Feedforward miR-181d degradation modulates population variance of methyl-guanine methyl transferase and temozolomide resistance

SUMMARY

Intratumoral heterogeneity plays a pivotal role in cancer evolution, providing the substrate for adaptation to selective pressures, including chemotherapy treatment. Here, we demonstrate that miR-181d modulates variability in methyl-guanine methyl transferase (MGMT) expression, contributing to this heterogeneity in glioblastoma, the most common form of adult primary brain tumor. Treatment with standard-of-care temozolomide (TMZ) chemotherapy triggers a feedforward loop that accelerates polyribonucleotide nucleotidyltransferase 1 (PNPT1)-dependent miR-181d degradation. This degradation requires the activation of ataxia-telangiectasia and Rad3-related (ATR) kinase. The degradation of miR-181d in glioblastoma cells increases the variance of MGMT expression in the cell population, contributing to acquired TMZ resistance. This resistance is suppressed by exogenously transfected miR-181d. These findings suggest that microRNA regulates intratumoral heterogeneity by modulating the transcriptional variability of key DNA repair enzymes, providing a compelling rationale for miRNA delivery as a platform for glioblastoma therapy.

Graphical Abstract

In brief

Singh et al. show that TMZ treatment in glioblastoma induces ATR- and PNPT1-dependent degradation of miR-181d. Loss of miR-181d de-represses PNPT1, establishing a feedforward loop that amplifies TMZ-induced miR-181d degradation. This circuit enhances MGMT expression and cell-to-cell variability, highlighting a mechanism driving resistance to alkylating chemotherapy.

INTRODUCTION

The concepts of mean and variance serve complementary roles in biological research. While the former captures the central tendencies of a trait in populations, the latter defines the degree of spread or dispersion of the data points relative to that central tendency.¹ Studies that aim to compare population differences in particular phenotypes (e.g., the expression of a protein before and after a perturbation) often emphasize the differences in means, using variance as a mathematical tool to interpret these differences.² However, the concept of variance holds independent biological meanings beyond this role. Variance is a key indicator of the complexity of the population and measures differences between the units of the population (e.g., cell-to-cell variability)³ and is a key determinant of intratumoral heterogeneity. Increasing gene expression variance within a population enhances phenotypic diversity to improve the overall fitness of a cell population in response to selective pressures, such as chemotherapeutic treatment.

MicroRNAs (miRNAs) are a class of small (19–24 bp), non-coding RNAs that play two key, post-transcriptional roles in modulating gene expression.⁴ miRNAs bind to complementary sequences on target mRNAs, which can accelerate their degradation and/or inhibition of translation.^5,6 These effects ultimately reduce the mean, steady-state expression of the target mRNA in a cell population. Because both RNA degradation and translational inhibition suppress the likelihood of stochastic bursts in transcription,^7,8 miRNA additionally decreases the variance in gene expression.^9–11 While such suppression enhances the robustness of a cell in maintaining constancy in gene expression, it also reduces the fitness of the population by suppressing transcriptional diversity.¹² The impact on population fitness is amplified when the miRNA target genes, pivotal for cellular survival, respond to selective pressure, such as DNA repair genes in response to DNA-damaging chemotherapy.^13–15

The DNA repair protein, O⁶-methyl-guanine methyl transferase (MGMT), encodes an evolutionarily conserved DNA repair protein that confers resistance to DNA alkylating agents. Temozolomide (TMZ) is a DNA alkylating agent that serves as the standard-of-care chemotherapy for glioblastoma, the most common form of primary brain cancer in adults.¹⁶ TMZ exerts its anti-tumoral effect predominantly by alkylating the O⁶-position of guanine.^17,18 The presence of O⁶-modified guanine in DNA induces DNA replication arrest and accumulation of single-stranded DNA (ssDNA) breaks,¹⁹ triggering DNA damage response (DDR) through activation of the ataxia-telangiectasia and Rad3-related (ATR) kinase.²⁰ MGMT restores O⁶-alkylated guanine to undamaged guanine by transferring the alkyl group to itself.^21,22 Each MGMT molecule serves as an acceptor for a single alkyl transfer.^17,23 Thus, high levels of MGMT are associated with glioblastoma resistance to TMZ.^24,25 Of note, MGMT is post-transcriptionally regulated by a network of miRNAs; the principal among these regulators is miR-181d.^26–28 Suppression of MGMT expression is dependent on specific interactions between miR-181d and the 3’ untranslated region (3′ UTR) of MGMT, abolished by mutagenesis of the miR-181d binding sites. Moreover, this suppression is rescued by a miR-181d-resistant MGMT construct.²⁹ In clinical glioblastoma specimens, high levels of miR-181d were associated with low MGMT expression and improved survival after TMZ treatment.^27,29

Cellular miRNA homeostasis is governed by the equilibrium between its biogenesis pathways and degradation mechanism.³⁰ Relative to the available literature on miRNA biogenesis,^31–33 studies of miRNA degradation remain sparse.^34–36 The human polyribonucleotide nucleotidyltransferase 1 (PNPT1) encodes a phosphorylase essential for the degradation of select miRNAs.³⁷ The encoded protein belongs to an evolutionarily conserved family of 3′-to-5′ exo-ribonucleases that use phosphate to catalyze RNA degradation.^38,39 All PNPase family members harbor two RNase-PH-domains (RPH), an S1-RNA binding domain, and a K-homology domain.³⁹ The crystal structure of bacterial PNPases revealed that the six RPH domains of the PNPase trimer form a doughnut-like structure, with a central channel through which the RNA to be degraded is threaded.⁴⁰ Beyond miRNA degradation, the human PNPT1 plays a pivotal role in the import of RNA into the mitochondria. Separate-of-function PNPT1 mutations have been reported for these distinct functions.^41,42

In this study, we demonstrate that TMZ treatment activates the ATR kinase to initiate a feed-forward PNPT1-miR-181d loop that expedites PNPT1-dependent degradation of miR-181d. The decreased miR-181d level triggered two independent mechanisms that contribute to acquired TMZ resistance: (1) increasing the population mean expression of MGMT, and 2) broadening the cell-to-cell variability in MGMT expression. Overexpression of miR-181d suppressed both resistance mechanisms and synergized with TMZ in tumoricidal activity against glioblastoma. These findings provide a compelling rationale for miRNA delivery-based therapy to suppress chemotherapy-induced genetic heterogeneity that underlies therapeutic resistance.

RESULTS

Decreased miR-181d expression following TMZ treatment

To characterize the changes in miRNA expression in response to TMZ therapy, two patient-derived glioblastoma lines BT-83⁴³ and CMK3⁴⁴ were treated with 500 μM TMZ or vehicle for 6 h. These cell lines were selected because their gene expression profile suggests they capture the two key glioblastoma cell states: BT-83 exhibits an expression pattern suggestive of the mesenchymal-like state, while CMK3 exhibits an expression pattern suggestive of the neural progenitor-like state.⁴⁵ RNA was extracted and profiled by RNA sequencing (Table S1).⁴⁶ In both cell lines, the TMZ treatment did not significantly alter the expression of >95% of the miRNAs (Figures 1A and S1A). miRNAs exhibiting a >2-fold change in expression following TMZ treatment in both lines were individually assessed to determine whether their expression differed in matched pre- and post-TMZ-treated clinical glioblastoma specimens (Figures 1B and S1B). Among the miRNAs evaluated, miR-181d was the only one consistently showing lowered levels in the post-TMZ specimens (Figure 1B, left panel). This finding recapitulated our prior profiling experiment, where miR-181d was downregulated in post-TMZ-treated clinical glioblastoma specimens.⁴⁷

Figure 1. TMZ induces the degradation of miR-181d and upregulation of MGMT.

