Functional Analysis of KSRP Interaction with the AU-Rich Element of Interleukin-8 and Identification of Inflammatory mRNA Targets^{▿ †}

Plasmids.

The expression vectors for β-globin (ptetBBB) carrying insertions of the IL-8 3′ UTR, for constitutively active MAP kinase kinase 6 (MKK6_2E) and MK2 (MK2_EE), and for negative interfering MK2 (MK2_K76R) have been described in detail previously (44, 46). ptetBBB-TNF_1315-1350 expresses an mRNA that contains the ARE of human TNF-α in its 3′ UTR. To express KSRP with an N-terminal Strep tag (stKSRP), a sequence encoding the Strep-tactin target peptide Strep tag III (29) was inserted into pcDNA3.1-HisC-KSRP (a kind gift from Ching-Yi Chen). The KSRP sequence in the resulting plasmid was replaced by the coding region of green fluorescent protein (GFP) to express Strep-tagged GFP (stGFP) or of human TTP to express Strep-tagged TTP (stTTP).

Transfection and determination of mRNA degradation.

HeLa cells constitutively expressing the tetracycline-controlled transactivator protein were transfected by the calcium phosphate method, and the degradation kinetics of plasmid-expressed mRNA was determined using the tet-off system as detailed earlier (44). Cells were stimulated with human recombinant IL-1α (a gift from A. Stern and P. T. Lomedico, Hoffman-La Roche) where indicated. The degradation kinetics for endogenous mRNA was determined by inhibiting transcription with actinomycin D (5 μg/ml). Total RNA was isolated and Northern blot analysis performed using digoxigenin-labeled antisense RNA probes. RNA half-lives were determined as described in reference 46, using a video imaging system and the Molecular Analyst program (Bio-Rad). siRNA duplexes (Qiagen) specifically interfering with expression of KSRP (18) or GFP (47) were transfected by the calcium phosphate method. Transfection was repeated 2 days later with the respective siRNA together with plasmids for reporter mRNA expression. Samples for determining KSRP protein levels and mRNA degradation were prepared the following day. Essentially similar increases in mRNA stability in KSRP knockdown cells were observed with a second siRNA duplex (modified from reference 3).

Analysis of poly(A) tail length.

To resolve heterogeneity in poly(A) tail length, RNA samples were separated on denaturating polyacrylamide gels (4% [wt/vol] acrylamide, 7 M urea) as described previously (44). Running in 1× Tris-borate-EDTA was performed at about 10 V/cm for variable times, depending on the size of the mRNA monitored. Deadenylated reference RNAs were prepared by RNase H digestion in the presence of oligo(dT) (44).

Strep tag affinity chromatography.

All steps were carried out in the cold. Cells were pelleted, resuspended at 10⁷ cells per 125 μl in lysis buffer (20 mM β-glycerophosphate [pH 7.4], 150 mM NaCl, 7.5% [vol/vol] glycerol, 1 mM EDTA, 1 mM sodium orthovanadate, 0.5% [vol/vol] Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml leupeptin, 200 units/ml RNase inhibitor [MBI Fermentas]) and kept on ice for 15 min. After centrifugation at 16,000 × g, the supernatant was collected (cytoplasmic extract). Aliquots were frozen for determination of stKSRP protein and input mRNAs. Strep-tactin agarose beads (IBA) preincubated for 30 min with tRNA from Escherichia coli (50 μg/ml) in 500 μl wash buffer (20 mM β-glycerophosphate [pH 7.4], 150 mM NaCl, 1 mM EDTA) were washed twice in the same buffer and added to the cytoplasmic extract. After 4 h with constant mixing, the beads were spun down and washed three times with wash buffer. Aliquots of the beads were heated to 95°C with sample buffer for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) for subsequent analysis of stKSRP amounts or were supplied with 10 μg of bacterial rRNA as carrier RNA and subjected to total RNA isolation (RNA isolation kit; Macherey & Nagel).

Reverse transcription-PCR (RT-PCR).

