PABP prevents the untimely decay of select mRNA populations in human cells

Abstract

Gene expression is tightly regulated at the levels of both mRNA translation and stability. The poly(A)‐binding protein (PABP) is thought to play a role in regulating these processes by binding the mRNA 3′ poly(A) tail and interacting with both the translation and mRNA deadenylation machineries. In this study, we directly investigate the impact of PABP on translation and stability of endogenous mRNAs in human cells. Remarkably, our transcriptome‐wide analysis only detects marginal mRNA translation changes in PABP‐depleted cells. In contrast, rapidly depleting PABP alters mRNA abundance and stability, albeit non‐uniformly. Otherwise stable transcripts, including those encoding proteins with constitutive functions, are destabilized in PABP‐depleted cells. In contrast, many unstable mRNAs, including those encoding proteins with regulatory functions, decay at similar rates in presence or absence of PABP. Moreover, PABP depletion‐induced cell death can partially be suppressed by disrupting the mRNA decapping and 5′–3′ decay machinery. Finally, we provide evidence that the LSM1‐7 complex promotes decay of “stable” mRNAs in PABP‐depleted cells. Taken together, these findings suggest that PABP plays an important role in preventing the untimely decay of select mRNA populations.

PABP depletion in human cells differentially alters abundance and stability of mRNA subsets, where normally stable transcripts are in particular affected.

Introduction

All eukaryotic cytoplasmic mRNAs possess a 5′‐cap structure and—with the exception of replication‐dependent histone mRNAs—maintain a 3′ poly(A) tail, elements that synergistically enhance mRNA translation in a number of systems (Hershey et al, 2012). These structures help to promote translation initiation by interacting with eukaryotic translation initiation factors, which in turn recruits the 40S ribosomal subunit. The mRNA 5′ cap binds the eukaryotic translation initiation factor complex eIF4F, which is comprised of the eIF4E cap‐binding protein, the ATP‐dependent RNA helicase eIF4A and eIF4G; a large scaffold protein binding both eIF4E and eIF4A (Pelletier & Sonenberg, 2019). In addition to binding eIF4E and eIF4A, eIF4G also interacts with eIF3, a translation initiation factor interfacing with the 40S small ribosomal subunit. Finally, in addition to these interactions, eIF4G directly interacts with the cytoplasmic poly(A)‐binding protein (termed PABP from here on), an interaction that is conserved from yeast to humans (Tarun & Sachs, 1996; Tarun et al, 1997; Imataka et al, 1998; Safaee et al, 2012).

PABP is highly conserved in eukaryotes, often binding to mRNA 3′ poly(A) tails as multimers that each cover roughly 25–30 adenosines (Baer & Kornberg, 1980). PABP contains four RNA recognition motifs (RRMs), a linker region and a C‐terminal domain facilitating several protein–protein interactions (Roy et al, 2002; Mangus et al, 2003; Smith et al, 2014). One such interactor is the PABP‐interacting protein 2 (Paip2), which when bound to PABP decreases PABP affinity for poly(A) RNA (Khaleghpour et al, 2001). The PABP C‐terminal domain also binds to GW182, a micro(mi)RNA‐associated scaffold that in addition to PABP interacts with Argonaute proteins and deadenylases to promote gene silencing (Jonas & Izaurralde, 2015). While budding yeast code for a single poly(A)‐binding protein (Pab1), the mammalian genome codes for several. These include the prototypical cytoplasmic PABP (PABPC1), as well as an embryonic PABP (PABPC1L), a testis‐specific PABP (PABPC2/3) and an inducible PABP (PABPC4), all containing highly similar domain architecture (Smith et al, 2014). PABP binding to eIF4G stimulates translation of reporter mRNAs in mammalian cell‐free extracts, and expression of an eIF4G mutant in Xenopus oocytes that cannot bind PABP impairs their progesterone‐induced maturation (Wakiyama et al, 2000; Kahvejian et al, 2005). The interaction between PABP and eIF4G has long been posited to facilitate communication between individual mRNA termini, often depicted as a “closed‐loop” model (Gallie, 1998; Wells et al, 1998). Indeed, PABP binding to eIF4G stabilizes the association of the eIF4F complex with the mRNA cap‐structure, as determined by the enhancement of eIF4E‐cap interaction in in vitro translation systems (Kahvejian et al, 2005).

In addition to interacting with translation factors, PABP is intimately linked to mRNA turnover. The major pathway in eukaryotic cells for mRNA decay is initiated by shortening of the 3′ poly(A) tail (deadenylation) via PAN2‐PAN3 and CCR4‐NOT deadenylase complexes (Goldstrohm & Wickens, 2008; Wiederhold & Passmore, 2010; Mugridge et al, 2018). While PABP stabilizes mRNAs, it also directly interacts with both deadenylase complexes to stimulate deadenylation and must be removed from the poly (A) tail for deadenylation to proceed. Once deadenylation has proceeded to the point where only a short oligo (A) tail remains, this stretch of adenosines often undergoes terminal uridylation by terminal uridyl transferases (TUT4/7) (Yu & Kim, 2020). This, in turn, provides a docking platform for the LSM1‐7 complex, which recruits mRNA decapping factors (i.e. DCP2) that hydrolyse the mRNA 5′ cap structure, thereby exposing the mRNA 5′ terminus to 5′–3′ decay by XRN1 exonuclease (Boeck et al, 1998; Tharun & Parker, 2001; Rissland & Norbury, 2009; Sharif & Conti, 2013).

Most studies on how mammalian PABP regulates mRNA translation and stability have used purified proteins and cell‐free in vitro translation systems. Using chromatography‐based methods, one can efficiently deplete endogenous PABP from lysates and functionally reconstitute with recombinant PABP (Svitkin & Sonenberg, 2004; Kahvejian et al, 2005; Svitkin et al, 2009). While these studies have been instrumental in determining some of the fundamental principles by which PABP can regulate mRNAs, how many of these principles hold true in mammalian cells for endogenous transcripts is unclear. Here, we sought to investigate how mammalian PABP regulates mRNA translation and decay in human cells. To accomplish this, we utilized CRISPR‐based genetic engineering coupled with degron tagging allowing for rapid depletion of endogenous PABP in human cells. Surprisingly, transcriptome‐wide analysis of PABP‐depleted cells by polysome profiling barely detect any changes in translation efficiency. In contrast, depleting PABP dramatically affected the abundance of specific mRNA subsets.

Results

PABPC1 and PABPC4 can cooperate to support mammalian cell viability

Studies using mammalian cell‐free extracts have shown that depleting PABP in vitro using affinity chromatography approaches dramatically inhibits cap‐ and poly(A) tail‐dependent translation of reporter mRNAs (Kahvejian et al, 2005). However, knock down of PABPC1 by RNAi has a limited impact on the translation of luciferase reporters (Yoshida et al, 2006) and no noticeable impact on cell viability (Fig 1A and B, upper right panel). PABPC1 is very abundant in mammalian cells (reported to be 4 µM in HeLa cells (Gorlach et al, 1994)). Moreover, the PABP inhibitor, Paip2 becomes ubiquitinylated and degraded in PABPC1‐depleted cells (Yoshida et al, 2006). Therefore, it is possible that even relatively low levels of PABPC1 following knockdown is sufficient to support mRNA translation. However, we successfully generated viable PABPC1 knockout HeLa cells using CRISPR/Cas9 (Fig EV1A and B), indicating that PABPC1 is not essential in HeLa cells. Based on these data, we hypothesized that other mammalian PABP genes are expressed at sufficient levels to support cell viability. A previous quantitative analysis of the HeLa cell proteome indicated that in addition to PABPC1, PABPC4 is also relatively abundant (0.9 µM (Hein et al, 2015)). As PABPC4 is almost 80% identical to PABPC1 in sequence, we hypothesized that PABPC4 may cooperate with PABPC1 in human cell lines to support cell viability and protein synthesis. To this end, we generated PABPC4 knockout (PABPC4^KO) HeLa cells using CRISPR/Cas9 (Fig 1A), and subsequently knocked down PABPC1 by RNAi. Similar to PABPC1 knockdown and knockout cells, PABPC4^KO cells are also viable (Fig 1B, bottom left panel). However, while knocking down PABPC1 in wild‐type HeLa cells is tolerated, depleting PABPC1 in PABPC4^KO HeLa cells was ultimately lethal (Fig 1B, bottom right panel). We also observed a dramatic drop in protein synthesis in PABPC1‐depleted PABPC4^KO cells using the SUnSET assay (Fig EV1C), where cells were treated with low concentrations of puromycin and puromycin‐labelled peptides were resolved by western blotting to evaluate the rate of protein synthesis. In addition to HeLa cells, we also observed PABPC1/PABPC4 cooperativity in HEK293T cells (Fig EV2A and B), where PABPC1 depletion in PABPC4^KO HEK293T cells dramatically impaired cell viability but had no noticeable impact on wild‐type cells. Taken together, these data suggest that PABPC1 and PABPC4 provide a collective pool of PABP to support cell viability in a number of human cell lines.

