Mammalian target of rapamycin positively regulates collagen type I production via a phosphatidylinositol 3-kinase-independent pathway.

The mammalian target of rapamycin (mTOR) is a multifunctional protein involved in the regulation of cell growth, proliferation, and differentiation. The goal of this study was to determine the role of mTOR in type I collagen regulation. The pharmacological inhibitor of phosphatidylinositol (PI) 3-kinase, LY294002, significantly inhibited collagen type I protein and mRNA levels. The effects of LY294002 were more pronounced on the collagen alpha1(I) chain, which was inhibited at the transcriptional and mRNA stability levels versus collagen alpha2(I) chain, which was inhibited through a decrease in mRNA stability. In contrast, addition of the PI 3-kinase inhibitor, wortmannin, did not alter type I collagen steady-state mRNA levels. This observation and further experiments using an inactive LY294002 analogue suggested that collagen mRNA levels are inhibited independent of PI 3-kinase. Additional experiments have established that mTOR positively regulates collagen type I synthesis in human fibroblasts. These conclusions are based on results demonstrating that inhibition of mTOR activity using a specific inhibitor, rapamycin, reduced collagen mRNA levels. Furthermore, decreasing mTOR expression by about 50% by using small interfering RNA resulted in a significant decrease of collagen mRNA (75% COL1A1 decrease and 28% COL1A2 decrease) and protein levels. Thus, mTOR plays an essential role in regulating basal expression of collagen type I gene in dermal fibroblasts. Together, our data suggest that the classical PI 3-kinase pathway, which places mTOR downstream of PI 3-kinase, is not involved in mTOR-dependent regulation of type I collagen synthesis in dermal fibroblasts. Because collagen overproduction is a main feature of fibrosis, identification of mTOR as a critical mediator of its regulation may provide a suitable target for drug or gene therapy.

A common characteristic of all fibrotic diseases, including scleroderma, is an abnormal accumulation of extracellular matrix proteins. Fibrotic lesions disrupt normal tissue architecture and contribute to organ failure. Type I collagen, the pri-mary component of fibrotic lesions, is a triple helix composed of two ␣1 chains and one ␣2 chain. These chains, although coordinately expressed, are not regulated via the same mechanisms. It is well established that collagen protein degradation, mRNA stability, and transcription are tightly regulated during collagen biosynthesis. Signals from external stimuli such as cytokines (1)(2)(3), nutrients (4), and cell interactions (5, 6) modulate these processes via several pathways, including TGF-␤ 1 / p38 (7,8), PKC (9), and stress-activated protein kinase/c-Jun N-terminal kinase (10,11). Despite these advances, the pathways controlling collagen biosynthesis are not fully characterized, prompting us to further examine potential signaling molecules involved in type I collagen regulation.
Prior studies suggested the involvement of phosphatidylinositol 3-kinase (PI 3-kinase), a ubiquitous lipid kinase, in collagen regulation. For example, Ivarsson et al. (12) described PI 3-kinase as a regulator of collagen production in attached, rounded fibroblasts. However, in the same study, the PI 3-kinase inhibitor LY294002 did not inhibit collagen production in spread dermal fibroblasts. Furthermore, in recent studies, Ricupero et al. (13) demonstrated that PI 3-kinase inhibitors (LY294002 and wortmannin) decreased collagen steady-state mRNA levels in lung fibroblasts. However, it has become clear that LY294002 and wortmannin have multiple targets, suggesting that PI 3-kinase may not be the relevant target. Other candidates include a family of proteins that share significant homology with the PI 3-kinase catalytic domain, a target of LY294002 and wortmannin. For the most part, these proteins act as checkpoints for DNA synthesis, protein translation, and mRNA quality and mediate cellular responses to stresses such as DNA damage and nutrient deprivation. For example, family members DNA-protein kinase (14), ataxia telangiectasia-mutated (ATM), ataxia telangiectasia-related genes are regulators of DNA repair, whereas the recently identified hSMG-1 (15,16) regulates nonsense-mediated decay. Consistent with other members of this family, the mammalian target of rapamycin (mTOR) (17,18) also appears to act as a checkpoint. mTOR regulates yeast growth (19,20), serves as a nutrient sensor (21), and is perhaps best known as a translational regulator. Although initially defined as a translational regulator that exerted its influence through p70 S6 kinase (p70 S6K ) (22) and elongation initiation factor 4E-binding protein (4EBP1) (23), recent evidence suggests that the influence of mTOR is more extensive. For instance, mTOR has also been implicated in the control of protein kinase C␦ (PKC␦) (24), PP2A (25), interleukin-3 mRNA stability (26), and eIF4G protein stability (27). Significantly, mTOR inhibition by rapamycin (28) decreases extracellular matrix deposition in a rat model of fibrogenesis, suggesting a role for mTOR in collagen regulation in vivo (29). Although mTOR regulates numerous genes, it in turn is regulated by a variety of stimuli and proteins. For example, c-Abl prevents mTOR phosphorylation in vitro and in vivo (30), and protein kinase B (or Akt) has been demonstrated to phosphorylate mTOR on Thr-2446 and Ser-2448, thereby linking mTOR to a mitogenic pathway (31,32). mTOR may also regulate its own activity by autophosphorylating on Ser-2481, and mutation of residues Asp-2338 or Asp-2357, within the catalytic domain, abolishes the autophosphorylation reaction (22). In addition to its kinase function, the presence of HEAT repeats (huntington, elongation factor 2, the A subunit of type 2A protein phosphatase (PP2A) and Tor) (reviewed in Ref. 33) and FAT/FATC domains (FRAP, ATM, and TRRAP C-terminal) and its large size (289 kDa) raise the possibility that mTOR also acts as a scaffolding protein. mTOR has also been shown to shuttle between the cytoplasm and the nucleus, although the function of this translocation is not clear. Surprisingly, however, no conventional nuclear import signal or nuclear export signal has been in found in the mTOR sequence. Nevertheless, this shuttling seems to be a necessary step in the phosphorylation of p70 S6K and 4EBP1 (34).
