Developmental Regulation of RNA Transcript Destabilization by A + U-rich Elements is AUF1-dependent*

The developmental immaturity of neonatal phagocytic function is associated with decreased accumulation and half-life (t 1 2 ) of granulocyte/macrophage colony-stimulating factor (GM-CSF) mRNA in mononuclear cells (MNC) from the neonatal umbilical cord compared with adult peripheral blood. The in vivo t 1 2 of GM-CSF mRNA is 3-fold shorter in neonatal (30 min) than in adult (100 min) MNC. Turnover of mRNA containing a 3′-untranslated region (3′-UTR) A + U-rich element (ARE), which regulates GM-CSF mRNA stability, is accelerated in vitro by protein fractions enriched for AUF1, an ARE-specific binding factor. The data reported here demonstrate that the ARE significantly accelerates in vitrodecay of the GM-CSF 3′-UTR in the presence of either neonatal or adult MNC protein. Decay intermediates of the GM-CSF 3′-UTR are generated that are truncated at either end of the ARE. Furthermore, thet 1 2 of the ARE-containing 3′-UTR is 4-fold shorter in the presence of neonatal (19 min) than adult (79 min) MNC protein, reconstituting developmental regulation in a cell-free system. Finally, accelerated ARE-dependent decay of the GM-CSF 3′-UTRin vitro by neonatal MNC protein is significantly attenuated by immunodepletion of AUF1, providing new evidence that this accelerated turnover is ARE- and AUF1-dependent.

Neonatal myelopoiesis is developmentally immature compared with that of adults (reviewed in Ref. 1). Although both the circulating levels and proliferative capacity of myeloid progenitor cells are elevated in neonatal humans and rats, total body neutrophil storage pools are significantly decreased in neonatal compared with adult rats (reviewed in Ref. 2). Neonatal rats cannot further increase their myeloid progenitor pool size or proliferative rate in response to experimental sepsis, whereas adult animals can. This impaired response leads to a further reduction in their already depleted neutrophil storage pools, a condition also associated with a fatal outcome in septic human newborns. These immaturities, coupled with diminished neonatal phagocytic function (3,4), appear to predispose newborns to depletion of mature effector phagocytes during overwhelming bacterial sepsis. Prophylactic (5) and therapeutic (6) administration of recombinant granulocyte/macrophage colony-stimulating factor (GM-CSF) 1 enhances the survival of neonatal rats during experimental sepsis. Recombinant human GM-CSF has also been shown to increase neutrophil, eosinophil, monocyte, and platelet production in very low-birthweight human newborns (7,8).
GM-CSF modulates myeloid proliferation, differentiation, and activation (reviewed in Ref. 9). The expression of GM-CSF mRNA is regulated post-transcriptionally by a destabilizing, adenylate ϩ uridylate (A ϩ U)-rich element (ARE) comprising AUUUA motifs (10 -12) within the 3Ј-untranslated region (3Ј-UTR). These elements are present in the 3Ј-UTR of a number of proto-oncogene and cytokine transcripts, including nearly all of the colony-stimulating factors, interleukins, and interferons (reviewed in Ref. 13). AREs are also evolutionarily conserved in the 3Ј-UTRs of diverse invertebrate and vertebrate transcripts (14). The degradation kinetics of these transcripts is dependent upon the presence of multiple AUUUA motifs (reviewed in Ref. 15) and, in particular, the nonameric sequence UUAUUUA(U/ A)(U/A) (16 -18). The human GM-CSF ARE consists of a cluster of eight AUUUA motifs including three overlapping nonamer sequences ( Fig. 1 and Ref. 19), characteristic of class II AREs, which initiate mRNA decay through processive deadenylation (15). ARE-targeted ribonuclease activities (20 -24) and many ARE binding factors (24 -33) have now been identified in mammalian cells, but the function of most is not yet clear. Two of these factors, AUF1 (34) and HuR (35), have been cloned and found to exhibit apparently opposing functional activities. HuR stabilizes ARE-containing mRNA when overexpressed in transfected fibroblasts (36,37), whereas AUF1-enriched protein fractions accelerate ARE-dependent in vitro mRNA decay (26). However, the exact mechanism by which transcript turnover occurs, including the explicit function(s) of AUF1, remains to be determined.