(A) Scatterplot representing the differential expression analysis of miRNAs from RNA sequencing data of patient-derived BT-83 glioblastoma cells treated with 500 μM TMZ for 6 h. miRNA expression levels in TMZ-treated cells were compared to DMSO-treated cells. miRNAs falling outside the diagonal lines indicate significant changes in expression following TMZ treatment. miR-181d, which shows a notable reduction, is highlighted in red.

(B) Quantification of miR-181d and MGMT expression in 10 matched pairs of newly diagnosed and recurrent formalin-fixed paraffin-embedded glioblastoma surgical specimens. miR-181d and MGMT expression were measured using RT-qPCR.

(C) Immunohistochemistry analysis of MGMT in paraffin-embedded human glioblastoma specimens from (B). MGMT staining intensity was scored and summarized in the table below. Hematoxylin and eosin (H&E) staining represents the even distribution of cells in both newly diagnosed and recurrent glioblastoma specimen sections. Scale bars, 4 μm. MGMT expression: no (−), low (+), intermediate (++), and high (+++).

(D) miR-181d and high MGMT expression in newly diagnosed and recurrent glioblastoma specimens (TCGA). Recurrent glioblastoma exhibited lower miR-181d expression (p = 0.0275) and higher MGMT mRNA expression (p = 0.0001).

(E and F) CMK3 glioblastoma cells were treated with 500 μM TMZ. RNA was isolated and analyzed by RT-qPCR for miR-181d and MGMT (E) as well as miR-181a, miR-181b, and miR-181c (F).

(G) CMK3 glioblastoma cells were transfected with exogenous miR-181d or miR-NT and treated with 500 μM TMZ. RNA was extracted and analyzed by RT-qPCR for miR-181d and MGMT expression. Data are represented as mean ± SD.

Since miR-181d downregulates MGMT expression,^26,29 we next characterized the MGMT mRNA expression in the matched pre- and post-TMZ-treated clinical glioblastoma specimen by quantitative reverse-transcription PCR (RT-qPCR). The lowered levels of miR-181d in the post-TMZ-treated specimens were associated with increased levels of MGMT expression (Figure 1B, right panel). MGMT immunohistochemical staining was also performed using formalin-fixed paraffin-embedded sections derived from three pairs of matched newly diagnosed (pre-TMZ-treatment) and recurrent (post-TMZ-treatment) clinical glioblastoma specimens. Increased MGMT expression was consistently observed in the recurrent tumor sections relative to the newly diagnosed tumor sections in the three matched pairs (Figure 1C).

To further establish the clinical significance of TMZ-induced changes in miR-181d and MGMT, we interrogated data from The Cancer Genome Atlas (TCGA). Our analysis revealed that recurrent glioblastoma specimens exhibited lower levels of miR-181d expression compared to newly diagnosed glioblastoma specimens (p = 0.0275, Figure 1D, left panel). This decrease in miR-181d was associated with an increase in MGMT expression (p = 0.0001, Figure 1D, right panel).

To characterize the effect of TMZ treatment on miR-181d expression in a cell-based system, CMK3 cells were treated with TMZ and analyzed by RT-qPCR. This analysis revealed a time-dependent reduction in miR-181d levels (Figure 1E, left panel), with a >50% decrease in miR-181d observed within 4 h of TMZ treatment. This observation was confirmed using the BT-83 glioblastoma line (Figure S2A). In both cell lines, the decrease in miR-181d level was associated with a concomitant increase in MGMT mRNA expression (Figure 1E (right panel) and S2B). As a control, the expression of another miRNA, miR-21, remained unaffected by TMZ treatment (Figure S2C). TMZ treatment did not reduce the expression levels of other miR-181 family members, namely miR-181a, b, and c (Figures 1F and S2D). Finally, exogenous expression of miR-181d mimic maintained miR-181d level after TMZ treatment (Figures 1G [left panel] and S2E) and suppressed TMZ-induced increase in MGMT expression (Figures 1G [right panel] and S2F). These findings support a key role for miR-181d in TMZ-induced MGMT upregulation. Notably, miR-181d transfection did not alter the MGMT promoter methylation status in CMK3 cells (Figure S2G).

PNPT1 is required for miR-181d degradation

A decrease in the steady-state level of miR-181d following TMZ treatment could result from either suppressed biogenesis or increased degradation. To distinguish between these possibilities, the effect of TMZ on both precursor (pre) and mature (mat) miR-181d was examined. The level of pre-miR-181d remained unchanged in response to TMZ treatment in CMK3 cells. In contrast, the level of mat-miR-181d significantly decreased after treatment (Figure 2A). Similar results were observed in the BT-83 cells (Figure S3A). These observations suggest that the reduced level of miR-181d resulted from a degradative process rather than suppressed biogenesis. Supporting this hypothesis, inhibition of transcription by pre-treatment with actinomycin D1^48,49 did not further reduce the miR-181d levels in response to TMZ (Figure 2B). As control, we show that actinomycin D at the applied concentration was sufficient to inhibit Myc mRNA biogenesis (Figure S3B).

Figure 2. PNPT1 is required for TMZ-induced degradation of miR-181d.

(A) CMK3 cells were treated with 500 μM TMZ for up to 24 h. RNA was isolated, and the precursor (pre) or mature (mat) miR-181d expression was analyzed using RT-qPCR.

(B) CMK3 cells were pretreated with actinomycin D1 (100, 250, and 500 nM) and exposed to TMZ (500 μM) for 24 h. Mature (mat) miR-181d expression was analyzed by RT-qPCR.

(C) CMK3 cells transfected with siExoSC4, siPNPT1, or siXRN1 and miR-181d expression were analyzed by RT-qPCR. Statistical significance, *p < 0.05.

(D and E) Exogenous expression of PNPT1 restored TMZ-induced miR-181d degradation. CMK3 WT or PNPT1-KO cells were transfected with pCMV6-AC-GFP-PNPT1 and exposed to TMZ or DMSO. miR-181d (D) or MGMT (E) expression was analyzed by RT-qPCR.

(F) CMK3 cells transfected with biotinylated (Bi)-miR-181d or miR-NT control mimic. Bi-miRNA was affinity pull-down (AP) with streptavidin-magnetic beads. The associated protein complex was eluted and analyzed by western blotting. Total cell lysate was used as input.

(G and H) PNPT1 protein was immunoprecipitated and incubated with miR-181d or miR-21 mimic (1 × 10⁵ copies). RNA was extracted and quantified by RT-qPCR for miR-181d (G) and miR-21 (H).