The PI 3-kinase inhibitors LY294002 and the structurally unrelated inhibitor wortmannin have helped to elucidate the role of PI 3-kinase in many cellular processes. However, the nonspecific nature of these inhibitors warrants caution. For this reason, rapamycin, which binds the FK506-binding protein 12 (FKBP12)-rapamycin binding domain (35) of mTOR, is useful in the characterization of PI 3-kinase-independent pathways. Based on previous work showing that PI 3-kinase and related molecules may be involved in collagen gene regulation, the aim of our study was to use these inhibitors and related reagents to ascertain the role of the PI 3-kinase pathway in collagen production. In this study, we show that mTOR inhibition decreases collagen mRNA stability via a PI 3-kinase-independent mechanism.  , TX). Tissue culture reagents, Dulbecco's modified Eagle's medium (DMEM), fetal calf serum (FCS), and 50-and 100-bp DNA ladders were obtained from Invitrogen. Primers were obtained from Sigma-Genosys (The Woodlands, TX). Nylon membranes were from Stratagene (La Jolla, CA). Actinomycin D, dimethyl sulfoxide (Me 2 SO), phosphatidylinositol (PI), and ATP were obtained from Sigma. Protease inhibitor mixture set III and pansorbin were obtained from Calbiochem. ECL reagent and anti-rabbit horseradish peroxidaselinked whole antibody (from donkey) were from Amersham Biosciences. D.C. protein assay reagent was obtained from Bio-Rad. Akt and phospho-Akt antibody were from Cell Signaling Technology (Beverly, MA). Anti-p85 PI 3-kinase antibody was obtained from Upstate Biotechnology, Inc. (Waltham, MA). Bicinchoninic acid protein assay reagent was from Pierce. The 4EBP1 antibody was purchased from Zymed Laboratories Inc. (South San Francisco, CA).

Materials
Cell Culture-Human fibroblasts were obtained from foreskins of healthy newborns from the Medical University of South Carolina hospital. Foreskins were treated with collagenase (Sigma) overnight at 37°C. Primary fibroblasts cultures were established in 25-cm 2 culture flasks in DMEM supplemented with 20% FCS and 2 mM glutamine.
Monolayer cultures were maintained at 37°C in 10% CO 2 until confluence. Cells were then passaged and used for experiments.
RNA Preparation and Northern Blot Analysis-Fibroblasts were grown to confluence in 10-cm 2 dishes in DMEM supplemented with 10% FCS. Confluent fibroblasts were serum-starved in DMEM containing 0.1% BSA for 24 h, followed by incubation with inhibitors at the indicated concentrations for 24 h. Control cells received the vehicle (Me 2 SO). Total RNA was extracted; 3.5 g of RNA was analyzed by Northern blotting as described previously (36). Nylon membranes were hybridized with 32 P-labeled cDNA probes for COL1A1 (kindly provided by Dr. E. Vuorio, Finland), COL1A2 (kindly provided by Dr. R. G. Crystal, New York), and TIMP-1 (as described previously (37)) and normalized with a probe for 18 S rRNA (5Ј-ACG GTA TCT GAT CGT CGA ACC-3Ј) as described previously (38). The membranes were scanned and quantitated by using a PhosphorImager (Amersham Biosciences).
siRNA Transfection-Fibroblast (1.25 ϫ 10 5 ) cells were plated in 60-mm 2 dishes for the measurement of mTOR, COL1A1, and COL1A2 mRNA levels. Cells were transfected by using OligofectAMINE (Invitrogen) and an mTOR-specific small-interfering RNA (siRNA) (Xeragon) (DNA target sequence, TGA GAG GAA AGG TGG CAT C) and a nonsilencing siRNA (Xeragon) (DNA target sequence, AAT TCT CCG AAC GTG TCA CGT). The final concentration of the siRNA was 200 nM. Cells were incubated for 72 h and then total RNA was isolated and treated with DNase I. RNA concentrations were measured by a spectrophotometer and then confirmed by RNA electrophoresis. RNA (500 ng) was amplified by using the Qiagen QuantiTect SYBR Green RT-PCR Kit on the Bio-Rad iCycler. Primers were utilized at a final concentration of 0.3 M, in a total volume of 25 l. Amplification was for 95°C for 15 min, followed by 38 cycles of 94°C for 30 s, 60°C for 30 s, and 73°C for 30 s, followed by a 5-min extension at 72°C. Endogenous mTOR levels were quantitated by using a standard curve. Endogenous human mTOR was amplified using upper primer (5Ј GGC CGA CTC AGT AGC AT 3Ј) and lower primer (5Ј CGG GCA CTC TGC TCT TT 3Ј), which yields a 119-bp product.
Electroporation and Rat mTOR-KD Amplification-Fibroblasts (3 ϫ 10 6 ) cells were resuspended in 750 l of DMEM. A 200g aliquot of pcDNA 3 or FLAG-tagged rat mTOR-kinase dead (D2338A) (mTOR-KD) in pcDNA 3 (kindly provided by Robert Abraham, Duke University) was added to the suspension. Suspensions were added to 0.4-cm gap electrocuvettes (Bio-Rad). Samples were electroporated at 0.36 kV and 0.5 microfarads (Gene Pulser® II Electroporation System with Capacitance Extender Plus, Bio-Rad). Samples were brought to a final volume of 3 ml in DMEM containing 10% FCS. Cells were allowed to recover 72 h before collection. Total RNA was isolated from electroporated cells and treated with DNase I. RNA (1 g) was converted to cDNA by using a random hexamer primer. Rat mTOR-KD was amplified by using upper primer (5Ј-CGC CAT TGC CAC AGA GGA C-3Ј) and lower primer (5Ј-TCT CAT TGC TCT CGG CTT CAC-3Ј) yielding a 383-bp product. Primers were utilized at a final concentration of 0.6 M and MgCl 2 at a final concentration of 1 mM in a total volume of 50 l. Amplification was for 95°C for 5 min, followed by 35 cycles of 95°C for 1 min, 59°C for 30 s, and 72°C for 45 s, followed by a 5-min extension at 72°C. In parallel, 3.5 g of RNA was used for collagen Northern blot analysis.
