro studies indicate that nuclear L1CAM affects genes with functions in DNA post-replication repair, cell cycle control, and cell migration and differentiation, but its in vivo role and how its nuclear levels are regulated is less well-understood. Here, we report that mutations in the conserved ankyrin-binding domain affect nuclear levels of the sole Drosophila homolog neuroglian (Nrg) and that it also has a noncanonical role in regulating transcript levels of the oncogene Myc in the adult nervous system. We further show that altered nuclear levels of Nrg correlate with altered transcript levels of Myc in neurons, similar to what has been reported for human glioblastoma stem cells. However, whereas previous in vitro studies suggest that increased nuclear levels of L1CAM promote tumor cell survival, we found here that elevated levels of nuclear Nrg in neurons are associated with increased sensitivity to oxidative stress and reduced life span of adult animals. We therefore conclude that these findings are of potential relevance to the management of neurodegenerative diseases associated with oxidative stress and cancer.

L1 cell adhesion molecule (L1CAM) is well-known for its importance in nervous system development and cancer progression. In addition to its role as a plasma membrane protein in cytoskeletal organization, recent in vitro studies have revealed that both transmembrane and cytosolic fragments of proteolytically cleaved vertebrate L1CAM translocate to the nucleus. In vitro studies indicate that nuclear L1CAM affects genes with functions in DNA post-replication repair, cell cycle control, and cell migration and differentiation, but its in vivo role and how its nuclear levels are regulated is less well-understood. Here, we report that mutations in the conserved ankyrin-binding domain affect nuclear levels of the sole Drosophila homolog neuroglian (Nrg) and that it also has a noncanonical role in regulating transcript levels of the oncogene Myc in the adult nervous system. We further show that altered nuclear levels of Nrg correlate with altered transcript levels of Myc in neurons, similar to what has been reported for human glioblastoma stem cells. However, whereas previous in vitro studies suggest that increased nuclear levels of L1CAM promote tumor cell survival, we found here that elevated levels of nuclear Nrg in neurons are associated with increased sensitivity to oxidative stress and reduced life span of adult animals. We therefore conclude that these findings are of potential relevance to the management of neurodegenerative diseases associated with oxidative stress and cancer.
L1-type cell adhesion molecules (CAMs) 4 are single-pass transmembrane glycoproteins belonging to the immunoglobu-lin family of receptors that are highly conserved from invertebrates to vertebrates (1). The structure of L1CAM consists of extracellular immunoglobulin and fibronectin type III domains, a transmembrane domain, and a cytoplasmic tail harboring an ezrin-binding FERM domain as well as an ankyrinbinding FIGQY domain (Fig. 1A). In its unphosphorylated state, the highly conserved FIGQY domain reversibly binds to ankyrin (Fig. 1A), which couples L1-type CAMs to actin. This interaction is known to mediate neuritogenesis, synapse growth, and stability (2)(3)(4). In contrast, phosphorylation of the tyrosine in the FIGQY domain inhibits ankyrin binding (5,6).
In addition to its function as a cytoskeleton-organizing protein at the plasma membrane, vertebrate L1CAM can be proteolytically cleaved with fragments translocating to the nucleus (7)(8)(9)(10)(11). The 200-kDa full-length L1CAM is cleaved proximal to the plasma membrane by metalloproteases to a 32-kDa fragment (12,13). The 32-kDa fragment is further cleaved by ␥-secretase/presenilin, releasing a 28-kDa cytosolic fragment containing the intracellular domain (ICD), which translocates to the nucleus (10,11). Similar to full-length L1CAM, recombinant expression of L1-ICD in nonneuronal cell lines affects gene expression of CRABPII and ␤3-integrin (11). In addition, it was shown that nuclear L1-ICD also led to up-regulation of NBS1 (Nijmegen breakage syndrome gene) via c-Myc, which plays a role in DNA damage checkpoint responses and renders human glioblastoma stem cells chemo-and radioresistant (7,14). However, little is known about the normal role of nuclear L1-type CAMs in the nervous system. In the nervous system, it was shown that an endocytosed sumoylated 70-kDa L1CAM transmembrane fragment is released from endosomes into the cytosol and translocates into the nuclei of cerebellar neurons with potential functions in development, regeneration, neurodegeneration, and synaptic plasticity (8). The 70-kDa L1CAM fragment can also be cleaved into a sumoylated 30-kDa cytosolic fragment (Fig. 1D) that translocates to the nucleus in neurons and glia, which was shown to promote migration and myelination in vitro (9). A mutation of Lys-1147 at the beginning of the conserved FERM binding domain (Fig. 1D) was shown to prevent the presence of the 70-and 30-kDa L1CAM fragments in chromatin fractions (8,9), although a second putative nuclear localization motif (NLS) adjacent to the FIGQY motif was unaltered (Fig. 1D).  cro ARTICLE However, it remains to be determined whether other motifs have a role in regulating nuclear levels of L1CAM, whether altered nuclear levels of L1-type CAMs also affect c-Myc transcript levels in neurons, and what the in vivo effects are of altered nuclear levels of L1-type CAMs.
Here, we use Drosophila as a model to address the above questions in the adult nervous system. The domain structure of the sole Drosophila homolog neuroglian (Nrg) is conserved, and the gene produces a neuron-specific isoform, Nrg180, and a ubiquitous isoform, Nrg167, via alternative splicing allowing for genetic alteration of the FIGQY motif for each individual isoform (3).
This study involves transgenic overexpression of Nrg180 and L1CAM constructs as well as animals carrying mutations specific for the neuronal isoform of Nrg (Nrg180) to determine the domain requirements and in vivo effects. We previously used nrg pacman mutants to determine the functional role of various domains in Nrg180 in nervous system development and maintenance (3,15). In these mutants, the loss-of-function allele nrg 14 was rescued with gene constructs that carry the entire WT nrg gene locus (nrg 14 ;Pac[WT]) or the nrg gene locus in which the neuron-specific exon was mutated (3). WT and mutant Nrg180 proteins are expressed at endogenous levels in these pacman mutants (3); thus, they are a valuable tool to determine domain-specific functions in regulation of nuclear levels of Nrg in vivo.

