Deadenylation-Dependent and -Independent Decay Pathways for α1-Tubulin mRNA in Chlamydomonas reinhardtii

Abstract

The α- and β-tubulin mRNAs of Chlamydomonas reinhardtii exhibit different half-lives under different conditions: when expressed constitutively, they degrade with half-lives of about 1 h, whereas when induced by deflagellation, they degrade with half-lives of only 10 to 15 min. To investigate the decay pathway(s) used under these two conditions, an α1-tubulin gene construct which included an insert of 30 guanidylate residues within the 3′ untranslated region was introduced into cells. This transgene was efficiently expressed in stably transformed cells, and the mRNA exhibited constitutive and postinduction half-lives like those of the α1-tubulin mRNA. Northern blot analysis revealed the occurrence of a 3′ RNA fragment derived from the poly(G)-containing α1-tubulin transcripts. The 3′ fragment was shown to accumulate as full-length mRNA disappeared in actinomycin D-treated cells, indicating a precursor-product relationship. Insertion of a second poly(G) tract upstream of the first resulted in accumulation of only a longer 3′ fragment, suggesting that the decay intermediate is generated by 5′-to-3′ exonucleolytic digestion. A translational requirement for generation of the 3′ fragment was demonstrated by experiments in which cells were deflagellated in the presence of cycloheximide. Analysis of fragment poly(A) length revealed that the fragments were, at most, oligoadenylated in nondeflagellated cells but had a long poly(A) tail in deflagellated cells. These findings suggest that the oligoadenylated fragment is a decay intermediate in a deadenylation-dependent, constitutive degradation pathway and that the requirement for deadenylation is bypassed in deflagellated cells. This represents the first example in which a single transcript has been shown to be targeted to different decay pathways under different cellular conditions.

Chlamydomonas reinhardtii is a biflagellated green alga capable of rapidly and synchronously regenerating amputated flagella. Complete regeneration requires a massive induction of flagellar protein synthesis, mediated by the accumulation of their mRNAs (23). Following flagellar regeneration, induced flagellar protein mRNAs are rapidly degraded, exhibiting half-lives of 5 to 20 min, thereby effectively returning the cells to their normal program of protein synthesis (see, e.g., references 3, 4, 14, and 31). We have exploited this induction as a model system to study the regulated stability of the tubulin mRNAs. The mRNAs encoding the α- and β-tubulins (the major flagellar proteins) are synthesized both constitutively and in response to deflagellation. Previous studies have shown that the postinduction half-lives of these mRNAs are accelerated about fourfold relative to their constitutive half-lives, although the transcripts are apparently identical (2). Whether all flagellar protein mRNAs exhibit these dual stability characteristics is not known, because the nontubulin mRNAs are present at only very low levels in nonregenerating cells. We are interested in understanding the nature of the postinduction degradation pathway and whether it differs qualitatively from the degradation process involved in the constitutive turnover of the tubulin mRNAs.

While significant progress is being made in mapping sequence elements that influence mRNA stability, determining exactly what effects those sequences exert on the degradation process has been difficult. Little is known about the pathway(s) by which the great majority of cytoplasmic mRNAs are degraded. One of the major reasons for this poor understanding is the rarity of mRNA degradation intermediates stable enough to accumulate in vivo. There are a few known exceptional cases in which naturally long-lived, discrete degradation intermediates are generated. These rare examples have provided evidence that the initial step(s) in the degradation of some mRNAs is one or more specific endonucleolytic cleavages (9, 10, 28, 32, 33, 35). On the other hand, degradation of the oat phytochrome A mRNA yields an array of intermediates, best explained by a combination of 5′-to-3′ and 3′-to-5′ degrading activities (17).

Remarkable progress toward defining mRNA decay pathways in yeast has resulted from the finding that insertion of a stretch of guanosine nucleotides into an mRNA yields a “trapped” 3′ degradation intermediate spanning from the poly(G) tract to the natural 3′ terminus of the mRNA (15). The following evidence indicates that the stable fragment that results is the product of mRNA decapping and impeded 5′-to-3′ exonuclease activity (reviewed in reference 11). Decapped, full-length products accumulate in cells that are deficient in Xrn1p activity, the major 5′-to-3′ exonuclease in yeast (26). 3′ decay fragments fail to accumulate in dcp1Δ strains, which are deficient in decapping activity (8). The major fragment that accumulates in cells expressing constructs with two poly(G) tracts is the one extending from the first poly(G) to the terminus (26). Analysis of the polyadenylation status of trapped decay intermediates has defined two distinct pathways leading to the onset of decapping and exonuclease digestion: one pathway, likely to be the common pathway for many mRNAs, requires deadenylation to an oligo(A) length (15, 26, 27), while the second pathway does not (25). To date, the only known substrates for the second pathway are aberrant mRNAs that contain premature nonsense codons, although there are likely to be others.

