Characterization of Ribonucleoprotein Complexes and Their Binding Sites on the Neurofilament Light Subunit mRNA*

Levels of neurofilament (NF) gene expression are important determinants of basic neuronal properties, but overexpression can lead to motoneuron degeneration in transgenic mice. In a companion study (Cañete-Soler, R., Schwartz, M. L., Hua, Y., and Schlaepfer, W. W. (1998) J. Biol. Chem. 273, 12650–12654), we show that levels of NF expression are regulated by altering mRNA stability and that stability determinants are present in the 3′-coding region (3′-CR) and 3′-untranslated region (3′-UTR) of the NF light subunit (NF-L) transcript. This study characterizes the ribonucleoprotein complexes that bind to the NF-L mRNA when cytoplasmic brain extracts are incubated with radioactive probes. Gel retardation assays reveal ribonucleoprotein complexes that are selectively competed with poly(C) or poly(U))/poly(A) homoribopolymers and are referred to as C-binding and U/A-binding complexes, respectively. The C-binding complex forms on the proximal 45 nucleotides of 3′-UTR, but its assembly is markedly enhanced by 23 nucleotides of flanking 3′-CR sequence. U/A-binding complexes form at multiple binding sites in the 3′-CR and 3′-UTR. A pattern of reciprocal binding suggests that the C-binding and U/A-binding complexes interact and may compete for common components or binding sites. Cross-linking studies reveal unique polypeptides in the C-binding and U/A-binding complexes. The findings provide the basis for probing mechanisms regulating NF-L mRNA stability and the relationship between NF overexpression and motoneuron degeneration in transgenic mice.

Neurofilaments (NFs) 1 are the principle constituent of the axonal cytoskeleton, so that levels of NF expression are believed to be a major determinant of axonal size. This view is supported by the simultaneous up-regulation of the light (NF-L), midsized (NF-M), and heavy (NF-H) NF subunit mRNAs (2) that accompanies the enlargement and myelination of axons during postnatal development of the nervous system (3). The postnatal surge in NF expression is a posttranscriptional event (4) that is mediated in part by stabilization of NF mRNAs (5). The extent of axonal enlargement, and presumably the levels of NF up-regulation, are determined by the nature of target cell innervations and can be reconstituted in transected nerve upon successful reinnervation of the target site (6). Moreover, the stability of NF mRNAs becomes dependent upon continuity of axons with target sites in that the transcripts are destabilized upon nerve transection or excision and transfer of parent neurons to primary culture (7).
Interestingly, NF mRNAs are stabilized during the same postnatal interval in which neurons acquire the ability to survive nerve transection and mount a regenerative response (8). The overlapping appearance of these phenomena raises the possibility that common components of posttranscriptional pathways regulate neuronal homeostasis and levels of NF expression. If so, then titration of regulatory components could account for: (i) the motoneuron degeneration in transgenic mice upon overexpression of a wild-type NF-L (9) or NF-H (10) transgene or modest expression of a mutant NF-L transgene (11), (ii) the preferential degeneration of NF-enriched neurons in mice bearing a mutant SOD-1 transgene (12), and (iii) the transient up-regulation of NF mRNA that precedes the spontaneous motoneuron in Wobbler mice (13). Mechanisms regulating levels of NF expression may therefore bear an important relationship to neurodegenerative states.
We have begun to map stability determinants that regulate steady-state levels of NF-L mRNA. Deletion of the 3Ј-untranslated region (3Ј-UTR) from a mouse NF-L transgene stabilizes the NF-L transcript in neuronal cell lines and alters the developmental up-regulation and axotomy-induced down-regulation of the transgene in transgenic mice (14). More recently, we have begun to apply an expression system with tetracyclineregulated promoter to map stability determinants in NF mRNAs. In a companion study (1), we demonstrate the presence of stability determinants in the 3Ј-coding region (3Ј-CR) and 3Ј-UTR of NF-L mRNA and have localized a stabilizing element at the junction of 3Ј-CR and 3Ј-UTR that may account for the stabilizing properties of the transcript.
The present study examines the ribonucleoprotein (RNP) complexes that assemble on NF-L mRNA when nascent radioactive fragments of the transcript are incubated with cytosolic brain extracts. The RNP complexes have been characterized by their abilities to be competed with homoribopolymers, thereby taking advantage of the high affinity and selective binding of many RNA-binding proteins for specific homoribopolymers (15). A similar approach has been very helpful in characterizing RNP complexes that are believed to stabilize ␣-globin mRNA (16,17). The goal of the present study is to correlate biochemical with functional studies (1) and thereby develop a working model for elucidating the mechanisms that regulate the stability of the mouse NF-L transcript.

