Human Pat1b Connects Deadenylation with mRNA Decapping and Controls the Assembly of Processing Bodies^{▿ †}

Plasmids.

The following plasmids have been described previously: pCMV-FLAG-NOT1 (45), pcDNA3-Myc-Ccr4, pcDNA3-Flag-Dcp1a, pcDNA3-Flag-Dcp2 (25), and pCIneo-RL (31).

To generate pcDNA3-HA (p2003), oligonucleotides G85/G86 (for primer sequences, see Table S1 in the supplemental material) were annealed and cloned into the HindIII-EcoRI sites of pcDNA3 (Invitrogen). For pcDNA3-YFP (p2168), yellow fluorescent protein (YFP) was amplified with oligonucleotides G243/G244 and then ligated into the HindIII-KpnI sites of pcDNA3-HA (p2003), thereby replacing the hemagglutinin (HA) tag.

For plasmid pEYFP-Pat1a (p2639), the human Pat1a coding sequence was amplified in three portions from the cDNA of HeLa cells and ligated sequentially into the BglII-XbaI sites of pEYFP-C1 (Clontech/BD Biosciences). The following primer pairs were used: G1332/G1222 (1st PCR) and G1332/G1198 (2nd PCR) for the Pat1a C-terminal portion, ligated as a KpnI-XbaI fragment; G1315/G1221 for the middle portion, ligated as a BglII-KpnI fragment; and G1219/G1340 (1st PCR) and G1353/G1340 (2nd PCR) for the Pat1a N-terminal portion, ligated as a BamHI-BglII fragment and thereby introducing an EcoRI site upstream of the start codon. For pcDNA3-HA-Pat1a (p2638), the human Pat1a coding sequence was excised from pEYFP-Pat1a (p2639) as an EcoRI-XbaI fragment and cloned into the same sites of pcDNA3-HA (p2003).

A human Pat1b cDNA clone (DKFZp451I053) fused at the C terminus to YFP was kindly provided by Stefan Wiemann (German Cancer Research Center, Heidelberg, Germany). In order to generate HA-tagged Pat1b constructs, fragments were amplified by PCR from human Pat1b cDNA and inserted into the KpnI-XhoI sites of pcDNA3-HA (p2003). The following primers were used: G1225/G1230 for pcDNA3-HA-Pat1b (p2516), G1225/G1226 for pcDNA3-HA-AN (p2511), G1225/G1228 for pcDNA3-HA-ANH (p2514), G1227/G1228 for pcDNA3-HA-H (p2512), G1227/G1230 for pcDNA3-HA-HC (p2515), G1229/G1230 for pcDNA3-HA-C (p2513), G1225/G1425 for pcDNA3-HA-A (p2591), G1408/G1226 for pcDNA3-HA-N (p2592), and G1408/G1230 for pcDNA3-HA-dA (p2675). For pcDNA3-HA-dH (p2542), the N- and C-terminal regions of Pat1b were first amplified as separate fragments with primers G1225/G1295 and G1296/G1230, respectively. The PCR products were then aligned, Pat1b lacking the H domain was amplified with primers G1225/G1230, and the PCR product was cloned into the KpnI-XhoI sites of pcDNA3-HA (p2003).

pcDNA3-YFP-Pat1b full length (p2521), pcDNA3-YFP-A (p2630), pcDNA3-YFP-AN (p2626), pcDNA3-YFP-N (p2627), pcDNA3-YFP-ANH (p2522), pcDNA3-YFP-H (p2632), pcDNA3-YFP-C (p2628), pcDNA3-YFP-HC (p2629), pcDNA3-YFP-dA (p2631), and pcDNA3-YFP-dH (p2633) were generated by replacing the HA tag in the corresponding HA-tagged Pat1b constructs with YFP excised as a HindIII-KpnI fragment from pcDNA3-YFP (p2168). For pcDNA3-HA-Pat1b-YFP (p2520), Pat1b-YFP was amplified by PCR with primers G1225/G1253, using clone DKFZp451I053 as a template, and inserted into the KpnI-XhoI sites of pcDNA3-HA-Pat1b (p2516).

In order to generate pcDNA3-HA-PP7cp-Pat1b (p2634), pcDNA3-HA-PP7cp-AN (p2658), pcDNA3-HA-PP7cp-HC (p2659), pcDNA3-HA-PP7cp-dA (p2663), pcDNA3-HA-PP7cp-dH (p2660), and pcDNA3-HA-PP7cp-Pat1a (2661), PP7cp (46) was amplified by PCR using primers G1501/G1502 and inserted into the KpnI site of corresponding pcDNA3-HA-Pat1b constructs and pcDNA3-HA-Pat1a (p2638). Orientation of the insert was checked by sequencing.

