Immunological Characterization of Tristetraprolin as a Low Abundance, Inducible, Stable Cytosolic Protein*

Tristetraprolin (TTP) is a zinc finger protein that can bind to AU-rich elements within certain mRNAs, resulting in deadenylation and destabilization of those mRNAs. Its physiological targets include the mRNAs encoding the cytokines tumor necrosis factor (cid:1) (TNF) and granulocyte-macrophage colony-stimulating factor. TTP was originally identified on the basis of its massive but transient increase in mRNA levels following mitogen stimulation of fibroblasts. It has been difficult to reconcile this transient mRNA profile with the presumed continuing “need” for TTP protein, for example, to reverse the effects of lipopolysaccharide (LPS)-stimulated TNF secretion. To investigate this and other questions con-cerning endogenous TTP protein in cells and tissues, we raised a high titer rabbit antiserum against full-length mouse TTP. TTP could be detected on immunoblots of mouse cytosolic tissue extracts; it was most highly expressed in spleen, but its concentration in that tissue was only about 1.5 n M . TTP could be detected readily in splenic macrophages and stromal cells from LPS-in-jected rats. In both LPS-treated RAW 264.7 macrophages and fetal calf serum-treated mouse embryonic fibroblasts, TTP protein was stable after induction, with minimal degradation occurring for several hours after treatment of the cells with

mals, three members of this family have been characterized: TTP (also known as ZFP36, TIS11, G0S24, and NUP475), ZFP36L1 (also known as TIS11b, cMG1, ERF1, BRF1, and Berg36), and ZFP36L2 (also known as TIS11d, ERF2, and BRF2) (1). Although they are encoded by different genes and their patterns of cell-and tissue-specific expression and agoniststimulated expression are quite different, they share certain properties: All have highly conserved tandem zinc finger domains, in which each C8xC5xC3xH zinc finger is preceded by the sequence (R/K)YKTEL, and the two fingers are separated by 18 amino acids (2); all are nuclear-cytoplasmic shuttling proteins (3,4); and all are capable of binding AU-rich elements (ARE) within single-stranded RNA (2,(5)(6)(7)(8)(9) and promoting the deadenylation and subsequent destruction of those transcripts, both in transfection studies and in cell-free experiments (2,7,8,10). In intact animals, TTP deficiency causes a profound inflammatory syndrome with erosive arthritis, autoimmunity, and myeloid hyperplasia, apparently due almost entirely to excessive production of tumor necrosis factor ␣ (TNF) and granulocyte-macrophage colony-stimulating factor (GM-CSF), both of whose mRNAs are direct targets of TTP and are stabilized in cells from the knockout (KO) mice (7,10,11). For these reasons, TTP can be thought of as an anti-inflammatory or arthritis-suppressor protein.
The cDNAs encoding TTP were originally cloned by three groups by virtue of its very rapid and dramatic transcriptional induction in fibroblasts in response to insulin, phorbol esters, and serum (12)(13)(14). In a typical example, TTP mRNA was undetectable in serum-starved, insulin-responsive 3T3-L1 mouse fibroblasts, but the transcript became detectable within 10 min of insulin stimulation, peaked at ϳ45 min, and returned to near baseline by 120 min (13). These dramatic but transient responses were also seen in cells in which TTP is thought to exert an effect in normal physiology, i.e. in the regulation of TNF mRNA stability in macrophages (7). Clearly, in most cell types there is a mechanism for the rapid transcription of the TTP gene (Zfp36 in the mouse) in response to various agonists, mediated by some well characterized and some relatively uncharacterized enhancer elements (15), and a mechanism for the rapid turning off of transcription while transcript degradation is occurring, leading quickly to a return to baseline mRNA levels despite the continued presence of agonist (1).
This rapid but transient increase in TTP mRNA levels is difficult to reconcile with the presumed continuing requirement for TTP protein in situations in which it might be needed to destabilize TNF mRNA, for example. For these and other reasons, it was important to establish the normal patterns of TTP protein expression in cells and tissues, as well as the protein response patterns in cells such as macrophages in which the protein is thought to play an important physiological role. However, studies of this type have been hampered by the lack of good antibodies to TTP, a deficiency that is partly due to difficulties in expressing the intact protein in soluble form in typical expression systems (6). In the present studies, we describe the development of a high titer antibody to mouse TTP, and its use in characterizing the normal patterns of TTP expression in mouse tissues and its pattern of induction in cultured cells. Among the interesting findings were that TTP seems to be an extraordinarily low abundance protein in normal mouse tissues; that its levels do not necessarily parallel the steady-state levels of its transcript in these tissues; that it is a very stable cytoplasmic protein once induced, at least in macrophages and fibroblasts; and finally, that it undergoes progressive phosphorylation during biosynthetic induction over a scale of hours.

MBP-mTTP Purification and Antibody
Production-Maltose binding protein-mouse TTP fusion protein (MBP-mTTP) and MBP were purified as described (6) and used to produce rabbit antisera according to standard procedures (Covance Research Products, Inc., Denver, PA). The antiserum that contained the highest titer anti-TTP activity was centrifuged at 10,000 ϫ g for 10 min, sodium azide was added to a final concentration of 0.02% (w/v), and the serum was stored at Ϫ70°C.