Total RNA was reverse transcribed with Moloney murine leukemia virus reverse transcriptase and amplified with Taq polymerase (both from MBI Fermentas) and specific primers (MWG Biotech) for CXCL3 (sense, 5′TCTCGCACAGCTTCCCGA; antisense, 5′CTAGAAAGCTGCTGTTCTC), 14-3-3ζ (sense, 5′AGGTCATCTTGGAGGGTC; antisense, 5′CTCCTTGGGTATCCGATG), IκBζ (sense, 5′CTCCTGTTGCACATCCGA; antisense, 5′CAGCTAACTTGAACTGTGTT), IL-6 (sense, 5′CTGGGCACAGAACTTATGTTG; antisense, 5′GGTAAGCCTACACTTTCCA), IL-8 (sense, ATCAAATATTTGTGCAAGAATTTGG; antisense, TTTCAGATAAACAATAATGT); KSRP (sense, 5′GAGAAGATAGCGAGATCTAA; antisense, 5′GAATGTTCCACCTCTAACTA); FUBP3 (sense, 5′AATCAAGCAGTTGCAGGAG; antisense, 5′CCGTGTATGTTATCTCCTG), and TTP (sense, 5′GATCTGACTGCCATCTACG; antisense, TCACTCAGAAACAGAGATGC).

Western blot analysis.

Proteins were separated by SDS-PAGE (10% acrylamide) and blotted onto polyvinylidene difluoride membranes (Millipore). Equal loading and protein transfer were controlled by staining with Coomassie brilliant blue. KSRP was detected by chemiluminescence using a specific monoclonal antibody (kindly provided by C.-Y. Chen) and alkaline-phosphatase-labeled secondary antibody.

Microarray analysis.

RNA from pulldown or siRNA-mediated knockdown experiments was purified and Cy3-labeled cRNA prepared as described previously (21). For analysis of KSRP-associated mRNAs on high-density arrays, RNA was isolated from Strep-tactin pulldown in four independent experiments which yielded essentially similar sets of mRNAs enriched by stKSRP compared to stGFP. Global mRNA expression of input samples was analyzed in parallel. Table S1 in the supplemental material shows results from one experiment where stringency was increased by raising the NaCl concentration to 700 mM in the first two washes of Strep-tactin agarose beads after incubation with cytoplasmic extract (see above). Cy3-labeled cRNA was generated using an Amino Allyl MessageAmp II with Cy3 kit according to the manufacturer's recommendations (AM1795; Applied Biosystems). Even amplification efficiency was controlled by spiking reverse transcription reactions with a mixture of 10 different E1A mRNA species (5188-5977; Agilent Technologies) that were each detected by 32 probes on the microarray. cRNA quality, yield, and labeling efficiency were analyzed by an Agilent 2100 bioanalyzer. To minimize technically caused variations, cRNA for each RNA sample was generated in duplicate and pooled prior to labeling and hybridizing. Labeled cRNAs were hybridized to the Whole Human Genome Microarray (G4112F, ID 014850; Agilent Technologies) containing 45,015 probes for 15,713 annotated genes and 15,269 uncharacterized transcripts. cRNA fragmentation, hybridization, and washing steps were performed exactly as directed by the manufacturer's One-Color Microarray-Based Gene Expression Analysis Protocol V5.0.1 manual. Slides were scanned on an Agilent G2505 B microarray scanner at high (100%) and low (5%) photomultiplier tube settings. Data extraction and intra-array normalization were performed with Feature Extraction V9.1.3.1 software (Agilent protocol file GE1-v5_91_0806.xml). Flagged spots were excluded from further analysis. Interarray normalization was performed according to E1A signals. Intensity values close to background levels were replaced by surrogate values, calculated from E1A signals, to compensate for elevated bias at the low end of the dynamic intensity range. Ratio data of probes directed against the same RNA were averaged by calculating the geometric mean.

In vitro transcription and electrophoretic mobility shift assays.

Radiolabeled RNA fragments corresponding to the IL-8 ARE and mutants thereof were synthesized with T7 RNA polymerase from linearized plasmids as described recently (44). The labeled RNA probes were purified using mini Quick Spin RNA columns (Roche Diagnostics) and incubated with cytoplasmic extract (prepared by Nonidet P-40 lysis as in reference 45) or with affinity-purified stKSRP eluted from the Strep-tactin agarose with desthiobiotin (10 mM; IBA). After digestion with RNase T1, samples were separated on nondenaturing polyacrylamide gels (44) or UV cross-linked with 900 mJ/cm² and separated by SDS-PAGE (45).