Figure 1. PABPC1 and PABPC4 cooperate to support HeLa cell viability.

1. Western blot analysis wild‐type or PABPC4^KO HeLa cells depleted of PABPC1 by siRNA‐mediated knockdown. siGFP represents a negative control.
2. Phase‐contrast images of wild‐type or PABPC4^KO HeLa cells depleted of PABPC1 by siRNA‐mediated knockdown. Scale bar, 100 μm.
3. Total numbers of viable cells in (B) were quantified with acridine orange and propidium iodide staining and graphed as a percentage of wild‐type HeLa cells. Error bars represent the SEM of three biological replicates.

Figure EV1. PABPC1 and PABPC4 cooperate to support protein synthesis in HeLa cells.

1. Western blot analysis of wild‐type or PABPC1^KO HeLa cells.
2. Phase‐contrast images of wild‐type and PABPC1^KO HeLa cells. Scale bar, 100 μm.
3. Wild type or PABPC4^KO HeLa cells depleted of PABPC1 by siRNA‐mediated knockdown. Seventy‐two hours post‐transfection, cells were incubated with a short pulse of puromycin or puromycin and cycloheximide (control). Cells were subsequently lysed, and equal protein amounts were resolved by SDS‐PAGE. Western blot analysis was performed using a monoclonal antibody against puromycin. Equal lane loading was confirmed by ponceau staining of the membrane (bottom panel).

Figure EV2. PABPC1 and PABPC4 cooperate to support HEK293 cell viability.

1. Western blot analysis wild type or PABPC4^KO HEK293 cells depleted of PABPC1 by siRNA‐mediated knockdown. siGFP represents a negative control.
2. Phase‐contrast images of wild type or PABPC4^KO HEK293 cells depleted of PABPC1 by siRNA‐mediated knockdown. Scale bar, 100 μm.
3. Temporal analysis of PABPC1 depletion by siRNA in HeLa cells. HeLa cells were transfected with siRNA targeting GFP (control) or PABPC1. Cells were subsequently harvested at 24‐, 48‐, 72‐ and 96‐h time points post‐transfection, resolved by SDS‐PAGE and probed with antibodies against PABPC1 and Actin.

An inducible system for rapidly depleting PABP in human cells

Robust PABPC1 depletion by RNAi takes roughly 72 h (Fig EV2C) and is most likely due to the very slow turnover rate of PABP (Gorlach et al, 1994). Unfortunately, by the time PABPC1 is depleted in PABPC4^KO cells, cell viability is already affected. Selecting a time‐point to investigate primary effects on gene expression depending on PABP is therefore difficult. We consequently wanted to generate a system where PABP is quickly depleted. To this end, endogenous PABPC1 in PABPC4^KO HeLa cells was C‐terminally tagged with dihydrofolate reductase (DHFR), a bacterial protein that is stable in the presence of the antibiotic Trimethoprim (TMP), but is ubiquitylated and rapidly degraded without it (Fig 2A) (Sheridan & Bentley, 2016). Therefore, DHFR has often been used as an inducible degron system for rapidly degrading eukaryotic proteins it is fused to. PABPC4^KO HeLa cells were subjected to CRISPR/Cas9 genome editing using repair templates incorporating DHFR at the C‐terminus of PABPC1 as well as antibiotic resistance markers for selection. Transfection of these constructs together with plasmids expressing two PABPC1‐specific guide RNAs and co‐expressing the Cas9 nickase yielded multiple antibiotic‐resistant colonies, hereafter referred to as PABP^DHFR. Western blotting of PABP^DHFR lysates confirmed homozygous targeting, shown by the higher molecular weight PABPC1 at levels similar to that of endogenous non‐tagged PABPC1 in wild‐type HeLa cells (Fig 2B). To test if activation of the degron tag would lead to depletion of endogenous PABPC1, we switched our PABP^DHFR cells to TMP‐free media and isolated cells at various time points post‐TMP withdrawal for western blotting. PABPC1 levels were reduced within 4 h of TMP removal and were nearly eliminated after 8 h (Fig 2B). This rate of PABPC1 elimination is roughly nine‐fold faster as compared to RNAi. In keeping with previous reports that PABP depletion leads to Paip2 degradation (Yoshida et al, 2006), we also observed a decrease in Paip2 levels upon PABP depletion in our degron cell line (Fig 2B). PABP^DHFR cells showed no noticeable defects when grown in the presence of TMP (Fig 2C, upper left panel). However, PABP^DHFR cells cultured in TMP‐free media ultimately died within 48 h (Fig 2D, lower left panel). To confirm that this lethality was solely due to the depletion of PABP, we transfected PABP^DHFR cells with a plasmid expressing wild‐type PABPC1 (Fig 2D). Cells were split 48 h post‐transfection into media containing or lacking TMP and cultured for an additional 48 h to maintain or deplete endogenous PABPC1. Ectopic PABPC1 rescued the viability of degron cells in the absence of endogenous PABP (Fig 2C, lower right panel as compared to upper right panel), indicating that depletion of endogenous PABP via degron activation underlies cell death. Therefore, our degron cell line provides a system to study effects of PABP depletion on protein synthesis at early time points, as well as the impact of PABP on cell viability at later time points.

Figure 2. Generation of PABP^DHFR HeLa cells.

1. Schematic diagram of strategy employed to C‐terminally tag PABPC1 in PABPC4^KO HeLa cells using CRISPR‐Cas9 mediated homology directed repair (HDR). Adapted from (Sheridan & Bentley, 2016). Homology arms flanked repair cassettes containing a DHFR sequence, preceded by a HA tag and separated from an antibiotic resistance gene (denoted as Ab^R and either Puro or Hygro) by a T2A cleavage site followed by a SV40 signal.
2. Western blot confirmation of homozygous PABPC1 tagging with DHFR and degron activity. Successful homozygous tagging of PABPC1 is indicated by the higher molecular weight migration (lane 2) as compared to untagged PABPC1 (lane 1). Degron activity was assessed by culturing PABP^DHFR cells in the absence of TMP for 4 or 8 h and assessing PABPC1 levels by western blotting.
3. Phase‐contrast images of PABP^DHFR cells, or PABP^DHFR cells transfected with a plasmid coding for wild‐type PABPC1, 48 h post‐TMP removal. Scale bar, 100 μm.
4. Western blot analysis of lysates derived from cells imaged in (C).