Preparation of Tissue Extracts-Wild-type (WT) C57B16 mice, and mice deficient in TTP, TTP/TNF␣ receptor 1 (TNFR1), or TTP/TNF␣ receptor 2 (TNFR2) (11,16) were maintained in microisolator cages in a barrier facility according to institutional guidelines. Tissues were collected from mice at ϳ6 months of age after euthanasia with CO 2 . The tissues were frozen in liquid nitrogen, stored at Ϫ70°C, pulverized with a mortar and pestle under liquid nitrogen, and then homogenized with an Overhead Strirrer (Wheaton Instruments, Millville, NJ) in 50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1.5 mM MgCl 2 , 250 mM sucrose, 100 mM NaF, 5 mM dithiothreitol, 2 mM sodium phosphate buffer, 2 mM ZnCl 2 , 1 mM phenylmethylsulfonyl fluoride, and 2 M leupeptin. Supernatants from centrifugations at 20,000 ϫ g for 10 min at 4°C were stored at Ϫ70°C after addition of 20% glycerol (v/v). A cytosolic fraction was separated from a membrane fraction by further centrifugation at 100,000 ϫ g for 1 h at 4°C.
Cell Culture-Mouse bone marrow-derived macrophages (BMM) were prepared from 6-to 8-month-old WT and TTP KO mice as described previously (17), then stimulated with 0.1 or 1 g/ml lipopolysaccharide (LPS, Sigma Chemical Co., St. Louis, MO) for various times. Primary mouse embryonic fibroblasts (MEF) were prepared from mouse embryos at embryonic day 14 and cultured in Dulbecco's modified Eagle's medium (Invitrogen, Gaithersburg, MD) as described (18). They were serum-deprived overnight and then stimulated with 10% FCS for various times. Mouse RAW 264.7 cells (American Type Culture Collection) were cultured as MEF, except in Eagle's minimum essential medium (Invitrogen), and treated with LPS (0.1 g/ml) for various times. In one experiment, cells were stimulated with LPS for 2.5 h, followed by LPS "washout" as described (19). In some experiments, 50 M cycloheximide (CHX) was added to inhibit protein synthesis.
Cell Extracts-After washing twice with PBS, cells were gently scraped with a rubber policeman into 5 ml of ice-cold PBS, and centrifuged at 1,000 ϫ g for 5 min at 4°C. Cells from each 10-cm dish were then resuspended and lysed in 0.5 ml of a lysis buffer containing 50 mM NaH 2 PO 4 , pH 7.6, 250 mM NaCl, 50 mM NaF, 10 mM imidazole, 0.5% Nonidet P-40, 1 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride. The cell lysate was left on ice for about 20 min and then centrifuged at 1,000 ϫ g for 5 min at 4°C. The supernatant was then centrifuged again at 10,000 ϫ g for 10 min at 4°C. The 10,000 ϫ g supernatant was mixed with glycerol to a final concentration of 20% (v/v), frozen in liquid nitrogen, and stored at Ϫ20°C for future use. The pellet from the 1,000 ϫ g centrifugation was used for the preparation of nuclear extracts as described previously (7). Alternatively, nuclear extract was prepared from the pellet by sonication and then centrifugation at 10,000 ϫ g for 10 min at 4°C. To concentrate proteins in the mouse fibroblast extracts, proteins in the 10,000 ϫ g supernatant from 2-4 dishes were precipitated with 25% (v/v) of trichloroacetic acid on ice for 2 h, washed twice with cold acetone, dried with a Speed-Vac (Thermo Electron Corp., Waltham, MA), and dissolved in 40 l of 10 mM Tris-HCl at pH 8.5. Dephosphorylation of mTTP from LPS-stimulated RAW 264.7 cellular extracts used calf intestine alkaline phosphatase (CIAP) (Stratagene) as described in the legend to Fig. 6D.
Size Exclusion Chromatography of TTP from Mouse RAW 264.7 Cells-Proteins in the 10,000 ϫ g supernatant from RAW 264.7 cells stimulated with 0.1 g/ml LPS for 2 h were precipitated with ammonium sulfate at 50% saturation, followed by separation on a Superose 6 HR 10/30 size exclusion column (Amersham Biosciences, Uppsala, Sweden) using a similar procedure to that described previously (20). Proteins were eluted with 50 mM NaH 2 PO 4 , 300 mM NaCl, 20 mM imidazole, 0.05% Tween 20, pH 8.0, and column fractions were analyzed by immunoblotting using anti-MBP-mTTP serum. The molecular mass of mTTP was determined by comparing its elution volume to a standard curve generated with protein standards separated on the same column. The protein standards (Amersham Biosciences) used were bovine pancreas ribonuclease A (13.7 kDa), bovine pancreas chymotrypsinogen (25 kDa), hen egg ovalbumin (43 kDa), bovine serum albumin (67 kDa), rabbit muscle aldolase (158 kDa), bovine liver catalase (232 kDa), horse spleen ferritin (440 kDa), and bovine thyroid thyroglobulin (669 kDa). The void volume of the column was determined with blue dextran (2000 kDa).
SDS-PAGE and Immunoblotting-Protein concentrations were determined by the Bradford method (Bio-Rad Laboratories, Hercules, CA), with bovine serum albumin (BSA) as the standard, with NaOH treatment as described (21). Proteins were separated by SDS-PAGE following standard protocols (22) and visualized with either Coomassie Blue (22) or silver staining (23). For Western blotting, proteins were transferred onto nitrocellulose membranes in transfer buffer containing 0.1% SDS and visualized with Ponceau S staining. The membranes were blocked with 5% nonfat dry milk in 0.05% Tween 20 in Trisbuffered saline (TTBS) for 30 -60 min and incubated with the primary antiserum at a 1:10,000 dilution in the blocking buffer for 1-18 h. After being washed with TTBS three times for 10 min each, the membranes were incubated with a 1:10,000 dilution of secondary antibody in TTBS for 1-6 h. Following three, 10-min washes with TTBS, membranes were incubated with SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL) and exposed to x-ray film. The primary antibodies used were anti-MBP serum (New England Biolabs, Beverly, MA) and the anti-MBP-mTTP antiserum described here and its pre-immune control serum. The secondary antibodies were affinity-purified goat anti-rabbit IgG (HϩL) horseradish peroxidase conjugate with human IgG absorbed (GAR-HRP) (Bio-Rad).