For pcDNA3-Bgl (p2001), the BglII site in pcDNA3 (Invitrogen) was removed by digestion with BglII and blunt-end religation, thereby introducing a novel ClaI site. pcDNA3-7B (p2308) expressing a T7-tagged rabbit β-globin gene was subsequently generated by amplifying β-globin by PCR with primers G18/G19 from plasmid puroMXbglobin (39). The fragment was digested with EcoRV and SalI and ligated into the BamHI/blunt-XhoI sites of pcDNA3-Blg. For pcDNA3-7B-PP7bs (p2314), six repeats of the PP7 binding site were excised as a BamHI-BglII fragment from pSP73-6xPP7bs (46) and introduced into the BglII site of pcDNA3-7B (p2308). For pFLB (p2524), firefly luciferase (FL) cDNA was amplified by PCR using primers G1256/G1257 from pGL3-Control (Promega) and cloned into the HindIII-KpnI sites of pcDNA3-7B (2308). From there, the 6× PP7bs repeats were cloned as an EcoRI-BglII fragment from pcDNA3-7B-PP7bs (p2314) into the corresponding sites of pFLB (p2524) to generate pFLB-PP7bs (p2646).

For TOpuro-Caf1a-mycSG (p2485), the human Caf1a cDNA was amplified by reverse transcription (RT)-PCR using primers G1122 and G1123 and cloned into the BamHI-XhoI sites of TOpuro-mycSG (p2484). For pTOpuro-Caf1a-AA-mycSG (p2737) containing the D40A/E42A mutation, a fragment amplified by PCR with primers G1730 and G1731 was inserted into the KpnI-EcoRI sites of TOpuro-Caf1a-mycSG (p2485). For pEGFP-TEV-Caf1a (p2793), Caf1a cDNA was cloned as a BamHI-XhoI fragment into the BglII-XhoI sites of pEGFP-N1 containing a tobacco etch virus (TEV) cleavage site. To generate pcDNA3-Flag-Dcp2-AA (p2798), the E147A/E148A mutation was introduced into the KpnI-EcoRV sites of pcDNA3-Flag-Dcp2 (25) using a PCR fragment amplified with primers G1883 and G1884.

Cell culture and transfection.

HeLa, U2OS, HEK293, and COS7 cells were cultured in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (PAA Laboratories), 2 mM l-glutamine, 100 U/ml penicillin, and 0.1 mg/ml streptomycin (all PAN Biotech) at 37°C/5% CO₂. Plasmids were transfected with Lipofectamine 2000 (Invitrogen) for protein interaction studies or with Fugene HD (Roche) for immunofluorescence analysis according to the manufacturer's recommendations. Alternatively, polyethyleneimine (Polysciences Europe; 1 mg/ml, pH 7.0) was also used as a transfection reagent for interaction and functional studies. siRNAs were transfected at a concentration of 100 nM with Lipofectamine 2000 twice over a time period of 4 days. Where indicated, cells were treated for 2 h with 5 μg/ml cycloheximide (Roth).

siRNAs.

siRNAs used in this study were synthesized by Ambion against the following sequences (sense strand): D0 (control), 5′-GCAUUCACUUGGAUAGUAAdTdT-3′; T2 (human Pat1b), 5′-GCAGAAAGGCUCAGUAAGAdTdT-3′; and T3 (human Pat1b), 5′-GGACGGAGGUGAUGUUCAUdTdT-3′.

Immunofluorescence microscopy.

Twenty-four hours after transient transfection, cells grown on glass coverslips were fixed in 4% paraformaldehyde for 10 min at room temperature. Cell membranes were then permeabilized in −20°C cold methanol for 10 min. Phosphate-buffered saline (PBS) containing 0.1% sodium azide and 5% horse serum was used for blocking and antibody dilution. DNA was visualized using Hoechst dye (no. 33342, 1 μg/ml; Sigma). After washing in PBS, cells were mounted onto glass slides using a solution of 14% polyvinyl alcohol (P8136; Sigma) and 30% glycerol in PBS. P bodies were counted on an upright epifluorescence microscope (BX60; Olympus). Images were acquired at the Nikon Imaging Center, Heidelberg, Germany, either on an upright epifluorescence microscope (90i; Nikon) or by spinning-disc confocal microscopy (Ultraview ERS [Perkin Elmer] on a TE2000 inverted microscope [Nikon]) using an electron-multiplying charge-coupled-device (EM-CCD) camera (Hamamastu).

Coimmunoprecipitation (co-IP) and Western blot analysis.