Depleting PABP alters mRNA levels with negligible effects on mRNA translation

To investigate how depleting PABP affects protein synthesis in mammalian cells, we cultured PABP^DHFR cells in TMP‐containing media to maintain PABP levels, or in TMP‐free media for 12 h to deplete PABP. Cell lysates were subsequently prepared and resolved by centrifugation on sucrose gradients for polysome profiling (Fig 3A). We observed a substantial decrease of heavy polysomes with a concomitant increase in 80S ribosomes in PABP‐depleted cells as compared to cells expressing PABP. Such a shift in polysomes is generally indicative of a block in mRNA translation at the initiation step (Hershey et al, 2012). A decrease in protein synthesis rates in PABP‐depleted cells was also observed by SUnSET assay (Fig 3B). This decrease in protein synthesis could be the result of PABP depletion activating stress responses. To assess whether depleting PABP induced cell stress, we evaluated phosphorylation status of the alpha subunit of eIF2 (eIF2α), a key translation initiation factor inactivated via phosphorylation as a result of stresses such as oxidative stress, heat shock, radiation, hypoxia and amino acid starvation (Holcik & Sonenberg, 2005). Treating PABP^DHFR cells with sodium arsenite (a mediator of oxidative stress) triggered eIF2α phosphorylation; in contrast depleting PABP via culturing PABP^DHFR cells in the absence of TMP did not lead to phosphorylation of eIF2α (Fig EV3). These data therefore suggest that reduced protein synthesis in the absence of PABP is not due to stress‐dependent repression of mRNA translation.

Figure 3. Depleting PABP alters transcriptome abundance with a minimal impact on mRNA translation.

• A
  Ribosome profiles from PABP^DHFR cells grown in the presence or absence of TMP for 12 h. Lysates were subsequently resolved by ultracentrifugation on 5–50% sucrose gradients and fraction were collected during which time UV absorbance at 254 nm (Abs 254 nm) was monitored for tracing. 40S and 60S subunits, monosomes (80S) and polysomes positions are indicated above tracings. PABP‐expressing and ‐depleted cells are denoted by a black and blue trace lines, respectively.
• B
  SUnSET assay of PABP^DHFR cells grown in the presence or absence of TMP for 12 h to maintain or deplete PABP. Cells were subsequently pulsed with either puromycin or puromycin and cycloheximide (control), lysed and equal protein amounts were resolved by SDS‐PAGE. Western blot analysis was performed using a monoclonal antibody against puromycin or actin (loading control).
• C
  Ribosome profiles from PABP^DHFR cells grown in the absence of TMP for 12 h to deplete PABP and resolved on sucrose gradients containing low KCl [(150 nM) traced in dark blue] or high KCl [(500 mM) traced in turquoise].
• D
  Scatter plot comparing log2 fold‐changes [(−) PABP versus (+) PABP] in polysome‐associated mRNA (y‐axis) to corresponding changes in total cytoplasmic mRNA. Transcripts identified as regulated via altered translation efficiency (orange and red) or abundance (light and dark green) according to anota2seq are visualized together with non‐regulated transcripts (grey).
• E, F
  Top ten Wikipathway (WP) terms (Slenter et al, 2018) significantly enriched among proteins coded for by mRNAs that were decreased (E) or increased (F) in abundance in PABP‐depleted cells relative to cells expressing PABP. The number above each column represents the number of genes associated with its corresponding term.

Figure EV3. PABP depletion does not lead to eIF2α phosphorylation.

PABP^DHFR cells were grown in the presence or absence of TMP and/or sodium arsenite (NaAsO₂) and harvested at select time points post‐TMP removal of TMP. Cells were subsequently harvested, resolved by SDS‐PAGE and probed with antibodies against PABPC1, total eIF2α, phospho‐eIF2α (eIF2α~P) and actin.

We next set out to determine if the increased level of 80S was comprised of monosomes (composed of a single ribosome on an mRNA) or monomers (mRNA‐free ribosomes). This was accomplished by analysing cell lysates on high salt (500 mM of KCl) sucrose gradients, which disrupts free ribosomes into their ribosomal subunits but has no effect on mRNA‐bound ribosomes (Martin & Hartwell, 1970; Martin, 1973). We observed that 80S ribosomes in PABP‐depleted cells dissociate into 40S and 60S subunits on high‐salt gradients, while polysome levels are similarly maintained (Fig 3C). These results suggest PABP depletion leading to an increase in mRNA‐free 80S monomers in human cells rather than 80S monosomes.

Next, we carried out a transcriptome‐wide analysis using polysome profiling to assess whether translation is altered in the absence of PABP in a transcript selective fashion. To this end, total cytoplasmic mRNA and fractions containing mRNAs associated with heavy polysomes (> 3 ribosomes) were isolated from PABP^DHFR cells cultured in the presence or absence of TMP for 12 h in three biological replicates followed by RNA sequencing. Anota2seq analysis (Oertlin et al, 2019) was then used to assess PABP‐dependent changes in mRNA translation (i.e. changes in heavy polysome‐associated mRNA without corresponding changes in total cytoplasmic mRNA) and mRNA abundance (congruent changes in total cytoplasmic and heavy polysome‐associated mRNA). Strikingly, depleting PABP led to major changes in mRNA abundance together with relatively minor changes in translation efficiency (Fig 3D; Dataset EV1). Thus, while depleting PABP decreases protein synthesis rates and decrease polysome‐associated mRNAs in cells, our data suggest that this is most likely not a result of global, stress‐dependent, nor selective alterations in mRNA translation. We note that the laboratory of David Bartel independently reached a similar conclusion by ribosome profiling of HeLa cells that had been transfected with siRNAs targeting both PABPC1 and PABPC4 that was published while this manuscript was being reviewed (Xiang & Bartel, 2021).

Multiple features characterize mRNAs that are downregulated in PABP‐depleted cells

Depleting PABP led to major changes in mRNA levels, with thousands of mRNAs decreasing and increasing in abundance (Fig 3D, dark and light green dots, respectively). We therefore sought to investigate if these two populations of mRNAs displayed any key distinguishing features. First, to assess if genes whose mRNAs changed at least 2‐fold following PABP depletion shared biological functions, we performed gene set enrichment analysis (Slenter et al, 2018). Among mRNAs reduced upon PABP depletion, we observed a strong enrichment of those coding for proteins involved in constitutive processes, including mRNA translation, mitochondrial functions and nucleotide processing (Fig 3E; Dataset EV2). In contrast, proteins encoded by mRNAs upregulated in PABP‐depleted cells were enriched for regulatory processes, including signalling pathways, morphogenesis and differentiation (Fig 3F; Dataset EV2).

As untranslated regions often contain elements impacting mRNA levels, we next determined if UTR length characterizes mRNAs suppressed in PABP‐depleted cells. Indeed, mRNAs whose abundance decreased following PABP depletion had significantly shorter 5′ and 3′UTRs as compared to non‐regulated (background) or up‐regulated transcripts (Fig 4A and B). Moreover, as PABP binds the mRNA 3′ poly(A) tail, we determined whether any correlation exists between poly(A) tail length and mRNAs whose steady‐state levels change in the absence of PABP. To this end, we used a published dataset on poly(A) tail lengths in HeLa cells determined using TAIL‐Seq (Yi et al, 2018). This revealed that mRNAs whose abundance decreased following PABP depletion were reported to have shorter poly(A) tails as compared to mRNAs that had higher steady‐state levels in the absence of PABP, with median tail lengths of 100 and 116, respectively (Fig 4C).

Figure 4. Characteristics of mRNAs whose abundance is altered following PABP depletion.

• A–C
  Boxplots of 5′UTR (A), 3′UTR (B) and poly(A) tail lengths for mRNAs whose abundance decreased (“DOWN”, dark green) or increased (“UP”, light green) in PABP‐depleted cells as compared to cells expressing PABP. Solid horizontal middle lines in box plots denote the median; lower and upper box limits correspond to first and third quartiles; whiskers extend from box limits to the most extreme values. Wilcoxon‐Mann‐Whitney tests (two‐sided) were used to determine differences between upregulated mRNAs and downregulated mRNAs relative to each other and to the background (mRNAs whose abundance did not change in the absence of PABP relative to PABP‐expressing cells).
• D, E
  Comparison of log2‐fold change in transcript abundances within PABP‐depleted cells relative to cells expressing PABP versus their reported P‐body log2‐fold enrichment status (Hubstenberger et al, 2017). The Pearson’s correlation coefficient (R = 0.73) and the corresponding P‐value (P < 2.2e‐16) are reported with a corresponding trendline (black) and 95% confidence interval (grey). (D) Points in dark green indicate mRNAs that are downregulated in PABP‐depleted cells and depleted from P‐bodies. Points in light green indicate mRNAs that are upregulated in PABP‐depleted cells and enriched in P‐bodies. Transcript numbers in each quadrant are listed. (E) Nuclear‐encoded mRNAs encoding cytoplasmic ribosomes (orange), mitochondrial ribosomes (blue) and proteins with mitochondrial functions (green) that are downregulated in PABP‐depleted cells relative to PABP‐expressing cells and are predominantly depleted from P‐bodies. Proteins with mitochondrial functions were defined based on an inventory of human mitochondrial proteins (Calvo et al, 2016).