(I) Affinity-purified PNPT1 was isolated from CMK3 cells with and without TMZ treatment and co-incubated with miR-181d. Statistical significance: ***p < 0.001 between the same treatment groups.

(J and K) CMK3 cells were treated with 500 μM TMZ or DMSO and fractionated into cytoplasmic and mitochondrial fractions. These fractions were analyzed for PNPT1 localization by western blotting. Alpha (α)-tubulin was used as the cytoplasmic, and succinate dehydrogenase (SDH) was used as the mitochondrial control (J, left). Densitometry quantitation of PNPT1 in cytoplasmic and mitochondrial fractions (J, right). (K) Total miRNA was isolated from these fractions and analyzed for miR-181d by RT-qPCR. Statistical significance: ***p < 0.001 between the same treatment groups. Data are represented as mean ± SD.

To elucidate the mechanism underlying TMZ-induced miR-181d degradation, a small interfering (si)RNA screen was conducted targeting RNases (ExoSC4, PNPT1, and XRN1) that have been previously implicated in miRNA degradation.³⁷ Among these RNase-targeting siRNAs, only those directed against PNPT1 caused an increase in the steady-state miR-181d levels (a 4-fold increase, Figure 2C). This effect was confirmed using silencing and CRISPR-Cas9 disruption of PNPT1. In both CMK3 and BT-83, PNPT1-silencing (using a siRNA distinct from that used in the screen) induced a 2-fold increase in steady-state miR-181d level (Figures S3C and S3E) and suppressed TMZ-induced upregulation of MGMT (Figures S3D and S3F). PNPT knockout (KO) (Figures S4A and S4B) caused a >100-fold increase in miR-181d level (Figure S4C) and downregulation of MGMT (Figure S4D). PNPT1-KO (Figure S4E) also prevented the TMZ-induced reduction in miR-181d levels (Figure S4F) and MGMT upregulation (Figure S4G).

Additionally, ectopic overexpression of PNPT1 in both wild-type (WT) or PNPT1-KO CMK3 cells reduced the steady-state levels of miR-181d (Figure 2D), with an associated rise in MGMT mRNA expression (Figure 2E). Moreover, such overexpression enhanced TMZ-induced decrease in miR-181d (Figure 2D), with an associated increase in MGMT (Figure 2E). To investigate the interaction between miR-181d and PNPT1, CMK3 cells were transfected with biotinylated (Bi)-miR-181d or Bi-miR-NT mimic. A streptavidin-biotin pull-down assay was performed, followed by detection of PNPT1 using western blotting. The results demonstrated that PNPT1 co-precipitated with Bi-miR-181d in the streptavidin pull-down (Figure 2F).

To determine whether PNPT1 is required for miR-181d degradation, an in vitro miRNA degradation assay was performed. The PNPT1 protein was immunoprecipitated (IP) using an anti-PNPT1 antibody from CMK3 lysates (Figure 2G) and then incubated with either mat-miR-181d or miR-21 mimic. The miRNAs were subsequently isolated from the reaction mixture and analyzed by RT-qPCR. The results demonstrated a time-dependent decrease in miR-181d level (Figure 2G). Such a decrease was not observed with miR-21 mimic (Figure 2H). Moreover, the reduction in miR-181d level was more pronounced when incubated with affinity-purified PNPT1 derived from CMK3 treated with TMZ for 24 h than when using affinity-purified PNPT derived from CMK3 treated with the vehicle (Figure 2I).

PNPT1 is an enzyme localized to both the mitochondrial and cytosolic compartments.⁵⁰ We next determined whether TMZ treatment alters the sub-cellular localization of PNPT1. Without TMZ treatment, PNPT1 localizes to both the cytoplasm and the mitochondria at a ratio of 70% mitochondria and 30% cytoplasm (Figure 2J, lanes 1 and 2). After TMZ treatment, this ratio is flipped with a ratio of 30% mitochondria:70% in cytoplasm (Figure 2J, lanes 3 and 4). These findings are consistent with a previous report demonstrating PNPT1 re-localization to the cytoplasm in response to environmental stress.⁵¹ This shift in sub-cellular localization was associated with a decrease in the cytoplasmic miR-181d level (Figure 2K). These findings suggest that TMZ-induced re-localization of PNPT1 into the cytoplasmic compartment facilitated the degradation of miR-181d (and subsequent de-repression of MGMT expression).

PNPT1 modulates TMZ sensitivity through miR-181d and MGMT regulation

The effect of PNPT1 on TMZ sensitivity was evaluated by the limiting dilution neurosphere-forming assays.^52,53 CMK3 cells were transfected with PNPT1-siRNAs and treated with TMZ (500 μM) or vehicle to determine the clonogenic potential. The results demonstrated that silencing of PNPT1 (Figure S5A) caused a nearly two orders of magnitude increase in TMZ sensitivity (Figure S5B, top panel). Similarly, transfection of miR-181d mimic or MGMT-siRNA into CMK3 caused an approximate two orders of magnitude increase in TMZ sensitivity (Figure S5B, middle and bottom panel). siPNPT1-mediated increase in TMZ sensitivity can be blocked by an anti-miR against miR-181d (Figure S5C, top panel) or MGMT overexpression (Figure S5C, bottom panel). These results support a genetic pathway in which PNPT1 acts on miR-181d to regulate MGMT expression levels and modulate TMZ sensitivity.

Mutations inactivating PNPT1 ribonuclease activity suppressed its effects on miR-181d

Next, we determined the expression of PNPT1 in clinical glioblastoma specimens. PNPT1 expression was elevated by 2- to 25-fold in tumor specimens relative to the adjacent brain tissue at the mRNA (Figure 3A) and protein levels (Figure 3B). PNPT1 is a multi-functional protein comprising two RNase-PH (RPH) domains, two COOH-terminal RNA binding domains: an S1-RNA binding and a K-homology domain.⁵⁴ Separation-of-function mutations between its two key activities, mitochondrial RNA import and exo-ribonuclease activity, have been described.⁴¹ Specifically, mutations S135G, Q387R, and S484A disrupted PNPT1 mitochondrial function without significantly affecting ribonuclease activity,^51,55,56 while mutations R445E, R446E, D538A, and D544G inactivated the exo-ribonuclease activity⁵⁵ (Figure 3C, top panel). To assess the impact of these mutations on miR-181d degradation, expression constructs of PNPT1-WT and mutants were transfected into CMK3 (as a control) or two independent clones of PNPT1-KO lines. Expression of WT-PNPT1 in the PNPT1-KO lines suppressed the steady-state miR-181d level by approximately an order of magnitude (Figure 3C, bottom panel). Transfection of PNPT1 constructs bearing the S135G, Q387R, or S484A (mutations that do not affect the exo-ribonuclease activity) resulted in suppression of steady-state miR-181d level comparable to WT-PNTP1. In contrast, constructs bearing the R445E, R446E, D538A, or D544G mutations (mutations that disrupted the exo-ribonuclease activity) did not significantly affect steady-state miR-181d level (Figure 3C, bottom panel).

Figure 3. Ribonuclease activity of PNPT1-dependent degradation of miR-181d.

PNPT1 expression analysis in clinical glioblastoma specimens. RNA (n = 5) and protein (n = 3) extracted from matched adjacent brain tissue specimens were analyzed for PNPT1.