Measurement of Collagen mRNA Message Stability-Fibroblasts were grown to confluence in 10-cm 2 dishes in DMEM supplemented with 10% FCS. Confluent fibroblasts were serum-starved in DMEM containing 0.1% BSA for 24 h, followed by incubation with inhibitors or Me 2 SO at the indicated concentrations for 12 h. Actinomycin D (8 M) was added; and cells were scraped before actinomycin addition and at the indicated time points. Total RNA was extracted; 3.5 g was analyzed by Northern blotting. Bands were normalized against 28 S and 18 S rRNA. The 0-h time point was set to 1 with the remaining time points relative to this time point. t1 ⁄2 was calculated as the amount of time required for the collagen message to decrease 50% from time 0 h. Bands were quantitated using NIH ImageJ software.
Measurement of COL1A1 and COL1A2 Transcription-Fibroblasts were grown to confluence in 10-cm 2 dishes in DMEM supplemented with 10% FCS. Confluent fibroblasts were serum-starved in DMEM containing 0.1% BSA for 24 h followed by incubation with inhibitors at the indicated concentrations for 24 h. Me 2 SO was added to control cells. Total RNA was isolated and treated with DNase I. RNA (1 g) was reverse-transcribed to cDNA by using a random hexamer primer. Transcription was measured using a RT-PCR-based method described previously (39). Unspliced COL1A1 (newly transcribed, heterogeneous RNA) was amplified using an upper primer to exon 46 (5Ј-AGG GTG AGA CAG GCG AAC AG-3Ј) and lower primer to intron 46 (5Ј-GCC CGA GGG TGT CCA TA-3Ј) and yielded a 300-bp product. Unspliced COL1A2 was amplified using an upper primer to exon 1 (5Ј-CGC GGA mTOR Regulates Type I Collagen CTT TGT TGC TGC TTG-3Ј) and lower primer to intron 1 (5Ј-GGC CCT TGA CTT CCT CCA CCA C-3Ј) and yielded a 393-bp product. 18 S rRNA was amplified using the QuantumRNA TM Classic 18 S Standards (Ambion) to a yield a 488-bp product. One microliter of cDNA for unspliced collagen and 0.5 l of cDNA for spliced and 18 S rRNA were used for amplification. Amplification used primers at a final concentration of 0.6 M and MgCl 2 at a final concentration of 1 mM in a total volume of 50 l. Unspliced collagens were amplified in the presence of 6% formamide. Amplification was for 95°C for 5 min, followed by cycles of 95°C for 1 min, 59°C for 30 s, and 72°C for 45 s followed by a 5-min extension at 72°C. Unspliced COL1A1, COL1A2, and 18 S rRNA were amplified for 34, 32, and 23 cycles, respectively. Product sizes were measured using a 100-bp DNA ladder.
Measurement of COL1A2 Poly(A) Tail Length-Fibroblasts were grown to confluence in 10-cm 2 dishes in DMEM supplemented with 10% FCS. Confluent fibroblasts were serum-starved in DMEM containing 0.1% BSA for 24 h. LY294002 (40 M) or Me 2 SO was added for 6 h. Actinomycin D (8 M) was then added, and cells were scraped at the indicated time points. Poly(A) tails were amplified based on the procedure of Salles and Strickland (40). Total RNA was isolated, and 2 g of RNA was converted to cDNA using the lower poly(dT) primer. Primers were utilized at a final concentration of 0.6 M and MgCl 2 at a final concentration of 1.5 mM in a total volume of 50 l. Amplification was for 5 min at 95°C and then 26 cycles of 95°C for 1 min, 59°C for 30 s, and 72°C for 30 s followed by a 5-min extension at 72°C. The 3Ј poly(A) tail was amplified by using upper primer (5Ј-TGT TCT TTG CCA GTC TCA TTT-3Ј) and lower poly(dT) primer (5Ј-GCG AGT CCG CTT TTT TTT TTT TTT T-3Ј). Product sizes were measured using a 50-bp DNA ladder.
Procollagen Analysis on SDS-Polyacrylamide Gel-Fibroblasts were seeded in 12-well plates and grown to confluence in DMEM supplemented with 10% FCS. Confluent fibroblasts were serum-starved in DMEM containing 0.1% BSA and 50 g/ml ascorbic acid for 24 h. LY294002 was then added at the indicated concentrations for 48 h in the presence of ascorbic acid. L-[2,3,4,5-3 H]Proline (20 Ci/ml, 3.66 TBq/mmol) was added during the last 24 h of incubation. Medium was collected from each well, and cells were trypsinized and counted with a Z TM series Coulter® cell counter from Beckman-Coulter Inc. (Miami, FL). Aliquots of medium normalized for cell numbers were concentrated using a Speed-vac, denatured by boiling in SDS sample buffer containing 15 mM dithiothreitol, and loaded on a 6% SDS-polyacrylamide gel. Media were then electrophoresed, and gels were enhanced using Fluoro-Hance (Research Products International, Mount Prospect, IL) and visualized by autoradiography. The nature of the collagen bands was verified previously by collagenase digestion (41).