Longer 3′UTRs may include more binding sites for miRNAs and RNA‐binding proteins (RBPs) that in turn regulate mRNA translation and stability (Sandberg et al, 2008). Moreover, many mRNAs targeted by miRNAs and RBPs are enriched in phase‐separated membraneless structures called P‐bodies (Hubstenberger et al, 2017). We therefore assessed correlations between mRNAs whose levels changed following PABP depletion and mRNA populations reported to be enriched in purified P‐bodies (Hubstenberger et al, 2017). Indeed, there was a strong correlation (R = 0.73, P < 2.2e‐16) between PABP‐dependent mRNA levels and P‐body enrichment (Fig 4D). Specifically, mRNAs whose levels decreased in PABP‐depleted cells were depleted from P‐bodies (Fig 4D, dark green) and included many nuclear transcripts encoding ribosomal and mitochondrial proteins (Fig 4E). In contrast, mRNAs upregulated upon PABP‐depletion were enriched in P‐bodies (Fig 4D, light green). Taken together, these data show that (i) UTR length, (ii) poly(A) tail length and (iii) P‐body enrichment associate with PABP‐dependent changes in mRNA levels.

PABP differentially impacts mRNA stability

PABP is tightly linked to mRNA turnover. We therefore sought to determine whether changes to mRNA decay rates could explain PABP‐dependent changes in mRNA levels. To this end, we first used published data of mRNA half‐lives in a human cell line (human umbilical vein endothelial cells) and compared it to PABP‐dependent changes in mRNA levels (Tiana et al, 2020). Indeed, mRNAs whose abundance decreased following PABP depletion were reported to have significantly longer half‐lives relative to mRNAs whose steady‐state levels increased following loss of PABP (Fig 5A). To directly explore how depleting PABP impacts mRNA turnover, we assessed stabilities in several mRNAs whose levels decreased in PABP‐depleted cells (Fig 5B–F). These included ACTB and GAPDH housekeeping genes, as well as mRNAs encoding proteins with mitochondrial functions (ATP5O and UQCC2) and a ribosomal protein (RPS6). Briefly, cells were treated with Actinomycin D (Act‐D) to inhibit de novo transcription; total RNA was isolated and mRNA levels were quantified by RT‐qPCR with spike‐in RNA normalization. In keeping with previously published data, these mRNAs were stable in PABP‐expressing cells (Tiana et al, 2020) but were markedly destabilized in PABP‐depleted cells, as evidenced by their accelerated decay rates (Fig 5B–F). We also assessed the decay rates for BCL2 and MCL1, mRNAs that did not decrease in abundance in PABP‐depleted cells and that contain long 3′UTRs (Fig 3D). Furthermore, BCL2 and MCL1 were reported to be post‐transcriptionally regulated by miRNAs and RNA‐binding proteins, and to be enriched in P‐bodies (Ishimaru et al, 2010; Hubstenberger et al, 2017; Cui & Placzek, 2018). In keeping with this, both BCL2 and MCL1 mRNAs rapidly decayed in PABP‐expressing cells (Fig 5G and H). However, in contrast to stable mRNAs, BCL2 and MCL1 decay kinetics was PABP‐independent. Finally, we sought to directly test if PABP impact the decay rate of a miRNA‐targeted transcript. To this end, PABP^DHFR HeLa cells were transfected with a plasmid coding for a Renilla luciferase (RL) reporter mRNA harbouring six let‐7 miRNA target sites in its 3′UTR (RL‐6xB) or a RL‐6xB reporter mRNA containing mutations in nucleotides complementary to the let‐7 seed sequences (RL‐6XBMUT) (Fig 6A). Transfected cells were split into TMP‐containing media to maintain PABP levels, or into TMP‐free media to deplete PABP, for 12 h and subsequently treated with Act‐D to assess reporter mRNA stability. Similar to MCL1 and BCL2 mRNAs, RL‐6xB mRNA decayed rapidly in a PABP‐independent fashion (Fig 6B). RL‐6xBMUT, which is not targeted by the miRNA machinery, decayed slowly in PABP‐expressing cells (Fig 6B). However, in contrast to RL6xB, depleting PABP accelerated the decay rate of RL‐6xBMUT. Collectively, these data suggest that PABP plays a critical role in maintaining the stability of mRNAs that are normally long‐lived, whereas mRNAs that have short half‐lives, including those targeted by miRNAs, may not be as reliant upon PABP to engender their rapid decay in cells.

Figure 5. Depleting PABP accelerates the decay of stable mRNAs.

• A
  Boxplots of mRNA half‐lives (hrs) assessed by metabolic labelling of normal human umbilical vein endothelial cells for mRNAs whose abundance decreased (“DOWN”, dark green) or increased (“UP”, light green) in PABP‐depleted cells as compared to cells expressing PABP comparing (Tiana et al, 2020). Solid horizontal middle lines in box plots denote the median; lower and upper box limits correspond to first and third quartiles; whiskers extend from box limits to the most extreme values but not further than 1.5× the inter‐quartile range (IQR).
• B–H
  PABP^DHFR cells were cultured in the presence or absence of TMP for 12 h. The stabilities of select mRNAs was assessed by using actinomycin D (5 μg/ml) for the indicated amounts of time. Total RNA was isolated, reverse transcribed with random hexamer oligonucleotide primers and quantified by qPCR. mRNA decay rates were normalized to an in vitro synthesized spike‐in RNA with the zero time point set at 1. Error bars represent the SEM of three independent experiments. First order exponential decay trend lines were generated by non‐linear regression analysis.

Figure 6. Depleting PABP does not impact the decay of a microRNA‐targeted reporter mRNA.

1. Schematic representation of the Renilla luciferase (Rluc) reporter mRNAs containing let‐7‐binding sites (RL‐6xB) or mutated seed sites (RL‐6xBMUT).
2. PABP^DHFR cells were transfected with plasmids encoding RL‐6xB or RL‐6xBMUT. Twenty‐four hours after transfections, the cells were split and cultured for an additional 12 h in the presence or absence of TMP. The stabilities of reporter mRNAs was assessed by using actinomycin D (5 μg/ml) for the indicated amounts of time. Total RNA was isolated, reverse transcribed with random hexamer oligonucleotide primers and quantified by qPCR. mRNA decay rates were normalized to an in vitro synthesized spike‐in RNA with the zero time point set at 1. Error bars represent the SEM of three independent experiments. First order exponential decay trend lines were generated by non‐linear regression analysis.