Nuclear localization of human L1CAM in Drosophila neurons
Cytosolic as well as transmembrane L1CAM fragments translocate to the nucleus in vertebrates (7)(8)(9)11). To test whether the mechanism for nuclear localization of human L1CAM is also conserved in Drosophila, we transgenically expressed L1CAM in the giant fiber (GF) neurons. The somas of the GFs are located in the brain (Fig. 1B) and, due to their large nuclear size (Ն5 m), allow for the determination of subcellular localization in vivo. In addition, we previously demonstrated that human L1CAM can rescue developmental loss of function phenotypes of Nrg in the GFs and that the 200-kDa full-length isoform is proteolytically cleaved into 65-70-kDa fragments in the fly nervous system (16,17).
We used the R68A06-Gal4 line to selectively drive expression of human L1CAM in the GFs (18). In addition to L1CAM labeling of the plasma membrane and vesicular compartments in the cytoplasm, we also observed L1CAM labeling in the nucleus (Fig. 1B, right panels, green). Whereas a single confocal slice of the GF soma intersecting the nucleus demonstrated the presence of L1CAM in the nucleus, the absence of immunohistochemical labeling in a soma that does not transgenically express L1CAM demonstrated the specificity of the antibody directed against the intracellular domain of L1CAM (Fig. 1B, right panels). We quantified L1CAM labeling in six somas for the entire nuclear volumes and observed on average 3.2 Ϯ 0.47 L1CAM puncta. In addition, we found that fluorescent sum intensity of the entire nuclear areas of L1CAM expression was statistically significant (p ϭ 0.0002, two-tailed Student's t test) from background labeling in nuclear areas that did not express L1CAM (n ϭ 5) and showed no specific L1CAM puncta in the nucleus. This suggests that L1CAM is imported to the nucleus in Drosophila neurons.
To further confirm that L1CAM translocates to nuclei in Drosophila and to determine the relevant isoforms, we employed two distinctive fractionation kits (Fig. 1C). The Thermo Fisher Scientific NE-PER kit (Fig. 1C, left) allows the separation of the cytoplasm and of the whole nucleus excluding the nuclear membrane with high yields. To further distinguish between soluble nuclear and chromatin-bound content, we used the Thermo Fisher Scientific Subcellular Protein Fractionation Kit (Fig. 1C, right). Although this kit also generates cytosolic, membrane, and cytoskeletal fractions, we only used the membrane-specific fraction as a nonnuclear control. In addition to WT L1, we also tested the effects of four alanine missense mutations (L1-4A) in the FERM domain on nuclear localization and chromatin binding (19). For fractionations of nervous systems, we used the nSyb-Gal4 and A307-Gal4 lines to pan-neuronally or broadly express L1CAM throughout the central nervous system, including the GF, respectively.
With both fractionation kits, we detected not only the 65-70-kDa L1CAM fragment but also the full-length 200-kDa L1CAM in the nuclear/chromatin fractions (Fig. 1C). Interestingly, both isoforms were detected in fractions (n ϭ 3) that contained the whole nucleoplasm but were absent in chromatin fractions in animals expressing L1-4A. The synaptic vesicleassociated cysteine string protein (Csp), nuclear cytoskeleton protein lamin, and histone were only detected in the respective membrane, soluble nuclear, or chromatin fractions, demonstrating subcellular specificity of the fractions (Fig. 1C). In summary, our findings suggest that nuclear import of L1CAM is conserved in Drosophila and that full-length L1CAM as well as L1CAM fragments translocate to the nucleus. In addition, our results suggest that the mutated amino acids in the FERM motif are required for chromatin binding in vivo.