Immunocytochemistry and Confocal Microscopy of RAW 264.7 Cells-Mouse RAW 264.7 cells were grown on glass slides or glass coverslips under the conditions described above. Following the addition of LPS (0.1 g/ml) for various times, the cells were washed with ice-cold PBS three times before being fixed at room temperature for 10 min in 4% (w/v) paraformaldehyde in PBS. Cells were again washed with PBS twice and then permeabilized with 0.2% (v/v) Triton X-100 in TBS at room temperature for 10 min. The cells were then incubated at 4°C overnight in anti-MBP-mTTP serum (1:8 -10,000 dilution) and 1% normal goat serum in 0.1 M sodium phosphate buffer (pH 7.4). Following two washes with 0.2% (w/v) Triton X-100 in TBS, the cells were incubated at room temperature for 30 min in goat anti-rabbit Alexa Fluor 488 (1:300 dilution, Molecular Probes) and 25% normal goat serum in TBS. The cells were rinsed twice with PBS for 10 min each time before being mounted with Prolong anti-fade (Molecular Probes). After drying overnight, the slides were examined and imaged with an LSM510 UV confocal microscope (Zeiss, Thornwood, NY).
Immunostaining of Rat Spleen Sections-We took advantage of the fact that mouse and rat TTP are 96% identical (compare GenBank TM accession numbers NP_035886 and NP_579824) to examine the patterns of immunoreactive TTP expression in rat spleen. Male Sprague-Dawley rats at ϳ12 weeks of age were maintained as described (24) and injected intraperitoneally with LPS (5 mg/kg in 5 ml/kg PBS) or the same volume of PBS as a control. 2 h later, spleens were collected, fixed in Bouin's fixative for 48 h, clarified in 70% ethanol, then embedded in paraffin and used for immunohistochemistry essentially as described (25). The sections were incubated with the anti-MBP-mTTP serum or preimmune serum (1:1000) in PBS containing 1% BSA at 37°C for 45 min, washed in PBS three times for 5 min each, and incubated with a biotin-labeled anti-rabbit antibody (1:200) (Vector Laboratories, Burlingame, CA) in PBS containing 1% BSA at 37°C for 20 min. The sections were washed as above, and the avidin/biotin complex was developed with 3,3Ј-diaminobenzidine staining, counterstaining, and dehydration as described (25) using the Vectastain ABC Elite kit (Vector Laboratories).

RESULTS
Production and Characterization of Anti-MBP-mTTP Serum-For antigen production, MBP-mTTP was expressed in Escherichia coli and purified to near-homogeneity as described (6). The results of this purification are shown after staining with Coomassie Blue (Fig. 1A) and immunoblotting with an anti-MBP antibody (Fig. 1B). The Mono Q fractions with the highest purity of MBP-mTTP (Fig. 1, lane 6) were pooled, concentrated, and used as antigen.
The resulting anti-MBP-mTTP serum at a 1:10,000 dilution was able to detect as little as 1 ng of the purified MBP-mTTP, using the SuperSignal detection reagent and a 5-s exposure of the blot (Fig. 1C). We also tested the antiserum using extracts from 293 cells expressing mouse TTP (3,4). The antiserum at 1:10,000 dilution was able to detect mouse TTP when as little as 1 g of total protein from the transfected 293 cells was loaded into the gel (Fig. 1C). This antiserum recognized mouse TTP but not human TTP or mouse Zfp36L1 under these immunoblotting conditions (data not shown). However, it could recognize human TTP when much more protein was used or when the blot was exposed to film for longer times (see below).
Identification of TTP in Mouse Cells and Tissues-Using the present antiserum at 1:10,000 dilution, TTP could be detected in Western blots of normal spleen when 5 mg of protein per gel lane and the SuperSignal detection reagent were used ( Fig.  2A). TTP in cells and tissues occurs as multiple bands of M r about 40 -50,000, probably a mixture of differentially phosphorylated species and degradation products (1, 7). The TTP bands were absent in spleen taken from TTP KO mice ( Fig. 2A). As a positive control, we used 2 g of protein extract from transfected 293 cells ( Fig. 2A).
The 20,000 ϫ g spleen supernatant was separated into supernatant and pellet fractions by centrifugation at 100,000 ϫ g. TTP was found primarily in the 100,000 ϫ g supernatant, with very little immunoreactivity in the membrane pellet (Fig. 2B). No TTP signal was detected in the 20,000 ϫ g pellet (data not shown). Because purified recombinant MBP-mTTP (10 ng, or 0.134 pmol of the 74.9-kDa fusion protein) was probed on the same blot as the spleen cytosol (1.7 mg of total protein), we used this blot to estimate the approximate concentration of TTP in normal mouse spleen. In these studies, the average spleen weight was 95.2 mg (n ϭ 14), the average soluble protein content per spleen was 16.5 mg/spleen (n ϭ 14), and adult mouse spleen water content was ϳ86% (26). As shown in Fig.  2B, the TTP immunoreactivity in 1.7 mg of spleen extract protein appeared to be about 10% of that of 10 ng (0.134 pmol) of the recombinant protein, or 0.0134 pmol. Therefore, TTP was expressed at ϳ0.0134 pmol per 1.7 mg/16.5 mg/spleen, or 0.1265 pmol/spleen. Because average spleen water content was 86% of an average wet weight of 95.2 mg/spleen, the final estimated TTP concentration was 0.1265 pmol/81.87 l, or about 1.5 nM. were probed with the anti-MBP-mTTP serum (1:10,000) and GAR-HRP (1:10,000) for 30 min each before being incubated in the detection reagent for 5 min and exposed to x-ray film for 30 s. The smaller immunoreactive bands seen both with the E. coli protein and the protein expressed in 293 cells are presumed to be proteolytic fragments.