PABP protects stable mRNAs from deadenylation‐independent mRNA decay

PABP directly interacts with both the CCR4‐NOT and PAN2‐PAN3 deadenylase complexes and must be removed from mRNA poly(A) tails for deadenylation‐dependent mRNA decay to proceed (Siddiqui et al, 2007; Wiederhold & Passmore, 2010; Yi et al, 2018). Depleting PABP by siRNA (Fig 1) or degron‐mediated destabilization (Fig 2) leads to cell death. As loss of PABP alters transcriptome abundance (Fig 3), we next investigated if depleting key mRNA decapping and decay factors had any impact on the viability of PABP‐depleted HeLa cells, as has been reported to be the case in yeast (Caponigro & Parker, 1995). To this end, we transiently transfected PABP^DHFR cells with control siRNA or siRNAs targeting the LSM1 subunit of the LSM1‐7 complex, the decapping protein DCP2 or the 5′–3′ exonuclease XRN1 (Fig 7A). PABP^DHFR cells were subsequently split 48 h post‐transfection and cultured in TMP‐containing or TMP‐free media for an additional 48 h prior to assessing their viability. We observed that PABP depletion‐induced lethality can be partially rescued by knocking down LSM1, DCP2 (which led to XRN1 being co‐depleted) or XRN1 (Fig 7B). We also tested if the loss of LSM1 suppresses PABP depletion‐induced cell death in a non‐degron cell line. LSM1/PABPC4 double knockout HeLa cells were generated by CRISPR/Cas9 and cells were transfected with siRNAs targeting GFP (control) or PABPC1 (Fig EV4A). Similar to above, knock‐down of PABPC1 in PABPC4^KO cells leads to cell death (Fig EV4B, upper right panel). However, LSM1/PABPC4 double knockout cells were insensitive to PABPC1 depletion (Fig EV4B, lower right panel). We also assessed if the deadenylase activity of the CCR4‐NOT complex plays a role in PABP depletion‐induced cell death. To test this, we transiently transfected PABP^DHFR cells with a plasmid encoding a catalytically inactive CNOT7 (D40A) mutant, which impairs deadenylation in the absence of PABP (Yi et al, 2018) (Fig EV5A). However, in contrast to knocking down mRNA decapping factors, inhibiting the CCR4‐NOT complex did not suppress PABP depletion‐induced cell death (Fig EV5B). Taken together, these data suggest that mRNA decapping and 5′–3′ decay factors, including the LSM1‐7 complex, DCP2 and XRN1, promote cell death in the absence of PABP.

Figure 7. Depleting mRNA decapping and decay factors suppresses PABP depletion‐induced cell death.

1. Western blot analysis PABP^DHFR cells depleted of LSM1, DCP2 or XRN1 by siRNA‐mediated knockdown. siGFP represents a negative control.
2. Phase‐contrast images of PABP^DHFR cells previously transfected with siRNAs targeting GFP (control), LSM1, DCP2 or XRN1 and cultured in the presence or absence of TMP for 48 h. Total numbers of viable cells, denoted in the upper right hand corner of images, were quantified with acridine orange and propidium iodide and calculated as a percentage of PABP^DHFR cells grown in the presence of TMP, along with the SEM. Scale bar, 100 μm.

Figure EV4. Knocking out LSM1 suppresses PABP depletion‐induced cell death.

1. Western blot analysis PABPC4^KO and PABPC4/LSM1^DKO HeLa cells depleted of PABPC1 by siRNA‐mediated knockdown.
2. Phase‐contrast images of PABPC4^KO and PABPC4/LSM1^DKO HeLa cells previously transfected with siRNAs targeting GFP (control) or PABPC1 cultured in the presence or absence of TMP for 48 h. Scale bar, 100 μm. Total numbers of viable cells in were quantified with acridine orange and propidium iodide and graphed as a percentage of PABPC4^KO and PABPC4/LSM1^DKO HeLa cells transfected with siRNAs targeting GFP.

Figure EV5. Impairing the CNOT7 deadenylase does not suppress PABP depletion‐induced cell death.

1. Western blot analysis of PABP^DHFR cells transiently transfected with a plasmid coding for V5‐tagged CNOT7^D40A.
2. Phase‐contrast images of PABP^DHFR cells previously transfected with a plasmid coding for V5‐tagged CNOT7^D40A and cultured in the presence or absence of TMP for 48 h. Scale bar, 100 μm.

The LSM1‐7 complex plays an important role in promoting mRNA decapping following deadenylation (Boeck et al, 1998; Tharun & Parker, 2001). This is in part accomplished by the LSM1‐7 complex binding to the short, often terminally uridylated, oligo (A) tail remaining following deadenylation. As PABP‐depleted cells retained a degree of viability in absence of LSM1, we next assessed how knocking down LSM1 affects mRNA stability in PABP‐depleted cells. PABP^DHFR cells were transiently transfected with siRNAs targeting GFP (control) or LSM1, cultured for 48 h to deplete LSM1 and subsequently cultured in TMP‐containing or TMP‐free media for another 12 h to maintain or deplete PABP (Fig 8A). Cells were then treated with Actinomycin D to inhibit de novo transcription, total RNA was isolated and select mRNAs were quantified by RT‐qPCR. As shown earlier (Fig 5C–E), depleting PABP led to a marked destabilization of GAPDH, UQCC2 and ATP5O mRNAs (Fig 8B). Strikingly, these mRNAs remained stable in PABP‐depleted LSM1 knockdown cells (Fig 8B). We also assessed the stability of MCL1 mRNA, which showed PABP‐independent decay (Fig 5G) and observed that it was still rapidly decayed in LSM1 knockdown cells (Fig 8B). Altogether, these data suggest that the LSM1‐7 complex promotes the decay of stable mRNAs in the absence of PABP.

Figure 8. Disrupting the LSM1‐7 complex prevents deadenylation‐independent decay in PABP‐depleted cells.

• A
  Western blot analysis PABP^DHFR cells depleted of LSM1 by siRNA‐mediated knockdown. Transfected cells were subsequently cultured in the presence or absence of TMP to maintain or deplete endogenous PABP.
• B
  PABP^DHFR cells were transfected with siRNAs targeting GFP (control) or LSM1. Forty‐eight hours later, transfected cells were split and cultured in the presence or absence of TMP for 12 h. The stabilities of select mRNAs was assessed by using actinomycin D (5 μg/ml) for the indicated amounts of time. Total RNA was isolated, reverse transcribed with random hexamer oligonucleotide primers and quantified by qPCR. mRNA decay rates were normalized to an in vitro synthesized spike‐in RNA with the zero time point set at 1. Error bars represent the SEM of three independent experiments.
• C, D
  Poly(A) tail analysis of GAPDH and ACTB mRNAs from total RNA isolated from PABP^DHFR cells described in (B) was determined by extension poly(A) tail (ePAT). PCR products were run on high‐resolution agarose gels. “A₁₂” represents the size of ePAT RT‐PCR amplicons derived from a mRNA with a fixed (A₁₂)‐tail.

As the LSM1‐7 complex normally plays an important role in promoting mRNA decay following deadenylation, we also wished to determine the poly(A) tail length of mRNAs whose abundance decreases in PABP‐depleted cells. To this end, we used the PCR‐based extension poly(A) test (ePAT) to measure the poly(A) tails of GAPDH and ACTB mRNAs in PABP‐expressing and PABP‐depleted cells (Fig 8C and D). ePAT analysis indicated that depleting PABP led to an increase in the poly(A) tail lengths of both ACTB and GAPDH mRNAs. Interestingly, while knocking down LSM1‐stabilized GAPDH in the absence of PABP (Fig 8B), GAPDH and ACTB mRNAs still maintained longer poly(A) tails (Fig 8C and D). All in all, these data suggest that mRNA decapping factors, including the LSM1‐7 complex, promote the decay of select mRNAs in the absence of PABP via a mechanism that bypasses deadenylation.

Discussion

Eukaryotic PABP has been studied extensively using in vitro translation systems to elucidate the modes by which it enhances mRNA translation (Svitkin & Sonenberg, 2004; Kahvejian et al, 2005; Amrani et al, 2008; Svitkin et al, 2009). However, many of the principles regarding how PABP stimulates translation have yet to be directly tested on endogenous mRNAs in mammalian cells. We examined this using genetically engineered HeLa cells allowing us to rapidly deplete PABP in hours as opposed to days. To our surprise, our transcriptome‐wide analysis by polysome profiling did not detect substantial selective changes in mRNA translation in PABP‐depleted HeLa cells. This matches well with ribosome profiling data from the Bartel lab, where depleting PABPC1 and PABPC4 from HeLa cells by RNAi had a minimal transcript‐selective impact on translation efficiency.