Mutations in the FIGQY motif affect nuclear levels of Drosophila Nrg180
We previously demonstrated a developmental role for Nrg180 in the GFs (3,16,17). To further investigate whether Nrg180, similar to vertebrate L1CAM, translocates to the nucleus in adult neurons, we analyzed the subcellular Nrg180 localization in GF somas using the Nrg180 cyto antiserum. Nrg180 cyto is directed against a C-terminal sequence of the neuronal Nrg180 isoform (Fig. 1D) (3). As a control for the specific labeling of antibodies, we used nrg 14 ; Pac[nrg180 ⌬C ] animals in which the neuron-specific exon of Nrg180 is deleted (3). On average, 7.5 (n ϭ 6) Nrg puncta ( Fig. 2A, green) were labeled with Nrg180 cyto in the GF nucleus ( Fig. 2A, magenta). Analysis of the sum fluorescent intensity labeled by Nrg180 cyto in the nuclei areas showed a significant 2-fold difference when compared with nrg 14 ; Pac[nrg180 ⌬C ] ( Fig. 2A), suggesting that Nrg is imported to the nucleus.
Canonical nuclear localization motifs typically contain a series of positively charge amino acids, such as arginine and lysine, and the FERM domain has previously been shown to play a role in nuclear localization of L1CAM (8,9). However, in three independent soluble nuclear fractions, we detected L1-4A Ankyrin-binding motif in regulating nuclear levels of Nrg ( Fig. 1C), suggesting that other domains play a role in regulating nuclear levels of L1-type CAMs. Therefore, to confirm that Nrg180 localizes to nuclei in the nervous system and to test for the involvement of other domains required for regulation of nuclear levels of Nrg180, we performed cellular fractions of available nrg pacman mutants in which the FIGQY motif and the c-terminal PDZ motifs of Nrg180 are deleted (3).
In WT (nrg 14 ;Pac[WT]) and nrg 14 ;Pac[nrg180 ⌬PDZ ] mutants but not in nrg 14 ; Pac[nrg180 ⌬C ] and nrg 14 ;Pac[nrg180 ⌬FIGQY ] animals, we observed full-length Nrg180 as well as smaller frag- B, subcellular localization of human L1CAM in the GF soma. Left, schematic representation of the giant fiber circuit in the Drosophila nervous system. The GF soma (gray square) is in the brain, and the GF terminals synapse with the tergo-trochanteral motoneuron (TTMn) in the ventral nerve cord (VNC). For visualization of the GF morphology, fluorescent dyes are injected into the GFs at the cervical connective (CvC). Right panels, UAS-mCherry-NLS was co-expressed with and without UAS-L1 in the GFs using the R68A06-Gal4 line. A single confocal slice intersecting the GF nucleus and probed with L1CAM antibodies against the cytoplasmic domain is shown. Anti-L1CAM labeling is shown with (green, top panels) and without (gray, bottom panels) nuclear labeling by mCherry-NLS (magenta). In addition to labeling of L1CAM at the plasma membrane and vesicular compartments in the cytosol, L1CAM puncta can also be seen in the nucleus (arrows). No specific L1CAM labeling is detected in GF somas that do not express L1CAM. C, subcellular localization of human L1CAM (L1) in the Drosophila nervous system using two distinct fractionation kits. Left, the A307-Gal4 driver was used to express UAS-L1 broadly in the nervous system inclusive of the GF circuit. The Thermo Fisher Scientific NE-PERா kit was used to obtain cytoplasmic and nuclear fractions. The cytoplasmic fraction of ϳ7 brains and the nuclear fraction of ϳ21 brains were loaded per lane. The full-length (200-kDa) as well as the proteolytically cleaved L1CAM (65-70-kDa) isoforms were detected in the cytoplasmic as well as nuclear fractions. Anti-lamin and anti-Csp labeling served as loading controls for the nuclear and cytoplasmic fractionation. Right, human WT L1CAM (L1) and L1CAM carrying four alanine missense mutations (L1-4A) in the FERM motif were panneuronally (nSyb-Gal4) expressed in Drosophila. The Thermo Fisher Scientific Subcellular Protein Fractionation Kit was used to obtain membrane, soluble nuclear as well as chromatin fractions. The membrane fraction of 10 heads, nuclear fraction of 100 heads and chromatin fraction of 100 heads were loaded per lane. Anti-Csp, anti-lamin, and anti-histone served as loading controls for the membrane, nuclear, and chromatin fractions, respectively. D, amino acid sequences of the ICDs of human L1, L1-4A, and Nrg180. The putative NLS and ankryin-binding motifs are labeled in green and red, respectively. Peptide sequence used to generate the Nrg cyto antiserum is indicated (3). ments of very low abundance in nuclear fractions, whereas WT and mutant Nrg180 proteins were detected in membrane fractions (Fig. 2B). We previously showed that expression levels of nrg 14 ;Pac[nrg180 ⌬FIGQY ] and nrg 14 ;Pac[WT] are equal in larval nervous systems (3). Here, we confirmed that total protein levels of Nrg180 and Nrg180 lacking its FIGQY motif in the mutant animals are also similar in protein extracts of adult nervous systems (Fig. 2C). Thus, these results suggest that the FIGQY motif but not the PDZ motif has a role in regulating nuclear levels of Nrg180.
To further investigate the role of the FIGQY motif in regulating nuclear levels of Nrg180, we took advantage of pacman mutants that carry missense mutations of the tyrosine residue within the FIGQY motif. These mutants mimic either constitutively phosphorylated (Y-D) or nonphosphorylated (Y-F) isoforms of Nrg180 (3). We previously showed that all rescue constructs express WT and mutant Nrg180 at endogenous protein levels in larval brains (3). Similarly, we find that total expression levels of WT and mutant Nrg180 proteins are comparable with control animals in the adult nervous system as well (Fig. 3A). With the NE-PER kit, we found that the nuclear levels of full-length Nrg180 in nrg 14 and nrg 14 ;Pac[nrg180 ⌬C ] but not in nrg 14 ;Pac[nrg180 Y-D ] mutants were significantly reduced when compared with control animals, whereas the Lamin loading control was similar for all genotypes (Fig. 3, B and D). Further analyses with the Thermo Fisher Scientific Subcellular Protein Fractionation Kit showed similar levels of Nrg180 ⌬FIGQY , Nrg180 Y-F , Nrg180 Y-A , and Nrg180 Y-D proteins in membrane fractions when compared with WT Nrg180 (Fig. 3, C and D). In contrast, significantly less or no Nrg180 was detected in the soluble nuclear and chromatin fractions of nrg 14 ;Pac[nrg180 Y-F ] and nrg 14 ; Pac[nrg180 ⌬FIGQY ] mutants, whereas no significant difference from control animals was observed for nrg 14 ; Pac[nrg180 Y-D ] animals (Fig. 3, C and D). Consistent with this, we found visibly more abundant labeling of Nrg180 Y-D than Nrg180 Y-F and Nrg180 ⌬FIGQY proteins in the GF nucleus (Fig. 3E). These results further support a critical role for the tyrosine in FIGQY motif in regulating nuclear levels of Nrg180.
However, Nrg180 Y-A protein was present in soluble nuclear and chromatin fractions, whereas its levels were only mildly reduced in both fractions (Fig. 3, C and D). Whereas a Y-A