EXPERIMENTAL PROCEDURES
Preparation of Cytosolic Extracts-Rat brain and liver (10 -15 g) were minced, washed in phosphate-buffered saline, and homogenized in * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  a Dounce homogenizer with two volumes of 50 mM potassium acetate, 3 mM magnesium acetate, and 2 mM dithiothreitol in 20 mM HEPES buffer, pH 7.4 (Buffer A), and a mixture of protease inhibitors (18). Crude supernatant from a microcentrifuge spin (10,000 ϫ g for 10 min) was centrifuged at 100,000 ϫ g for 60 min, yielding cytosolic extracts with 7-9 or 15-20 mg/ml protein from brain and liver, respectively. Glycerol (5%) was added, and the extracts were aliquoted, snap-frozen, and stored at Ϫ80°C. All operations were conducted on ice or at 4°C.
Multiple dishes (Ͼ10 8 cells) of confluent P19, N2a, or L cells, originally obtained from the American Type Culture Collection, were washed twice in phosphate-buffered saline, detached with a scraper, pelleted in a clinical centrifuge (1,000 ϫ g for 5 min), and homogenized in a Dounce homogenized with two volumes of Buffer A containing 0.5% Nonidet P-40. Crude supernatants from a microcentrifuge spin (10,000 ϫ g for 10 min) were centrifuged at 100,000 ϫ g for 60 min, yielding cytosolic extracts (protein of 5-8 mg/ml). Extracts were admixed with glycerol, aliquoted, and stored at Ϫ80°C. Some extracts were also concentrated 3-fold in a Centricon-10 (Amicon Corp.).
Preparation of RNA Probes-Templates for RNA probes (Fig. 1) were prepared by PCR with restriction sites or a T7 promoter sequence in the primers to facilitate cloning or to generate RNA products directly from PCR templates. The parent A680 template (ϩ1482/ϩ2161) was cloned into HindIII/BamHI sites of the pBluescript II SKϩ vector (Stratagene) and was cut with BamHI (at ϩ2161), HincII (at ϩ1779), or BspMI (at ϩ1712) to yield the full-length A680 template (ϩ1482/ϩ2161), the A680/H template (ϩ1482/ϩ1779), or the A680/B template (ϩ1482/ ϩ1712). The SalI (and HincII) site in the polylinker was mutated to preserve the continuity of T7 promoter sequence with the A680/H template. The A680/del template was generated by a second PCR reactions that fused upstream and downstream PCR products with sequence between ϩ1712 and ϩ1779 deleted. Probes 1-5 were cloned into the SKϩ vector, and probes A, B, and X were generated off PCR templates. Templates were sequenced to confirm junctional and mutational sites.
RNA probes were uniformly labeled with [ 32 P]UTP using T7 polymerase (7). Full-length probes were excised from denaturing acrylamide gels, eluted overnight into 0.5 M NH 4 acetate, 0.1% SDS, and 1 mM EDTA, and ethanol-precipitated. Probes were diluted to 2.5 ϫ 10 4 cpm/l in Buffer A immediately prior to use.
Gel Retardation and Cross-linking Studies-Homoribonucleotides (Sigma) and antisense oligonucleotides were solubilized in Buffer A. Radioactive probes (5 ϫ 10 4 cpm), cytosolic extract (120 g), and varying amounts of homoribonucleotides or antisense oligonucleotides were incubated for 30 min at 20°C in 20 l, digested for 10 min with 2 units of RNase T1 (Sigma), dissociated with heparin (5 mg/ml), and electrophoresed on a 5% acrylamide gel in 0.5ϫ Tris-borate-EDTA buffer. Dried gels were subjected to autoradiography. For cross-linking analyses, the complexes were cross-linked on ice for 30 min at 3 cm distance underneath a UV light (4 ϫ 10 6 J/cm 2 ) and further digested for 15 min at 20°C with RNase T1 (10 units) or RNase A (10 g). Radioactive cross-linked polypeptides were solubilized by boiling for 5 min in SDS sample buffer, separated by SDS-polyacrylamide gel electrophoresis, and identified by autoradiography of dried gels.