While a role for poly(A) shortening in triggering the degradation of some mRNAs has been suspected for some time (reviewed in references 1 and 34), Parker and colleagues provided the first direct demonstration of this relationship. The extent to which this mechanism occurs in other organisms is not yet clear. There is a convincing compilation of evidence that the onset of degradation of mRNAs encoding c-Fos and other early-response proteins in mammalian cells requires deadenylation (reviewed in references 1 and 12). On the other hand, analysis of the length of the poly(A) sequence on the few naturally long-lived 3′ decay intermediates identified in plants and animals (see above) provides no evidence for a poly(A) shortening prerequisite for these mRNAs. Recently, Couttet et al. (13) have published data indicating that deadenylation precedes decapping for at least four mammalian mRNAs. Whether 5′-to-3′ digestion follows their decapping and whether this is a major or minor decay pathway for these mRNAs could not be addressed by the methods used in that study.

In this study, we have used the approach developed by Decker and Parker (15) to trap degradation intermediates of α1-tubulin mRNA in Chlamydomonas reinhardtii. These studies indicate that α1-tubulin mRNA is normally subject to a deadenylation-dependent, 5′-to-3′ exonucleolytic decay process but that its accelerated postinduction decay occurs via a different pathway. This is the first demonstration that a single transcript can be targeted to different decay pathways under different cellular conditions.

MATERIALS AND METHODS

Plasmids.

Plasmid ptubHApG was derived from plasmid p853, which contains the entire α1-tubulin gene with 5′- and 3′-flanking genomic sequences and sequences encoding a 12-amino-acid influenza virus hemagglutinin epitope tag (CYPYDVPDYASL) adjusted for Chlamydomonas codon bias (21). p853 was a gift from Joel Rosenbaum (Yale University). To begin constructing ptubHApG, a unique NcoI site was introduced into p853 8 bp downstream of the translational termination codon by site-directed mutagenesis. Mutagenesis was accomplished by the method of Kunkel et al. (22) with the mutagenic oligonucleotide sequence 5′-CCCTGATGCCATCCATGGAGTCTAGTAC-3′ (underlining indicates sites of base substitutions) to generate p853/NcoI. Two oligonucleotides composed of poly(G)₃₀ and poly(C)₃₀ tracts containing NcoI recognition sequences on both ends were annealed and subsequently digested with NcoI. This fragment was then ligated into NcoI-linearized p853/NcoI to generate ptubHApG. Plasmid ptubHApG² was constructed by insertion of a second poly(G)₃₀ tract in frame at a BstEII site at codon 366 within the α1-tubulin coding region of plasmid ptubHApG. The poly(G)₃₀ tracts of both plasmids were sequenced to confirm their orientation.

Cell culture, transformation, and screening.

Chlamydomonas strains 125M⁺ and nit1-305 were obtained from the Chlamydomonas Genetic Stock Center (Duke University). Strain 5C12G12, which expresses the α1-tubulin gene containing the HA epitope tag (tubHA), was also kindly provided by the Rosenbaum laboratory (21). The cells were cultured in minimal medium and deflagellated by mechanical shearing as previously described (3). Cycloheximide (CX; Sigma Chemical Co., St. Louis, Mo.) was added to cultures at a final concentration of 20 μg/ml. Actinomycin D (Act-D; Sigma Chemical Co.) was used at a final concentration of 160 μg/ml (3). It is unknown why such a high concentration of this inhibitor is required for efficient inhibition of transcription in Chlamydomonas; however, the cells tolerate it well, since they swim vigorously throughout the experiments. Transformation was accomplished by the glass bead vortexing method of Kindle (19), using the Chlamydomonas gene for nitrate reductase as the selectable marker for transformation (20). Prior to transformation, the cell walls were removed by autolysin treatment (16). Transformed colonies were expanded on selective medium (in replicate experiments) and screened for expression of tubHApG or tubHApG² protein by Western analysis. Sodium dodecyl sulfate (SDS)-solubilized cell extracts were prepared by the method of Rochaix et al. (29). For Western blot analysis, protein samples were separated by SDS-polyacrylamide gel electrophoresis and electroblotted to nitrocellulose (Schleicher & Schuell, Keene, N.H.). Efficient transfer was monitored by Ponceau S staining. The blots were blocked with 1% nonfat dry milk in TTBS (0.5 M NaCl, 20 mM Tris-HCl [pH 7.5], 0.1% Tween 20) and probed with a polyclonal antiserum (HA.11; Berkeley Antibody Co., Berkeley, Calif.) against the HA epitope. Secondary antibody coupled to alkaline phosphatase was detected with the Western Blue 5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium (BCIP/NBT) chromogenic substrate (Promega, Madison, Wis.).

RNA preparation and Northern blot analysis.