For collagen assays utilizing RNA interference (siRNA), fibroblasts (6 ϫ 10 4 ) were plated in 12-well plates. Cells were transfected with the mTOR-siRNA (400 nM) using OligofectAMINE (Invitrogen) 24 h after plating. Twenty four hours before each time point, applicable wells were placed in serum-free media containing [ 14 C]proline (20 Ci/ml) and ascorbic acid (50 g/l). The medium was collected after 48, 72, and 96 h, and cell numbers were counted. Media normalized for cell number were concentrated using Ultrafree-MC centrifugal filter devices (Millipore), denatured by boiling in SDS sample buffer containing 15 mM dithiothreitol, loaded on a 6% SDS-polyacrylamide gel, and electrophoresed. Gels were enhanced using Fluoro-Hance (Research Products International, Mount Prospect, IL) and visualized by autoradiography.
[ 35 S]Methionine-labeled Total Protein Synthesis-Fibroblasts were grown to confluence in 10-cm 2 dishes in DMEM supplemented with 10% FCS. Confluent fibroblasts were serum-starved in DMEM containing 0.1% BSA for 24 h followed by the addition of LY294002 (80 M) for 24 h. Cells were then incubated for 30 min in methionine-free DMEM. [ 35 S]Methionine (100 Ci/ml) was added for 3 h. Cells were then washed in phosphate-buffered saline and collected by scraping. An aliquot of cells was counted to normalize for cell number. Cells were lysed, and protein, normalized for cell number, was electrophoresed on a 12% SDS-polyacrylamide gel. After electrophoresis, the gel was dried and exposed to autoradiography film for 1 h.
Akt Western Blot-Fibroblasts were grown to confluence in 60-mm 2 dishes in DMEM supplemented with 10% FCS. Inhibitors or Me 2 SO, as a control, were added for 24 h. Cells were then scraped and lysed in ice-cold Lysis buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 10% glycerol, 5 mM EDTA, protease inhibitor mixture set III (2 l/ml), 10 mM NaF, 10 mM NaPP i , 0.1 mM phenylmethylsulfonyl fluoride, 0.2 mM Na 3 VO 4 ). Cell lysates were normalized for protein concentrations by using the D.C. protein assay reagent (Bio-Rad). Lysate (30 g) was electrophoresed on a 12% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane. The membrane was blocked with 4% milk in Tween 20-TBS (TTBS) for 1 h and then incubated with either anti-phospho-Akt or anti-Akt antibody (1:1000) overnight in (TTBS). After being washed in TTBS three times for 5 min, membranes were incubated with horseradish peroxidase-labeled secondary antibody (1:2500) for 1 h. After washing in TTBS three times for 5 min, bands were visualized using ECL reagent. Membranes were stripped as instructed in the ECL protocol and reprobed with the next primary antibody as above.
4EBP1 Western Blot-Inhibitors or Me 2 SO, as a control, were added to confluent fibroblasts in DMEM ϩ 10% FCS for 2 h before scraping. Western blotting was performed as above. Briefly, primary antibody for 4EBP1 was used at a 1:1000 dilution (Zymed Laboratories Inc.). Specific binding of antibody was detected with a secondary anti-rabbit antibody (1:2000) (Amersham Biosciences). Phosphorylation states were observed as changes in electrophoretic mobility. The ␣ band represents the least phosphorylated form, whereas the ␦ band is the most highly phosphorylated form.
Statistical Analysis-The Student's t test analysis using GraphPad InStat Statistics software (version 1.12) was performed to determine statistical significance. Values of less than or equal to 0.05 were considered statistically significant.

LY294002 Selectively Decreases Collagen Protein and mRNA
Levels-We investigated the ability of PI 3-kinase or related proteins to regulate collagen production by using LY294002, a PI 3-kinase inhibitor. LY294002 (16 and 32 M) was added to cells that were serum-starved for 24 h and labeled with [ 3 H]proline. In Fig. 1A, LY294002 decreased types I and III collagen protein levels, suggesting that PI 3-kinase or a related protein regulates collagen production. However, the possibility that LY294002 was a general inhibitor of protein synthesis remained. Further experiments showed that total protein levels were not decreased after LY294002 (80 M) addition (Fig.  1B). Taken together, these experiments suggest that LY294002 selectively inhibits collagen synthesis. Additional experiments were designed to investigate the affected point in type I collagen biosynthesis. PI 3-kinase has been shown to be involved in the sorting and transport of some lysosomal proteins (42,43). Therefore, it was possible that a block in collagen transport to the extracellular space might result in the decrease in collagen production observed in our assays. Thus, to define further the role of LY294002 in collagen biosynthesis, steady-state mRNA levels were analyzed. Northern blot analysis with 32 P-labeled cDNA probes for COL1A1 and COL1A2, normalized to 18 S rRNA, showed that 24 h after LY294002 (20 -80 M) addition, COL1A1 and COL1A2 mRNA decreases in a dose-dependent manner (Fig. 1C). The inhibition of COL1A1 was more pronounced than that of COL1A2, whereas TIMP-1 mRNA showed increased steady-state mRNA levels. This evidence suggests that LY294002 decreases type I collagen protein levels by decreasing its mRNA levels.

mTOR Regulates Type I Collagen
To define further how LY294002 decreases type I collagen steady-state mRNA levels, the amounts of unspliced (newly transcribed) collagen message were measured. A reduction in steady-state mRNA levels could signify a decrease in either transcription or message stability. Hence, the influence of LY294002 on COL1A1 and COL1A2 transcription was explored. To measure COL1A1 and COL1A2 transcription, we utilized a previously described method based on the PCR (39). Cells were treated for 24 h with or without LY294002 (40 M).