(A) PNPT1 mRNA expression by RT-qPCR. Statistical significance: *p < 0.05 and ***p < 0.001.

(B) PNPT1 protein expression by western blotting.

(C) Left: PNPT1 protein structure labeled with numbers represents the specific position for the amino acid substitution mutation. Right table: summary of amino acid substitution within PNPT1 domains. Bottom: CMK3 WT and PNPT1-KO clones (PNPT1-KO1 and PNPT1-KO2) were transfected with either WT or mutated PNPT1 constructs. miR-181d expression was quantified by RT-qPCR.

(D) In vitro miR-181d degradation assay by immunoprecipitated WT or mutant PNPT1. GFP-tagged PNPT1 protein was immunoprecipitated using an anti-GFP antibody from WT or D538A mutant PNPT1-expressing lines and incubated with miR-181d. RNA was extracted and quantified by RT-qPCR. Data are represented as mean ± SD.

To further establish the involvement of PNPT1’s exo-ribonuclease activity in miR-181d degradation, GFP-tagged WT-PNPT1 or the D538A mutant was affinity-purified from stably expressing CMK3 PNPT1-KO cells and incubated with mature miR-181d mimic. The level of miR-181d remained unchanged when incubated with the affinity-purified fraction from PNPT1-D538A, while a significant reduction of miR-181d was observed when incubated with affinity-purified fraction from WT-PNPT1 (Figure 3D). These results suggest that PNPT1 exo-ribonuclease activity contributes to the degradation of miR-181d.

PNPT1 is post-translationally regulated by miR-181d

Sequence analysis of the PNPT1–3′ UTR revealed three potential miRNA response elements (MREs) for miR-181d, depicted as MRE1 (705–729 bp), MRE2 (1250–1274 bp), and MRE3 (1464–1487 bp) (Figure 4A). This raises the possibility that PNPT1 is regulated by miR-181d in a feedforward loop, where PNPT1 expression increases following TMZ-induced miR-181d degradation. Supporting this hypothesis, PNPT1 expression increased in response to TMZ treatment at both the RNA (Figure 4B) and protein (Figure 4C) levels. Further supporting this hypothesis, increased PNPT1 mRNA expression was observed in matched pre- and post-TMZ-treated clinical glioblastoma specimens (Figure 4D).

Figure 4. miR-181d regulates PNPT1 expression.

(A) miR-181d miRNA response elements (MREs) in PNPT1 3′ UTR. Pink block: predicted MREs of miR-181d (MRE1: 705–729, MRE2: 1250–1274, and MRE3: 1464–1487 bp). miR-181d binding sites within the MREs and mutated-MREs that disrupted miR-181d are depicted below the schema.

(B) CMK3 cells were treated with TMZ or DMSO. Total RNA was extracted, and PNPT1 expression was analyzed using RT-qPCR. GAPDH was used as the control.

(C) PNPT1 protein expression by western blotting in CMK3 cells used in (B).

(D) PNPT1 mRNA expression was quantified in 10 matched pairs of newly diagnosed and recurrent glioblastoma specimens using RT-qPCR.

(E and F) PNPT1 mRNA (E) and PNPT1 protein expressions (F) were assessed after CMK3 cells were transfected with miR-181d or miR-NT mimic. **p < 0.01 indicates statistically significant differences compared to miR-NT.

(G) CMK3 cells transfected with anti-miR-181d or anti-miR-NT. The cell lysate was analyzed for PNPT1 expression by western blotting.

(H) CMK3 cells transfected with Bi-miR-181d or -NT mimic. The lysate was affinity-purified with streptavidin-magnetic beads. Isolated RNA was analyzed for PNPT1 by RT-qPCR. Statistical significance **p < 0.01.

(I) Luciferase reporter assay of miR-181d MREs and its mutants in PNPT1 3′ UTR with miR-181d or -NT mimic. Statistical significance *p < 0.05. Data are represented as mean ± SD.

To further substantiate miR-181d regulation of PNPT1, CMK3 cells were transfected with miR-181d or miR-NT mimic. 48 h post-transfection, RT-qPCR detected ∼80% reduction in PNPT1 mRNA (Figure 4E) and protein expression (Figure 4F). In addition, transfection of anti-miR-181d into CMK3 cells (expressing high levels of miR-181d) increased PNPT1 protein expression. This effect was not observed after the transfection of a control anti-miR-NT (Figure 4G). Next, we determined whether PNPT1 mRNA binds to miR-181d. CMK3 cells were transfected with Bi-miR-181d or Bi-miR-NT mimic. A streptavidin-biotin pull-down assay was performed, followed by the detection of PNPT1 mRNA by RT-qPCR. The analyses detected approximately a 5-fold enrichment of PNPT1 mRNA with the Bi-miR-181d pull-down relative to the Bi-miR-NT pull-down (Figure 4H).

To confirm that PNPT1 is a target of miR-181d, each of the three miR-181d MREs was cloned into a pSiCheck-2 reporter construct. Luciferase activity of a pSiCheck-2 reporter construct bearing the first miR-181d MRE1 was unaffected by the introduction of miR-181d. However, the construct harboring the miR-181d MRE2 and miR-181d MRE3 exhibited ∼20% and ∼30% reduction in luciferase activity, respectively, upon introduction of miR-181d (Figure 4I). Mutations in these MRE sites (mut-MRE2 and mut-MRE3) that disrupted complementarity to miR-181d abolished these effects.

Our data suggested that miR-181d downregulates PNPT1 expression by direct binding of its 3′ UTR. Supporting this hypothesis, exogenous expression of a cDNA encoding a Myc-FLAG-tagged PNPT1 (lacking the PNPT1-3′ UTR) rescued the PNPT1 suppressive effect of miR-181d (Figure 5A, compare lanes 3 and 2, Myc-PNPT1 transcripts). The expression of Myc-FLAG-PNTP1 was assessed using both RT-qPCR (Figure 5A) and FLAG western blotting (Figure 5B). Importantly, Bi-miR-181d pulled down only endogenous PNPT1 (with WT 3′ UTR) and not the Myc-FLGA-tagged PNTP1 (without the 3′ UTR, Figure 5C, comparing lanes 3 and 4). PNPT1 expression in a cell line that exogenously expressed Myc-FLAG-PNPT1 with MRE2 added to the 3′ UTR and transfected with miR-181d is lower than in a similarly treated cell line exogenously expressing Myc-FLAG-PNPT1 cDNA without the MRE2 sequence (Figure 5A [compare lanes 4 and 3] and 5B [compare lanes 4 and 3]). This finding suggests that the MRE2 sequence interacts with miR-181d to target the Myc-FLAG-PNPT1 transcript for degradation. PNPT1 expression is restored when MRE2 is mutated to abolish miR-181d binding (Figure 5A [compare lanes 4 and 5] and 5B [compare lanes 4 and 5]). Importantly, Bi-miR-181d pulled down only Myc-FLAG-PNPT1-MRE2 and not the Myc-FLAG-PNPT1-mutated-MRE2 (Figure 5D). A similar pattern of results was observed for MRE3. PNPT1 expression in a cell line that exogenously expressed Myc-FLAG-PNPT1 with MRE3 added to the 3′ UTR and transfected with miR-181d is lower than in a similarly treated cell line exogenously expressing Myc-FLAG-PNPT1 cDNA without the MRE3 sequence (Figures 5A [compare lanes 6 and 3] and 5B, [compare lanes 6 and 3]), suggesting that the MRE3 sequence interacts with miR-181d to target the Myc-FLAG-PNPT1 transcript for degradation. PNPT1 expression is restored when MRE3 is mutated to abolish miR-181d binding (Figures 5A [compare lanes 6 and 7] and 5B [compare lanes 6 and 7]). Bi-miR-181d pulled down only Myc-FLAG-PNPT1-MRE3 and not the Myc-FLAG-PNPT1-mutated-MRE3 (Figure 5E). These rescue and mutagenesis experiments support specific interactions between miR-181d and the 3′ UTR of PNPT1.