Twenty-four hours after transient transfection, HEK293 cells from a confluent 10-cm dish were collected and lysed in 400 μl ice-cold hypotonic lysis buffer (10 mM Tris [pH 7.5], 10 mM NaCl, 10 mM M EDTA, 0.5% Triton X-100 with freshly added protease inhibitors [Complete; Roche]). Nuclei were removed by centrifugation at 500 × g for 5 min at 4°C. The cytoplasmic lysate was precleared by the addition of 30 μl protein A/G-agarose beads (Pierce or Santa Cruz) for 1 h at 4°C and incubated with 1 μg of antibody for 2 h. A total of 30 μl of protein A/G beads was added for an additional 2 h and washed six times in NET2 buffer (50 mM Tris [pH 7.5], 150 mM NaCl, 0.5% Triton X-100). Protein complexes were eluted with 25 μl SDS sample buffer with or without 100 mM dithiothreitol (DTT). Where indicated, RNase A (Sigma; 0.1 mg/ml) was added to the lysates after preclearing. Proteins were resolved on 10% polyacrylamide gels and transferred onto a 0.2-μm-pore-size nitrocellulose membrane (Peqlab) for Western blotting. Horseradish peroxidase-coupled secondary antibodies (Jackson Immunoresearch) in combination with Western Lightning enhanced chemiluminescence substrate (Perkin Elmer) were used for detection. The green fluorescent protein (GFP)-binder was used for immunoprecipitation (IP) of YFP-tagged proteins as described previously (33).

Antibodies.

Polyclonal antibody Ab387 against human Pat1b was produced at Eurogentec by immunizing rabbits with peptide QGPEDDRDLSERALPR. The following antibodies were used for IP and Western blotting, and immunofluorescence analysis: mouse monoclonal antibodies anti-HA (HA.11; Covance) and anti-Edc3 (Abcam; ab57780-100); a cross-reacting mouse monoclonal phospho-S6-kinase antibody (sc-8416; Santa Cruz) for detection of Hedls; polyclonal chicken antibodies against Lsm1 (15-288-22100F; Genway Biotech) and Lsm4 (19101; Abcam); polyclonal rabbit antibodies against Xrn1 (A300-461A; Bethyl Laboratories), p54/Rck (A300-443A; Bethyl Laboratories), eIF4G (sc-11373; Santa Cruz), 14-3-3 (sc-629; Santa Cruz), and HA (sc-805; Santa Cruz); and a polyclonal goat antibody against eIF3B (sc-16377; Santa Cruz). Polyclonal rabbit anti-Caf1a was kindly provided by Ann-Bin Shyu (University of Texas, Houston, TX).

Northern blot analysis.

For RNA decay experiments, actinomycin D (Applichem) was added to cell cultures at 5 μg/ml, and total RNA was extracted using the Genematrix RNA purification kit (Eurx, Roboklon). A total of 7 to 15 μg of RNA was resolved by 1.1 or 1.6% agarose-2% formaldehyde-MOPS (morpholinepropanesulfonic acid) gel electrophoresis and blotted overnight with 8× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) onto Hybond-N+ Nylon membranes (Amersham, GE). Membranes were hybridized overnight with digoxigenin-labeled RNA probes at 55°C and washed twice with 2× SSC-0.1% SDS for 5 min and twice with 0.5× SSC-0.1% SDS for 20 min at 65°C. Alkaline phosphatase-labeled anti-DIG Fab fragments and CDP-Star substrate (both Roche) were used for detection according to the manufacturer's instructions. The following primers were used to generate templates for Sp6 probes by PCR: G1000/G1001 for the probe against exon 1 and 2 of rabbit β-globin, G78/G1008 for the probe against human RPS7, and G083/G1009 for the probe against nucleolin. Primer sequences are listed in Table S1 in the supplemental material.

Luciferase assay.

One third of a 6-cm dish of transiently transfected HeLa cells was lysed in 150 μl of passive lysis buffer (dual-luciferase reporter assay system; Promega) and frozen at −20°C. After thawing, nuclei were removed by centrifugation for 2 min at 13,000 rpm at 4°C. A total of 20 μl of the supernatant was mixed with 50 μl of substrates from the dual-luciferase reporter assay system, diluted 1:3. Firefly and Renilla luciferase activities were measured on a Fluostar Optima (BMG Labtech) plate reader.

Identification of two human Pat1 homologs.