GM-CSF is produced by human peripheral blood mononuclear cells (MNC) activated by phorbol myristate acetate (PMA) ϩ phytohemagglutinin (PHA) (38). Expression of GM-CSF is differentially regulated depending upon whether its cellular source is neonatal or adult. For example, GM-CSF protein and mRNA are expressed at 7-and 4-fold lower levels in activated neonatal umbilical cord blood-derived MNC than in adult pe-ripheral blood-derived MNC (39,40). Because the rate of gene transcription is comparable in neonatal and adult MNC (41), the 4-fold lower level of GM-CSF mRNA in neonatal MNC is most likely because of its 3-fold shorter half-life (t1 ⁄2 ) in neonatal MNC (39). Several other ARE-containing cytokine transcripts, including macrophage colony-stimulating factor (42), granulocyte colony-stimulating factor (43), macrophage inflammatory protein-1␣ (44), interleukin-3 (43), interleukin-8 (44), and interleukin-12 (45), are also less abundant in neonatal than in adult MNC. These differences may also be the result of decreased mRNA stability in neonatal MNC (41,42,44,45). Taken together, these data suggest that increased turnover of cytokine transcripts in neonatal MNC could account for their reduced production and, in part, for the dysregulation of neonatal phagocytic immunity.
To explore the molecular basis for this increased turnover, we previously compared interaction of the GM-CSF 3Ј-UTR with neonatal versus adult MNC proteins, including AUF1 (46). The results of these studies suggest that increased ARE binding by specific AUF1 isoforms may contribute to the developmentally increased rate of GM-CSF mRNA turnover in neonatal MNC. The present study was designed to determine the functional roles of both the ARE and AUF1 in regulating the turnover of GM-CSF mRNA by neonatal versus adult MNC. A cell-free system (47) was employed to assay mRNA degradation activity in MNC protein extracts. Our data indicate that the GM-CSF ARE directs rapid in vitro RNA decay, characterized by the generation of discrete RNA decay intermediates. Moreover, decay is significantly more rapid in the presence of neonatal compared with adult MNC protein. Finally, immunodepletion of AUF1 from neonatal MNC protein extracts significantly stabilizes RNA containing the GM-CSF ARE. These results indicate that accelerated in vitro RNA decay in the presence of neonatal MNC protein is ARE-directed and AUF1-dependent.

EXPERIMENTAL PROCEDURES
Human MNC Isolation and Cell Culture-Human MNC were isolated from normal term umbilical cord or donor adult peripheral blood and cultured as described previously (42). After incubation overnight at 37°C in 5% CO 2 , cultures were stimulated for 6 h with 20 ng/ml PMA (Sigma) ϩ 2 g/ml PHA (Life Technologies, Inc.).
In Vitro RNA Synthesis and 32 P Labeling-The human GM-CSF 3Ј-UTR, a 305-base pair MscI-XhoI fragment derived from pXM:GM-CSF (courtesy of Dr. G. G. Wong, Genetics Institute), was cloned into the pBluescriptII:KS(Ϫ) transcription vector (Stratagene) digested with SmaI and XhoI ( Fig. 1) to create pBsGM305. Internally, 32 P-labeled RNA transcripts were synthesized from pBsGM305 by linearization at the restriction sites shown (XhoI for 8AU and MboII for 0AU; Fig. 1), followed by in vitro transcription with T7 RNA polymerase (48) in the presence of 60 Ci of [␣-32 P]UTP (800 Ci/mmol, Amersham Pharmacia Biotech) ϩ 50 M UTP, or 0.2 Ci [␣-32 P]UTP ϩ 500 M UTP for trace-labeled transcripts. A 32 P-labeled, 335-nt RNA transcript derived from the pBluescriptII:KS(ϩ) vector was synthesized by linearization with PvuII, followed by in vitro transcription as described above. Unincorporated [␣-32 P]UTP was removed by NICK-column chromatography (Amersham Pharmacia Biotech) and transcript yield was determined by scintillation counting of acid precipitates (49,50). Tracelabeled in vitro transcripts were 5Ј-end 32 P-labeled using the T4 polynucleotide kinase exchange reaction with 30 Ci of [␥-32 P]ATP (6000 Ci/mmol, Amersham Pharmacia Biotech). All enzyme reagents were obtained from Life Technologies, Inc. and used according to their recommendations.