For each time point, cells were resuspended in RNA lysis buffer (0.3 M NaCl, 5 mM disodium EDTA, 50 mM Tris-HCl [pH 8.0], 2% SDS) at 2 × 10⁷ cells/ml and quick-frozen. Phenol-chloroform extractions were performed as previously described (2). For Northern blots, total nucleic acid was run on 1.4% formaldehyde–agarose gels and vacuum blotted to Nytran membrane (Schleicher & Schuell) in 4× SSC (1× SSC is 0.15 M NaCl plus 15 mM sodium citrate). Methylene blue staining of the blots was routinely performed to evaluate the relative RNA loading and integrity. The filters were hybridized with ³²P-labeled riboprobes specific for the rbcS2 (ribulose bisphosphate carboxylase small-subunit 2) mRNA and α1-tubulin 3′ untranslated region (UTR) as previously described (5). Oligonucleotide probes were 5′-end labeled with ³²P by the method of Sambrook et al. (30). The sequence of the oligonucleotide probe used to detect the influenza virus HA epitope sequence is 5′-GGCGTAGTCGGGCACGTCGTAGGGGTA-3′ (21). The HA oligonucleotide probe was hybridized in 5× SSC–5× Denhardt’s reagent–0.5% SDS–100 μg of sonicated salmon sperm DNA per ml at 60°C overnight. These blots were washed three times at room temperature in 1× SSC–0.1% SDS and once at 60°C in 0.1× SSC–0.1% SDS for 15 min each. The poly(C)₃₀ oligonucleotide probe was hybridized in 6× SSC–10× Denhardt’s reagent–0.1% SDS at 60°C overnight. The blots were washed in 6× SSC–0.1% SDS three times at room temperature and once at 70°C for 20 min each.

High-resolution Northern blot poly(A) length analysis.

3′ mRNA fragments were generated by oligonucleotide-directed RNase H cleavage as previously described (4). Briefly, an α1-tubulin-specific oligonucleotide was incubated with total nucleic acid, and Escherichia coli RNase H (Promega) was added to cleave the RNA-DNA hybrids. Oligo(dT) was included in some reactions to completely deadenylate the mRNA fragments. The digests were then run on 6% acrylamide–7 M urea gels in 1× TBE (90 mM Tris-borate, 2 mM disodium EDTA [pH 8.0]) and subsequently electroblotted to Nytran membranes. Radiolabeled RNA size markers, prepared by in vitro transcription, were also included in these gels (24). Blots were hybridized, washed, and autoradiographed as described above.

RESULTS

A poly(G)-containing α1-tubulin mRNA exhibits stability characteristics like those of the natural α1-tubulin mRNA.

The tubulin mRNAs in Chlamydomonas are degraded with no detectable accumulation of decay intermediates, as is the case for the majority of mRNAs in other organisms. The success of using a poly(G) insert to trap exonucleolytic decay intermediates of yeast mRNAs (15) led us to try this approach in Chlamydomonas. We first stably transformed cells with the ptubHApG construct illustrated in Fig. 1. The plasmid contained the entire α1-tubulin genomic sequence, including both 5′- and 3′-flanking regions needed for proper transcriptional induction and 3′-end formation (21). A tract of 30 guanidylate residues was inserted 8 bp downstream of the termination codon within the 3′ UTR. The construct also contains a 12-codon influenza virus HA epitope tag immediately upstream of the stop codon. This inserted tag was used to screen for transformants by Western analysis as well as to provide a unique sequence by which to distinguish transgenic from endogenous α1-tubulin mRNA. The mRNA produced from this construct is referred to as the tubHApG mRNA.

FIG. 1.

Schematic representation of the transcribed regions of the ptubHApG and ptubHApG² constructs and position of relevant probes used in this study. Plasmid ptubHApG contains the entire α1-tubulin gene including 2 kb of upstream genomic sequence and 1.3 kb of flanking downstream sequence. It has two inserted sequences: a 12-codon influenza virus HA epitope tag immediately upstream of the termination codon and a 30-bp poly(G) tract 8 bp downstream of the termination codon. Above the construct is a restriction map of the indicated region. The ptubHApG² construct contains a second poly(G)₃₀ tract within the α1-tubulin coding region (in frame at codon 366). The positions of the three probes used in this study are shown below the construct. The probes include an in vitro-transcribed riboprobe complementary to 142 bases within the α1-tubulin 3′ untranslated region and two antisense oligonucleotide probes complementary to the poly(G)₃₀ tract and the HA epitope sequence.

Figure 2 compares the induction kinetics and stabilities of the α1-tubulin mRNA in wild-type cells and the tubHApG mRNA in transformed cells. Figure 2A, C, and E shows Northern blotting results for deflagellated cells. Following deflagellation, a transient burst of RNA synthesis led to an accumulation of α1-tubulin mRNA (Fig. 2A). Peak levels were reached by 30 min after deflagellation, and the transcript then decayed with a (maximum) half-life of about 15 min. Figure 2C shows that the tubHApG mRNA was fully inducible by deflagellation and that the kinetics of accumulation and decay were nearly identical to those of the endogenous α1-tubulin mRNA. These results are presented graphically in Fig. 2E. Lanes 1 and 2 of Fig. 2C demonstrate the specificity of the antisense HA oligonucleotide probe; there was no detectable signal from RNA extracted from wild-type cells 30 min after deflagellation, while the probe did recognize a transcript from a cell strain expressing a tubHA mRNA [containing the HA epitope sequence but no poly(G)]. All the blots were rehybridized with a probe detecting the constitutively expressed, stable rbcS2 mRNA as a gel-loading and transfer control.