Figure 5. miR-181d binds to the PNPT1 3′ UTR and regulates PNPT1 expression.

(A and B) CMK3 cells were transfected with Myc-FLAG tagged PNPT1 cDNA construct with or without miR-181d MRE2, MRE3, mutated-MRE2, or mutated-MRE3 (disrupting miR-181d binding) as 3′ UTR. Total RNA and protein were extracted 24 h after transfection. PNPT1 mRNA expression was analyzed by RT-qPCR using primers specific to the endogenous (Endo)-PNPT1 or Myc-FLAG-PNPT1 (A). The expression of Myc-FLAG-PNPT1 was analyzed by western blotting. GAPDH used as loading control (B).

(C) CMK3 cells expressing Myc-FLAG-PNPT1 were transfected with Bi-miR181d or Bi-miR-NT. The lysate was affinity-purified with streptavidin-magnetic beads. Isolated RNA was analyzed for endo-PNPT1 or Myc-FLAG-PNPT1 mRNAs. Statistical significance ***p < 0.001.

(D and E) CMK3 cells expressing Myc-FLAG-PNPT1 cDNA construct containing MRE2, MRE3, mut-MRE2, or mut-MRE3 were transfected with Bi-miR-181d mimic. The lysate was affinity-purified with streptavidin-magnetic beads. Isolated RNA was analyzed for Myc-FLAG-PNPT1 transcripts. (D) MRE2 and mut-MRE2

(E) MRE3 and mut-MRE3. Statistical significance ***p < 0.001. Data are represented as mean ± SD.

ATR is required for TMZ-induced miR-181d degradation

TMZ treatment initiates DDR by activating the ATR kinase.⁵⁷ To investigate whether ATR activation is necessary for TMZ-induced miR-181d degradation, CMK3 cells were pretreated with the ATR inhibitor VE-821,⁵⁸ followed by TMZ exposure, and miR-181d expression levels were assessed. While TMZ treatment induced a time-dependent decrease in miR-181d, this reduction was suppressed by VE-821 (Figure 6A).

Figure 6. PNPT1 required ATR for TMZ-induced miR-181d degradation.

(A) CMK3 cells were pretreated with ATR inhibitor (VE-821, 10 μM, 12 h) before TMZ treatment. miR-181d level was measured at the indicated time points using RT-qPCR.

(B and C) Doxycycline (Doxy)-inducible shATR-expressing with or without PNPT1 overexpression in CMK3 cells were treated with TMZ or DMSO. miR-181d (B) and MGMT (C) expressions were analyzed by RT-qPCR.

(D) siATR or siNT was transfected into CMK3 WT or PNPT1-KO cells. miR-181d expression was analyzed by RT-qPCR.

(E) Doxy-induced shATR were transfected with Bi-miR-181d mimic. Cell lysates were affinity-purified with streptavidin and analyzed by western blotting.

(F) shATR induced CMK3 cells treated with TMZ or DMSO. PNPT1 was immunoprecipitated from the cell lysate and incubated with miR-181d mimic for the degradation assay. miR-181d was quantified by RT-qPCR. Data are represented as mean ± SD.

Next, we tested the interaction between ATR and PNPT1 in their regulation of miR-181d. To this end, we generated a CMK3 line harboring the doxycycline-inducible ATR-silencing (shATR, short hairpin ATR) construct that overexpresses PNPT1. ATR-silencing suppressed TMZ-induced miR-181d degradation (Figure 6B, lane 4 versus lane 3), while PNPT1 overexpression enhanced miR-181d degradation phenotype (Figure 6B, lane 6 versus lane 2). Notably, ATR-silencing in the cell line that overexpressed PNPT1 showed no evidence of TMZ-induced miR-181d degradation (Figure 6B, lane 8 versus lane 7), suggesting that ATR is required for PNPT1-mediated miR-181d degradation, with associated increased MGMT expression (Figure 6C). To further evaluate the genetic interaction between PNPT1 and ATR, CMK3, or CMK3-PNPT1-KO lines were transfected with siATR or siNT control. In CMK3 cells, siATR transfection resulted in an order of magnitude increase in miR-181d than in siNT-transfected cells, while such a difference was not observed in the PNPT1-KO line (Figure 6D). These results suggest that the effect of ATR on miR-181d requires PNPT1.

To determine whether ATR is required for binding of PNPT1 to miR-181d, the CMK3 line harboring the doxycycline-inducible shATR was transfected with Bi-miR-181d or Bi-NT and treated with TMZ or vehicle. Streptavidin pull-down was performed using lysate derived from doxycycline-treated or control-treated cells, followed by western blotting with an anti-PNPT1 antibody. Without doxycycline induction of ATR-silencing, PNPT1 was detected in the Bi-miR-181d pull-down, with or without TMZ treatment (Figure 6E). However, upon doxycycline treatment-induced ATR-silencing, PNPT1 was not detected in the Bi-miR-181d pull-down (Figure 6E), suggesting that ATR is required for PNPT1-miR-181d interaction.

To determine whether ATR is essential for PNPT1-mediated degradation of miR-181d, we treated CMK3 harboring doxycycline-inducible shATR and GFP-PNPT1 with either doxycycline or DMSO. After 24 h, the doxycycline or DMSO cells were subjected to TMZ or DMSO treatment. Extracts were then prepared from cells treated with these four conditions (DMSO+DMSO, DMSO+TMZ, doxycycline + DMSO, and doxycycline + TMZ), and GFP-PNPT1 was affinity-purified as described for Figure 3D. The affinity-purified GFP-PNPT1 was then incubated with synthetic miR-181d to determine its degradative activity. A lower level of miR-181d was detected after incubation with affinity-purified GFP-PNPT1 isolated from DMSO+DMSO-treated cells relative to the reactions incubated with IgG pull-down (Figure 6F, lane 1: GFP versus IgG), suggesting that the PNPT1 immunoprecipitate harbored miR-181d degradative activity. The level of miR-181d was further lowered in the reaction with GFP-PNPT1 immunoprecipitate isolated from DMSO+TMZ-treated cells, suggesting that TMZ treatment enhanced the degradative activity of the PNPT1 immunoprecipitate (Figure 6F, lane 2: GFP versus lane 1). In the reactions involving GFP-PNPT1 immuno-purified from doxycycline-treated cells (i.e., ATR-silenced), we did not observe a decrease in miR-181d level relative to the reaction with IgG pull-down (Figure 6F, lane 3: GFP and lane 4: GFP), suggesting that ATR is required for PNPT1-mediated degradation of miR-181d.