Primers to exon 46 and intron 46 were used to measure unspliced (newly transcribed, heterogeneous nuclear RNA) COL1A1 mRNA. Exon 1 and intron 1 primers were utilized to measure unspliced COL1A2 mRNA (Fig. 1D). Quantification of the bands by NIH image densitometry software and normalization to 18 S rRNA showed that COL1A1 unspliced mRNA levels were decreased after LY294002 treatment (average 91.3% decrease Ϯ 6%), pointing to transcriptional regulation. However, COL1A2 mRNA levels were not significantly dimin- FIG. 1. A, LY294002 decreases procollagen synthesis. LY294002 (20 -80 M) was added for 24 h to serum-starved foreskin fibroblasts followed by addition of [ 3 H]proline for an additional 24 h. Media were collected, normalized for cell number, and then electrophoresed on a 6% SDS-polyacrylamide gel. Gels were enhanced, dried, and then exposed to autoradiography film. Gel is representative of three independent experiments. Lane C, control. B, LY294002 does not decrease total protein synthesis. LY294002 (80 M) was added for 24 h to serum-starved foreskin fibroblasts. Cells were incubated in methionine-free media for 30 min. [ 35 S]Methionine (100 Ci/ml) was added to the media for 3 h. An aliquot of cells was counted, and then protein, normalized for cell numbers, was electrophoresed. Gels were enhanced, dried, and exposed to autoradiography film for 1 h. Experiment was performed in duplicate. C, LY294002 decreases COL1A1 and COL1A2 mRNA levels but increases TIMP-1 mRNA. Left panel, LY294002 was added to serum-starved confluent fibroblasts for 24 h. Total RNA was extracted and analyzed by Northern blot with 32 P-labeled cDNA probes for COL1A1, COL1A2, TIMP-1, and 18 S rRNA. A representative gel is shown. Lane C, control. Right panel, bands were quantitated with a PhosphorImager. Bars represent the average of three independent experiments with standard deviation. D, LY294002 (LY) inhibits COL1A1 but not COL1A2 transcription. RT-PCR was performed after treatment with LY294002 (40 M) for 24 h by using primers for unspliced COL1A1 and COL1A2 mRNA. Lane C, control. 18 S rRNA primers were used to normalize for loading. Gels were stained with ethidium bromide and photographed, and bands were quantitated with NIH image. A representative gel is shown.
ished, suggesting no decrease in transcription (average 26% decrease Ϯ 16%) (Fig. 1D). These results suggest that although LY294002 partially regulates COL1A1 at the transcriptional level, COL1A2 is regulated post-transcriptionally. However, an increase in unspliced RNA resulting from a reduction in the splicing capabilities of the cell cannot be eliminated.
Collagen Message Stability Is Decreased by LY294002-To explore additional post-transcriptional mechanism altered by LY294002, we examined the effect on collagen message stability. It has been shown previously that both actinomycin D (44) and pulse labeling (45) yield similar results when measuring collagen message stability. Prior evidence that mRNA stability plays a role in collagen regulation also exists. For example, amino acid deprivation (4) and a lack of cysteine (46) cause a decrease in collagen mRNA stability, whereas TGF-␤ has been implicated in the stabilization of collagen mRNA (44). To measure collagen message stability, cells were serum-starved for 24 h and then treated with LY294002 (40 M) for 12 h. Fibroblasts were collected at 0, 4, 6, 8, and 10 h after the addition of actinomycin D (8 M). RNA was electrophoresed, and the membranes were then hybridized with 32 P-labeled cDNA probes for COL1A1 and COL1A2. The results showed that LY294002 significantly decreased COL1A1 and COL1A2 message stability by 2.02-and 2.46-fold, respectively (Fig. 2, A and B). Therefore, LY294002 can decrease collagen type I production through destabilization of the collagen message.
Numerous factors, including changes in protein binding, activation of mRNA surveillance, and changes in poly(A) tail length may regulate mRNA stability. Poly(A) tails serve to stabilize mRNAs by protecting them from nuclease attack, and deadenylation of the poly(A) tail precedes the degradation of most mRNAs (47). To confirm further that LY294002 alters COL1A2 stability, we measured poly(A) tail lengths. The COL1A2 message contains five different polyadenylation sites (48,49), but for simplicity only the 3Ј-most poly(A) tail was amplified. Primers were designed to encompass the end of the 3Ј-UTR of COL1A2 and the poly(A) tail (Fig. 2C). Reverse transcription using a poly(dT) primer with a GC anchor, followed by PCR with 3Ј-UTR and poly(dT) primers, allowed visualization of the COL1A2 poly(A) tail length (ϳ75 adenosines at 0 h). Specific amplification was demonstrated by the release of a 45-bp fragment after HindIII digestion (data not shown). Cells were treated with or without LY294002 for 6 h; transcription was then halted with actinomycin D (8 M). It was found that the poly(A) tail of COL1A2 decreased at a faster rate in LY294002-treated cells than in control cells consistent with destabilization of the COL1A2 message (Fig. 2D).
Collagen Steady-state mRNA Levels Are Decreased after the Addition of a LY294002 Analogue but Not Wortmannin-To evaluate further the role of PI 3-kinase in collagen production, wortmannin, a structurally unrelated inhibitor of PI 3-kinase, was used. In contrast to the expected decrease in collagen synthesis, we observed no consistent effect on COL1A1 or COL1A2 mRNA levels at concentrations reported to decrease PI 3-kinase activity (10 -1000 nM) (Fig. 3A).
Based on these results, we tested the effects of LY303511, an inactive LY294002 analogue. LY303511 contains a one-atom substitution that abolishes its ability to inhibit PI 3-kinase activity (50). Its actions on other signaling molecules have not been described. At concentrations of 20 -80 M, LY303511 de- mTOR Regulates Type I Collagen creased COL1A1 and COL1A2 steady-state mRNA levels in a dose-dependent manner (Fig. 3B). These results suggest that a PI 3-kinase-related molecule that is not inhibited by wortmannin is responsible for the inhibitory action of LY294002.