Complex formations were compared on probes that were denatured at 95°C for 2 min and rapidly or slowly cooled to 20°C prior to the addition of extract. Probes were also heat-denatured with antisense oligonucleotide in excess of EDTA (5 mM) and incubated with RNase H (1 unit) for 20 min at 37°C in excess magnesium (3 mM) prior to addition of extract.

RNP Complexes Bind to the NF-L mRNA and Are Selectively
Competed with Specific Homoribopolymers-An NF-L mRNA probe (A680) that spans the entire 3Ј-UTR (427 nt) and distal 253 nt of 3Ј-CR (see Fig. 1) was used to identify binding factors and binding sites that could account for stability determinants in these regions of the transcript (1). When brain extracts were incubated with the radioactive A680 probe and electrophoresed on non-denaturing gels, multiple gel-shifted bands were detected (Fig. 2, lane 3). 1) or when the extract was treated with protease K (lane 2) or heated to 65°C (data not shown).
C-binding Complexes Require the Proximal 45 nt of 3Ј-UTR and Distal 23 nt of 3Ј-CR for Binding-To localize binding sites, gel shift assays were conducted with probes that extended to varying distances into the 3Ј-UTR (Fig. 3). The Cbinding complex (solid arrowheads) formed on the full-length A680 probe extending to ϩ2161 (lanes 1-3), to a much lesser extent on probe A680/H extending to ϩ1779 (lanes 4 -8), but not on probe A680/B extending to ϩ1712 (lanes 9 -12) or on probe A680(del) in which sequence between ϩ1712 and ϩ1779 is deleted (lanes [13][14][15]. Variations in relative intensities of gel-shifted bands of the U/A-and C-binding complexes occurred when examined with different probes. In addition, the C-binding complex was markedly enhanced on probe A680/H when U/A-binding complex was competed off with poly(U) or poly(A) (compare lanes 4 with lanes 6 -8). The findings indicate that formation of the C-binding complex requires sequence between ϩ1712 and ϩ1779. This sequence comprises the distal 23 nt of 3Ј-CR (ϩ1712/ϩ1734) and the proximal 45 nt of 3Ј-UTR (ϩ1735/1779).
To confirm the location of the binding site for the C-binding complex, gel shift studies were conducted with probe A (ϩ1712/ ϩ1779), which contained the 3Ј-CR and 3Ј-UTR components of the putative binding site, and with probe X (ϩ1735/ϩ1779), which contained only the 3Ј-UTR component of the binding site. Fig. 4 shows that the C-binding complex (solid arrowhead) readily formed on probe A (lane 1), but not very well on probe X (lane 6). Upon prolonged exposures of autoradiograms, similar but markedly reduced amounts of C-binding complex can be identified on probe X (data not shown). Low (20 ng), medium (200 ng), and high (2000 ng) amounts of poly(C) competed the C-binding complex from probe A (lanes 2-4) and probe X (data not shown). Note that the addition of poly(C) not only abolished the C-binding complex but also enhanced the formation of the slow-migrating U/A-binding complex (open arrowhead) on probe A (lanes 2-4) and probe X (data not shown). Additional gel-shifted bands formed on probe A and, especially, on probe X, but were not competed with poly(C) or poly(U), and their relationship to the C-and U/A-binding complexes is unclear. Their presence does not obscure the principle finding, i.e. that the C-binding complex binds weakly to the proximal 45 nt of 3Ј-UTR and that this binding is markedly enhanced by the addition of upstream flanking sequence.
Further evidence that 3Ј-CR is instrumental in the formation of the C-binding complex is seen in gel shift assays using a series of 5Ј-deletion probes extending from ϩ2064 (probe 1), ϩ1870 (probe 2), ϩ1768 (probe 3), ϩ1735 (probe 4), and ϩ1519 (probe 5) to ϩ2161 (see Fig. 1). Whereas probes 1-4 contained increasing amounts of 3Ј-UTR, only probe 5 extended beyond the 3Ј-UTR and into the 3Ј-CR. Fig. 5 shows that addition of poly(C) abolished the set of fast-migrating gel-shifted bands (solid arrowheads) that formed on probe 5 (compare lanes 1 and 3) but did not abolish the set of fast-migrating gel-shifted bands that formed on probe 4 (compare lanes 4 and 6), probe 3 (compare lanes 7 and 9), probe 2 (compare lanes 10 and 12) or probe 1 (compare lanes 13 and 15). The ability to be competed away with poly(C) identifies the fast-migrating set of bands on probe 5 as the C-binding complex. This complex was also enhanced in the presence of poly(U) (compare lanes 1 and 2). Gel-shifted bands of similar migration also formed on probes 1-4, but were not competed with poly(C), so that their relationship to the C-binding complex is unclear. The findings provide additional evidence that formation of the C-binding complex is dependent upon sequences in the proximal 3Ј-UTR and distal 3Ј-CR.