Cell-free RNA Decay Assay-RNA degradation activity in the cytoplasmic extracts was assayed by modification of methods previously established for polysomal extracts, using radioactive substrates (47). Decay reactions containing 2 g of cytoplasmic protein and 0.5 ng of 32 P-labeled 8AU or 0AU RNA (200 -1000 kcpm/ng) in 25 l of decay buffer (10 mM Tris-HCl, pH 7.6, 100 mM KOAc, 2 mM Mg(OAc) 2 , 2 mM dithiothreitol, 0.1 mM spermine, 1 mM ATP, 0.4 mM GTP, 10 mM phosphocreatine (Sigma), 1 g creatine phosphokinase (Sigma), 1 unit/l RNasin) were incubated at 37°C for 30 to 120 min, after which reactions were terminated with 400 l of urea lysis buffer (10 mM Tris-HCl, pH 7.5, 350 mM NaCl, 10 mM EDTA, 2% SDS, 7 M urea). After the addition of 20 g of Escherichia coli transfer RNA (Sigma) as a carrier, phenol:chloroform extraction, and ethanol precipitation, radioactive pellets were resuspended in 10 l of distilled H 2 O ϩ 10 l of denaturing RNA gel loading buffer (80% formamide, 10 mM EDTA, 1 mg/ml bromphenol blue (Sigma), 1 mg/ml xylene cyanol (Sigma)) and electrophoresed on denaturing 7 M urea-5% (1:29 bis:acrylamide) polyacrylamide gels. Following autoradiography, the relative signal strength of the 32 P-labeled RNA was quantified by two-dimensional densitometric scanning with the Bio-Image Model 50S (Millipore) or the Gel-Pro CCD Analyzer (Media Cybernetics) imaging systems. Quantitative decay of 32 P-labeled transcripts was calculated as the percentage of signal remaining compared with substrate incubated in decay buffer alone for the entire time course of the respective assay.
Immunodepeletion and Immunoblotting of AUF1-AUF1 immunodepletion was accomplished using protein A-Sepharose CL-4B (Sigma) incubated with a 1:10 volume of rabbit anti-human AUF1-immune or preimmune serum (34) for 30 min at room temperature in cytoplasmic extraction buffer. After three washes, equal volumes of MNC extract and antibody-protein A-Sepharose beads were incubated for 60 min at 4°C twice in succession, using microcentrifugation (12,000 ϫ g for 15 s) to sediment beads.
To verify AUF1 depletion, cytoplasmic extract proteins were electrophoresed on denaturing SDS-polyacrylamide gels (10%, 1:29 bis:acrylamide) along with low range, prestained SDS-PAGE standards (Bio-Rad). Proteins were electroblotted onto nitrocellulose membranes using a MiniTrans-Blot Electrophoretic Transfer Cell (Bio-Rad) according to the manufacturer's recommendations. Electroblot membranes were incubated with 5% nonfat milk in PBS for 60 min at room temperature, followed by 1:3000 AUF1-immune serum overnight at 4°C and 1:3000 horseradish peroxidase-conjugated goat anti-rabbit IgG (Bio-Rad) for 2 h at room temperature. Membranes were washed with antibodybinding buffer (2% nonfat milk in PBS) for 15 min at room temperature between incubations and with 0.1% Triton X-100 in PBS three times for 10-min each before development with luminol/peroxide chemiluminescent substrate (Pierce) and exposure to x-ray film.