In order to identify a human homolog of yeast Pat1, we compared Pat1 sequences of different fungi and observed a highly conserved central region of about 40 amino acids (aa), hereinafter termed the homology (H) domain. Using this domain, BLAST analysis was able to identify homologous sequences in various metazoa, including Caenorhabditis elegans, Drosophila melanogaster, Danio rerio, and Xenopus laevis (see Fig. S1 in the supplemental material). Two proteins sharing the H domain were found in mice and humans, indicative of a gene duplication in mammals. Outside the H domain, the only detectable sequence conservation between the fungal and metazoan proteins lies in a highly acidic stretch of 50 to 70 aa at the very amino terminus, hereinafter referred to as the acidic domain (A). We termed the two human proteins Pat1a and Pat1b. Pat1b is identical to PatL1 previously identified by Scheller et al. (34). Pat1a matches the partial sequence that was previously suggested as the human Pat1 ortholog (9) and differs from the predicted PatL2 sequence at the N terminus (34). An alignment between the human Pat1a and Pat1b sequence (see Fig. S2 in the supplemental material) shows that Pat1a lacks a region of about 150 aa in the amino-terminal part following the A domain.

Human Pat1b, but not Pat1a, localizes to P bodies.

We first examined the localization of human Pat1a and Pat1b by expressing yellow fluorescent protein (YFP)-tagged fusion proteins in COS7 cells. Epifluorescence microscopy showed that YFP-Pat1a is expressed throughout the cytoplasm, where it has a reticular distribution, but does not accumulate in P bodies (Fig. 1B). YFP-Pat1b, however, strongly accumulates in small cytoplasmic foci (Fig. 1C to E). These foci correspond to P bodies as determined by colocalization with different markers, including the enhancer of decapping Hedls (Fig. 1C), the helicase Rck (Fig. 1D), and Lsm1 (Fig. 1E). The localization of Pat1b in P bodies was also observed with a C-terminal fusion of YFP and with N-terminal HA and Flag tags (data not shown). We confirmed the localization of Pat1b in P bodies of human HeLa (data not shown) and U2OS cells (see Fig. S3 in the supplemental material). Thus, human Pat1b, but not Pat1a, localizes in P bodies. This is in line with the previously observed localization of Pat1b (PatL1) in HeLa cells (34).

In order to determine whether endogenous Pat1b would also localize in P bodies, we raised a peptide antibody in rabbit (Ab387). This antibody was able to detect overexpressed Pat1b by Western blot analysis (see Fig. S4A in the supplemental material). By immunofluorescence microscopy, Ab387 gave a punctate staining in both the nucleus and the cytoplasm of untransfected COS7 (Fig. 1F and G) and U2OS cells (see Fig. S3F and G in the supplemental material). The larger foci in the cytoplasm perfectly overlap with Hedls and Lsm1, indicating that endogenous Pat1b also localizes to P bodies.

Human Pat1b induces P-body formation.

We noticed that cells transfected with YFP-Pat1b showed strongly increased numbers of P bodies compared to control cells transfected with YFP or YFP-Pat1a (compare Hedls staining in Fig. 1A to C). By counting P bodies using Hedls as a marker, we found that most COS7 cells overexpressing YFP-Pat1b have >20 P bodies per cell compared to an average of about 9 P bodies per cell in untransfected or YFP-expressing cells (Fig. 1H). Overexpression of YFP-Pat1a, however, did not increase the number of P bodies. The strong increase in P-body numbers by Pat1b overexpression was confirmed with an HA-tagged version, and the same was also observed with other cell lines, such as HeLa and U2OS (data not shown). Since Pat1b overexpression causes a dramatic amplification of P bodies, many of which are very small, Pat1b may play an active role in nucleating the assembly of P bodies.

To further address the role of Pat1b in P-body assembly, we knocked down endogenous Pat1b in U2OS cells. Two different siRNAs, T2 and T3, reduced Pat1b mRNA levels to 5% and 6%, respectively, compared to the expression level in cells transfected with control siRNA (see Fig. S5A in the supplemental material). Indeed, knockdown of Pat1b led to a nearly complete loss of P bodies (Fig. 1I; see also micrographs shown in Fig. S5B in the supplemental material). This result supports the notion that Pat1b nucleates the assembly of P bodies.

Pat1b domains control P-body assembly.

Next, we wanted to examine the contribution of different Pat1b domains, schematically depicted in Fig. 2A, to its intracellular localization. Fragments and deletions of Pat1b were expressed as YFP fusion proteins in COS7 cells and examined by spinning-disc confocal microscopy. Localization data are summarized in Table S2 in the supplemental material, and examples are shown in Fig. 2. As stated above, YFP-Pat1b forms many small- or medium-size foci, reminiscent of a “salt”-like appearance, and these foci colocalize with all P-body markers tested, including Hedls, Rck, Lsm1, and Lsm4 (Fig. 1C to E and 2B; see also Fig. S6E in the supplemental material). Thus, overexpressed Pat1b localizes to and strongly amplifies the number of P bodies. Deletion of the homology domain (YFP-dH) did not change this localization pattern (Fig. 2C). Deletion of the acidic region (YFP-dA), however, caused the protein to localize to much smaller and more numerous foci, reminiscent of a “pepper”-like appearance. These foci colocalize with Hedls (Fig. 2D) and with Rck (data not shown). Thus, YFP-dA triggers the formation of small P bodies, yet deletion of the acidic domain seems to prevent small aggregates from growing or merging into larger P bodies.