Ankyrin-binding motif in regulating nuclear levels of Nrg
mutation is often used to mimic the "unphosphorylated" form, we previously found that this is not the case for binding of the FIGQY motif to ankyrin; in contrast to the Y-F mutation, the Y-A mutation severely disrupted ankyrin binding (3). Because an alanine is structurally very distinctive from a phenylalanine and because it does not carry a negative charge, Nrg180 Y-A proteinmaynotactastheunphosphorylatedorphos-phorylated isoform in vivo. This may be a cause for its inconsistent subcellular localization.

C-terminal GFP tag but not HA tag affects nuclear levels of full-length Nrg180
Overexpression of full-length L1CAM or expression of the intracellular domain of L1CAM in nonneuronal cell lines Ankyrin-binding motif in regulating nuclear levels of Nrg results in altered target gene expression, such as the oncogene c-Myc (7,11). To further study the functional role of Nrg in the nucleus and to determine whether it also affects Myc transcript levels in the Drosophila nervous system in vivo, we generated a UAS construct that allows the expression of the Nrg180 intracellular domain with an N-terminal HA tag (hereafter referred to as (Nrg ICD-HA ). In addition to Nrg ICD-HA , we also analyzed the subcellular localization of C-terminally HA-and GFPtagged full-length Nrg180 (hereafter referred to as Nrg HA and Nrg GFP ). We were able to detect panneuronally (nSyb-Gal4) expressed Nrg HA and Nrg ICD-HA in nuclear fractions of 80 Drosophila brains (Fig. 4A), and the HA-tagged Nrg180 proteins were also detected in the nucleus when selectively expressed in the GF neurons with the R68A06 Gal4 driver (Fig. 4C). In contrast to Nrg HA , in nuclear fractions of 80 brains, Nrg GFP was not detected (Fig. 4B), and immunohistochemical labeling of Nrg-GFP in the GF somas showed that it was present in the cytoplasm and the perinuclear region but not inside the nucleus (Fig. 4C). This suggests that the presence of a GFP but not an HA tag at the C terminus prevents or significantly reduces localization of Nrg in the nucleus.

Nrg180 FIGQY motif affects Myc transcript levels in the Drosophila nervous system
Expression of full-length L1CAM or of the intracellular domain of L1CAM resulted in up-regulation of Myc transcript levels in vitro (7,11). Here, we used quantitative real-time PCR to determine whether Nrg180 gain of function and mutations in the FIGQY motif affect Myc transcript levels also in nervous systems. We observed reduced Myc transcripts in nrg 14 Fig. 4D). These results correlate well with the nuclear but not the membrane levels of Nrg180 in the different mutants (Fig. 3). Similar to the in vitro studies with L1CAM, we observed the opposite in Nrg gain of function animals (7,11). The Myc transcript levels were increased 3-5-fold when Nrg ICD-HA , Nrg HA , or untagged Nrg180 was overexpressed panneuronally (Fig. 4D). However, overexpression of Nrg GFP did not affect Myc transcript levels, which is consistent with our observation that Nrg GFP was not detected in the nucleus (Fig. 4D). Although we cannot exclude the possibility that the failure of Nrg GFP to not affect Myc transcript levels is not derived from a dysfunction at that plasma membrane that is distinct from its failure to localize to the nucleus, our results nevertheless suggest a novel noncanonical function for the FIGQY motif in regulating Myc transcript lev-els and that this function is likely to be associated with Nrg levels in the nucleus.