FIG. 2.

Comparison of expression and stability of tubHApG mRNA and α1-tubulin mRNA. (A to D) Cell cultures of a wild-type strain (wt) and a strain transformed with ptubHApG (csHApG) were deflagellated or treated with Act-D at time zero, and total nucleic acid was extracted at various time points for Northern blot analysis. Values above each lane in autoradiograms indicate minutes after deflagellation (DF) (A and C) or after addition of Act-D (B and D). For Northern blots, nucleic acid samples (4 to 6 μg) were run on 1.4% formaldehyde–agarose gels, transferred to a nylon membrane, and hybridized with ³²P-labeled probes hybridized with an α1-tubulin 3′ UTR probe (A and B) or with an antisense HA oligonucleotide probe (C and D). Each blot was rehybridized with a probe that detects the constitutively expressed, stable rbcS2 mRNA. (E and F) Plots of Northern blot data shown above each graph, quantified by densitometry. ○ and •, α1-tubulin mRNA in wild-type cells; □ and ▪, tubHApG mRNA in transformed cells.

Figure 2B, D, and F shows mRNA levels in nondeflagellated cells treated with Act-D. Figure 2B and D shows that the constitutively expressed α1-tubulin mRNA and the tubHApG mRNA both disappear relatively slowly in the presence of Act-D. Although this experiment was not carried out long enough to obtain an accurate half-life, an approximate half-life of 70 to 75 min was calculated, consistent with previous determinations for α-tubulin mRNA (2, 5). We know that Act-D was effectively inhibiting transcription in these studies, because rehybridization of the blots with a probe for a short-lived mRNA showed that it disappeared rapidly (data not shown).

These results indicate that the nucleotide changes in the tubHApG mRNA did not alter its dual stability properties relative to those of the unmodified α1-tubulin mRNA. It is therefore most likely that the nucleotide changes did not alter the operative degradation pathway(s) for this mRNA.

A 3′ fragment of the poly(G)-containing α1-tubulin mRNA accumulates in both uninduced and deflagellated cells.

Figure 3A and B shows Northern blots of total RNA from uninduced and deflagellated cells expressing the tubHApG mRNA (lanes 3 to 8). The blot shown in Fig. 3A was hybridized with an α1-tubulin 3′ UTR riboprobe (which detects both α1-tubulin and tubHApG mRNAs). A low-molecular-weight RNA species was detected at all time points; it accumulated during induction and disappeared as induced transcript levels returned to basal levels. Its presence in uninduced cells is difficult to detect in these Northern blots but is obvious in Fig. 3C. This low-molecular-weight species was absent from both wild-type cells (lane 1) and tubHA-expressing cells (lane 2) 30 min after deflagellation. Based on the estimated size of the fragment (∼300 nucleotides) and the results of previous studies with yeast, we hypothesized that this RNA species was a decay intermediate spanning from and including the poly(G) tract to the 3′ end of the transcript. To show that it contained at least part of the poly(G) sequence, we hybridized a Northern blot of the same set of RNAs with a poly(C)₃₀ oligonucleotide probe and detected the same-sized fragment (Fig. 3B). The poly(C) probe did not detect RNA of the correct size in wild-type or tubHA-expressing cells (lanes 1 and 2). (The signal in lanes 1 and 2 occurs because this probe sticks to 18S rRNA, which migrates slightly higher than the α1-tubulin and tubHApG mRNAs. This sticking is also evident in Fig. 3C, lanes 4 and 5.) Evidence that the fragment extends to the natural 3′ terminus of the mRNA is presented later in this report.

FIG. 3.

Accumulation and decay of 3′ fragments derived from tubHApG and tubHApG² mRNAs. Stably transformed cell strains expressing the tubHApG mRNA, the tubHApG2 mRNA, or the tubHA mRNA [no poly(G) insert] and wild-type (wt) cells were deflagellated, and nucleic acid was prepared at the time points indicated. Northern blots were prepared as described in the legend to Fig. 2. FL, full-length transcript; DI, decay intermediate. (A) Northern blot of RNAs from wild-type cells (lane 1) and tubHA-expressing cells (lane 2) 30 min after deflagellation and tubHApG-expressing cells before and after deflagellation (lanes 3 to 8). Numbers above lanes 4 to 8 are minutes after deflagellation (DF); 0, nondeflagellated cells. The blot was probed with the α1-tubulin 3′ UTR riboprobe. (B) Northern blot of RNAs from the same cell strains as in panel A but probed with the poly(C) oligonucleotide. (C) Northern blot of RNAs from Act-D-treated, nondeflagellated cells, probed with the poly(C) oligonucleotide. Numbers above each lane are hours in Act-D. (D) Northern blot of RNAs from the cell strains indicated prepared before deflagellation (lanes 1, 3, 5, and 7) or at 30 min after deflagellation (lanes 2, 4, 6, and 8), probed with the poly(C) oligonucleotide.