While depleting PABP did not lead to widespread changes in mRNA translation, mRNA abundances were dramatically altered. mRNA stability assays also demonstrate that a number of mRNAs that decrease in abundance in PABP‐depleted cells also display accelerated decay rates. Recently, Xiang and Bartel identified mRNAs with short poly(A) tails as being preferentially destabilized in PABP‐depleted cells (Xiang & Bartel, 2021). Our data match well with this, in that mRNAs that are decreased in abundance in PABP‐depleted cells have been reported to maintain shorter poly(A) tails (Fig 4C). Moreover, a number of the transcripts we identified with lower abundance following PABP depletion, such as ribosomal protein‐encoding mRNAs, often maintain short poly(A) tails (Lima et al, 2017). However, our analysis suggests that several other features characterize mRNAs that are destabilized in the absence of PABP. Specifically, we observe that mRNAs decreasing in abundance in PABP‐depleted cells generally contain shorter UTRs and have longer half‐lives than mRNAs whose levels do not decrease following loss of PABP. Interestingly, a number of mRNAs decreasing in abundance have been previously referred to as “antitargets”, non‐targeted mRNAs with 3′UTRs under evolutionary pressure to avoid miRNAs and code for proteins involved in basic cellular processes (Stark et al, 2005). These include mRNAs encoding ribosomal proteins and nuclear‐encoded mitochondrial proteins. It is therefore plausible that PABP plays a role in stabilizing “anti‐target” mRNAs that lack RNA elements (e.g. miRNA‐ and RBP‐target sites) in their 3′UTRs. In support of this hypothesis, our analysis suggests that mRNAs upregulated in absence of PABP have been reported to be enriched in P‐bodies, which contain miRNA‐ and RBP‐targeted mRNAs (Hubstenberger et al, 2017), whereas numerous mRNAs with decreased abundance following PABP depletion are not. Moreover, we observed that depleting PABP did not alter the stability of a miRNA‐targeted reporter mRNA, whereas a stable reporter mRNA containing mutated miRNA target sites displayed accelerated turnover in PABP‐depleted cells (Fig 6). Notwithstanding that PABP depletion did not alter the turnover of a miRNA‐targeted reporter, we cannot rule out that PABP does not play a role in miRNA‐mediated mRNA decay. Indeed, human PABP directly interacts with the miRNA machinery by binding to GW182 (Jinek et al, 2010; Kozlov et al, 2010). Moreover, depleting PABP from mammalian cell‐free extracts abrogates miRNA‐mediated deadenylation (Fabian et al, 2009). Thus, it is possible that PABP‐GW182 contact promotes the decay of a subset of miRNA targeted mRNAs that depend upon this interaction.

Eukaryotic mRNA turnover most often proceeds via deadenylation‐dependent mRNA decay (Goldstrohm & Wickens, 2008). Deadenylation machineries (i.e. PAN2‐3 and CCR4‐NOT) directly interface with PABP and the poly(A) tail, and PABP must be displaced from the poly(A) tail for deadenylation to proceed. The LSM1‐7 complex normally binds the remaining oligo(A) tail following deadenylation, which has often been modified by terminal uridyl transferases (TUT4/7), and subsequently promotes mRNA decapping and decay (Boeck et al, 1998; Tharun & Parker, 2001; Sharif & Conti, 2013). PABP is an essential gene in yeast; however, PABP depletion‐induced lethality can be suppressed by knocking out specific mRNA decay factors, including XRN1 (Caponigro & Parker, 1995). Our data indicate that PABP is critical for mammalian cell viability. However, in contrast to the single PABP gene in yeast (Pab1), our data suggest that mammalian cells can utilize multiple cytoplasmic PABP genes (PABPC1 and PABPC4) to generate a collective pool of PABP in order to support their viability and protein synthesis. In keeping with early studies in yeast, our data also show that depleting mRNA decapping and decay factors (e.g. LSM1, DCP2 or XRN1) partially suppresses PABP depletion‐induced lethality in mammalian cells. In contrast, inhibiting the deadenylase activity of the CCR4‐NOT complex did not rescue cell viability in the absence of PABP. Importantly, GAPDH and ACTB mRNAs maintain longer poly(A) tails in the absence of PABP, a result that matches well previous studies in yeast and mammalian cells, where depleting PABP leads to an increase in poly(A) tail length, as well as an increase in terminal uridylation (Sachs & Davis, 1989; Yi et al, 2018; Xiang & Bartel, 2021). Importantly, GAPDH mRNA remains stable in PABP‐depleted cells where LSM1 has been knocked down, yet still maintains an extended poly(A) tail. This is in contrast to depleting TUT4 and TUT7 in PABP‐depleted cells, which has been reported to restore the GAPDH poly(A) tail to its shorter length (Xiang & Bartel, 2021). As the LSM1‐7 complex binds with high affinity to oligo(U) flanked by short oligo(A) stretches (Rissland & Norbury, 2009), it is possible that in absence of PABP, TUT4/7 urdiylate the termini of mRNAs, thereby providing a docking platform for the LSM1‐7 complex to bind and subsequently engender mRNA decapping and decay in a deadenylation‐independent manner. In conclusion, our analyses suggest that PABP plays an important role in stabilizing select mRNA populations, in particular those that encode proteins with constitutive functions, by preventing their deadenylation‐independent decay.