Nrg180 overexpression increases sensitivity to oxidative stress and reduces life span
Expression of the intracellular domain of L1CAM in glioblastoma stem cells up-regulates c-Myc expression, which in turn was shown to result in up-regulation of the DNA damage checkpoint response gene NBS1, suggesting that nuclear L1CAM renders these tumor cells radioresistant (7). Although these findings imply a protective function in cell survival in vitro in cancer cells, the impact of excess nuclear L1-type CAMs has not been studied in vivo or in noncancerous cells. Numerous studies suggest that overexpression of c-Myc increases the reactive oxygen species levels sufficiently to induce DNA damage and apoptosis (20 -24). Therefore, we analyzed both the sensitivity to oxidative stress and the life span of flies overexpressing Nrg ICD-HA , Nrg HA , Nrg GFP , or untagged Nrg180 panneuronally. We observed that the survival rate was inversely proportional to the Myc transcript levels, when flies were exposed to 20 mM paraquat. Significantly fewer Nrg HA or Nrg180 gain-of-function animals were alive after 72 h when compared with WT controls (nSyb-Gal/ϩ) or Nrg GFP -expressing flies (Fig. 5B). Although not significant, expression of Nrg ICD-HA also increased the sensitivity to oxidative stress. Similarly, we found that the life span of animals overexpressing Nrg180, Nrg ICD-HA , or Nrg HA was significantly shorter than that of control animals (nSyb/ϩ and Nrg GFP ; Fig. 5C). Alterations of Myc transcript levels, sensitivity to oxidative stress, and longevity correlated with the overexpression of Nrg constructs that localized to the nucleus but not with an Nrg180 construct (Nrg GFP ) that did localize to the nucleus. This suggests that the impact of Nrg180 gain of function on an animal's viability is likely due to elevated levels of nuclear Nrg180 and the consequence of increased reactive oxygen species associated with elevated Myc transcript levels.

Discussion
Previous studies showed that cytoplasmic as well as transmembrane fragments of L1CAM translocate to the nucleus in vertebrate neurons and glia (8,9). Here, we found that, in addition to fragments, full-length Nrg180 and L1CAM also localize to the nucleus in Drosophila nervous systems. Whereas nuclear translocation of the proteolytically generated intracellular domains of receptors is very common, there is accumulating evidence showing that numerous holoreceptors, including tyrosine kinase receptors, epidermal growth factor receptors,

Ankyrin-binding motif in regulating nuclear levels of Nrg
and fibroblast growth factor receptors, translocate to the nucleus and potentially serve as transcription activators (25)(26)(27)(28)(29)(30)(31). A mechanism for the release of the sumoylated transmembrane 70-kDa L1CAM fragment from an endosomal compartment to the cytosol and its subsequent nuclear import has been described (8). It is thus conceivable that full-length L1-type CAMs use a similar mechanism.
The FERM-binding motif contains a putative nuclear localization motif (Fig. 1D), and in vitro studies showed that the lysine (Lys-1147) proximal to the transmembrane domain is Ankyrin-binding motif in regulating nuclear levels of Nrg required for the presence of the sumoylated 70-and 30-kDa vertebrate L1CAM fragments in the chromatin fraction (8,9). We identified four other amino acids in the FERM-binding motif (19) that were also required for chromatin binding of full-length transmembrane fragments of L1CAM but did not affect the presence of L1CAM in soluble nuclear fractions (Fig.  1C). This suggests that other motifs play a role in regulating nuclear levels of L1-type CAMS via canonical or noncanonical mechanisms (32)(33)(34)(35), and here, we reveal a novel role for the ankyrin-binding FIGQY motif in the regulation of nuclear levels of full-length Nrg180.
One mechanism by which the FIGQY motif can affect nuclear levels of Nrg is by regulating nuclear import directly, as a noncanonical NLS motif, or indirectly. Phosphorylation has been well-documented to regulate nuclear translocation and transcription activation (34). A prior study suggests that the ICD of L1CAM exists in an open and folded conformation and that phosphorylation of Tyr-1151 or Tyr-1229 within the FERM domain and FIGQY motif promotes opening of the cytoplasmic domain to mediate inside-out signaling of L1CAM (36). Therefore, phosphorylation of the FIGQY motif may have an indirect role in regulating nuclear levels of Nrg180 by inducing a conformation change of the intracellular domain as a mechanism to unmask motifs that regulate nuclear import or export. Alternatively, to regulating nuclear import, it is also conceivable that the FIGQY motif has a role in nuclear export. Here, the lack of the FIGQY motif or the unphosphorylated motif would promote Nrg180 nuclear export via an unknown mechanism.
Intriguingly, we find that the lack of the C-terminal PDZ motif or the addition of a small HA tag had no effect on nuclear levels of Nrg, but the addition of a GFP tag did (Figs. 2 and 4). It is likely that the large size of the GFP tag (ϳ27 kDa) or its tendency to dimerize (37) prevents posttranslational modification, such as phosphorylation or sumoylation required for nuclear trafficking sterically, or that it affects the normal conformation of the intracellular domain, which promotes such modifications.
The finding that decreased Myc transcript levels correlated with decreased Nrg180 levels in the nucleus but not in the membrane, which is unaltered in Nrg mutants (Figs. [3][4][5] suggests that Nrg180 exerts its effect on Myc transcript levels in the nucleus and has a promotive effect on Myc transcript abundance directly or indirectly. A role for Nrg in up-regulation of Myc transcript levels is further supported by the finding that, like for L1CAM and L1-ICD in vitro (7), the overexpression of full-length or ICD Nrg constructs that do localize to the nucleus had the reverse effect and increased Myc transcript levels (Figs. 4 and 5).
The fact that Nrg GFP overexpression did not affect Myc transcript levels, while also not localizing to the nucleus, provides further evidence that up-and down-regulation of Nrg180 levels in the nucleus alters Myc transcript levels. Therefore, Nrg GFP is also a valuable additional control to assess whether in vivo effects of Nrg180 gain-of-function constructs correlate with Myc transcript levels and nuclear levels of Nrg180.
In contrast to previous studies suggesting that elevated nuclear levels of L1CAM in glioblastoma stem cells promote cell survival by the activation of a DNA damage checkpoint gene, which renders tumor cell radio-resistant (7, 38), we found that only expression of Nrg180 constructs that localized to the nucleus and were associated with elevated Myc transcript levels