Distribution of TTP in Mouse Tissues-
The anti-MBP-mTTP serum (1:10,000 dilution) cross-reacted with proteins of the appropriate size in spleen, lung, liver, large intestine, and thymus, and with proteins of different sizes from several other tissues (Fig. 3A). To distinguish between true positive and false positive signals, we analyzed tissues from mice deficient in either TNFR1 and WT for TTP (Fig. 3B) or deficient in both TNFR1 and TTP (Fig. 3C). These genotypes were used, because the absence of the TNFR1 largely prevents the development of the inflammatory TTP-deficiency syndrome, which might have altered the expression of nonspecific bands in the tissues. When compared with tissues from the TTP KO mice (Fig. 3C), true positive TTP signals were seen in spleen, lung, liver, and large intestine (Fig. 3B), whereas nonspecific immunoreactive bands of various sizes were detected in brain, heart, pancreas, and skin from both the TTP WT and KO mice (Fig. 3, compare  B and C). These data showed that spleen expressed the highest concentration of TTP of any tissue examined.
Induction of TTP in Cultured Cells-To address the specificity of this antiserum in cultured cells, BMM from WT and TTP KO mice were exposed to LPS (1 g/ml) for 2 h, and proteins from 10,000 ϫ g supernatants were separated by SDS-PAGE and probed with the antiserum at a 1:10,000 dilution. The multiple bands of TTP were clearly detected when 200 g of cellular protein was used per gel lane from the LPS-induced WT BMM, but were not seen in the macrophages from the TTP KO mice (Fig. 4A). Similar results were obtained using 500 g of protein per lane in the 10,000 ϫ g supernatants from WT but not KO MEF, which had been stimulated with 10% FCS for 2 and 3 h (Fig. 4B). As also shown in Fig. 4B, TTP was readily detected in RAW 264.7 cell extracts when only 50 g of protein was used per lane from the 10,000 ϫ g supernatant of cells stimulated with 0.1 g/ml LPS.
Time Course of Induction and Stability of TTP in Cultured Cells-To investigate the patterns of TTP protein accumulation in cultured cells, BMM were collected following LPS (0.1 g/ml) stimulation. TTP was undetectable at time 0 but readily detectable after 2 h, reaching a peak of expression at about 4 h (Fig. 5A). Immunoreactive TTP in 500 g of BMM protein was compared with only 50 g of protein from RAW 264.7 cells stimulated with LPS for 1.5 and 2 h (Fig. 5A).
Similar induction kinetics were seen in MEF stimulated with 10% FCS (Fig. 5B). Again, TTP was undetectable in the serumdeprived cells at time 0, but accumulated dramatically to reach peak levels at ϳ2-3 h. These peak levels, detected with 200 g of cellular protein per lane, were comparable to the levels seen in LPS-stimulated RAW 264.7 cells when only 10 g of protein was loaded into the gel lane (Fig. 5B). Interestingly, there was a continued shift upward in the apparent molecular weight of the protein with still longer times of exposure to FCS (Fig. 5B), compatible with increasing phosphorylation. Protein was still readily detectable at 5-6 h.
To estimate the stability of the newly synthesized protein in these cells, the cells were stimulated for 2 h with 10% FCS and then treated with CHX (50 M) for a further 4 h (Fig. 5B). Remarkably, there was very little apparent disappearance of the protein over 4 h following treatment with CHX (Fig. 5B), at concentrations that completely inhibited protein synthesis in this cell type (data not shown).
Because RAW 264.7 cells produced much more TTP than the other cell types we tested, we analyzed the induction kinetics in more detail in these cells. As shown in Fig. 5 (C and D), TTP FIG. 2. Identification of endogenous TTP in mouse spleen. A, 2 g of protein from 293 cell extracts expressing HA-mTTP, and 5 mg of protein from 20,000 ϫ g supernatants of spleen homogenate from WT and TTP KO mice, were separated by SDS-PAGE and transferred onto nitrocellulose membranes for Western blotting. The membrane was incubated in anti-MBP-mTTP serum (1:10,000) overnight and with GAR-HRP (1:10,000) for 1 h and exposed to x-ray film for 1 min. B, the 20,000 ϫ g supernatant from mouse spleen was further centrifuged at 100,000 ϫ g for 1 h to separate microsomal membranes (Membranes) from the cytosol (Soluble). Equal amounts of protein (1.7 mg) from the soluble and membrane fractions were used for Western blotting as above, along with MBP-mTTP (10 ng) from the amylose resin column and HA-mTTP in the transfected 293 cell extracts (1 g of total soluble protein). In this case, the membrane was exposed to x-ray film for 10 min.
was undetectable in unstimulated cells and in cells stimulated with LPS for 15 and 30 min, using 50 g of cytosolic protein and a relatively short autoradiographic exposure (Fig. 5, C and D). TTP could be detected in the 0.5-h samples if much more protein and/or much longer autoradiographic exposure were used (data not shown). TTP was detectable following induction for 45 min (Fig. 5C). At 1 h, TTP was detected as two major bands of about M r 38,000 and 42,000 (Fig. 5D). As TTP continued to accumulate, there were continued increases in the apparent molecular weight to 40 -50,000, which stabilized by 2 h (Fig. 5, C and D). Significant amounts of immunoreactive TTP were still detectable in samples collected after 24 h of LPS induction (Fig. 5D). We also calculated TTP concentration in the stimulated RAW 264.7 cells with 0.1 g/ml LPS for 2 h using known concentrations of the purified MBP-mTTP (1-100 ng). The immunoreactivity in 50 g of RAW 264.7 cell 10,000 ϫ g supernatant was approximately the same as that of 10 ng (0.134 pmol) of the recombinant protein. This represents about 340 times the concentration in normal spleen, or ϳ0.5 M.