If the fragment is an intermediate in decay, a precursor-product relationship should be demonstrable. Figure 3C shows a kinetic analysis of the disappearance of the full-length tubHApG mRNA and accumulation and disappearance of the 3′ fragment in Act-D-treated, nondeflagellated cells. By 1 h, the level of full-length mRNA had decreased by half while the level of the fragment had increased. It is clear, however, that the fragment itself was not very stable and had disappeared almost completely by 3 h. This study indicates that the 3′ fragment is a decay intermediate of the tubHApG mRNA in the constitutive decay pathway.

By analogy to the results of studies in yeast, it seemed likely that the 3′ fragment is generated as a product of 5′-to-3′ exonuclease activity. However, it could also be generated by an endonucleolytic cut near the poly(G), possibly stimulated by the poly(G) sequence itself. To try to distinguish between these possibilities, a second poly(G) tract was added upstream of the first, within the coding region of the gene (Fig. 1, ptubHApG²). Figure 3D compares transcripts and 3′ decay fragments observed in cells expressing the single poly(G) construct (lanes 5 and 6) and those expressing the double poly(G) construct (lanes 7 and 8). The size of the decay intermediate arising from the tubHApG² mRNA is consistent with its extending from the upstream poly(G) tract to the 3′ terminus of the mRNA. No RNA species corresponding to a smaller fragment extending from the second poly(G) tract to the 3′ terminus could be detected, even after a long exposure of the blot. This result indicates that there is not an endonuclease target site just upstream of the original poly(G) in the tubHApG mRNA and that artifactual cleavage caused by poly(G) tracts is unlikely. These data strongly suggest that the observed 3′ decay fragments arise via blockage of a 5′-to-3′ exonuclease. Consistent with this interpretation, we could not detect an RNA species which would correspond to a 5′ mRNA fragment extending to the poly(G) by a probe complementary to the complete α1-tubulin mRNA (data not shown).

tubHApG decay intermediates are oligoadenylated in nondeflagellated cells and polyadenylated in deflagellated cells.

If the 3′ fragments of the tubHApG mRNA represent bona fide intermediates in the decay of this transcript, their accumulation provides the opportunity to address whether deadenylation is required for the onset of degradation. If the onset of decay requires deadenylation, the 3′ fragments should bear only short or no poly(A) tails. If deadenylation is not required, the fragments should bear a distribution of poly(A) lengths similar to those of the intact mRNA population. These 3′ fragments will be referred to hereafter as decay intermediates to distinguish them from in vitro-generated fragments (discussed below).

The polyadenylation status of the decay intermediates from both constitutive and induced tubHApG mRNAs is shown in Fig. 4A. This high-resolution Northern blot revealed that the decay intermediate arising from constitutively expressed tubHApG mRNA (lane 3) was smaller than those arising from deflagellation-induced tubHApG mRNA (lanes 4 to 8). The decay intermediate from constitutive tubHApG mRNA (lane 3) migrated identically to an in vitro-deadenylated decay intermediate from induced tubHApG mRNA [lane 9, a 30-min postdeflagellation sample treated with oligo(dT) and digested with E. Coli RNase H]. Thus, it appears that these fragments terminate at the normal 3′ mRNA end and that the intermediate in deflagellated cells is polyadenylated while the intermediate in nondeflagellated cells has, at most, an oligo(A) tail.

FIG. 4.

Poly(A) status of tubHApG and tubHApG² mRNAs and their decay intermediates in nondeflagellated and deflagellated cells. For all of the Northern blots shown, 15-μg nucleic acid samples were run on 6% polyacrylamide gels and transferred electrophoretically to nylon membranes. (A) Northern blot showing tubHApG decay intermediates in nondeflagellated cells (lane 3) and in deflagellated cells (lanes 4 to 8). This blot was hybridized with the α1-tubulin 3′ UTR riboprobe. Lanes 1 and 2 contain RNA from wild-type (wt) and tubHA-expressing cells, respectively, isolated 30 min after deflagellation (DF). Lane 9 contains deflagellated cell RNA (30-min sample) deadenylated in vitro by incubation with oligo(dT) and RNase H. Brackets depict the poly(A) length distribution observed for the decay intermediate (DI) (A⁺, longest length; A⁻, deadenylated). The deadenylated 3′ fragment migrates slightly faster than predicted relative to a 333-nucleotide RNA marker (data not shown). (B) Northern blot showing tubHApG² mRNA decay intermediates in nondeflagellated cells (lane 1) and at 30 min after deflagellation (lane 2). Lane 3 shows the same sample as in lane 2 after in vitro deadenylation as described above. (C) Northern blot showing the poly(A) status of intact tubHApG mRNA in deflagellated cells (lanes 2 to 6) and nondeflagellated (NDF) cells (lane 8). Full-length tubHApG mRNAs were cleaved into 5′ and 3′ fragments by oligonucleotide-directed RNase H digestion. The blot was probed with ³²P-labeled antisense oligonucleotide complementary to the HA sequence tag, which recognizes the in vitro-generated 3′ cleavage fragments but not the decay intermediate. Lanes 1 and 7 show RNA samples deadenylated in vitro by including oligo(dT) in the RNase H digestion. (D) Northern blot showing the poly(A) status of the tubHApG decay intermediate (DI) and of in vitro-generated 3′ mRNA fragments from full-length mRNAs (α1-tub, α1-tubulin and tubHApG mRNAs combined). All RNA samples (except that shown in lane 7) were incubated with an oligonucleotide that targets the full-length mRNAs, but not the decay intermediate, for RNase H cleavage. This blot was probed with the α1-tubulin 3′ UTR riboprobe. Lane 1 shows the migration of in vitro-deadenylated fragments of α1-tubulin mRNA from wild-type cells (∼400 nucleotides), and lane 7 shows in vitro-deadenylated decay intermediates (∼315 nucleotides).