Statistical Analysis-Data are presented as the mean Ϯ S.E. Probability of significant differences (p Ͻ 0.05) was calculated using Student's t test on InStat for Windows (GraphPad Software). Curve fitting was accomplished by regression analysis using Prism for Windows (GraphPad Software).

Decay of GM-CSF 3Ј-UTR RNA In Vitro Is ARE-dependent-
To determine whether the GM-CSF ARE conferred instability to an RNA substrate in the in vitro system, decay of the 32 P-labeled GM-CSF 3Ј-UTR was evaluated, with (8AU) and without (0AU) the ARE (diagrammed as dashed lines in Fig. 1). The "Free" samples shown in Fig. 2 (lanes 1 and 4) and throughout the study were incubated in decay buffer alone for the entire time course of the assay. In the presence of neonatal MNC extract protein, significantly more 32 P-8AU than 32 Fig. 2, lane 3 versus 6), but the difference remained significant (p Ͻ 0.008). A 32 Plabeled, 335-nt transcript derived from the pBluescriptII: KS(ϩ) vector was also as stable as 32 P-0AU in the presence of either neonatal or adult MNC protein (data not shown).
To determine the kinetics of ARE-dependent RNA turnover, time course decay assays were performed with the 32 P-8AU and 32 P-0AU substrates. Decay of 32 P-8AU by neonatal MNC protein was significantly faster than that of 32 P-0AU (p Ͻ 0.03) during the first 60 min, after which 32 P-8AU became nearly undetectable (Fig. 3A). Decay of 32 P-8AU by adult MNC protein was also more rapid than that of 32 P-0AU through 120 min, but the difference was significant (p Ͻ 0.008) only after 45 min (Fig. 3B). These results indicate that the ARE has a greater effect on accelerating GM-CSF 3Ј-UTR turnover in the presence of neonatal MNC protein.
Decay of 8AU RNA is More Rapid in the Presence of Neonatal than Adult MNC Extract Protein-When ARE-dependent decay activity was compared between the two MNC extract sources, 32 P-8AU was found to be significantly less stable (p Ͻ 0.005) in the presence of neonatal versus adult MNC protein during the first 60 min (Fig. 3C). The first-order kinetics for 32 P-8AU decay showed a significantly shorter (p Ͻ 0.0001) t1 ⁄2 of 19 Ϯ 6 min for neonatal MNC protein compared with 79 Ϯ 7 min for adult MNC protein (Fig. 3C). Decay of 32 P-8AU was, therefore, four times as rapid with neonatal as with adult MNC protein based on the half-lives derived from regression analysis. These data indicate that the shorter in vivo t1 ⁄2 of GM-CSF mRNA in neonatal MNC is reflected by a significantly shorter in vitro t1 ⁄2 of the GM-CSF 3Ј-UTR in the presence of neonatal MNC protein.
Decay of 8AU RNA Generates Discrete Intermediates-Lower molecular weight 32 P-labeled products were also observed as a consequence of 32 P-8AU in vitro decay. Quantitative analysis of four time course decay assays with three sets of MNC extracts found that generation of two of these products was reproducibly dependent upon incubation with MNC protein. Decay of the 360-nucleotide (nt) 32 P-8AU substrate by neonatal MNC protein generated significant levels (p Ͻ 0.03) of 320-and 245-nt 32 P-labeled products (Fig. 2, lane 2; 8AU Ϫ 40 and 8AU Ϫ 115, respectively) after 45 min. Although these products did not accumulate to the same extent during the slower decay of 32 P-8AU with adult MNC protein, they remained detectable (Fig. 2, lane 3) with significant levels (p Ͻ 0.02) of 8AU Ϫ 40 generated after 45 min. To unambiguously locate the cleavage sites responsible for generating these degradation products, in vitro transcribed 8AU was 32 P-labeled exclusively at the 5Јend. Appearance of the 320-and 245-nt products after incubation of this 5Ј-end 32 P-labeled 8AU substrate with neonatal MNC protein confirmed that 8AU Ϫ 40 and 8AU Ϫ 115 resulted from cleavage at the 3Ј-end of the 3Ј-UTR (Fig. 4, lanes  2-4). Compared with the levels of these products generated by rapid decay in the presence of neonatal MNC protein, accumulation of 8AU Ϫ 40 and 8AU Ϫ 115 in the presence of adult MNC protein was not significant (Fig. 4, lanes 5-7). The 3Ј-ends of these decay products map in the A ϩ U-rich region, 40 and 115 nt upstream of the polyadenylation site within 20 nt from either end of the ARE (designated by vertical arrows in Fig. 1). These results suggest that ribonucleolytic cleavage on either side of the ARE plays a role in the rapid turnover of GM-CSF mRNA by neonatal MNC.