KSRP is essential for rapid degradation of IL-8 mRNA.

Rapid turnover of IL-8 mRNA is largely dependent on an ARE characterized in detail recently (44). To find out if the ARE-binding protein KSRP is involved in the function of the IL-8 ARE, HeLa cells were transfected with siRNAs specific for KSRP mRNA. Amounts of KSRP were strongly diminished compared to those in cells transfected with siRNA against GFP as a control (Fig. 1A). Degradation of β-globin reporter mRNAs carrying the IL-8 ARE as well as the TNF ARE was slowed down in the KSRP knockdown cells compared to that in the control cells (Fig. 1B). Endogenous IL-8 mRNA is barely detectable in unstimulated HeLa cells but strongly and rapidly induced by the proinflammatory cytokine IL-1. Two hours after addition of IL-1, the stability of endogenous IL-8 mRNA was strongly increased when KSRP expression was suppressed by siRNA (Fig. 1C).

Suppression of KSRP impairs deadenylation of IL-8 mRNA.

Evidence has been presented for recruitment of the deadenylating enzyme PARN by KSRP (18). To compare the poly(A) tail lengths of IL-8 mRNA in control and KSRP knockdown cells, the RNA samples were separated by PAGE. This revealed primarily short poly(A) tails in control cells after 2 h of stimulation with IL-1, as expected from previous work in which we showed that IL-1 induces rapid and transient IL-8 mRNA expression with long poly(A) tails at the 1-h maximum, with rapid poly(A) tail shortening and decrease in RNA amounts ensuing thereafter (20). In the cells treated with siRNA against KSRP, IL-8 transcripts with intermediate and long poly(A) tails were prominent (Fig. 2A). The weighted arithmetic mean length of poly(A) tails increased from 58 nucleotides (nt) in control cells to 120 nt in KSRP knockdown cells. Furthermore, in the latter cells amounts and poly(A) tail lengths remained fairly constant for 1 hour after inhibiting transcription. Similarly, steady-state levels and poly(A) tail lengths of CCL20 mRNA were also increased in the KSRP knockdown cells (Fig. 2B) (mean length 136 nt versus 96 nt in the control cells). The mRNA of IκBα is short lived but does not contain a typical ARE with overlapping AUUUA motifs in its 3′ UTR. Amounts and poly(A) tail lengths are found to be hardly affected when cells treated with control or KSRP-specific siRNA are compared (Fig. 2B) (mean lengths of 97 and 110 nt, respectively), indicating that KSRP knockdown selectively affects only certain mRNAs. A stable mRNA for a housekeeping enzyme, GAPDH (glyceraldehyde-3-phosphate dehydrogenase), also remains unaffected by suppressing KSRP levels (mean lengths of 107 and 104 nt, respectively, in control and KSRP knockdown). These results provide evidence for a role for KSRP in the rapid and specific deadenylation and degradation of endogenous ARE-containing mRNAs.

Both domains of the IL-8 ARE are involved in its interaction with KSRP.

In UV cross-linking experiments, radiolabeled RNA corresponding to the IL-8 ARE forms protein-RNA complexes with cytoplasmic extract (Fig. 3A). Formation of the highest-molecular-weight complex is increased with cytoplasmic extract from cells expressing stKSRP. Furthermore, IL-8 ARE RNA can be UV cross-linked to purified stKSRP, resulting in a product of the expected molecular weight in SDS-PAGE. Complex formation of stKSRP and the IL-8 ARE is also detected in nondenaturing gels (Fig. 3A, right).

Interaction of KSRP with the IL-8 ARE was confirmed in pulldown assays with stKSRP. First, stKSRP plasmid was transfected in different amounts and stKSRP and total KSRP were detected by Western blot analysis (transfected and endogenous KSRP could not be separated in SDS-PAGE). In the range of plasmid amounts tested, strong increases in total KSRP were not observed (Fig. 3B). Amounts that corresponded to 7 μg in the setting of Fig. 3B were chosen for further experiments. In Northern analysis of mRNAs associated with stKSRP after pulldown from cytoplasmic extract with Strep-tactin agarose beads, β-globin mRNA without an ARE was not detected (Fig. 3C). β-Globin mRNA with the ARE-containing region of IL-8 mRNA was selectively detected in pulldown from cells expressing stKSRP but not in pulldown from control cells. Importantly, endogenous IL-8 mRNA was also found to associate with stKSRP. GAPDH mRNA, which exhibits very high signal intensity in total RNA, was detected in spurious and not significantly differing amounts in both the specific and the control pulldown preparations.