It has been reported that wortmannin is unstable in media for extended periods (51). Thus, it was possible that PI 3-kinase activity returned to normal levels during the course of the experiment, resulting in no effect on collagen mRNA steadystate levels. To test whether the PI 3-kinase pathway remained suppressed after incubation with the various inhibitors, we examined the phosphorylation of Akt, a downstream target of PI 3-kinase. Inhibitor addition for 24 h showed that LY294002 and wortmannin decreased Akt phosphorylation to a similar extent (Fig. 3C). Furthermore, an in vitro PI 3-kinase assay verified that LY294002 and wortmannin inhibit PI 3-kinase activity, but neither rapamycin nor LY303511 inhibit its activity (Fig. 3D). Therefore, these observations confirm that although wortmannin and LY294002 efficiently block PI 3-kinase activity in our experiments, only LY294002 specifically decreases type I collagen production. Thus, it is apparent that LY294002 inhibits a molecule other than PI 3-kinase that is refractory to wortmannin but regulates type I collagen.
Because our data suggested an alternative target for LY294002, we examined the effect of inhibitor treatment on 4EBP1 phosphorylation, a translation regulator downstream of mTOR (Fig. 3E). Within a cell, increases in phosphorylation of 4EBP1 dissociate it from its target eIF4E thus increasing capdependent translation. Changes in 4EBP1 phosphorylation can be observed as alterations in electrophoretic mobility. The more highly phosphorylated form was arbitrarily assigned as ␦ and the least phosphorylated form as ␣. LY294002 addition resulted in the greatest reduction in 4EBP1 phosphorylation to almost exclusively the ␣ form, whereas rapamycin and wortmannin diminished the ␦ form of 4EBP1 and amplified the less phosphorylated ␥ form. In addition, the ␤ form is now present in the rapamycin-and wortmannin-treated cells, although more substantially in the rapamycin-treated cells. Together these results show wortmannin and rapamycin block mTOR activity to a lesser degree than LY294002. This is consistent with the observation that rapamycin is less efficacious toward mTOR kinase activity as compared with LY294002 and clearly demonstrates differences in their effects on mTOR and its downstream targets. The strong effect of LY294002 on 4EBP1 dephosphorylation may also be a result of the activation of a phosphatase not affected by wortmannin or rapamycin.
mTOR Regulates Collagen mRNA Levels-Because LY294002 can inhibit other PI 3-kinase-related molecules such as mTOR/ FRAP, ATM, ataxia telangiectasia-related, and DNA-protein ki- FIG. 3. A, wortmannin does not alter COL1A1 and COL1A2 mRNA steady-state levels. Left panel, cells were treated with inhibitor (10 -1000 nM) and analyzed as above. A representative gel is shown. Lane C, control. Right panel, bars represent the average of at least three independent experiments with standard deviation. B, LY303511 decreases COL1A1 and COL1A2 mRNA steady-state levels. Left panel, cells were treated with inhibitor (20 -80 M) and analyzed as above. A representative gel is shown. Lane C, control. Right panel, bars represent the average of at least three independent experiments with standard deviation. C, Akt phosphorylation after LY294002, LY303511, rapamycin, and wortmannin addition. Cells were incubated with indicated inhibitors for 24 h. Protein (30 g) from each condition was analyzed by Western blots using anti-phospho-Akt and anti-Akt antibodies. A representative gel is shown (inset). Graph represents quantitation of three independent experiments with standard deviation. *, p Ͻ 0.05. D, effect of LY294002, LY303511, rapamycin, and wortmannin on in vitro PI 3-kinase activity. Immunoprecipitated PI 3-kinase was incubated with the indicated inhibitors for 5 min before addition of cold ATP, PI, and [␥-32 P]ATP. PI 3-kinase activity was assayed by measuring the conversion of PI to PI[ 32 P] by TLC and visualization by autoradiography. E, 4EBP1 phosphorylation after inhibitor treatment. Reagents were used at the following concentrations: LY294002, 40 M; rapamycin, 10 nM; and wortmannin, 100 nM. Cells were incubated in the presence of inhibitors for 2 h followed by scraping and Western blot analysis with 4EBP1 antibody. mTOR Regulates Type I Collagen nase, we explored the possibility that mTOR is the relevant target of LY294002. The immunosuppressant rapamycin, in complex with FKBP12 (FK506-binding protein 12), inhibits mTOR at nanomolar concentrations. Therefore, the effects of rapamycin on collagen were examined. Type I collagen mRNA levels were significantly inhibited (50%) by rapamycin (10 nM) (Fig. 4A). It is not FIG. 4. A, rapamycin decreases COL1A1 and COL1A2 mRNA steadystate levels. Left panel, cells were treated with the indicated concentrations of inhibitor and analyzed as described under "Experimental Procedures." Lane C, control. Right panel, bars represent the average of at least three independent experiments with standard deviation. B, mTOR-siRNA decreases endogenous mTOR levels. Left panel, cells were transfected with an mTOR-specific interfering RNA or non-silencing (NS) interfering RNA for 72 h. Endogenous mTOR levels were measured by quantitative real time RT-PCR. Right panel, bars represent the average of at least three independent experiments with standard deviation. C, decrease in endogenous mTOR levels reduces COL1A1 and COL1A2 mRNA levels. Left panel, cells were transfected with an mTOR-specific interfering RNA or non-silencing interfering RNA for 72 h. Experiments were performed in parallel with the RNA remaining from the real time RT-PCR experiments. Northern blots were hybridized to COL1A1 and COL1A2 probes, and mRNA levels were normalized to 28 S and 18 S rRNA. Right panel, bars represent the average of at least three independent experiments with standard deviation. D, collagen type I protein levels decrease with addition of mTOR-siRNA. For collagen protein levels, mTOR-siRNA was added for the indicated times. [ 14 C]Proline and ascorbic acid were added to cells 24 h before media collection. Media normalized to cell numbers were electrophoresed and visualized by autoradiography. E, mTOR-KD decreases collagen mRNA levels. Left panel, mTOR-KD mRNA from electroporated samples was amplified using rat mTORspecific primers. Right panel, cells were electroporated with 20 g of mTOR-KD or pcDNA 3.0 and allowed to recover for 72 h. Type I collagen mRNA levels were analyzed by Northern blot. mTOR Regulates Type I Collagen surprising, however, that rapamycin and LY294002 did not exhibit the same efficacy toward collagen. These differences are likely a reflection of their dissimilar mechanisms of action. Specifically, LY294002 competes for ATP binding to mTOR (50), whereas the rapamycin-FKBP12 complex, through steric hindrance, partially occludes the ATP-binding site of mTOR (17,18,22,52,53). However, the effect of rapamycin on mTOR function is still not precisely defined. For example, Peterson and co-workers (22,54) reported that rapamycin does not inhibit the autophosphorylation of mTOR but instead may affect its translocation. Although rapamycin is generally thought to be specific for mTOR, decreases in other proteins have been shown after rapamycin addition, but this may be due to the key role of mTOR as a translational regulator (55). Regardless, our results point to the possibility that mTOR regulates collagen production.