Localization of the YFP-AN fragment was very heterogeneous: in about half of the cells, YFP-AN showed a diffuse localization in the cytoplasm, and in these cells endogenous P bodies were strongly suppressed (see Table S2 in the supplemental material). In other cells, YFP-AN formed aggregates apparent as large round foci (Fig. 2E) or irregular patches (not shown). Many of these round foci and patches did not contain other P-body markers, suggesting that the AN fragment had the intrinsic ability to form aggregates. This aggregation-prone behavior was also observed with the YFP-N fragment lacking the A region (Fig. 2F). In contrast, the YFP-A fragment had a diffuse localization indistinguishable from YFP alone (see Table S2). The YFP-HC fragment primarily showed a diffuse localization in the nucleus and the cytoplasm. In some cells, YFP-HC also accumulated weakly in P bodies (Fig. 2G) yet did not affect the size or number of P bodies. Taken together, these results suggested that the HC fragment is passively recruited to P bodies, whereas the N fragment contains an active aggregation-prone domain. In the context of full-length Pat1b, the N region can strongly promote the formation of P bodies. The idea that Pat1b actively promotes P-body assembly is supported by our observation that Pat1b interacts with itself in a coimmunoprecipitation (co-IP) assay (Fig. 2H).

We then tested whether P bodies induced by YFP-Pat1b would be resistant to treatment with the translation inhibitor cycloheximide (CHX). CHX was shown to inhibit the formation of P bodies (10), presumably because mRNAs are trapped in polysomes by CHX. After 2 h of CHX treatment, P bodies were nearly absent in untransfected cells, whereas large numbers of foci were still present in YFP-Pat1b-expressing cells (see Fig. S6A, D, and F in the supplemental material). Colocalization of Hedls, Rck, Xrn1, and Lsm4 indicated that these structures had the typical composition of P bodies. Interestingly, Pat1b-induced foci persisted even after 6 h of CHX treatment (data not shown). Although we cannot tell whether these P bodies are functional in the presence of CHX, this results indicates that Pat1b not only nucleates P bodies but also stabilizes P bodies once assembled. Aggregates induced by the expression of YFP-AN were also resistant to CHX treatment (see Fig. S6B in the supplemental material). These aggregates showed distinct colocalization with Rck but not with Hedls, suggesting that Rck may interact with the AN fragment.

Tethering of Pat1b accelerates mRNA degradation.

Since P bodies have been implicated in the control of both mRNA translation and decay, we went on to test if human Pat1b plays a role in the posttranscriptional control of gene expression. To this end, we tethered Pat1b to a reporter mRNA using a heterologous RNA-protein interaction of the Pseudomonas aeruginosa bacteriophage PP7 (23, 46). The PP7 coat protein (cp) binds as a dimer with high affinity to a stem loop in the bacteriophage RNA, the PP7 binding site (bs). We cloned six copies of the PP7bs into the 3′ UTR of a rabbit β-globin reporter gene containing firefly luciferase (FL) (FLB-PP7bs). A fusion protein containing an HA tag, PP7cp, and Pat1b, as shown schematically in Fig. 3A, brings Pat1b in close proximity to the reporter mRNA. When expressed together in HeLa cells, HA-PP7cp-Pat1b strongly suppressed the expression level of FLB-PP7bs mRNA compared to HA-PP7cp or HA-Pat1b alone (Fig. 3B, lanes 4 to 6). In contrast, HA-PP7cp-Pat1b had only a very weak effect on the FLB mRNA lacking PP7bs (Fig. 3B, lanes 1 to 3). Similar to HeLa cells used for Fig. 3, Pat1b was also found to suppress the expression of the tethered reporter mRNA in U2OS cells (data not shown).

We then compared the effect of tethering Pat1a, Pat1b, and fragments of Pat1b to the FLB-PP7bs reporter mRNA. We measured, as a readout, both firefly luciferase activity and reporter mRNA levels (Fig. 3C). The experiment showed that the tethering of Pat1a had no significant effect compared to PP7cp alone. The tethering of full-length Pat1b had the strongest suppressive effect, reducing FL activity by about 3-fold and mRNA levels by about 5-fold. The tethering of the AN fragment, the dH deletion mutant, or the dA deletion mutant resulted in strong inhibition of the FLB-PP7bs reporter, though at the mRNA level the inhibition was less pronounced than with full-length Pat1b. Since the tethering of the HC fragment had a much weaker effect, the AN fragment of Pat1b appears to be most important for the suppressive effect on reporter mRNA expression. In general, there was good agreement between the relative FL activities and mRNA abundance, indicating that tethered Pat1b acts primarily at the level of mRNA expression.