The stability of TTP in RAW 264.7 cells was further evaluated following LPS stimulation for 2 h (0.1 g/ml), followed by treatment with CHX (50 M). No TTP was detected in RAW 264.7 cell extracts 1, 1.5, 2, 3, 4, or 5 h after both LPS and CHX were added to the culture medium at the same time, indicating that this concentration of CHX was effective at inhibiting protein synthesis under these experimental conditions (data not shown). Despite the inhibition of protein synthesis, there was only a modest decrease of TTP immunoreactivity over the next 4 h, although there was a continued shift to apparently greater molecular weight between 2 and 3 h after LPS, even in the presence of CHX (Fig. 5E). Similar slow rates of protein disap-pearance were observed in cells stimulated with LPS for 2.5 h followed by LPS "washout" and CHX treatment (data not shown).
Monomeric Nature of TTP in RAW 264.7 Cells-The 10,000 ϫ g supernatant from RAW 264.7 cells stimulated with LPS (0.1 g/ml) for 2 h was subjected to size exclusion chromatography following ammonium sulfate concentration. TTP could be detected in a range of column fractions, but the peak of TTP was detected in fraction #31 (Fig. 6A), which corresponded to a monomer size of about 40 kDa when compared with a standard curve (Fig. 6B). Small amounts of TTP were detected in earlier fractions (Fig. 6A), suggesting that a small proportion of the protein might exist in RAW 264.7 cells as oligomers or in complexes with other proteins under these conditions.
When LPS-stimulated RAW 264.7 cells were fractionated into cytosolic and nuclear fractions, most of the TTP was in the cytosolic fraction, with little if any associated with the nuclear fraction (Fig. 6C); there was no evidence of higher molecular weight oligomers or aggregates under these conditions. There was also some degree of cross-reactivity with the human TTP protein expressed in and purified from E. coli (Fig. 6C, lane 5). This figure also illustrates the extent of the inhibited SDS-PAGE migration caused by, presumably, phosphorylation in the RAW 264.7 cells. Human TTP has a predicted molecular weight of 34,086, and the protein purified from E. coli migrated as a single band at M r ϳ36,000, as estimated with protein standards (lane 5), as we have shown previously (6). However, the mouse protein in LPS-stimulated RAW 264.7 cells, with a calculated molecular weight of 33,613, migrated as multiple bands of apparent molecular mass 40 -50 kDa. The migration FIG. 3. Expression of TTP in mouse tissues. A supernatant (10,000 ϫ g) was prepared from tissues from WT and TTP KO mice and was used for immunoblotting. Transfected 293 cell extracts expressing HA-mTTP (1 g of total protein) were used as positive controls. A, in each lane, 1.7 mg of total protein in the supernatant from WT mouse tissue was used, and the blot was exposed to x-ray film for 4 min. The broad band or doublet corresponding to TTP is indicated. B, TNFR1 KO (TTP WT); C, TNFR1/TTP double knockout (TTP KO) tissue extracts. 1 mg of total protein in the tissue supernatants was used, and the blots were exposed to x-ray film for 1 h in the same cassette. The bands seen in B but not in C represent specific TTP immunoreactivity; the bands shown in both blots represent nonspecific protein reactivity.
of TTP was increased following dephosphorylation of mTTP in the RAW 264.7 cell 10,000 ϫ g supernatant with CIAP treatment (Fig. 6D). Incubation with longer time or with more CIAP resulted in a faster migration of the protein, although the sizes of those dephosphorylated proteins were still larger than that of hTTP purified from E. coli cells (data not shown). This change in SDS-PAGE migration appears to be largely due to phosphorylation, because the multiband complex collapses into a single protein band of lower apparent molecular weight upon dephosphorylation with alkaline phosphatase (27).
Immunostaining of TTP in Intact Cells and Tissues-To further localize TTP in RAW 264.7 cells, they were treated with 0.1 g/ml LPS for 3 h, followed by immunostaining with anti-MBP-mTTP (1:8,000 dilution). As shown in Fig. 7A, confocal microscopy revealed bright cytosolic immunofluorescence in LPS-treated RAW 264.7 cells, whereas minimal fluorescence was detected in controls treated in parallel with LPS induction and pre-immune serum; without LPS induction but with anti-MBP-mTTP serum; or without LPS treatment but with preimmune serum. The cytosolic staining was in a vesicular pattern (Fig. 7A). There were occasional foci of nuclear staining, but these were seen also in the cells not treated with LPS; they were more prominent with immune serum than in the other two negative controls (Fig. 7A), suggesting that they may represent true foci of nuclear TTP. Serial sections of a single RAW 264.7 cell following LPS induction showed that the vesicular pattern of fluorescence was almost exclusively cytosolic, with little if any signal in the nucleus (Fig. 7B).
Consistent with the immunoblotting results, TTP was barely detectable under unstimulated conditions but peaked in the cytosol by 2-3 h following LPS stimulation (Fig. 7C). Significant immunostaining was still visible in the RAW 264.7 cells after 5 h, although the signal was somewhat decreased relative to the earlier time points (Fig. 7C). Although there was minimal detectable fluorescence in the unstimulated cells in Fig. 7  (A and C), we occasionally noticed increased fluorescence in the cytosol of cells undergoing division, even in the absence of LPS (Fig. 7D).
We also evaluated the use of the antibody in rat spleen, after the intraperitoneal injection of LPS (5 mg/kg). 2 h after the injection, there was prominent TTP staining in macrophages and stromal cells in the spleen periphery (Fig. 8A), whereas no specific staining was seen in spleen from a PBS-injected rat (C) or with preimmune serum staining of spleen from either an LPS-injected (B) or PBS-injected (D) rat. The light microscopic images in A-D were obtained and processed under identical conditions. Of interest was the negative staining of the white pulp in the center of Fig. 8A (arrow) (see "Discussion"). This immunostaining was seen with the use of Bouin's fixed, paraffin-embedded sections, but not with spleens fixed in 4% paraformaldehyde and processed for frozen sections (data not shown).