U  10 and 11), and probe 1 (lanes 13 and 14). Gel-shifted bands of U/A-binding complexes on probes 1-5 differed in their relative abundance. Likewise, similar gel-shifted bands in different proportions comprised the U/A-binding complex that formed on probes containing only 3Ј-CR sequence (A680/B), with the addition of proximal 3Ј-UTR (A680/H) or with the addition of the full 3Ј-UTR (A680) (see Fig. 3). U/Abinding complexes were also present on short probes (see Fig.  4) and were generally composed of fewer and fainter bands. Moreover, the U/A complexes were sometimes competed more effectively with poly(U) or poly(A), depending on the size and position of the probe. Finally, in many instances, competing off the U/A-binding complex with poly(U) or poly(A) enhanced the formation of the C-binding complex. We conclude that slowmigrating U/A-binding complexes of varying composition form at multiple sites in the 3Ј-CR and 3Ј-UTR of NF-L mRNA and that formation of U/A-binding complexes may adversely affect the formation of C-binding complexes.
Formation of C-binding Complexes Requires Multiple C-rich Sequences in the Proximal 3Ј-UTR-The sequence necessary for the formation of the C-binding complexes extends from ϩ1712 to ϩ1779 of the NF-L cDNA and spans the distal 23 nt of 3Ј-CR and proximal 45 nt of 3Ј-UTR, as follows.
To assess specific sequence requirements for the formation of the C-binding complex, gel shift assays were conducted with brain extracts and probe A (ϩ1712/ϩ1779) that had been preincubated with 15-mer antisense oligonucleotides to the proximal (oligo P), middle (oligo M), or distal (oligo D) 3Ј-UTR sequence in probe A (as schematically shown above). Fig. 6 shows that formation of the C-binding complex was abolished by preincubation with antisense probes P or M and markedly reduced by antisense probe D. Preincubation with the same amounts (100 ng) of 15-mer oligos to sequence in the SKϩ vector (oligo SK) or 5Ј-flanking region of NF-L (oligo 5Ј) did not

FIG. 4. Gel shift assay of the C-binding complex (closed arrowhead) and U/A-binding complex (open arrowhead) that form on probe A (؉1712/؉1779) and on probe X (؉1735/؉1779) after incubation with brain extract in the presence of 0 (؊), 20, (؉), 200 (؉؉), or 2000 (؉؉؉) ng of poly(U) or poly(C).
Other RNP complexes that form on the probes are not competed with homoribonucleotides. affect the formation of C-binding complexes. The findings indicate that assembly of the C-binding complexes can be disrupted at multiple sites, thereby requiring an extended stretch of intact sequence in the proximal 3Ј-UTR that is enriched in cytosine and pyrimidine residues.

Components for Assembly of C-binding Complexes Are Present in Cytosolic Extracts from Neuronal and Non-neuronal Cell Lines but Are Enriched in Neuronal
Tissues-Cytosolic extracts from neuronal (P19 and N2a) and non-neuronal (L cells) cell lines were tested in gel shift assays and were found to generate C-binding complexes (data not shown). Studies were then conducted to compare the C-binding complexes from neuronal (brain) and non-neuronal (liver) tissues. Fig. 7 shows that very similar C-binding complexes formed when brain (lanes [3][4][5][6][7][8] or liver (lanes 9-14) extracts were gel-shifted with probe A. However, at least a 10-fold larger amount of C-binding complex formed with brain extracts, based upon radioactivity per microgram of protein in the cytosolic extract.