Next, we wanted to determine whether Pat1b would affect the stability of the tethered mRNA. We expressed a β-globin reporter mRNA containing 6 copies of PP7bs (7B-PP7bs) together with the HA tag alone, HA-PP7cp, HA-PP7cp-Pat1b, or HA-Pat1b in HeLa cells (Fig. 3D). Transcription was blocked by the addition of actinomycin D in order to monitor degradation of the reporter mRNA. The analysis showed that 7B-PP7bs mRNA was stable in the presence of coexpressed HA or HA-Pat1b or when HA-PP7cp alone was tethered to the mRNA. In contrast, the tethering of HA-PP7cp-Pat1b caused rapid degradation of the reporter mRNA. Overall signal intensities were quantified to obtain mRNA half-lives (Fig. 3D, bottom). Measurement of the signal intensity along the length of the reporter mRNA shows that the mRNA was deadenylated as a result of Pat1b tethering (Fig. 3D, middle).

Human Pat1b interacts with Rck, Hedls, Xrn1, and Lsm1.

To further explore the connection between human Pat1b and mRNA degradation, we tested whether Pat1b interacts with different mRNA decay factors that localize to P bodies. By expressing HA-tagged Pat1b in HEK293 cells, we found that Pat1b coimmunoprecipitates with the 5′-3′ exoribonuclease Xrn1 and with the activators of decapping Rck, Hedls, Lsm1, and Lsm4 (Fig. 4A, lane 5) as well as Edc3 (Fig. 4D). These interactions were also observed with a YFP-tagged form of Pat1b (HA-Pat1b-YFP) (Fig. 4B). eIF3B and eIF4G, translation initiation factors that do not localize in P bodies, did not interact with Pat1b (Fig. 4A and B). As a negative control, the HA tag alone did not co-IP with any of these proteins. Given that mRNAs are also a component of P bodies, we next tested whether the interaction of Pat1b with its partners is RNA dependent. The addition of RNase A to the IP did not affect the interaction of HA-Pat1b with any of its partners (Fig. 4B). RNA extracted from the unbound fraction showed that RNase A treatment had efficiently degraded rRNA (Fig. 4B, lanes 10 and 11). These results suggest that the association of Pat1b with Rck, Hedls, Xrn1, and Lsm1 is based on protein-protein interactions.

Since Pat1b contains an aggregation-prone domain (fragment N) (Fig. 2F), we tested whether Pat1b may aggregate and precipitate unspecifically in our co-IP experiments. From the same HA-Pat1b-containing lysate (input), we carried out the IP procedure with and without the HA antibody (Fig. 4C). In the absence of HA antibody, neither HA-Pat1b nor associated proteins were precipitated (Fig. 4C, lane 2). This shows that precipitation by the antibody was specific.

To further characterize the binding of Pat1b to its partners, we subjected the complexes to increasing salt concentrations (Fig. 4D). Whereas the interaction with Hedls, Edc3, and Xrn1 was severely reduced at a NaCl concentration of 300 mM, the interaction with Rck and Lsm4 was resistant to 300 mM NaCl. Lsm1-Pat1b was the most stable of these interactions and could not be separated by 500 mM NaCl. This indicates that Lsm1/Lsm4 and Rck are more tightly associated with Pat1b than Hedls, Edc3, and Xrn1. Since Xrn1 forms a stable complex with Lsm1-7 in yeast (6), the interaction of human Xrn1 with Pat1b may occur through Lsm1-7. From these co-IP experiments, however, it is not possible to tell whether Pat1b directly binds to Rck and Lsm1.

Interestingly, Lsm1 co-IPs in similar amounts with Pat1a and Pat1b, yet the association of Lsm4 and Xrn1 with Pat1a is much weaker (Fig. 4A, lane 6). Thus, it is conceivable that Pat1a may preferentially bind to “free” Lsm1 that is not associated with the other Lsm proteins or Xrn1.

To obtain additional evidence for the observed interactions, inverse IPs were carried out in HEK293 cells transiently transfected with HA-Pat1b (Fig. 4E). The Xrn1 antibody was able to co-IP HA-Pat1b and Lsm1 (Fig. 4E, lane 2), the Hedls antibody could co-IP HA-Pat1b, Xrn1, and Lsm1 (lane 3), and the Rck antibody could co-IP HA-Pat1b and Lsm1 (lane 4). The Lsm1 antibody was not suitable for IP (Fig. 4E, lane 5), and a 14-3-3 antibody served as a negative control (lane 6). Taken together, these results confirmed the association of Xrn1, Hedls, and Rck with both Pat1b and Lsm1 and provides evidence for multiple interactions within this group of P-body proteins.