DISCUSSION
Despite more than 13 years of information about the transcriptional induction of TTP in response to a variety of growth factors, cytokines, and mitogens, relatively little is known about the behavior of the TTP protein itself under similar circumstances (1). This lack of information is due in part to difficulties in expressing the recombinant protein, making preparation of high affinity antibodies more difficult (6). However, the most difficult problem to overcome has been the apparent extraordinary scarcity of the protein in most cells and tissues. We have begun to approach some of these issues by developing a high titer antibody to the recombinant mouse TTP protein expressed in E. coli as a MBP fusion protein.
One of the fundamental questions we hoped to address with this new antiserum was whether TTP protein was relatively stable in cells after its induction by mitogens and cytokines. This question arose because of the archetypal immediate-early response gene characteristics of the TTP mRNA after its induction by insulin, serum, or other mitogens in cultured fibroblasts FIG. 4. TTP expression in primary mouse macrophages, RAW 264.7 cells, and MEF. A, primary mouse macrophages from TTP WT and KO mice were treated with LPS (1 g/ml) for 2 h, and soluble cellular extracts were prepared. Equal amounts of protein from these extracts (200 g) were used for immunoblotting, using the anti-MBP-mTTP serum (1:10,000). The blot was exposed to x-ray film for 30 s. The position of immunoreactive TTP is shown in the WT but not the KO cells. B, RAW 264.7 cells and MEF. RAW 264.7 cells (RAW) were stimulated with LPS (0.1 g/ml) for 1.5 and 2 h, and used for the preparation of 10,000 ϫ g supernatants. Similarly, MEF from WT and TTP KO mice were stimulated with 10% FCS for 2 and 3 h. Supernatants from RAW (50 g) and MEF (500 g) were used for immunoblotting with the anti-MBP-mTTP serum (1:10,000). The blot was exposed to x-ray film overnight. The position of TTP is indicated; note its absence in the KO MEF.
(1). For example, in mouse 3T3-L1 adipocytes, TTP mRNA was essentially undetectable by Northern blot in the basal, serumdeprived G 0 state. However, within 10 min of exposure to insulin, there was a detectable level of mature TTP transcript, and this continued to accumulate for the first 45 min, after which there was a rapid decline in transcript levels to near basal concentrations by 120 min (13). Similar induction and turnover kinetics were observed in what is believed to be a physiologically relevant cell type and response, i.e. the induction in macrophages by agents such as LPS or TNF (7). In both situations, the rapid and dramatic induction of transcription was followed by a similarly rapid shut-off of transcription accompanied by mRNA lability, such that steady-state levels of mRNA had returned to near-basal levels despite the continued presence of the inducing stimulus (1).
This rapid and transient transcript response is characteristic of prototypical immediate early response genes such as c-fos. However, this type of response did not fit with the proposed physiological function of the TTP protein, i.e. to bind to and destabilize mRNAs such as those encoding TNF and GM-CSF (1). For example, TNF mRNA is rapidly and dramatically induced by LPS in macrophages, followed by the production of secreted TNF (7); how can we reconcile the rapid but transient TTP transcript response with the presumed continued requirement for TTP protein to cause degradation of the TNF transcript and reversal of TNF secretion?
In the present study, we addressed this apparent discrepancy by evaluating the induction patterns of TTP protein and its stability in cultured MEF after stimulation with serum, and in cultured mouse RAW 264.7 macrophages after stimulation by LPS. In the latter case, TTP protein was readily detectable as multiple bands on Western blots of a crude cytoplasmic fraction from the cells after induction, although the protein was barely detectable in the unstimulated cells, as noted previously (28). These multiple bands were shown previously to be multiple phosphorylated species, because treatment of the protein in extracts from 293 cells with alkaline phosphatase caused collapse of the multiple bands into a single, presumably dephosphorylated species (27). In LPS-stimulated RAW 264.7 cells, TTP protein accumulated to reach near steady-state levels after 90 -120 min; remarkably, these high levels of protein remained nearly constant for many hours thereafter, and were still readily detectable after 24 h. In keeping with this pattern, treatment of the cells with CHX 2 h after LPS stimulation, at concentrations shown to completely inhibit protein synthesis in these cells, revealed a very stable protein, with minimal decline FIG. 5. Time course of induction and stability of TTP in mouse cells. RAW 264.7 cells and primary mouse macrophages were stimulated with 0.1 g/ml LPS, and MEF were stimulated with 10% FCS, for various times as indicated. To measure protein stability, the cells were stimulated for 2 h before being treated with CHX (50 M) for the indicated times. A supernatant (10,000 ϫ g) was prepared from the cells and used for immunoblotting using the anti-MBP-mTTP serum (1:10,000). A, data from RAW 264.7 cells (RAW) (50 g of protein per lane) were compared with data from primary macrophages (500 g per lane). The blot was incubated in the primary antiserum for 2 h and the secondary antibody for 1 h, then exposed to x-ray film overnight. B, data from RAW 264.7 cells stimulated by LPS for 2 h (10 g per lane) were compared with data from MEF (fibroblasts) stimulated with FCS with or without the later addition of CHX (200 g of protein per lane). The blot was incubated in the primary antiserum overnight and the secondary antibody for 2 h, then exposed to x-ray film for 2 min. C, time course of TTP induction in RAW 264.7 cells (50 g of protein per lane). The blot was incubated in the primary antiserum for 18 h and the secondary antibody for 1 h. The blot was exposed to x-ray film for 5 s. D, time course of TTP induction in RAW 264.7 cells without CHX (50 g of protein per lane). The blot was incubated in the primary antiserum for 1.5 h and the secondary antibody for 1 h. The blot was exposed to x-ray film for 30 s. E, TTP stability in RAW 264.7 cells (50 g of protein per lane). The cells were first stimulated with LPS for 2 h, followed by treatment with CHX for the indicated times. The blot was incubated in the primary antiserum for 1.5 h and the secondary antibody for 1 h, then exposed to x-ray film for 30 s. in protein levels observed for several hours, until generalized CHX toxicity began to develop. Similar patterns were observed in serum-treated mouse fibroblasts, again with the formation of a very stable protein. Taken together, these data make physiological sense, in that the continued presence of the newly synthesized protein would be expected to continue to exert its negative regulatory effects on ARE-containing transcripts for several hours after the initial stimulation.