miR-181d degradation enhanced cell-to-cell variability and TMZ resistance

miRNAs suppress transcriptional variance in gene expression.⁵⁹ To determine whether this suppression influences cell-to-cell variability, we treated BT-83 glioblastoma cells with TMZ or DMSO. Single-cell RT-qPCR was performed on approximately 80 cells per treatment group to assess miR-181d and MGMT. Consistent with our hypothesis that TMZ induces degradation of miR-181d, the population mean of miR-181d expression is reduced after TMZ treatment (Figure 7A, top panel). This reduction is associated with an increased population mean of MGMT expression (Figure 7A, bottom panel). Notably, while the variance in miR-181d expression remained unchanged in response to TMZ, the variance of MGMT increased by a factor of five after TMZ treatment (p = 0.001). Transfection of a control miRNA did not alter the distributional mean or variance of miR-181d expression (Figure 7B, top panel). On the other hand, transfection of miR-181d mimic increased the distributional mean of miR-181d expression without significantly changing its variance (Figure 7B, top panel). In terms of MGMT expression, miR-181d transfection reduced the distributional mean without affecting the variance of MGMT expression (Figure 7B, bottom panel). Together, these results suggest that miR-181d modulates the population variance of MGMT expression.

Figure 7. miR-181d-mediated cell-to-cell variation of MGMT increased TMZ resistance.

(A) TMZ reduced miR-181d expression and increased the variation in MGMT expression. BT-83 cells were treated with TMZ or DMSO. Single-cell RT-qPCR was performed on approximately 80 cells per treatment group to assess miR-181d and MGMT, with the data presented as a distribution. Black: DMSO and Red: TMZ treatment.

(B) miR-181d transfection suppressed TMZ-induced variability in MGMT expression. The cells were transfected with miR-181d before TMZ treatment and single-cell RT-qPCR.

(C) Kaplan-Meier survival curves of nude mice bearing intracranial BT-83 cells with low-variance (LV) and high-variance (HV) in MGMT expression. The mice underwent the intraperitoneal administration of TMZ at 50 mg/kg/day for 5 days, followed by a 23-day treatment interruption. TMZ treatment was initiated 7 days post-tumor implantation. Each group consisted of 10 mice.

(D and E) The low- and high-variance cells were transfected with miR-181d and implanted into nude mice. The mice were treated with TMZ as described (C). The study was continued for 90 days. Low- (D) and high-variance (E). Data are represented as mean ± SD.

Next, we tested whether variability in MGMT expression translates to altered TMZ resistance. To this end, we isolated BT-83 subclones that varied in MGMT expression and pooled these subclones to generate populations with comparable MGMT expression but with either low- or high-variance in MGMT expression. For the low-variance group, we pooled seven clones exhibiting similar MGMT expression levels (Figure S6A). The high-variance group was created by pooling three clones with MGMT expression levels comparable to the low-variance group, along with two clones exhibiting higher and two clones with lower MGMT expression (Figure S6A). RT-qPCR analysis confirmed that the mean MGMT mRNA expression levels were comparable between the low- and high-variance pools (Figure S6B). However, western blotting shows that MGMT protein expression is slightly higher in the low-variance group (Figure S6C). The miR-181d expression in the two pools is comparable (Figure S6D). The baseline clonogenic potential of the high- and low-variance pool was comparable (Figure S6E). However, the high-variance pool showed increased TMZ resistance relative to the low-variance pool (Figure S6E).

To validate these results in vivo, the high- and low-variance pools were orthotopically implanted into nude mice. The mice were treated with TMZ to assess the impact of MGMT variance on therapeutic response. Without TMZ treatment, the mice bearing low-variance and high-variance MGMT expression pools exhibited similar median survival of 38 and 37 days, respectively (Figure 7C). After TMZ treatment, the median survival for mice implanted with the low-variance MGMT expression pool extended to 72 days. In contrast, those implanted with the high-variance pool demonstrated a significantly shorter median survival of 53 days (p < 0.0001), indicating increased resistance to TMZ in the high-variance group (hence prolonged survival) relative to the low-variance MGMT pool. This finding is notable given that the populational MGMT expression level in the low-variance group is slightly higher than that of the high-variance group (Figure S6C), supporting a role for MGMT expression variance in TMZ resistance.

The general survival pattern observed in Figure 7C closely correlated with that observed in bioluminescence studies of tumor burden dynamics (Figure S7A), low- and high-variance pooled BT-83 were transduced with a luciferase construct and orthotopically implanted. Tumor burden dynamics were followed using bioluminescence. Without TMZ treatment, the low- and high-variance pools exhibited comparable time-dependent increase in bioluminescence, suggesting a similar rate of growth in vivo. In response to TMZ treatment, the low-variance pool exhibited a slower rise in bioluminescence relative to the high-variance pool (p = 0.02195 on day 14 and p = 0.00043 on day 21), suggesting increased TMZ resistance of the high-variance pools.

Next, we tested the effect of miR-181d transfection on the TMZ sensitivity of the low-variance population (Figure 7D). Mice implanted with the low-variance pool transfected with miR-181d had a median survival of 52 days, significantly longer than in mice transfected with control miRNA (37 days, p = 0.00135). This finding supports previous reports that miR-181d suppressed gene expression required for survival and proliferation, including K-Ras and BCL-2.⁶⁰ TMZ treatment of those mice transfected with miR-NT increased median survival to 68 days (p = 0.00118). TMZ treatment in the miR-181d-transfected group further prolonged survival, with all but one mouse surviving the planned study endpoint of 90 days (p < 0.0001 compared to TMZ-treated mice implanted with the miR-NT-transfected tumor).

Further, we assessed the impact of miR-181d transfection on TMZ sensitivity of the high-variance populations. Mice implanted with the high-variance pool transfected with miR-181d exhibited a median survival of 70 days, significantly longer than those transfected with miR-NT (42 days, p < 0.0001, Figure 7E). TMZ treatment of those mice transfected with miR-NT increased median survival to 61 days (p < 0.0001). Notably, TMZ treatment in the miR-181d-transfected group further prolonged survival, with all but one mouse surviving the planned study endpoint of 90 days (p < 0.0001 compared to TMZ-treated mice implanted with miR-NT-transfected tumor).

Since the experiment using the low-variance pool was conducted separately from the experiment using the high-variance pool, a direct, quantitative comparison between the two experiments is not possible. However, it is notable that the median survival of the cohort with TMZ-treated low-variance population is longer than that of the cohort treated with miR-181d. In contrast, the median survival of the cohort with TMZ-treated high-variance population is approximately the same as that of the cohort treated with miR-181d. These results are consistent with our observation that the high-variance pool is more refractory to the cytotoxic effect of TMZ than the low-variance pool.