Human Pat1b interacts with enhancers of decapping via separate domains.

Since the above analysis revealed that Rck and Lsm1 are tightly associated with Pat1b, we determined which domains of Pat1b are involved in these interactions. To this end, we expressed different fragments of Pat1b as depicted in Fig. 5A. Figure 5B shows that the AN fragment efficiently co-IPs with Rck (lane 9), whereas the carboxy-terminal (HC) fragment co-IPs Lsm1 (lane 11). Similar to Lsm1, the more loosely associated proteins Hedls, Edc3, and Xrn1 were found to be connected to Pat1b via the HC domain (see Fig. S7 in the supplemental material). As shown in Fig. 4B for full-length Pat1b, the interactions mediated by the AN and HC fragments were resistant to RNase A treatment and are thus likely to be RNA independent (Fig. 5B).

We further analyzed the association of Rck with the amino-terminal half of Pat1b (Fig. 5C). Removal of the acidic domain (dA) from full-length Pat1b was sufficient to abrogate the interaction with Rck (Fig. 5C, lane 9). The same was observed when the N fragment was expressed alone in comparison to the AN fragment (Fig. 5C, lanes 10 and 11). Indeed, the HA-tagged acidic domain alone, although not visible by Western blot analysis due to its small size, efficiently co-IPs with Rck (Fig. 5C, lane 12).

We then examined in more detail the association of HC with Lsm1. Deletion of the conserved homology domain H from full-length Pat1b (dH) abolished the interaction with Lsm1 (Fig. 5D, lanes 8 and 9). Similarly, comparison of the HC and C fragments indicated that the H domain is required for association with Lsm1 (Fig. 5D, lanes 10 and 11). The ANH fragment, however, did not co-IP with Lsm1 (Fig. 5D, lane 12), indicating that the H domain alone is not sufficient to mediate this interaction. We concluded that Lsm1 binding requires both the H and C fragment of Pat1b.

Human Pat1b interacts with the Dcp2-Dcp1a decapping enzyme.

Since we found Pat1b to associate with a number of decapping enhancers (Rck, Hedls, Edc3, and Lsm1), we wanted to know if Pat1b would also interact with the decapping enzyme Dcp2 and its closely associated activator Dcp1a. To this end, we switched to YFP-tagged Pat1b proteins, since the use of GFP-binder (33) overcomes detection problems related to the presence of immunoglobulin in the IP. Full-length Pat1b tagged with YFP was found to co-IP efficiently with Flag-Dcp2 and Flag-Dcp1a (Fig. 6A and B, lanes 8). Interestingly, the two proteins associate with both the N and the HC region of Pat1b (Fig. 6A and B, lanes 10 and 12). The association with Dcp2 and Dcp1a was found to be independent of RNA (see Fig. S8 in the supplemental material). Moreover, the two proteins did not dissociate from Pat1b at 1 M NaCl (Fig. 6C), indicating that Dcp2 and Dcp1a are tightly connected to Pat1b. Taken together, the results show that Pat1b associates with all enhancers of decapping through at least three separate domains: the A fragment associates with Rck, the N fragment with Dcp2 and Dcp1a, and the HC fragment with Dcp2/Dcp1a, Lsm1/Lsm4, Hedls, and Edc3.

Human Pat1b interacts with the Ccr4-Caf1-Not deadenylation complex.

Since we found that the tethering of Pat1b causes mRNA deadenylation (Fig. 3D), we tested for interactions of Pat1b with the Ccr4-Caf1-Not deadenylation complex. Notably, this complex is also concentrated in P bodies (27, 41, 47). Indeed, we found that YFP-Pat1b co-IPs efficiently with Flag-Not1 (Fig. 7A) and myc-Ccr4 (Fig. 7B), as well as HA-Caf1a and HA-Caf1b (Fig. 7C). In all cases, RNase A treatment did not interfere with binding. We then confirmed the interaction of YFP-Pat1b with endogenous Caf1a (Fig. 7D, lane 8). Analysis of subdomains indicated that Caf1a can interact with the N fragment of Pat1b alone (Fig. 7D, lane 10), yet the acidic domain A enhances Caf1a binding (Fig. 7D, lane 11). Interestingly, the A domain was even more important for association with Flag-Not1 and myc-Ccr4 since only the AN fragment showed a robust interaction with these two proteins (see Fig. S9A and B in the supplemental material). Although Flag-Not1 and myc-Ccr4 also co-IP weakly with the HC domain (Fig. S9A and B), we concluded from these experiments that the Ccr4-Caf1-Not complex associates primarily with the AN domain.