Another interesting aspect of this protein induction pattern is the gradual increase in apparent phosphorylation that accompanied the increase in steady-state protein levels. Both maximal phosphorylation, as indicated by maximal increase in apparent molecular weight on SDS gels, and maximal steadystate protein levels were achieved by ϳ4 h after stimulation in both macrophages and fibroblasts. This interesting pattern is different from a more typical protein phosphorylation paradigm, in which previously synthesized protein is phosphoryl-ated within a few minutes by an activated protein kinase. In this unusual case, phosphorylation and biosynthesis appeared to occur in parallel, raising interesting questions about the types of protein kinases involved, their regulation, and their effects on the protein. To date, sites for the broad family of mitogen-activated protein kinases have been identified in the protein (18), as well as sites for the MKK2 kinase that appear to be involved in binding of the protein to 14-3-3 protein (29,30). Nonetheless, many more phosphorylation sites remain to be identified on this protein. It is interesting to speculate that the increase in phosphorylation that accompanied the increased biosynthesis of the protein might be involved in conferring stability to the protein, as noted previously for p53, for example (31,32). Obviously, other regulatory possibilities of this phosphorylation exist, including the regulation of the nuclear to cytoplasmic shuttling of TTP, its binding affinity for its RNA targets, its association with other proteins, and others.
where V e , V o , and V t are the elution volume of the protein determined by the experiment, the void volume determined with blue dextran, and the bed volume of the column provided by the manufacturer, respectively. The peak fraction containing TTP (#31) corresponded to a molecular size of 40 kDa (diamond). C, cytosolic localization of TTP in RAW 264.7 cells. RAW 264.7 cells were fractionated into cytosolic and nuclear fractions in two separate experiments after stimulation with LPS (0.1 g/ml) for 4 and 6 h. Equal amounts of protein from the cytosolic and nuclear fractions (50 g) were used for immunoblotting with anti-MBP-mTTP serum (1:10,000). Human TTP (no fusion partner) purified from E. coli (20 ng) was used to demonstrate the electrophoretic mobility shift of TTP in RAW 264.7 cells following LPS induction. This blot was incubated in the primary antiserum for 1.5 h and the secondary antibody for 1 h. The blot was exposed to x-ray film for 10 s. D, dephosphorylation of mTTP in LPS-stimulated RAW 264.7 cell extracts. RAW 264.7 cells were stimulated with LPS (0.1 g/ml) for 0, 1, 2, 3, 4, and 6 h. The 10,000 ϫ g supernatant (100 g of total protein) was incubated with or without CIAP (35 units) at 23°C (room temperature) for 1 h. Half of the mixture was used for immunoblotting detection. This blot was incubated in the primary antiserum for 18 h and the secondary antibody for 1 h. The blot was exposed to x-ray film for 10 s. "hTTP" refers to the same protein as in C, lane 5, above.
The studies described here are in general agreement with previous studies of TTP expression and subcellular localization. For example, Taylor et al. (18) used a mouse aminoterminal anti-peptide antibody and immunoprecipitation to demonstrate the induction of TTP in mouse fibroblasts and the decreased gel mobility upon cell stimulation with various mitogens, due in part to phosphorylation by one or more mitogenactivated protein kinases. They also used the same antiserum to show that TTP in stably transfected serum-deprived fibroblasts was mostly in the nucleus but translocated to the cytosol within 5 min of serum stimulation (33). Carballo et al. (7) found a massive increase of TTP in the cytosol of primary mouse macrophages 4 h after LPS or TNF stimulation, using radioactive labeling and immunoprecipitation with the same antipeptide antiserum. Lai et al. (8) were able to immunoprecipitate mouse macrophage TTP cross-linked to radiolabeled TNF RNA and found that most overexpressed TTP was present in the cytosol of 293 cells. Rigby and colleagues (34) used an affinity-purified polyclonal antibody directed against a carboxyl-terminal peptide from human TTP to show that most immunoreactive TTP was in the cytosol of LPS-stimulated THP-1 cells and could be induced rapidly and transiently in those cells after LPS stimulation (35). Interestingly, their data from THP-1 cells suggested a more transient response of the TTP protein to LPS stimulation than seen in the mouse cells described here. The protein turnover rate was not commented on in those studies, nor was the apparent size change upon stimulation noted, probably because most of their protein assays involved fluorescence-activated cell sorting.