Ankyrin-binding motif in regulating nuclear levels of Nrg
correlated with an increase of an animal's sensitivity to oxidative stress and with reduced life span. Our in vivo findings are consistent with numerous previous studies in Drosophila and in vertebrates that show that increased expression of Myc is associated with an increase in reactive oxygen species, induction of DNA damage and apoptosis, and reduced life span (20, 23, 39 -43). One possible cause for the differences between our in vivo studies and the previous in vitro studies may stem from the Warburg effect; metabolism of cancer cells is distinct from WT cells (44,45).
In conclusion, aberrant expression of L1CAM correlates with poor prognosis of various cancers, including ovarian, pancreatic, glioma, and nonsmall cell lung cancer (46 -48), and high nuclear L1CAM levels play an important mechanistic role in rendering tumor cells radioresistant (7,38). Our findings that altered nuclear levels of Nrg180 associated with altered Myc transcript levels result in increased sensitivity to oxidative stress and reduced longevity (Fig. 5) suggest that abnormally high nuclear L1-type CAM levels may contribute to the pathology of neurodegenerative diseases, whereas low levels in noncancerous cells are likely to reduce oxidative stress and thereby promote cell survival. Thus, post-developmental knockdown of L1CAM selectively in the nucleus may be an effective approach to increase chemo-and radiosensitivity of tumor cells while providing protection against oxidative stress in noncancerous cells. Thus, the identification of a novel role for the FIGQY motif in regulating nuclear L1-type CAM levels and Myc transcript levels allows the exploration of new avenues to identify putative therapeutic reagents for cancer and degenerative diseases associated with oxidative stress.

Fly stocks
The  14 ;Pac[nrg180 ⌬PDZ ] pacman stocks and the 10ϫ pUAST Nrg180, Nrg180-GFP, and Nrg180-HA (inserted at attP40) lines have been described previously (3). UAS-L1 and UAS-L1-4A lines have been described previously (16,17,19). Da-Gal4 and nSyb-Gal4 were used to drive expression ubiquitously or pan-neuronally, respectively. A307 and R68A06 Gal4 lines were used to drive expression broadly, including the Giant Fiber circuitry or more exclusively in the GF, respectively (18,49). All genetic crosses were performed on standard fly media at 25°C, and 2-5-day-old flies were used for all of the experiments unless indicated otherwise.

Generation of HA-tagged Nrg ICD construct
The intracellular domain of Nrg180 (nucleotides 3761-4207) was amplified with an HA tag included in the forward primer (forward primer, CAC CAT GTA CCC ATA CGA TGT TCC AGA TTA CGC TCG ACG CAA TCG TGG CGG AAA; reverse primer, TAG ACG TAG GTG GCC ACG GC (Integrated DNA Technologies, Coralville, IA)) and inserted into the pENTR vector (Invitrogen) via TOPO cloning. The construct was subsequently recombined into pUASTattB-10xUAS destination vector (3) using the standard gateway cloning protocol. The UAS-Nrg ICD-HA construct was verified by sequencing (Genewiz LLC). For comparability with other Nrg180 UAS constructs, transgenic lines were established at the attP40 site (BestGene Inc.).