We further studied the ARE-KSRP interaction by investigating the importance of the four AUUUA motifs and of the two functionally distinct domains in the ARE that we had previously defined (44). β-Globin reporter RNAs containing the ARE or derivatives thereof in the 3′ UTR were expressed (Fig. 4A). Compared to the wild-type ARE with an intact auxiliary domain and core domain (AD+CD), a mutant in which the third AUUUA motif of the core domain was changed into AUGUA (M3) showed slightly diminished interaction with KSRP in pulldown assays (Fig. 4B). Interaction was strongly reduced for a mutant in which all four AUUUA motifs were destroyed (M1234). Comparably strong impairment of KSRP interaction was observed for an mRNA that contained only the core domain. An RNA containing only the auxiliary domain showed the weakest interaction. The results revealed a close correlation between the interaction of the mRNAs with KSRP and their half-lives in intact cells (Fig. 4B) (see also reference 44). Importantly, the auxiliary domain strongly contributes to interaction of KSRP with the ARE, suggesting that this is the cause for its enhancing effect on the moderate destabilization exerted by the core domain alone.

The relationship between the destabilizing function of the IL-8 ARE and its interaction with KSRP was further probed in in vitro binding experiments. Complexes were analyzed in nondenaturing gels to avoid the problem of different cross-link efficiency rather than affinity when derivatives of the ARE were compared in their abilities to interact with KSRP. As shown in Fig. 4C, interaction was hardly affected if one of the four AUUUA motifs was mutated (M3). Strong decrease in interaction was observed for the mutant in which all motifs were destroyed (M1234). The core domain alone, which contains all four AUUUA motifs, showed markedly reduced interaction, and for the auxiliary domain alone, the interaction was reduced even stronger. Taken together, these data show a close correlation between the importance of the two domains for ARE function and interaction with KSRP.

Interaction of KSRP with the IL-8 ARE is impaired by p38 MAP kinase activation without involvement of MK2.

Rapid degradation of several ARE-containing mRNAs, including IL-8 mRNA, is impaired in cells exposed to IL-1 in a manner involving p38 MAP kinase and its substrate MK2 (46). In cells expressing stKSRP and β-globin mRNA containing the IL-8 ARE, significantly less of the mRNA was associated with stKSRP after treatment of the cells with IL-1 than in untreated cells (Fig. 5A). Inhibition of p38 MAP kinase using the specific inhibitor SB203580 largely reversed this effect of IL-1. Selective activation of the p38/MK2 pathway by expressing a constitutively active form of the p38-selective MAP kinase kinase MKK6 also impaired KSRP-RNA interaction (not shown). These results are in accordance with a decreased interaction of phosphorylated KSRP with ARE-containing RNA demonstrated in vitro (3). On the other hand, a kinase-inactive mutant of MK2 which interferes with IL-1-induced mRNA stabilization (46) could not reverse impairment of KSRP-mRNA interaction. Furthermore, expression of a constitutively active mutant of MK2 (MK2_EE), which induced stabilization (46) (Fig. 5C), did not impair interaction (Fig. 5B). This and the observation by Briata et al. that MK2 does not phosphorylate KSRP indicates that the contribution of MK2 to stabilization of ARE-containing mRNAs is not reflected on this level.

We also tested whether MK2 has an indirect role in mRNA stabilization, e.g., facilitating export of active p38 MAP kinase from the nucleus to the cytoplasm (2, 14), where p38 MAP kinase might phosphorylate its target KSRP. This is unlikely, however: as shown in Fig. 5C, constitutively active MK2 induces mRNA stabilization (compare “control” in upper panel [vector] with “control” in middle panel [MK2_EE], which is independent of p38 MAP kinase activity (middle, compare “control” and SB203580). Stabilization induced by the upstream kinase MKK6 was inhibited by SB203580 as expected.