Figure 4B shows a Northern blot of decay intermediates arising from the double-poly(G)-containing transcript. Again, the fragment in nondeflagellated cells is oligoadenylated while the fragment in induced cells is polyadenylated.

Figure 4C shows the poly(A) length distributions of the full-length tubHApG mRNAs in both deflagellated (lanes 2 to 6) and nondeflagellated (lane 8) cells. This Northern blot shows 3′ fragments of the tubHApG mRNA generated by oligonucleotide-directed RNase H cleavage. The blot is hybridized with an antisense HA probe that recognizes the fragment derived from the intact tubHApG mRNA but not the decay intermediate. This blot shows that the poly(A) lengths exhibited by the tubHApG mRNA are typical of those exhibited by the endogenous tubulin mRNAs (4, 5). The induced tubulin mRNAs are synthesized in a transcriptional burst that peaks within 15 min and is essentially over by 30 min after deflagellation (2, 3). Like the endogenous mRNAs, the newly synthesized tubHApG mRNAs (lane 2) have poly(A) tails in the range of 90 to 120 adenosine nucleotides. By 1 h after induction, substantial shortening had occurred to a modal value of about 60 adenosine nucleotides. Lane 8 shows that poly(A) tails on steady-state tubHApG mRNA in nondeflagellated cells have a broad distribution but are predominantly long. Thus, the oligoadenylated state of the constitutively produced decay intermediate is not representative of the poly(A) status of the steady-state tubHApG mRNA. We can conclude that the degradation process leading to accumulation of the 3′ decay intermediate does not begin until after extensive deadenylation has occurred. In contrast, the accumulation of the polyadenylated fragments in deflagellated cells demonstrates that deadenylation is not required for the degradation process that leads to the same 3′ decay intermediates.

The poly(A) length distribution of the degradation intermediates in deflagellated cells does not appear to be a perfect reflection of the poly(A) status of the intact tubHApG mRNA at all time points. Rather, the decay intermediate tails appear to remain uniformly long (Fig. 4A) while the tails on the full-length tubHApG mRNA shorten with time (Fig. 4C). This difference is readily visualized in the Northern blot shown in Fig. 4D, which shows the poly(A) status of both the full-length mRNAs (upper band; tubHApG and α1-tubulin mRNAs combined) and the decay intermediate (lower band) as a function of time after deflagellation. The upper fragment was generated by oligonucleotide-directed RNase H cleavage with an oligonucleotide which targets both the endogenous α1-tubulin mRNA and the tubHApG mRNA but not the decay intermediate. The blot was hybridized with the 3′-UTR probe, which recognizes all three RNA species. Possible interpretations of this finding are presented in Discussion.

Inhibition of protein synthesis blocks both tubHApG mRNA degradation and formation of the decay intermediate in deflagellated cells.

Previous studies have shown that the tubulin and other flagellar protein mRNAs accumulate normally when deflagellation occurs in the presence of CX but that their subsequent rapid degradation is blocked (3, 4). To further confirm that the 3′ fragment is an authentic decay intermediate, we asked whether inhibition of protein synthesis altered its accumulation. Figure 5A shows a Northern blot of RNA prepared from cells deflagellated in the absence or presence of CX and hybridized with the poly(C) probe. This blot shows that the decay intermediate accumulates to readily detectable levels in cells deflagellated in the absence of CX (lanes 3 through 8) but not in the presence of CX (lanes 10 through 13). The decay intermediate is clearly present in the predeflagellation time point (lane 9). These cells were incubated in CX for 15 min before deflagellation, indicating that exposure to CX does not, by itself, result in rapid loss of the constitutive decay intermediate. Figure 5B shows a high resolution Northern blot of the same CX-containing samples shown in Fig. 5A, hybridized with the tubulin 3′ UTR probe. This image confirms that the 3′ decay intermediate does not accumulate in cells deflagellated in the presence of CX. It also shows that incubation in CX did not alter the oligoadenylated state of the constitutive decay intermediate (lane 4). We have observed in multiple experiments that the oligoadenylated decay intermediate present at steady state completely disappears after deflagellation (in the presence or absence of CX) and that its disappearance occurs within 15 min. This could be due to activation of a new degradation mechanism in deflagellated cells or to the normal decay of this fragment in the absence of new accumulation.