Materials and Methods

Reagents and Tools table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Gene (Homo Sapiens)	PABPC1		HGNC:HGNC:8554
Cell line (Homo Sapiens)	293T	ATCC	CRL‐3216	Cell line maintained in DMEM + 10% FBS, 50 U/ml of penicillin and 50 µg/ml of streptomycin
Cell line (Homo Sapiens)	HeLa	ATCC	CCL‐2	Cell line maintained in DMEM + 10% FBS, 50 U/ml of penicillin and 50 µg/ml of streptomycin
Antibody	ACTIN	Cell Signalling	4967S	WB (1:30,000)
Antibody	PABPC1	Cell Signalling	4992S	WB (1:1,000)
Antibody	PABPC1	Abcam	ab75855	WB (1:1,000)
Antibody	PABPC4	Bethyl	A301‐467A‐M	WB (1:1,000)
Antibody	LSM1	Abcam	ab229316	WB (1:1,000)
Antibody	DCP2	Bethyl	A302‐597A	WB (1:1,000)
Antibody	XRN1	Bethyl	A300‐443A‐M	WB (1:1,000)
Antibody	HA tag	Bio‐Legend	901513	WB (1:1,000)
Antibody	V5 tag	Cell Signalling	13202S	WB (1:1,000)
Antibody	PUROMYCIN	Sigma‐Aldrich	MABE343	WB (1:25,000)
Antibody	Eif2a	Cell Signalling	L57A5	WB (1:1,000)
Antibody	EIF2a‐P	Abcam	AB32157	WB (1:1,000)
Recombinant DNA reagent	pLEX_307	Addgene	#41392	For all PABPC1 ectopic expression
Recombinant DNA reagent	pcDNA‐DEST40	Thermo Fisher	12274015	For expression of gateway cloned V5‐tagged CNOT7
Recombinant DNA reagent	pDONR221	Thermo‐Fisher
Sequence‐based reagent	gRNA PABPC4	This paper	PCR Primer	FWD: 5′TGTTTGCCCATGGCCTCCCTGTACGT REV: 5′AAACACGTACAGGGAGGCCATGGGCA
Sequence‐based reagent	gRNA PABPC1	This paper	PCR Primer	FWD: 5′CACCGATTTGTACACTTTGAGACGC REV: 5′AAACGCGTCTCAAAGTGTACAAATC
Sequence‐based reagent	PABPC1 degron gRNA1	This paper	PCR Primer	FWD: 5′TGTTTGAAGCAGTTAACAGTGCCAC REV: 5′AAACGTGGCACTGTTAACTGCTTCA
Sequence‐based reagent	PABPC1 degron gRNA2	This paper	PCR Primer	FWD: 5′TGTTTGAAAGCTCACTTTAAACAGT REV: 5′AAACACTGTTTAAAGTGAGCTTTCA
Sequence‐based reagent	DNA donor internal sequences	This paper	PCR Primer	FWD1: 5′CCAGAAAGCAGTTAACAGTGCCACCGGTGTTCCAACTGTTGGAGGCGGTTACCCATAC FWD2: 5′CTGTAGCTGTACTACAAGCCCACCAAGCTAAAGAGGCTGCCCAGAAAGCAGTTAACAGTG FWD3: 5′A*A*CCAGTTTATCTGTTTTCCCCTTTTTTAGGTTGATGAAGCTGTAGCTGTACTACAAGCCREV1: 5′CAAGCTTAAAACAACAAACCAGA GGGAAAAGCTCACTTTAGTCGACTGATCATAATCAGC REV2: 5′TCAAAATATGCTCAACAAACTTTATAAAAGATGAAGAAAACAAGCTTAAAACAACAAACC REV3: 5′A*T*TTAAATGAACTGTAAAATGATCTTTTGCTATGTACATTTCAAAATATGCTCAACAAAC
Sequence‐based reagent	PATL1 gRNA	This paper	PCR Primer	FWD: 5′CACCGGCATCTTCATCTTCATCCAG REV: 5′AAACCTGGATGAAGATGAAGATGCC
Sequence‐based reagent	LSM1 gRNA1	This paper	PCR Primer	FWD: 5′CACCGCTTAGTGCTACATCAGACTG REV: 5′AAACCAGTCTGATGTAGCACTAAGC
Sequence‐based reagent	PABPC1 cloning primers into pDONR221	This paper	PCR Primer	FWD: 5′ggggacaagtttgtacaaaaaagcaggctaccatgggcaaccccagtgcccccagctac rev: 5′ggggaccactttgtacaagaaagctgggtcttaaacagttggaacaccggtggca
Sequence‐based reagent	ePAT Template primer	This paper	RT primer	REV: 5′GCGAGCTGGCGCCGGCGCTTTTTTTTTTTT
Sequence‐based reagent	TVN Template primer	This paper	RT primer	REV: 5′GCGAGCTGGCGCCGGCGCTTTTTTTTTTTTVN
Sequence‐based reagent	ePAT universal primer	This paper	PCR Primer	REV: 5′GCGAGCTGGCGCCGGCGC
Sequence‐based reagent	GAPDH ePAT F	This paper	PCR Primer	FWD: 5′GGACCACCAGCCCCAGCAAG
Sequence‐based reagent	ACTIN ePAT F	This paper	PCR Primer	FWD: 5′AGGGGAGGTGATAGCATTGC
Sequence‐based reagent	human ACTB	This paper	qPCR Primer	FWD: CACCATTGGCAATGAGCGGTTC REV: AGGTCTTTGCGGATGTCCACGT
Sequence‐based reagent	human GAPDH	This paper	qPCR Primer	FWD: 5′catgagaagtatgacaacagcct rev: 5′agtccttccacgataccaaagt
Sequence‐based reagent	human BCL2	This paper	qPCR Primer	FWD: 5′ATCGCCCTGTGGATGACTGAGT REV: 5′GCCAGGAGAAATCAAACAGAGGC
Sequence‐based reagent	human MCL1	This paper	qPCR Primer	FWD: 5′CCAAGAAAGCTGCATCGAACCAT REV: 5′CAGCACATTCCTGATGCCACCT
Sequence‐based reagent	human ATP5O	This paper	qPCR Primer	FWD 5′CTCTCTTCCCACTCGGGTTT REV: 5′TGACCACAGAGGTACTGAAGCA
Sequence‐based reagent	human RPS6	This paper	qPCR Primer	FWD: 5′GCCACAGAAGTTGCTGCTGACG REV: 5′GGTCAAGACACCCTGCTTCATG
Sequence‐based reagent	human NDUFS6	This paper	qPCR Primer	FWD: 5′TGGAGACTCGGGTGATAGCGTG REV: 5′GTGGTGCTGTCTGAACTGGAGC
Sequence‐based reagent	human UQCC2	This paper	qPCR Primer	FWD: 5′CCTGTGATCAGATGTACGAGAGC REV: 5′CTGTGGACAGGATCAGCTTGTAC
Sequence‐based reagent	PABPC1 siRNA	This paper (Yoshida et al, 2006)	CTM‐644746	Sense: 5′G.G.U.G.G.U.U.U.G.U.G.A.U.G.A.A.A. A.U.dT.dT3′ Antisense: 5′A.U.U.U.U.C.A.U.C.A.C.A.A.A.C.C.A.C.C.dT.dT
Sequence‐based reagent	LSM1 siRNA	This paper	L‐005124‐00‐0005	SMARTpool
Sequence‐based reagent	GFP siRNA	Dharmacon	D‐001940‐01‐05	Accell eGFP control siRNA
Sequence‐based reagent	DCP2 siRNA	Dharmacon	L‐008425‐01‐0005	SMARTpool
Sequence‐based reagent	XRN1 siRNA	Dharmacon		Sense: 5′G.A.G.G.U.G.U.U.G.U.U.U.C.G.C.A.U.U.A.U.U Antisense: 5′U.A.A.U.G.C.G.A.A.A.C.A.A.C.A.C.C.U.C.U.U
Commercial assay or kit	GoTaq qPCR Master Mix	Promega	A6001	Reagent for all qPCR assays
Chemical	Trimethoprim	Sigma‐Aldrich	T7883
Chemical	Puromycin	Sigma‐Aldrich	P9620
Chemical	Cycloheximide	Sigma‐Aldrich	C4859
Chemical	Actinomycin D	Sigma‐Aldrich	A1410
Commercial assay or kit	Phusion	ThermoFisher	F630S
Commercial assay or kit	Taq polymerase	Bio Basic	B0089
Commercial assay or kit	RNA extraction KIT	Bio Basic	BS784
Chemical	TRIzol RNA Isolation Reagents	ThermoFisher	0296010
Commercial assay or kit	Klenow polymerase	NEB	M0212S
Commercial assay or kit	SuperScript III	ThermoFisher	18080‐044

Methods and Protocols

Antibodies

Antibodies were purchased from Cell signalling (PABPC1, Actin, V5, eIF2α), Sigma‐Aldrich (Puromycin), Abcam (PABPC1, LSM1, phosphor‐ eIF2α) and Bethyl laboratories (PABPC4, DCP2, XRN1).

DNA constructs

gRNAs targeting human PABPC1, PABPC4 and LSM1 were cloned in LentiCRIPSR v2 (Addgene Plasmid #52961) plasmid by conventional molecular cloning techniques using BsmB1 restriction enzyme sites. To generate PABPC1^DHFR cells knock‐in, two separate gRNAs targeting the sense and antisense strands of PABPC1 were separately cloned into a vector expressing a Cas9 nickase (D10A)‐GFP plasmid (a kind gift from Dr. Jerry Pelletier Lab) by conventional molecular cloning techniques using BsmB1 restriction enzyme sites. A V5‐tagged PABPC1 expression clone was generated by Gateway^® cloning (ThermoFisher) of pDONR‐PABPC1 into a pLEX_307vector (Addgene). Homology directed repair products for generating PABPC1^DHFR cells were generated using pAc5 HA3‐eDHFR‐T2A‐PURO and pAc5 HA3‐eDHFR‐T2A‐HYGRO (Sheridan & Bentley, 2016). pcDNA™‐DEST40 CNOT7 dominant negative construct was generated by introducing D40A/E42A mutations in the pDONR‐CNOT7‐WT using site directed mutagenesis.

Cell lines

Human embryonic kidney HEK293T cells and epithelioid carcinoma HeLa cells were obtained from ATCC. Cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum, 50 U/ml of penicillin and 50 µg/ml of streptomycin. PABPC1^KO HeLa cells and PABPC4^KO HeLa and HEK293 cells were generated by transiently transfecting the WT cells with PABPC1 LentiCRIPSR v2 and PABPC4 LentiCRIPSR v2, respectively. Puromycin (2 µg/ml) was added 24 h post transfection and 48 Hrs later cells were recovered in fresh medium. One week later, monoclonal cells were generated from the pool of cells using 96‐well plates. Medium was changed every 2–3 days. Two weeks later, clones were checked and selected by Western blotting. LSM1/PABPC4^DKO HeLa cells were generated by transiently transfecting the PABPC4^KO HeLa cells with LSM1 LentiCRIPSR v2. The transfection and the selection were done as describe previously. PABPC1^DHFR cells were generated by transiently transfecting PABPC4^KO HeLa cells with PCR repair products, along with two plasmids encoding Cas9 nickases along with PABPC1‐specific gRNAs. Primers used to generate HDF donors are listed in supplementary material.