The IL-8 ARE is a target of TTP.

MK2 can phosphorylate the ARE-binding protein TTP, which can impair its destabilizing activity toward its target mRNAs (16). To test whether TTP can contribute to regulation of IL-8 mRNA stability, stTTP was expressed in HeLa cells and pulldown performed after induction of IL-8 mRNA by IL-1. IL-8 mRNA was enriched in the stTTP pulldown fraction (Fig. 6A). Furthermore, purified TTP interacted with the IL-8 ARE in vitro (not shown). Since detection of a destabilizing effect of TTP might be obscured by the predominant destabilization of KSRP, the effect of TTP on the degradation of the IL-8 ARE-containing reporter RNA in KSRP knockdown cells was determined. Coexpression of stTTP resulted in a marked decrease in the steady-state level and the stability of the reporter RNA (Fig. 6B). Coexpression of constitutively active MK2 (MK2_EE) partly reverted this effect. These results indicate that TTP can participate in the control of IL-8 mRNA stability and may represent the MK2-sensitive component of rapid IL-8 mRNA degradation.

Microarray-based identification of mRNAs regulated by KSRP.

To obtain information on the population of mRNAs potentially regulated by KSRP, HeLa cells were stimulated with IL-1. A stimulation period of 2 hours was chosen because at this time IL-1-induced mRNA stabilization has ceased (20, 46) but expression of transiently induced mRNAs is still high enough for analysis. Two parameters were determined for RNAs from these cells, (i) enrichment in pulldown with KSRP and (ii) increase in amount upon knockdown of KSRP (Fig. 7A). RNA was analyzed on an oligonucleotide array covering 30,982 human genes. After pulldown of stKSRP, 1,734 mRNAs (i.e., 11% of all functionally annotated genes that were detected by the Agilent array) were enriched more than twofold in the stKSRP fraction (see Table S1 in the supplemental material). Determination of amounts in the cytoplasmic extract (input) excluded the possibility that this was due to different starting levels (not shown). Enrichment in the stKSRP fraction was verified by RT-PCR for several mRNAs (Fig. 7B).

mRNA from two independent KSRP knockdown experiments was analyzed to estimate the impact of KSRP on gene expression. The cells were stimulated with IL-1 for 2 hours, and RNA was isolated immediately or after additional incubation for 3 hours with actinomycin D to obtain information on changes in mRNA degradation. RNAs of 427 annotated genes showed a mean increase of twofold or higher in the KSRP knockdown cells compared to the level for GFP knockdown cells (see Table S2 in the supplemental material). For 1,674 of the 1,734 transcripts enriched in pulldown, complete data sets from both knockdown experiments were available. One hundred mRNAs were enriched in the stKSRP pulldown, and their levels also increased upon KSRP knockdown (Table 1). Forty-eight of them had short half-lives. Of these, 10 mRNAs, including that of IL-8, decayed more slowly in the KSRP knockdown cells (Table 1). This indicates that these mRNAs represent target transcripts for the destabilizing function of KSRP. Of note, 80% of them and 36% of all mRNAs in Table 1 were found in the ARED database (1), which lists mRNAs with AU-rich sequences in their 3′ UTRs as defined by a bioinformatic approach.

Possible autoregulation of KSRP.

Interestingly, among the transcripts associated with stKSRP was that of its homolog FUBP3 and also the KSRP mRNA itself (see Table S1 in the supplemental material). To exclude the possibility that the latter result merely reflects a higher input level (due to the additional mRNAs derived from the stKSRP vector), RT-PCR was performed with primers that amplify 3′ UTRs not contained in the vector (Fig. 7B). The result clearly demonstrates an association of stKSRP with endogenous KSRP mRNA. The plasmid-derived KSRP mRNA contains part of the 3′ UTR. Since that mRNA was also enriched, while a KSRP mRNA without 3′ UTR sequences was not (not shown), a binding site for KSRP protein was further localized by expressing the β-globin mRNA with 3′ UTR sequences from KSRP mRNA. Association with stKSRP in pulldown assays was found to be mediated by a non-ARE region of 104 nt 3′ of the stop codon in the KSRP mRNA (Fig. 7C).