C-binding Complexes Contain 18-and 36-kDa Polypeptides, whereas the U/A-binding Complexes Contain 72-and 80-kDa
Polypeptides-Components of the C-binding and U/A-binding complexes were identified by UV cross-linking of the complexes that form on radioactive probes, then digested away the radioactive residues that are not cross-linked to protein and using the cross-linked radioactivity to detect polypeptides in autoradiograms after their separation by SDS-polyacrylamide gel electrophoresis. When polypeptides in rat brain extracts were cross-linked to probe B (ϩ1670/ϩ1779) and separated on a 15% acrylamide gel, polypeptides with migrational rates of 39-and 18-kDa were identified as components of the C-binding complexes (Fig. 8A). Their association with the C-binding complexes was attributed to their ability to be competed with poly(C). The 39-kDa polypeptide was competed with low levels (20 ng), whereas loss of the 18-kDa polypeptide required higher levels (2000 ng) of poly(C). Cross-linking of the 18-kDa and, especially, the 39-kDa polypeptides was enhanced by the addition of poly(U). Digestion of the cross-linked complexes using RNase A (in addition to T1 nuclease) increased the migrational rate of the 39-kDa polypeptide to that of a 36-kDa polypeptide. Addition of poly(U) also competed away the cross-linking of an 80-kDa polypeptide.
When polypeptides in brain extracts were cross-linked to probe A (ϩ1712/ϩ1779) and separated on a 10% gel (Fig. 8B), 80-and 72-kDa polypeptides were competed away by the addition of poly(U) (compare lanes 2 and 4) and were slightly increased by the addition of poly(C) (compare lanes 2 and 3).  1 and 2). The findings identify 18-and 36-kDa core polypeptide components of the C-binding complex and 80-and 72-kDa components of the U/A-binding complex.

DISCUSSION
With this study, we have begun to identify the RNP binding factors and binding sites that regulate the stability of NF-L mRNA. Competition with homoribopolymers (15)(16)(17) has been very helpful in characterizing specific RNP components and in localizing their binding sites. In particular, the ability to be selectively competed with poly(C) has identified a prominent C-binding complex that binds to the proximal 3Ј-UTR of NF-L mRNA. In a companion study (1), we have mapped the stability determinants in the NF-L transcript and have shown that the major determinant of NF-L mRNA stability is localized to the binding site of the C-binding complex. The identification and characterization of RNP complexes on the NF-L transcript are therefore directly relevant to the mechanisms that regulate the stability of the NF-L transcript.
The C-binding complex is unusual, in that its binding site is located in the proximal 45 nt of 3Ј-UTR but the binding reaction is markedly enhanced by the addition of the 23-nt flanking upstream sequence in the 3Ј-CR. The ability of distal 3Ј-CR to enhance formation of C-binding complexes could be due to its participation in secondary (or tertiary) structure. Computer modeling based on the free energy minimization algorithm of Zuker et al. (19) indicates that the distal 3Ј-CR forms a stable stem structure with the proximal 3Ј-UTR. Alternatively, it is also possible that other factors bind to the 3Ј-CR and facilitate formation of the C-binding complex. Immediate upstream flanking sequences are necessary for binding to a 29-nt desta- bilizing element in the 3Ј-UTR of human amyloid precursor protein mRNA (20). Remote upstream sequences may also be required to form an essential stem structure (21).
The binding site for the C-binding complexes in the proximal 3Ј-UTR of NF-L mRNA is phylogenetically conserved (Table I). Several pyrimidine-rich stretches are present in proximal 45 nt of the 3Ј-UTR and may be instrumental for binding of the complex. Disruption of RNA structure by incubating probes with antisense oligonucleotides shows that the C-binding complex requires the integrity of sequences spanning the proximal, middle and distal pyrimidine-rich sites. Flanking and intervening sequences are also conserved (Table I) and may also be important for the formation of the C-binding complex.
Pyrimidine-rich stretches in the proximal 3Ј-UTR are also assembly sites for RNP complexes that regulate stability of other mRNAs (16,17,(22)(23)(24). The pyrimidine composition of some of these sites is C-rich and complex formation is also inhibited by poly(C) homoribopolymers (16,22). Moreover, pyrimidine-rich sites often contain tandem repeats of sequencespecific motifs that are not functionally redundant in that complex formation is markedly impaired if a single motif repeat is disrupted (16,22,24). Whereas point mutations in some residues can be very disruptive, complex formation may tolerate or even be enhanced by exchanging the C and U pyrimidine residues in the binding site (22,25). Interestingly, the sequence as well as the recognition site in the cognate binding factors have diverged during evolution so that the complex on the ␣-globin mRNA binds to a C-rich motif in the human but C/U-rich motif in the mouse (17). Splicing factors that bind to polypyrimidine tracts of intervening sequences have distinctive but overlapping sequence specificities (26 -28). Indeed, competition between multiple trans-acting factors for binding to polypyrimidine tracts is believed to be instrumental in splicesome assembly and in the selection of the 3Ј-splice sites.