By challenging these interactions with increasing salt concentrations, we found that Flag-Not1, myc-Ccr4, and HA-Caf1a/b are very stably associated with YFP-Pat1b (Fig. 7E). At a concentration of 1 M NaCl, these proteins remained in a complex with YFP-Pat1b, whereas binding of Lsm1 was reduced and binding of Rck was lost. Thus, it appears that Pat1b associates tightly with the Ccr4-Caf1-Not deadenylase complex and may act as a physical bridge between deadenylases and decapping enzymes.

To provide further evidence for such a bridging function, we tested whether Pat1b would promote the association between Caf1a and Dcp2. By IP of GFP-Caf1a, we found that the co-IP of Flag-Dcp2 is strongly enhanced by coexpression of HA-Pat1b (Fig. 7F, lane 3). This result shows that Pat1b can simultaneously interact with Caf1a and Dcp2 and thereby promote the association between the deadenylation-and-decapping complex.

Tethering of Pat1b causes both deadenylation and mRNA decapping.

In order to test whether both deadenylation and decapping activities were associated with Pat1b, we again turned to the tethering assay. A dominant negative Caf1a-AA (D40A/E42A) deadenylase similar to a mutant described by Zheng et al. (47) was generated. In comparison to the vector control, coexpression of Caf1a-wt slightly accelerated the degradation rate of the reporter mRNA tethered to Pat1b (Fig. 8A and B). Quantification of the mRNA decay rates is shown at the bottom of Fig. 8 and summarized in Table S3 in the supplemental material. In contrast to that of Caf1a-wt, coexpression of dominant negative Caf1a-AA prevented deadenylation of the mRNA (Fig. 8C and G). The lack of the A⁻ mRNA population is clearly visible in the middle panels of Fig. 8C and G, in which the signal intensity was quantified along the length of the mRNA. This result suggests that the tethering of Pat1b causes Caf1-dependent mRNA deadenylation. Interestingly, the A⁺ mRNA still underwent degradation without shortening of its size, suggesting that tethered Pat1b could promote decapping of the A⁺ mRNA independently of deadenylation.

To test this possibility, we generated Dcp2-AA (E147A/E148A) in analogy to a published mutation that inactivates the decapping enzyme Dcp2 (44). Whereas coexpression of Dcp2-wt did not have a significant effect on mRNA degradation (Fig. 8D), Dcp2-AA caused accumulation of the A⁻ mRNA (Fig. 8E). Strong accumulation of A⁻ mRNA was also visible when we cotransfected Caf1a-wt and Dcp2-AA (Fig. 8F). From this we concluded that the tethering of Pat1b also leads to Dcp2-dependent mRNA decapping. Only by coexpressing both Caf1a-AA and Dcp2-AA did we achieve full stabilization of the Pat1b-tethered mRNA (Fig. 8H). Taken together, these experiments demonstrate that Pat1b is tightly associated with both deadenylation and decapping activities in the cell.

N is the primary effector domain of Pat1b.

To determine which domain of Pat1b mediates deadenylation and decapping, we tested the effect of tethering Pat1b fragments to the reporter mRNA. The tethering of AN (Fig. 9C) or the N domain alone (Fig. 9F) very efficiently caused mRNA deadenylation and decapping, comparable to the effect of tethering full-length Pat1b (Fig. 9B). The mRNA decay rates are quantified in Fig. 9, bottom, and summarized in Table S4 in the supplemental material. In contrast to that of AN or N, the tethering of the HC fragment (Fig. 9D) or the A domain (Fig. 9E) did not induce mRNA decay and was similar to the tethering of PP7cp alone (Fig. 9A). These data suggest that N is the most potent fragment of Pat1b, and its tethering causes both deadenylation and decapping. This is in good agreement with our observation that the N fragment interacts with both Caf1a (Fig. 7D) and Dcp2 (Fig. 6A). We finally tested the effect of removing the A and the H domain (Fig. 9G and H). Compared to full-length Pat1b, the tethering of either the dA or dH mutant led to a significantly reduced mRNA decay rate. Both mutants caused mRNA deadenylation but resulted in the accumulation of A⁻ mRNA. This suggests that the A domain, possibly by interacting with Rck (Fig. 5C), and the H domain, possibly by enhancing the binding of Lsm1 (Fig. 5D), activate decapping of the tethered mRNA. These enhancing activities require the presence of the N domain since the A and HC fragments on their own were inactive in the tethering assay.