In the present study, TTP was essentially undetectable in unstimulated RAW 264.7 macrophages, although slightly higher levels of punctate nuclear staining were observed in these cells stained in parallel with the immune serum compared with pre-immune serum. This result is compatible with a previous study in serum-deprived cultured fibroblasts stably expressing TTP driven by a metallothionein promoter, in which unstimulated cells expressed predominantly nuclear TTP, but the protein had shifted into the cytosol within 5 min of serum stimulation (33). We also noted in the current study that there was detectable cytoplasmic expression in dividing cells, even without LPS stimulation. Within an hour or so of LPS staining in the cultured macrophages, clear-cut cytoplasmic staining was observed, which remained readily detectable in an interesting vesicular pattern for several hours. Cellular fractionation of similarly treated cells confirmed the cytoplasmic localization of almost all immunoreactivity in the LPS-treated cells. The nature of the vesicular pattern observed in the cytoplasm is not clear, but most immunoreactivity did not sediment with a membrane fraction centrifuged at 100,000 ϫ g for 1 h. Costaining with markers of other cellular organelles previously associated with TTP, such as exosomes (36) and stress granules (37), as well as other organelles will be of interest in this regard.
We also used this antiserum to probe the expression of TTP protein in normal tissues of adult animals. This required extraordinarily large amounts of protein to be loaded into the gel wells so that Western blots could be performed, with comparable samples from TTP KO mice used to ensure the specificity of the observed bands. When this was done, protein was detectable as multiple bands in similar tissues to those previously characterized as expressing the most mRNA by Northern blot-ting; these include spleen, thymus, lung, and liver, with lower or essentially undetectable expression in other tissues (13). One somewhat surprising finding was the relatively high level expression in large intestine, with much lower levels in small intestine; this is in contrast to previous Northern blot data, in which high level mRNA expression was observed in small intestine. 2 This is also interesting given the fact that the TTP KO mice on a C56Bl6 background do not seem to develop a Crohn'slike colitis, in contrast to mice in which the ARE of the TNF transcript has been removed, resulting in a stable transcript and TNF overexpression (38). Identification of the cell types expressing TTP under normal circumstances and in models of experimental colitis will be of interest in this regard.
In a very recent report in which a commercial antibody was used, the authors described a very different pattern of TTP tissue distribution in the mouse (39). For example, they detected the highest level of expression in liver, with significant expression in testis and ovary, with TTP being represented as a single sharp band of M r 43,000; in contrast, we could not detect immunoreactivity in the testis. In addition, they reported minimal to undetectable expression in spleen, thymus, lung, and intestine, in contrast to our results showing readily 2 G. A. Taylor and P.J. Blackshear, unpublished data.
FIG. 8. TTP immunostaining in rat spleen. Rats were injected intraperitoneally with LPS (5 mg/kg; A and B) or PBS (5 ml/kg, C and D), and spleens were removed and used for immunostaining of TTP with either the anti-MBP-mTTP serum (I) or pre-immune serum (PI) as described under "Materials and Methods." The reddish brown staining in A represents TTP immunoreactivity, whereas none was seen in the other three panels. The arrows in A point to the white pulp of the spleen, which is unstained by the TTP antibody. See the text for other details. detectable expression in those tissues. However, certain critical controls were omitted from their report, including the use of pre-immune serum, overexpressing cells as a positive control, and tissues from KO mice as negative controls, casting some doubt on the reliability of their expression data.
We used these Western blot data to calculate an approximate intracellular concentration of TTP in spleen of normal adult mouse and in LPS-stimulated RAW 264.7 cells. This was done by comparing the expression levels of a known amount of recombinant protein to that contained in a known amount of spleen cytosol and RAW 264.7 cells. This calculation revealed an overall concentration in spleen of about 1.5 nM, and in maximally stimulated RAW 264.7 cells of about 0.5 M. The extraordinarily low value in spleen may account for previous difficulties in determining protein levels in normal tissues and high background of nonspecific bands on Western blots reported here. The dramatic stimulation of TTP in RAW 264.7 cells by LPS may explain why TTP could be easily detected in the stimulated cells and the Western blots are much cleaner than those using normal tissues. Obviously, concentrations in specific cell types within the spleen are likely to be higher. One paradoxical finding is that, in previous studies of TTP transcript localization by in situ hybridization histochemistry, there was apparent high level transcript expression in the white pulp containing almost exclusively lymphocytes, whereas the rest of the spleen expressed much lower levels. 3 However, in the present studies, no protein immunoreactivity was detected in the white pulp of rat spleen, either under normal circumstances or 2 h after injection of LPS. Instead, the immunoreactive cells were in the periphery of the spleen and appeared to be macrophages and stromal cells; even in these cells, there was minimal detectable specific immunoreactivity in the saline-injected rat spleen. Obviously, there are differences in sensitivity of the two techniques, but the finding of relatively high transcript levels in the white pulp of the spleen (and other lymphoid organs of the adult mouse) in the setting of undetectable protein levels raises the possibility of a translational block under such circumstances. Alternative explanations include the possibility that the protein is labile in the unphosphorylated and/or nuclear state, resulting in the failure to accumulate significant amounts of protein in the unstimulated state. These uncertainties should be resolvable in cultured lymphocytes or lymphoid cell lines. In any case, the LPS stimulus in intact rats did not seem to overcome this failure of lymphoid protein expression, because the white splenic pulp remained antibody-negative.
A final point concerns the oligomeric nature of the induced endogenous protein in RAW 264.7 cells. When TTP is overexpressed in 293 cells, it is common to observe apparent dimers on SDS gels, either by Western blotting or by cross-linking with 32 P-labeled ARE probes (7,8). This finding, as well as our repeated identification of TTP as a binding partner for itself in yeast two-hybrid analyses, 4 suggest that a substantial fraction of TTP might exist in the cell as dimeric or oligomeric species. The current data suggest that the predominant species of endogenous, induced TTP in RAW 264.7 cells is a monomer; this was supported by the Western blotting data, which did not identify the same types of putative dimers as seen in the overexpression studies. Nonetheless, although there was no obvious higher molecular weight peak of immunoreactivity on the gel filtration column, substantial immunoreactive protein was detected at larger sizes. This could conceivably represent dimers or oligomers of TTP, or possibly hetero-oligomers with putative TTP-interacting proteins.