The observations made in the murine survival models closely correlated with those observed in bioluminescence studies of tumor burden dynamics (Figures S7B and S7C). The time-dependent rise in bioluminescence observed in both low- and high-variance pools was attenuated by either TMZ treatment or miR-181d transfection. While the high-variance group exhibited a faster rise in bioluminescence relative to the low-variance pool, the combined miR-181d+TMZ intervention produced comparable inhibitory effects in both high- and low-variance pools.

DISCUSSION

Transcriptional variability contributes to phenotypic diversity that underlies populational fitness to selective pressures, such as exposure of cancer cells to chemotherapy.⁶¹ miRNAs suppress such transcriptional variability.^11,62,63 Results from this study demonstrate that DNA-damaging chemotherapy treatment induced a feedforward loop to expedite miRNA degradation, enhancing cell-to-cell variability in the expression of a critical DNA repair protein. Specifically, treatment of glioblastoma cells with the chemotherapy, TMZ, triggers an ATR⁶⁴ and PNPT1-dependent degradation of miR-181d. Since PNPT1 is itself a target of miR-181d, miR-181d degradation de-repress PNPT1, leading to a feedforward loop that amplifies TMZ-induced miR-181d degradation, resulting in both an increased mean expression of MGMT as well as the cell-to-cell variability of MGMT expression. The importance of the former to TMZ resistance is well-established.⁶⁵ Our study demonstrates that the latter also contributes to acquired TMZ resistance. This result is intuitive since the higher MGMT-expressing clones in the high-variance pool are more likely to survive TMZ treatment and proliferate subsequently. Moreover, our results indicate that both mechanisms of TMZ resistance (increased mean and variance of MGMT expression) can be suppressed by miR-181d overexpression, suggesting potential as a therapeutic strategy. As therapeutic resistance is multifaceted, the proposed strategy targets only one component of a broader regulatory network.

The varied levels of MGMT expression in glioblastoma cells bear implications for intratumoral heterogeneity.¹⁵ MGMT restores TMZ-induced O⁶-methylated guanine that accumulates due to the alkylating activities associated with TMZ. O⁶-methyl-guanine is both a cytotoxic and a mutagenic substrate. By mispairing with thymine, O⁶-methyl-guanine induces a G to A transition if the mispair is not corrected by the mismatch repair (MMR). This ubiquitous repair process corrects non-Watson-Crick paired duplexes.⁶⁶ MMR involves an excision and resynthesis mechanism for correcting the DNA mismatch.⁶⁷ When the quantity of O⁶-methyl-guanine exceeds the cellular capacity to complete this process, ssDNA and O⁶-methyl-guanine: thymine mispairs accumulate. The former lesion induces cytotoxicity, while tolerance of the latter leads to mutagenesis. Thus, cancer evolution in response to TMZ represents a dynamic balance between these two lesions.^68,69 Each MGMT molecule transfers the methyl moiety from O⁶-methyl-guanine into itself before the methylated MGMT is targeted for degradation,⁶⁸ and the repair of O⁶-methyl-guanine by MGMT is stochiometric.⁷⁰ Thus, MGMT expression level is a key determinant in this dynamic interplay. The variability in MGMT expression associated with miR-181d degradation broadens the range of this interplay, which increases intratumoral heterogeneity.

The stability of most miRNAs (typically exceeding 12–24 h) is primarily attributed to the binding of Argonaute (AGO), which protects the ends and the phosphate backbone of the miRNA from nucleolytic degradation.^71,72 While our understanding of miRNA degradation remains incomplete, the target-directed miRNA degradation (TDMD) mechanism proposes that the binding of miRNA to a highly complementary sequence, termed “trigger RNAs,” induces a ternary complex that recruits E3 ubiquitin ligase and an E2 ubiquitin-conjugating enzyme. This process results in the proteasomal processing of AGO, releasing the deprotected miRNA for nucleolytic degradation.^73,74 Whether and how the roles of ATR and PNPT1 in miR-181d degradation interface with TDMD remains an intriguing area of future research.

Corroborating with our findings, a growing body of peer-reviewed studies employing matched clinical specimens has demonstrated elevated MGMT expression in promoter-unmethylated glioblastomas following TMZ treatment.^75–77 However, the temporal dynamics of this increase remain undefined, as clinical sampling within hours of TMZ administration is not feasible in standard practice. Instead, clinical samples are typically taken at the time of recurrence, which occurs months to years after the initiation of TMZ treatment. Our study suggests that TMZ exposure immediately initiates a race between glioblastoma kill and acquired resistance, and the relative kinetics of these two processes ultimately determine the clinical outcome. The feedforward loop between miR-181d and PNPT1 leads to rapid degradation of miR-181d, increasing the mean MGMT expression and the variance of MGMT expression. The increased variation in MGMT expression is expected to facilitate mutagenesis-related intratumoral heterogeneity, which enhances population fitness in response to selective pressures related to future therapeutics. This framework offers one explanation for the poor prognosis associated with patients afflicted with MGMT-expressing glioblastomas.^29,78

Our study suggests that the therapeutic delivery of miR-181d combined with TMZ should suppress the mean expression of MGMT and the variance of this expression, tilting the balance toward tumor kill and clinical efficacy. This strategy offers promise where other therapies have failed for MGMT-expressing glioblastomas.²⁹ It is important to note that this approach is unlikely to apply to glioblastomas lacking MGMT expression, underscoring the need for genotype-specific therapies in future clinical translations.

Limitations of the study

Our results suggest a complex regulatory mechanism for the ATR-PNTP1 axis. The finding that ATR-silencing abolished the binding of Bi-miR-181d to the PNPT1 mRNA (Figure 6E) suggests a potential regulatory mechanism. However, Figure 6F also suggests that ATR either directly or indirectly influenced the exo-ribonuclease activity of PNTP1. In interpreting these experimental results, it is worth noting that affinity-purified PNTP1 is limited in purity, and PNTP1-interacting proteins in the affinity pull-down may exert a more proximal effect on miR-181d than those discussed above. As such, the data presented in this study place ATR and PNPT1 only in a general pathway framework. While mechanistic interactions between ATR and PNTP1 may be inferred based on the data presented here, substantial work will be required to assess these inferences.

RESOURCE AVAILABILITY

Lead contact

Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Gatikrushna Singh (gsingh@umn.edu).

Materials availability

All reagents generated in this study will be made available by the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines

Human glioblastoma cell lines BT-83⁴³ and CMK3⁴⁴ were grown as neurosphere in NeuroCult media (Gibco) supplemented with heparin, human epidermal growth factor (EGF), human fibroblast growth factor (FGF) and 10% fetal bovine serum (FBS) in ultra-low attachment flasks (Nunc). All cell lines were kept in a humidified atmosphere at 37°C incubator with 5% CO₂ and regularly tested for mycoplasma contamination using Mycoplasma detection kit (ATCC).

Mice

Six-week-old healthy male and female wild-type nude mice (Strain #002019) were obtained from The Jackson Laboratory. Mice were housed in a specific pathogen-free environment within the Research Animal Resources (RAR) accredited animal facility of University of Minnesota, under veterinary supervision and in accordance with Institutional Animal Care and Use Committee (IACUC) guidelines. These animals were not involved in any prior studies. All experimental procedures were performed under protocols approved by the IACUC, University of Minnesota.