An unusual feature of the C-binding complex is that its assembly is effected by the presence of other RNP complexes on the transcript. Gel shift studies suggest that the C-and U/A-binding complexes have reciprocal binding interactions, as if the two sets of complexes compete for the same components or use the same or nearby sites. These binding features were most readily observed with short probes that spanned the C-binding and nearby U/A-binding sites. Moreover, reciprocal binding phenomena were readily apparent in cross-linking experiments, suggesting that interactions between the complexes alter core (i.e. RNA contacting) binding components of the respective complexes.
The assembly of similar U/A-binding complexes on probes to different regions (e.g. probes A, 1, and A680/B) of the NF-L transcript indicates that the complexes bind to multiple sites in the 3Ј-UTR and 3Ј-CR. Such a multiplicity of binding sites in the 3Ј-UTR and 3Ј-CR is reminiscent of some adenylate/uridylate-rich elements (ARE) that were originally described as destabilizing determinants in short-lived mRNAs of proto-oncogene, cytokines, and transcription factors (32). For example, destabilization of ␤-interferon mRNA is mediated by multiple AREs that compete with each other and with poly(U) or poly(A) FIG. 8. A, polypeptide in mouse brain extracts cross-linked to probe B (ϩ1640/ ϩ1779) in the presence of 0 (Ϫ), 20 (ϩ), or 2000 (ϩϩϩ) ng of poly(U) or poly(C), with (ϩ) and without (Ϫ) further digestion with RNase A or T1, and separated on a 15% SDS-polyacrylamide gel. Sizes of radioactive polypeptides (left-hand margin) are estimated by their rates of comigration relative to those of known standards. B, polypeptides in mouse brain extracts cross-linked to probe A (ϩ1712/ϩ1779) in the presence of 0 (Ϫ) or 2000 (ϩϩϩ) ng of poly(U) or poly(C), digested with RNase T1, and separated on a 10% SDS-polyacrylamide gel. Polypeptide sizes (lefthand margin) are estimated from the comigration of known standards. Control samples without extracts are shown in lanes 1.

TABLE I
Comparison of the sequences in the proximal 3Ј-UTR (lowercase) and adjacent 3Ј-CR (uppercase) of the mouse (38), human (39), rabbit (40)  in the binding of a 65-kDa polypeptide (33). Agonist-or hypoxia-induced stabilization of mRNAs may be mediated by masking destabilizing AREs (34) or by a different set of AREs (35). Binding of the same set of factors to multiple sites in the 3Ј-UTR is also believed to regulate the stability of the neuronal GAP-43 mRNA (31). Moreover, the GAP-43 binding factors are neuron-specific, including a factor that was recently identified as an Elav-like protein (36). Interestingly, Elav-like proteins are essential for the differentiation and maintenance of neurons and have been identified as the target immunogens that mediate autoimmune neurodegenerative disease (37).
The extent to which different components of the C-and U/A-binding complexes are neuron-specific is presently unknown. It is quite possible, for example, for neuron specificity to be mediated by altering (e.g. phosphorylation) a common component, by changing the concentration of a critical component(s), or by adding a novel component to the RNP complex in a neuronal setting. The 18-and 36-kDa polypeptide core binding components of the C-binding complexes are novel RNP components, which are enriched in neuronal tissues. They could represent subunits with similar properties or monomer/ dimer components of a common subunit. In the latter instance, the ability of poly(C) to compete preferentially with the binding of the 36-kDa polypeptide would indicate a greater effect on the assembly of the dimer in the complex. In either case, the subunits are smaller than those of other RNP components that bind to C-rich sites and are competed with poly(C) (16, 22, 29 -31).
The role of the C-binding and U/A-binding complexes in regulating the stability of the NF-L mRNA is presently unknown. The identification of the complexes and their binding sites will now enable further characterization of the RNP components and their role in stabilizing NF-L transcript. It is quite conceivable that factors regulating NF-L mRNA stability could also play a major role in the post-transcriptional regulation of neuronal metabolism. In particular, the findings could provide important insights into the relationship between overexpression of an NF-L transgene and the selective degeneration of motor neurons in transgenic mice.