Immunodepletion of AUF1 Stabilizes 8AU RNA-The effect of endogenous AUF1 on stability of the GM-CSF 3Ј-UTR substrates was examined by immunodepleting the MNC extracts of AUF1 prior to assaying in vitro decay. Immunodepletion of AUF1 was verified by immunoblot analysis of MNC extracts (Fig. 5). The in vitro decay of 32 P-8AU was significantly diminished by immunodepletion of AUF1 from neonatal MNC protein extracts (Fig. 6, lanes 2 and 4). Only 11 Ϯ 6% of 32 P-labeled 8AU remained after 30 min incubation in neonatal MNC extracts precipitated with pre-immune serum, whereas 81 Ϯ 14% remained after 30 min incubation in neonatal MNC extracts precipitated with AUF1-immune serum (p Ͻ 0.05, Table I). Conversely, 32 P-labeled 8AU degradation activity of adult MNC extracts was not affected by precipitation with AUF1immune (84 Ϯ 4% remaining) compared with pre-immune (85 Ϯ 11% remaining) serum (Fig. 6, lanes 3 and 5; Table I). Immunodepletion of AUF1 also had no significant effect on the decay of 32 P-0AU (Fig. 6, lanes 6 -10; Table I). These results indicate that depletion of AUF1, and possibly AUF1-associated proteins as well, significantly increases the in vitro stability of ARE-containing RNA in the presence of neonatal MNC protein. DISCUSSION The present study was designed to test the hypothesis that depletion of AUF1 would attenuate ARE-dependent RNA decay in vitro. The binding affinity of AUF1 for a particular ARE is known to reflect the transcript-destabilizing potency of that ARE in vivo (52), and AUF1-enriched protein fractions have been shown to accelerate ARE-dependent in vitro mRNA decay (26). Previous studies have found that the t1 ⁄2 of GM-CSF mRNA is shorter in neonatal MNC (30 min) than in adult (100 min) MNC (39). We have also demonstrated previously that the levels of AUF1 binding activity and the p37 AUF1 and p40 AUF1 isoforms are significantly increased in neonatal MNC extracts (46). Of the four AUF1 isoforms, p37 AUF1 has the highest ARE binding affin-ity (53). Furthermore, expression of AUF1 mRNA and protein is higher in fetal compared with adult murine liver (54). These findings suggested that increased levels of the p37 AUF1 and/or p40 AUF1 isoform(s) might promote ARE binding and transcript destabilization in developmentally immature tissues.