FIG. 5.

Effect of cycloheximide addition on decay intermediate accumulation. Cells expressing the tubHApG mRNA were deflagellated in the absence or presence of CX (20 μg/ml added 15 min before deflagellation). (A) Autoradiogram of a Northern blot probed with the poly(C) oligonucleotide. Numbers above the lanes are minutes after deflagellation (DF), in the absence (lanes 4 to 8) or presence (lanes 10 to 13) of CX. Lane 9 (0 min) shows RNA from nondeflagellated cells exposed to CX for 15 min. Lanes 1 and 2 show RNA from deflagellated wild-type (wt) and tubHA-expressing cell strains. The same blot was rehybridized with the rbcS2 probe. FL, full-length; DI, degradation intermediate. (B) High-resolution Northern blot analysis of RNA samples from CX-treated cells (lanes 4 to 8), hybridized with the 3′-UTR riboprobe. Lanes 3 and 9 show the migration of the in vitro-deadenylated decay intermediate.

Most importantly, the finding that both tubHApG mRNA degradation and 3′-fragment accumulation are blocked by CX provides additional evidence that the tubHApG 3′ fragment is a decay product of the normal postdeflagellation degradation process that is blocked by CX.

DISCUSSION

Evidence for deadenylation-dependent decay of α1-tubulin mRNA in Chlamydomonas.

This study demonstrates that a poly(G)₃₀ insert in the 3′ UTR of α1-tubulin mRNA of Chlamydomonas leads to the accumulation of a 3′ fragment of that mRNA. Experiments comparing fragment sizes before and after oligo(dT)-directed RNase H digestion indicate that the fragment terminates at the normal 3′ end in oligo(A) or poly(A). Hybridization of Northern blots with a poly(C) probe shows that the fragment contains at least some of the poly(G) tract, and its size is consistent with the poly(G) tract occurring at the 5′ terminus. RNA fragments beginning with the poly(G) tract and terminating with oligo(A) accumulate in yeast cells expressing poly(G)-containing mRNAs (15, 26, 27). In yeast, the fragments almost certainly result from the inability of a 5′-to-3′ exonuclease (Xrn1p) to proceed efficiently through the poly(G), since in cells lacking Xrn1p activity, full-length (decapped) mRNAs accumulate while the 3′ fragment is absent or much reduced (18, 26). There is strong evidence to support the notion that the XRN1 nuclease is a (or the) major mRNA-degrading enzyme in yeast (reviewed in reference 11), arguing that the observed fragments are trapped intermediates of the normal decay process. The accumulation of poly(G)-containing 3′ RNA fragments in Chlamydomonas, in particular the accumulation of only the longer fragment in cells expressing the tubHApG² mRNA, strongly suggests that a 5′-to-3′ exonuclease might operate to degrade mRNAs in this organism too.

In yeast, the events preceding the rapid 5′-to-3′ digestion of mRNA have been delineated: poly(A) is shortened to an oligo(A) length, which permits or stimulates decapping, exposing the 5′ end to the exonuclease activity (reviewed in references 11 and 34). In this report we have demonstrated that the tubHApG and tubHApG² 3′ fragments that accumulate in nondeflagellated cells occur only in a deadenylated form, indicating that the constitutive decay pathway for the α1-tubulin mRNA also requires deadenylation as a first step. Whether deadenylation is followed by decapping, as in yeast, or by some internal cleavage is not known. Recent evidence that at least some fraction of oligoadenylated mammalian mRNAs undergo decapping (13) suggests that this decay pathway may be a universal one.

The question whether the tubHApG 3′ fragment is an authentic decay intermediate, reflective of the normal decay process of α1-tubulin mRNA, is important. For example, it is possible that the poly(G) tract artifactually targets RNase activity to this altered version of α1-tubulin mRNA. The recent finding that a mouse homolog of the yeast Xrn1p exhibits a preference for G4-tetraplex substrates, though not necessarily simple poly(G) tracts, makes this issue a matter of concern (6). Evidence supporting the position that the poly(G) insert has trapped a normal degradation intermediate of the α1-tubulin mRNA includes the following. (i) Neither the constitutive nor the postinduction half-life of the tubHApG mRNA is altered relative to that of the α1-tubulin mRNA. (ii) The degradation process leading to accumulation of the fragment in nondeflagellated cells is selective for deadenylated mRNAs, while the parallel degradation process in deflagellated cells is not, indicating regulation of the observed degradation process. (iii) The stabilization of induced tubulin and tubHApG mRNAs by CX treatment is accompanied by failure to generate the fragment. Beelman and Parker (7) have shown that CX treatment inhibits the decapping reaction in yeast cells, and it is possible that the same reaction is inhibited under these conditions in Chlamydomonas.