Transfected cells were cultured in 10 μM TMP (Sigma). Puromycin (2 µg/ml) and Hygromycin (400 µg/ml) were added 72 h after transfection. Single clones were subsequently isolated and screened by Western blotting.

Transfections and siRNAs

Plasmid transfections were performed using polyethyleneimine (PEI). siRNAs (Dharmacon) were used against GFP (control), PABPC1, LSM1, XRN1 and DCP2. siRNAs were transfected were performed using Lipofectamine 2000 (Invitrogen) at a final concentration of 150 nM into HeLa and HEK293 cells for either GFP, PABPC1, LSM1, PATL1, DCP2 and XRN1. Twenty‐four hours post‐transfection, cell culture media was replaced and cells were permitted to grow for an additional 48 h before being harvested.

mRNA stability assays and RT‐qPCR analysis

PABP^DHFR cells were washed twice with PBS and grown in presence or absence of TMP for 12 h to maintain or deplete PABP. Cells were subsequently treated with Actinomycin D (5 µg/ml) for defined periods of time and total RNA was extracted using a EZ‐10 Spin Column RNA Miniprep Kit (Biobasic). Purified RNA was treated Turbo DNase (Invitrogen). 500 ng of RNA was then mixed with 100 pg of an in vitro synthesized RNA (spike‐in for qPCR normalization) prior to being random hexamer‐primed and reverse transcribed using Maxima H Minus Reverse Transcriptase (Thermofisher). qPCR on the generated cDNA was carried out with GoTaq qPCR Master Mix (Promega) and primers (IDT) against endogenous mRNAs and the spike‐in RNA.

Poly(A) tail length analysis

To measure poly(A) tail lengths of ACTB and GAPDH mRNAs, we utilized the extension poly(A) test (ePAT) assay as described previously (Jänicke et al, 2012). The anchor primer used in the reverse transcription step is (GCGAGCTGGCGCCGGCGCTTTTTTTTTTTT). The (dT)₁₂VN primer used in the reverse transcription to generate a size marker for the shortest possible TNV‐PAT product is (GCGAGCTGGCGCCGGCGCTTTTTTTTTTTTVN). One microgram of purified RNA was used for each RT reactions using Superscript III (Invitrogen), which was performed as described in (Jänicke et al, 2012). PCR reactions were then performed using Taq polymerase (Biobasic), resolved on 2% agarose gels stained with ethidium bromide and visualized using an Imagequant LAS 4000 imager (GE Healthcare).

Polysome profiling

PABPC1^DHFR HeLa cells (or 48 h post siRNA (siGFP or siLSM1) transfected PABPC1^DHFR HeLa cells) were seeded in a 15‐cm dish and cultured in the presence or absence of TMP for 12 h, at which point they reached 80% confluency. Cells were treated with 100 µg/ml of cycloheximide (CHX) for 5 min at 37°C. Cells were subsequently washed with PBS including 100 µg/ml CHX and harvested by cell scraping in PBS +100 µg/ml of CHX and collected by spinning down at 350 g for 5 min. PBS was removed and cell pellets were lysed in hypotonic lysis buffer (5 mM of Tris–HCl, pH 7.5; 2.5 mM of MgCl₂, 1.5 mM of KCl; 100 μg/ml of cycloheximide; 2 mM of DTT; 0.5% Triton; 0.5% sodium deoxycholate). Polysome profiling was carried out as described (Gandin et al, 2016). Gradient fractions were collected and RNA was extracted using TRIzol reagent.

Preparation of RNAseq libraries, sequencing and analysis

Biological triplicate libraries were prepared from total cytoplasmic and heavy polysome RNA fractions from PABP^DHFR HeLa cells cultured in the presence of TMP or in the absence of TMP for 12 h. RNA was depleted of ribosomal RNA prior to library generation. Sequencing reactions were carried out by single‐end 75 bp sequencing on a Nextseq500 platform (Genomics Platform at the Institute for Research in Immunology and Cancer, Montreal). Alignment (GRCh38, using HISAT2 2.2.0) and quantification (default parameters) of RNAseq reads were performed using the nf‐core/rnaseq version 3.3 pipeline available at https://github.com/nf‐core/rnaseq (Ewels et al, 2020). The reads were summarized using the featureCounts function in Rsubread version 2.0.1. After removing genes with 0 counts in at least one sample, 14,714 unique genes were represented in the RNAseq data set. The data were normalized and transformed using default settings in anota2seq (Oertlin et al, 2019). During anota2seq analysis, the biological replicate was used as a batch (using the batchVec argument in anota2seq). Changes in translational efficiencies between PABP‐depleted cells and PABP expressing cells were assessed using the anota2seqRun function. Genes were considered significantly regulated when passing default filtering criteria ([maxPAdj < 0.15], [selDeltaPT > log2(1.2)], [minSlopeTranslation > −1], [maxSlopeTranslation < 2], [selDeltaTP > log2(1.2)], [selDeltaP = NULL], [selDetaT = NULL]). The anota2seqRegMode function was used to report genes regulated by altered translation or abundance together with non‐regulated genes (background). The genes regulated via abundance were divided into up‐ or down‐regulated subsets, which were used as input together with the background when characterizing the following features: 5′UTR and 3′UTR lengths (from RefSeq database (O'Leary et al, 2016); when multiple UTR variants are present a random variant is selected); Poly(A) tail length (published mTAIL‐seq data from HeLa cells (Yi et al, 2018); mRNA half‐life (published data set of human RNA half‐lives in Human umbilical vein endothelial cells (HUVEC) (Tiana et al, 2020).

Fluorometric cell viability analysis

Cells were harvested, stained with acridine orange and propidium iodide and live cell numbers were assessed using a CellDrop FL Cell Counter (Denovix).

SUnSET assay

The targeted cells were seeded at 40% confluency a day before the treatment. HeLa cells were treated with 2 µg/ml of puromycin for 30 min, washed twice with PBS and lysed for western blotting. PABP^DHFR HeLa cells were incubated with 10 µg/ml of puromycin prior to lysis. For control reactions, cells were treated with 100 µg/ml of cycloheximide for 5 min prior to treat the cells with puromycin to inhibit mRNA translation elongation. Lysates were resolved by SDS‐PAGE and western blotting was carried out with an anti‐Puromycin antibody.

Gene Ontology analysis

Gene set enrichment analyses for Wikipathway (WP) terms (Slenter et al, 2018) enriched among mRNAs down or upregulated in PABP‐depleted cells were performed using the g:Prolifer online platform (Raudvere et al, 2019).

Author contributions

Sam Kajjo: Conceptualization; Formal analysis; Investigation; Methodology; Writing—original draft; Writing—review & editing. Sahil Sharma: Formal analysis; Writing—original draft; Writing—review & editing. Shan Chen: Formal analysis; Investigation; Methodology; Writing—review & editing. William R. Brothers: Formal analysis; Investigation; Writing—original draft; Writing—review & editing. Megan Cott: Validation; Investigation. Benedeta Hasaj: Formal analysis; Investigation. Predrag Jovanovic: Methodology. Ola Larsson: Formal analysis; Supervision; Investigation; Methodology; Writing—original draft; Writing—review & editing. Marc Fabian: Conceptualization; Formal analysis; Supervision; Investigation; Writing—original draft; Writing—review & editing.

In addition to the CRediT author contributions listed above, the contributions in detail are:

MRF, SK, SS, WRB and BH designed research; SK, SS, MC, PJ and BH performed experiments; SC and OL carried out bioinformatic analyses. All authors contributed to writing and editing the manuscript.

Supporting information

Expanded View Figures PDF

Dataset EV1

Dataset EV2

The EMBO Journal (2022) 41: e108650.

Data availability

RNAseq data generated and analysed in this study are available in the NCBI Gene Expression Omnibus repository under accession number GSE187450 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE187450).