Real-time PCR
Total RNA was isolated from 15 heads for each genotype using the RNAeasy Mini Kit (Qiagen, Valencia, CA) and quantified using NanoDrop TM . RNA was then reverse-transcribed to cDNA according to the manufacturer's protocol for the Bio-Rad iScript cDNA synthesis kit. Quantitative expression was performed with SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) using the Bio-Rad CFX96 Real-Time PCR Machine 7300 Gene Expression System. Primers against Drosophila Myc (forward primer, GAG CAA CAA CAG GCC ATC GAT ATA G; reserve primer, CCT TCA GAC TGG ATC GTT TGC G) and GAPDH (forward primer, AGC CAT CAC AGT CGA TTC; reverse primer, CCG ATG CGA CCA AAT CCA T) were used. GAPDH expression was used to normalize each sample. Each experiment was done in triplicates with a nontemplate control run simultaneously. The statistical significance (p Յ 0.05) across various genotypes was assessed by multiple-comparison ANOVA (GraphPad Prism version 7.02; GraphPad Software Inc., La Jolla, CA).

Behavioral studies
The paraquat assay was performed similarly to previously described protocols (50 -53). Twenty 3-5-day-old flies were transferred to vials containing standard cornmeal media with a final concentration of 20 mM paraquat (methyl viologen hydrate (98%), ACROS Organics). The survival of 100 flies (20 flies/vial) per genotype was determined after 72 h, and the percentages of survivors were plotted. The statistical significance (p Յ 0.05) of the survivability across various genotypes was assessed by multiple-comparison ANOVA (GraphPad Prism version 7.02, GraphPad Software).
For life span studies, 20 males (10 vials/genotype) were kept in vials containing standard cornmeal medium at 25°C. The surviving flies were transferred to the new vials every 2 days, and the number of dead flies was recorded. The percentages of surviving flies were plotted on the x axis versus the age in days on the y axis. For statistical analyses, Kaplan-Meier survival analyses with pairwise comparison using the Holm-Sidak method was used (GraphPad Prism version 7.02; GraphPad Software).

GF preparation and immunohistochemistry
The procedure for adult Drosophila nervous system dissection and subsequent dye injection has been described previously in detail (54,55). To visualize the localization of protein in the giant fiber nucleus, either a 10 mM Alexa Fluor 555 hydrazide (Molecular Probes) in 200 mM KCl or TRITC-dextran (Invitrogen) in 2 M potassium acetate was injected into the GF axons by passing hyperpolarizing or depolarizing current, respectively. Alternatively, UAS-mCherry-NLS was expressed with the R68A06 Gal4 driver. The samples were fixed with 4% paraformaldehyde and incubated with the following primary and secondary antibodies: L1CAM (Santa Cruz Biotechnology, c-20, 1:50 dilution), Nrg180 cyto (1:500 dilution), GFP (ab13970, Abcam, 1:2000 dilution), donkey anti-goat Cy2 (Jackson Immu-noResearch, 1:500 dilution), goat anti-rabbit Cy2 (Jackson ImmunoResearch, 1:200 dilution), donkey anti-rabbit Alexa Fluor 488 (Invitrogen, 1:1000), or anti-HA (3F10 Roche Applied Science, 1:100 dilution). Samples were scanned at a resolution of 1024 ϫ 1024 pixels, ϫ2.5 zoom, and 0.2-m step size with a Nikon C1 or a Nikon A1R confocal microscope using a Plan Apo ϫ60 oil immersion objective. Images were processed using Nikon Elements Advanced Research version 4.4 and Adobe Photosuite CS5 software. For analysis of nuclear labeling, the Binary Editor of Nikon Elements Advanced Research 4.4 was used to trace and three-dimensionally reconstruct the GF nuclei, and the ND Images Arithmetic function was used to extract Nrg180 cyto labeling. The sum fluorescent intensity was normalized to the trace nuclei area, and the average for each genotype with the S.E. was plotted. Two-tailed Student's t test was used to determine statistically significant differences.