Evidence for a deadenylation-independent decay pathway for α1-tubulin mRNA in deflagellated cells.

Most interestingly, we show that the 3′ fragments derived from deflagellation-induced tubHApG and tubHApG² mRNAs carry long poly(A) tails. Thus, the deadenylation prerequisite for the degradation process leading to 3′ decay intermediates is bypassed during the induction event. Deflagellation-induced α- and β-tubulin mRNAs exhibit a three- to fourfold-reduced half-life relative to the same transcripts in nondeflagellated cells (2). Bypassing the deadenylation step could contribute to, or be wholly responsible for, this shortened half-life. A deadenylation-independent mRNA decay pathway has also been described in yeast. The only known substrates of this so-called nonsense-mediated decay pathway are mutant or unspliced mRNAs containing premature nonsense codons; accessing this pathway results in accelerated degradation (25).

The observation that the tubHApG 3′ fragments appear to carry only long poly(A) tails, even at time points when the tails on the full-length mRNAs are measurably shorter, requires explanation. It precludes the simple scenario in which the accumulated fragments represent all products of a random targeting process. Moreover, the kinetics of accumulation and decay of the 3′ fragment in deflagellated cells are not entirely consistent with its being a product of decay of the whole population of induced mRNAs; specifically, it does not accumulate to its highest levels after 60 min when massive degradation is occurring.

Possible interpretations of this finding are (i) that the degradation process leading to polyadenylated decay intermediates is nuclear; (ii) that this degradation process is cytoplasmic but operative only during the first 15 to 30 min following deflagellation, when poly(A) tails are still long; (iii) that long poly(A)-tailed mRNAs, perhaps newly transported, are selectively targeted for this degradation process; or (iv) that only long poly(A)-tailed decay intermediates are stable enough to accumulate in deflagellated cells.

The finding that CX blocks both degradation of the tubHApG mRNA and the appearance of the polyadenylated fragment suggests that the degradation process leading to the fragment is cytoplasmic, not nuclear. We know that tubulin mRNAs induced in the presence of CX are predominantly cytoplasmic because (i) they are polysomal, by the criterion of sedimentation in polysomal regions of sucrose gradients in an EDTA-releasable form, and (ii) they are subject to poly(A) shortening within minutes after their synthesis, a process only known to occur in the cytoplasm (4, 5). Thus, it is most likely that CX is blocking a cytoplasmic process.

The occurrence of a cytoplasmic degradation process, operative only during the first 30 min or so of induction, could also explain the long poly(A) lengths of the fragments. If this explanation applies, we must assume that this transient degradation process is superseded by a different or more potent degradation process, which leaves no 3′ fragments. If only newly synthesized and transported mRNAs were substrates for this decay process, as in the third suggestion above, it could be proposed that two decay processes are functioning simultaneously—one process that leads to the polyadenylated 3′ fragments and another that leaves no trapped intermediate. With regard to the fourth explanation, there is no reason to presume that RNAs with full-length poly(A) tails should constitute a distinctly stable class.

While the full nature of the postinduction degradation process remains obscure, it is clearly qualitatively different from that responsible for the constitutive turnover of the α1-tubulin mRNA, in that at least one component is deadenylation independent. In fact, postinduction degradation is probably completely deadenylation independent, since the bulk of the induced tubulin mRNA is degraded well before poly(A) tails shorten below 40 Å (5). We suggest that decay of α1-tubulin mRNA can proceed by at least two different degradation modes, illustrated in Fig. 6 for the poly(G)-containing tubHApG mRNA: (i) a constitutive pathway requiring deadenylation to presumably some oligo(A) length prior to decay and (ii) a postinduction pathway in which mRNA decays in a deadenylation-independent manner. The absence of poly(G)-trapped decay intermediates bearing shortened poly(A) tails implies the existence of a third pathway that leaves no detectable poly(G)-trapped intermediates. It will be interesting to determine whether the deadenylation-independent decay pathway that we observe in deflagellated cells is specific for the tubulin mRNAs or specific for all flagellar protein mRNAs or whether all cellular mRNAs are transiently subject to this form of decay in deflagellated cells.

FIG. 6.

Deadenylation-dependent and -independent pathways for the degradation of α1-tubulin mRNA in Chlamydomonas. Analysis of poly(G)-trapped decay intermediates provides evidence for at least two decay pathways for the α1-tubulin mRNA. In the deadenylation-dependent constitutive pathway, poly(A) tails must be shortened before the onset of decay, as evidenced by the accumulation of only oligoadenylated intermediates in nondeflagellated cells. In the deadenylation-independent postinduction pathway, poly(A) shortening is not a prerequisite for decay, as evidenced by the accumulation of long poly(A)-tailed intermediates. Both pathways probably involve a 5′-to-3′ exonuclease which is impeded by the poly(G) tract. The failure to detect decay intermediates bearing the full range of poly(A) tail lengths in deflagellated cells suggests the occurrence of a second decay pathway that leaves no detectable 3′ decay intermediates.