To directly measure RNA degradation activity in the same extracts that were analyzed for ARE binding and AUF1 activity (46), the cell-free mRNA decay assay (47) was adapted for use with whole cytoplasmic rather than fractionated polysomal extracts. The cell-free system has been previously demonstrated to reconstitute the rank order, response to stimuli, and decay products observed during in vivo mRNA decay (reviewed in Ref. 55). In the present study, a 32 P-labeled GM-CSF 3Ј-UTR RNA transcript (8AU; Fig. 1) was utilized as a substrate to focus on the ARE region known to interact more strongly with neonatal than with adult MNC extract protein (46). The 3Ј-UTR ARE can confer instability independent of flanking RNA sequences (reviewed in Ref. 56). Initial characterization of the cell-free RNA decay system also established that RNA length alone does not affect the rate of decay (57), as evidenced by comparable stabilities of the 212-nt 0AU and 335-nt pBlue-scriptII:KS(ϩ) vector transcripts in the presence of either neonatal or adult MNC protein (data not shown). A GM-CSF ARE deletion mutant (58) similar to 0AU, as well as small segments of the tumor necrosis factor-␣ 3Ј-UTR ARE (24,58), have been effective as substrates for assaying ARE-specific decay activity in comparable cell-free systems. Transcript decay is a multi-  32 P-8AU RNA labeled at the 5Ј-end was incubated with 2 g of cytoplasmic protein from PMAϩPHA-stimulated MNC at 37°C for 0 -60 min. Purified RNA was fractionated by denaturing 7 M urea-5% PAGE, and 32 P-labeled products were detected by autoradiography. Lane 1 shows 32 P-labeled 8AU RNA substrate after incubation in decay buffer alone for 60 min. 32 1, 3, and 5) and adult (Ad, lanes 2, 4, and 6) MNC was assayed by AUF1 immunoblotting before (Control, lanes 1 and 2) and after immunodepletion with preimmune (lanes 3 and 4) or AUF1-immune (lanes 5 and 6) serum. Locations of AUF1 isoforms (p37, p40, p42, and p45) and residual goat anti-rabbit-reactive IgG␥ are as indicated. Results shown are representative of three separate assays. step process in which deadenylation precedes decay of the mRNA body (15, 59 -61). Therefore, nonadenylated substrates were utilized in order to focus on the subsequent ARE-directed decay of the RNA body. A GM-CSF 3Ј-UTR substrate with the ARE deleted (0AU; Fig. 1) was employed to assay ARE dependence because its inability to interact with MNC extract protein had been previously characterized (46). Acceleration of in vitro RNA decay by the ARE in the presence of neonatal or adult MNC extract protein demonstrated the validity of this system to assay ARE-dependent turnover (Figs. 2 and 3). A non-significant increase in ARE-independent 0AU degradation was observed, additionally, for neonatal compared with adult MNC extracts (Fig. 2). This increase may have been the result of a loss of specificity by the cell-free decay system or to an increase in ARE-independent exoribonuclease activity. Nonetheless, the destabilizing effect of the ARE remained significant (Fig. 3, A  and B).
The lower molecular weight, 32 P-labeled products (8AU Ϫ 40 and 8AU Ϫ 115) observed following in vitro decay of 32 P-8AU in the presence of neonatal MNC protein (Figs. 2 and 6) apparently result from specific cleavages within the A ϩ U-rich region (Fig. 4). The 3Ј-ends of these decay products map within 20 nt from either end of the ARE (designated by vertical arrows in Fig. 1). Although the same products were detectable during the slower decay of 32 P-8AU with adult MNC protein, they did not accumulate to the same extent as with neonatal MNC protein (Figs. 2, 4, and 6). This finding suggested that the same endoribonucleolytic decay pathway might be present in both but is more active in the neonatal MNC. Similar intermediates have been reported previously to result from the in vitro decay of endogenous human c-myc mRNA by polysomal extracts (60) and of an exogenous c-myc 3Ј-UTR substrate by polysomal supernatants (62). The in vitro decay products of the c-myc transcript were also truncated between the ARE and the polyadenylation site, and they were recently found to be generated in vivo as well (63). Thus, generation of discrete degradation products may be an intermediate step in the rapid decay of the GM-CSF mRNA body by neonatal MNC protein.