We sought to confirm further the link between mTOR and collagen and to provide independent evidence for the role of mTOR in collagen gene expression. Therefore, we utilized an siRNA to specifically down-regulate mTOR levels. Decreased mTOR mRNA levels were confirmed by real time RT-PCR (Fig.  4B). In parallel, COL1A1 and COL1A2 mRNA levels were determined by Northern blot analysis from cells treated with non-silencing or mTOR-specific siRNA. Fig. 4C demonstrates that mTOR mRNA inhibition (56%) (p ϭ 0.08) results in a decrease in COL1A1 (75%) (p ϭ 0.0003) and COL1A2 (28%) (p ϭ 0.0176) mRNA levels. These results are consistent with the differential effect of LY294002 on COL1A1 and COL1A2 mRNA levels as shown in Fig. 1. We extended these studies to determine the effect of decreased mTOR levels on collagen protein levels. We determined that mTOR-siRNA addition causes a marked reduction in types I and III collagen protein levels during a 48 -96-h time period (Fig. 4D). ␤-Actin and TIMP-1 levels were not significantly affected (data not shown). The greater reduction in the protein levels as compared with the mRNA levels may be a reflection of the effect of mTOR at translational and post-translational levels in collagen biosynthesis. These results are consistent with the protein data obtained after LY294002 addition.
We also assessed the importance of the kinase domain of mTOR. Dermal fibroblasts were electroporated with a kinasedead mTOR (D2338A) (28). This construct should compete with native mTOR proteins for their substrate but not initiate the phospho-transfer reaction. Electroporation of a green fluorescent protein containing plasmid performed in parallel revealed that ϳ30% of the cells received plasmid (data not shown). Electroporation of the mTOR-KD plasmid into cells was confirmed by specific amplification (Fig. 4E, left panel). Northern blot analysis measuring endogenous gene levels showed a significant decrease in type I collagen mRNA levels (10 -20%, p ϭ 0.0025, n ϭ 7) after mTOR-KD electroporation (Fig. 4E, right  panel). The relatively modest decrease in collagen mRNA after mTOR-KD electroporation, as compared with either rapamycin or siRNA, may reflect the low efficiency of plasmid delivery into primary cells and the possibility that endogenous mTOR function was not completely blocked. However, taking into account that only a proportion of fibroblasts received plasmid, the observed decrease in endogenous COL1A1 and COL1A2 levels suggests the importance of the kinase domain of mTOR. Taken together, these data provide evidence for the involvement of mTOR in type I collagen regulation. DISCUSSION This study demonstrates a positive role for mTOR in the regulation of type I collagen mRNA synthesis. We base this conclusion on our results showing that, in addition to LY294002, rapamycin, a mTOR-specific interfering RNA, and a kinase-dead mTOR decrease collagen steady-state mRNA lev-els, thereby identifying mTOR as a potential new regulator of the collagen type I gene (Fig. 4). In addition, even though the traditional linear PI 3-kinase pathway places mTOR downstream of PI 3-kinase, we demonstrate that mTOR regulates COL1A1 and COL1A2 mRNA synthesis in a PI 3-kinase-independent manner. Our results show that although both LY294002 and wortmannin inhibit PI 3-kinase, only LY294002 inhibits collagen production ( Fig. 1 and Fig. 3, A and D). Furthermore, inhibition of type I collagen steady-state mRNA levels by LY303511, an LY294002 analogue that does not inhibit PI 3-kinase activity, argues against a PI 3-kinase-dependent pathway (Fig. 3B).
Significantly, earlier reports have found that wortmannin and LY294002 inhibit other PI 3-kinase-like proteins. For example, wortmannin inhibits PI 4-kinase (56, 57) and hSMG-1 (15), and wortmannin and LY294002 inhibit DNA-protein kinase (14) and mTOR (28). Although both agents inhibit PI 3-kinase and mTOR, higher concentrations of wortmannin are required for inhibition of mTOR, whereas LY294002 inhibits mTOR and PI 3-kinase with nearly identical potency (22,28). Therefore, the disparity between type I collagen inhibition by LY294002 and wortmannin may exist due to their different efficacies toward mTOR.