The more rapid turnover of 8AU with neonatal (t1 ⁄2 ϭ 19 min) versus adult (t1 ⁄2 ϭ 79 min) MNC protein (Fig. 3C) corresponded directly to the more rapid turnover of GM-CSF mRNA in neonatal (t1 ⁄2 ϭ 30 min) versus adult (t1 ⁄2 ϭ 100 min) MNC (39). Reproducing the relative rates of GM-CSF transcript decay observed during myeloid development demonstrates the ability of the cell-free system to also reconstitute developmental regulation in vitro. Hence, this assay appeared applicable as an in vitro system for identifying the molecular components of accelerated ARE-dependent mRNA turnover in neonatal MNC. This capability was utilized to evaluate the effect of AUF1 immunodepletion on decay of 8AU in the presence of MNC extract protein. The in vitro decay of 8AU was significantly diminished by immunodepletion of AUF1 (Table I) from neonatal but not adult MNC extracts (Fig. 6). The inability of AUF1 immunodepletion to attenuate adult MNC protein 8AU decay activity, coupled with our previous finding of 35-fold less AUF1 binding activity in adult MNC extracts (46), suggests that AUF1-dependent mRNA destabilization is greater in neonatal MNC.
Decay of E. coli mRNA is mediated by a ribonucleoprotein complex known as the degradosome (64,65). Characterization of the degradosome has demonstrated a central role for the endoribonuclease E (RNase E) component (66,67). Two apparently distinct human RNase E homologs have also been identified (22,23). Both these and bovine endoribonuclease V (21) specifically cleave A ϩ U-rich RNA containing AUUUA motifs. Thus, an RNase E-like activity could play a role in cleaving the GM-CSF transcript within the ARE. Although the machinery for degrading the mRNA body in eukaryotes has not been as thoroughly characterized, a multiprotein complex known as the exosome mediates 3Ј 3 5Ј exonucleolytic mRNA decay in yeast (68,69). At least one exoribonuclease subunit, Rrp4p, has a human homolog that is also found in a multiprotein complex (68) and may be related to an ARE-targeted murine ribonuclease activity (20). Additionally, another multiprotein complex known as the proteosome has recently been implicated in AREdirected mammalian mRNA decay. Purified murine proteosomes specifically bind and endonucleolytically cleave AUUUA motifs in the tumor necrosis factor-␣ 3Ј-UTR (24). Inhibiting human proteosome activity stabilizes GM-CSF ARE-containing transcripts and causes the accumulation of polyubiquitinated AUF1 (70). Because a major function of the proteosome is the degradation of polyubiquitinated proteins, it is possible that proteosomes are targeted to the 3Ј-UTR by ARE-bound polyubiquitinated AUF1, after which the proteosome itself can bind and cleave the ARE. In short, it seems likely that both endoribonuclease(s), possibly RNase E-like (22,23) and associated with proteosomes (70), and exoribonuclease(s) like Rrp4p (68)  binding factors (24 -33) within multimeric complexes (68,70) remains an active area of investigation. At least six proteins are known to co-immunoprecipitate with AUF1 (34,70), four of which have recently been identified as heat shock protein 70, heat shock cognate protein 70, eukaryotic translation initiation factor-4G, and poly(A)-binding protein (70). Previous data demonstrating that either purified (26) or recombinant 2 AUF1 can still bind with but can no longer accelerate in vitro decay of ARE-containing mRNA suggests that these protein cofactors are likely contributors to accelerated ARE-dependent RNA turnover in neonatal MNC. In the present study, depletion of AUF1 and, possibly AUF1-associated protein from neonatal MNC extracts, increased ARE-dependent in vitro RNA stability, supporting the proposition that AUF1 plays a role in destabilizing ARE-containing transcripts. ARE-dependent RNA transcript stabilization in vitro by specific immunodepletion, thus, represents a novel and highly significant new finding.
Taken together, our results indicate that accelerated in vitro RNA decay in the presence of neonatal MNC protein is AREand AUF1-dependent. Earlier studies also demonstrated an associated increase in the binding of a neonatal MNC protein complex containing AUF1 (46). Further investigation into the possible roles of AUF1 binding cofactors (34,70) and the enzymes (22)(23)(24)70) responsible for the formation of RNA decay intermediates (60,62,63) should clarify the mechanism for ARE-dependent mRNA destabilization in neonatal MNC. This knowledge is essential for understanding the development of phagocytic immunity.