Recent evidence further supports the existence of a PI 3-kinase-independent pathway leading to mTOR. Fang and colleagues (58) recently determined that mitogenic activation of mTOR by phosphatidic acid occurs independent of PI 3-kinase. Moreover, it was reported that cAMP inhibited mTOR activity via a PI 3-kinase-independent pathway (32). Most important, cAMP elevations correlate with reductions in collagen synthesis (59,60). In addition, amino acid activation of mTOR occurs independent of Akt/protein kinase B and PI 3-kinase activation (31). Furthermore, prior evidence also exists for contrasting outcomes following LY294002 and wortmannin addition. For example, rapamycin and LY294002 inhibited nitric oxide production in Raw 264.7 cells, whereas wortmannin is ineffective (61). Likewise, studies in several different cell types have revealed that wortmannin and rapamycin exhibit distinct effects on the translation of individual mRNAs (62). Thus, it appears that there are multiple pathways leading to mTOR activation, which are utilized in different cellular processes.
On the other hand, previous studies (13) have also shown a role for PI 3-kinase in collagen regulation. Runyan et al. (63) determined that TGF-␤-mediated collagen stimulation involves PI 3-kinase, but PI 3-kinase alone is not sufficient for collagen up-regulation in mesangial cells. Also, although our studies and those by Ricupero and et al. (13) conclude that LY294002 inhibits collagen production via a mechanism affecting message stability, we disagree on the pathway implicated. Ricupero et al. (13) proposed that a PI 3-kinase-dependent pathway regulates collagen based on the observation that both LY294002 and wortmannin inhibit collagen steady-state levels in lung fibroblasts. However, our data indicate that wortmannin does not inhibit collagen steady-state levels in dermal fibroblasts (Fig. 3A). Nonetheless, the literature provides numerous examples describing distinct pathways regulating the same cellular process in different cell types. It should also be noted that although the literature describes wortmannin as an unstable inhibitor, based on the phosphorylation of Akt after both LY294002 and wortmannin exposure for 24 h, we conclude that its capacity to inhibit PI 3-kinase in fibroblasts is comparable with LY294002 (Fig. 3C).
Published reports further substantiate our results that mTOR regulates collagen and provide additional connections between mTOR and collagen production. For example, recent studies linked mTOR to the signaling pathway of the ␣ V ␤ 3 integrin (64). ␣ V ␤ 3 integrin, a receptor for vitronectin, is the primary molecule required for attachment and spreading of fibroblasts on cell culture dishes (65). It is well documented that collagen message stability decreases after seeding fibroblasts in collagen matrices, whereas plating cells on plastic increases collagen synthesis (66). It has also been demonstrated that collagen transcription and mRNA stability decrease after incubation of cells in amino acid-free media in lung fibroblasts (4). Thus, by taking into account that mTOR is a nutrient sensor, these data indirectly support the involvement of mTOR in collagen regulation. It may also be relevant that cAMP elevation inhibits mTOR activity in vitro (32) as well as collagen gene expression in vitro and in vivo (59,60). However, the mTOR and collagen studies were performed in different experimental systems.
In the context of collagen regulation, it is unlikely that mTOR inhibits type I collagen via p70 S6K or 4EBP1. COL1A2 possesses neither a polypyrimidine tract (TOP) motif nor significant secondary structure in its 5Ј-UTR (⌬G Ϫ55.3 kcal/mol; internet address: bioinfo.math.rpi.edu/ϳzukerm/rna/). Moreover, although type I collagen mRNA decreases after the addition of the LY294002 analogue, 4EBP1 phosphorylation does not diminish (data not shown). Additionally, 3-methyladenine and LY294002 are both inhibitors of autophagy, and 4EBP1 and 3-methyladenine did not inhibit collagen synthesis (67,68). 2 Thus, the most likely candidates for the downstream mediator of mTOR signaling involved in collagen production are PP2A, PKC␦, or another mTOR-associated protein. The recent findings of Jimenez et al. (9) showing that inhibition of PKC␦ causes a substantial reduction in COL1A1 mRNA levels supports this theory. Also relevant are the findings that PKC␦ is phosphorylated on Ser-662 and Thr-505. Although both phosphorylations are sensitive to LY294002, rapamycin only blocks phosphorylation on Ser-662 (69). This suggests a possible explanation for the differential effects of LY294002 and rapamycin on collagen and provides an additional link between PKC␦ and collagen production. Another likely explanation for the disparity between the effects of rapamycin and LY294002 involves the role of FKBP12, the rapamycin partner, in the TGF-␤ pathway. FKBP12 blocks TGF-␤ signaling by binding to TGF-␤RI, thus inhibiting TGF-␤RI phosphorylation by TGF-␤RII. Rapamycin reverses this effect (70,71). Therefore, based on these data and given that TGF-␤ increases collagen message stability, removal of FKBP12 by rapamycin likely counteracts some of the negative effects of rapamycin. This hypothesis is further substantiated by the mTOR siRNA studies. Fig. 4 demonstrates that a decrease in mTOR levels causes a decrease in COL1A2 levels and a substantial reduction in COL1A1 mRNA levels, which is mirrored at the protein level.
In summary, our study demonstrates that mTOR positively regulates type I collagen synthesis, independent of PI 3-kinase in human fibroblasts. Relevant to our findings is the recent clinical trial showing that sirolimus (rapamycin)-coated stents prevented restenosis (72,73). Rapamycin was also effective in preventing liver fibrogenesis in an animal model (29). It is possible that an alteration in the cell environment, such as hypoxia, activates mTOR leading to collagen up-regulation under fibrotic conditions. Although a tentative link has been formed between LY294002, mTOR, and collagen stability regulation, further studies are needed to establish a direct link between mTOR and type I collagen message stability. A direct link could have implications in fibrosis treatment because an increase in collagen mRNA stability is a primary mechanism of collagen up-regulation in activated hepatic stellate cells and contributes to elevated collagen levels in scleroderma fibroblasts (74,75). Therefore, mTOR may be a suitable target for drug or gene therapy, useful in the treatment of fibrotic diseases.