METHOD DETAILS

Stable and CRISPR-Cas9 knockout cell line generation

Stable cell lines constitutively expressing miR-181d were established by transfecting pCMV6-Entry-miR-181d miRNA expression vector (OriGene) into CMK3 and BT-83 cells selected with neomycin as described previously.²⁷ Single clones were isolated and named CMK3 (miR-181d), and BT-83 (miR-181d), respectively. The corresponding negative control cell lines were generated by transfecting an empty control vector (OriGene) into CMK3 or BT-83 cells and selected with neomycin.

PNPT1 CRISPR lentivirus was generated by the transfection of pLenti-CRISPR v2-sgPNPT1-2 lentiviral vector into HEK293T cells with psPAX2 and pCMV-VSV-G (Addgene) at a ratio of 10:9:1. Culture supernatant containing lentivirus was collected after 48 h of transfection, transduction to CMK3 cells and selection with puromycin. The single KO clones were isolated and confirmed by DNA sequencing. A negative control cell line was generated by transfecting the corresponding empty expression vector.

TMZ-resistance clone generation

BT-83 (0.5 ×10⁶) cells were seeded in a 6-well plate. The cells were treated with 500 μM TMZ or 1% DMSO for 48 h. Post-treatment, the cells were washed and replenished with fresh culture medium without TMZ. After 4 weeks, the live cells started forming colonies. The colonies were collected, single-cell suspension was prepared, serially diluted and inoculated into 96-well plates at one cell/well densities. Cells were allowed to grow for 4 weeks. Each well was then examined for the single cell clone, and the isolated clones were quantified for MGMT, miR-181d expression.

Plasmids

PNPT1-MRE1 or MRE2 or MRE3 or mut-MRE2 or mut-MRE3 oligonucleotide fragment was inserted into psiCHECK-2-vector (Promega) using Sgf-I and Not-I. In addition, MRE2 or MRE3 or mut-MRE2 or mut-MRE3 oligonucleotide fragment was sub-cloned into the pCMV6-Entry-PNPT1 plasmid (OriGene) using Mlu-I and Not-I restriction sites. PNPT1 point mutations (S135G, Q387R, R445E, R446E, S484A, D538A and 544G) were generated by the site-directed mutagenesis of pCMV6-AC-GFP-PNPT1 plasmid using QuikChange Lightning Site-Directed mutagenesis kit (Agilent). ATR-shRNA oligonucleotide fragment was inserted into pINDUCER20_3xFlag vector (Addgene) using Sal-I and Xho-I. All clones were confirmed by DNA sequencing.

RNA isolation and quantitative RT-PCR (RT-qPCR)

Total RNA was isolated from cells using miRNeasy Kit (Qiagen) following the manufacturer’s protocol. cDNA was synthesized using Omniscript RT kit (Qiagen). miRNA and mRNA transcripts were quantified using SYBR Green (Bio-Rad) and target-specific primers (Qiagen) on the Bio-Rad Chromo 4 DNA Engine Thermal Cycler.

Newly diagnosed or and at the time of recurrence from the same patients’ glioblastoma surgical specimens were subjected to RNA isolation, reverse-transcribed, and subjected to qPCR of miR-181a, miR-181b, miR-181c, miR-181d, miR-21, miR-939-5p, miR-3689d, miR-382-3p, miR-4513, miR-4519, miR-124-3p, miR-4323, miR-6657-3p, miR-6867-3p, miR-548aq-3p, miR-8071 and miR-5589-3p with miRCURY LNA primers (Qiagen) according to the manufacturer’s instructions.

miRNA, siRNA, or plasmid transfection

CMK3 cells were transfected with human miR-181d (Qiagen, MIMAT0026608) and miR-21 mimic (MIMAT0000076), Bi-miR-181d (Qiagen, MIMAT0002821) or the non-targeting (NT) control (Qiagen, MIMAT0000010) using HiPerfect transfection reagent (Qiagen) following the manufacturer’s instructions.

siRNA transfections were carried out by HiPerfect transfection reagent (Qiagen). ON-TARGETplus siRNAs (Dharmacon) targeting human genes were used: RRP41 or ExoSC4 (L-013760-00-0010), PNPT1 (L-019454-01-0010), XRN1 (L-013754-01-0010), and non-targeting control (D-001810-01-10).

CMK3 PNPT1-KO cells were transfected with the pCMV6-AC-GFP vector (OriGene) encoding human WT-PNPT1 or point mutants (RPH1 domain: D135G; RPH2 domain: Q387R, R445E, R446E, R484A and catalytic sites: D538A and D544G) or empty vector by lipofectamine 2000 (Invitrogen).

RNA sequencing and data processing

BT-83 or CMK3 cells were treated with 500 μM TMZ or DMSO (1%) for 6 h. The cells were harvested, and RNA was isolated (miRNeasy Kit, Qiagen). The RNA integrity was determined by Bioanalyzer RNA 6000 Nano kit (Agilent). The RNA samples were submitted to the University of Minnesota Genomic Center (UMGC) for library preparation using TruSeq Small RNA Library protocol and short-read sequencing using HiSeq 2500 High Throughput Sequencer (Illumina).

All datasets were analyzed using a uniform processing workflow. Raw sequencing data (FASTQ files) underwent preprocessing and quality control with miRTrace.⁷⁹ Reads were discarded if fewer than 50% of nucleotides had a Phred score above 20 nucleotides. Adapter sequences at the 3′end was trimmed, and sequences shorter than 18 nucleotides or consisting of repetitive elements were removed. Samples were excluded if fewer than 25% of reads were 20–25 nucleotides in length, more than 75% of reads were removed during QC, or fewer than 10% of reads were classified as miRNA. Studies were omitted if over half of their datasets failed these QC thresholds or if substantial contamination was detected. Following QC, read alignment and gene quantification were performed with miRge3.0,⁸⁰ using MirGeneDB2.0.⁸¹

Clinical glioblastoma specimen collection and immunohistochemistry (IHC)

The research protocol (STUDY00012599) was approved by the Institutional Review Board (IRB) of the University of Minnesota. All the patients gave informed consent before the study commenced. Only specimens that harbored wtIDH and umMGMT were collected for this study. Newly diagnosed or at the time of recurrence from the same patients’ glioblastoma surgical specimens were formalin-fixed and paraffin-embedded (FFPE). The FFPE sections were mounted on the slides and stained with MGMT antibody or Hematoxylin and Eosin (H&E). For MGMT detection, 4 μm sections were placed onto positively charged glass slides. The slides were deparaffinized and rehydrated. Antigen retrieval was performed. Endogenous peroxidase was blocked using 3% hydrogen peroxide, followed by Dako Protein Serum Block. Primary MGMT antibody (Abcam cat# ab39253, 1:100 dilution) was incubated for 30 min at RT. Detection was achieved using Rabbit EnVision+ Kit (cat# K4003, Dako) and developed using diaminobenzidine (Dako) chromogen. Slides were counterstained with Mayer’s Hematoxylin. The IHC sections were evaluated and scored based on a previously published scoring system.⁸² In brief, for each MGMT stained section, ten random 400X fields were evaluated (5 random intratumoral and five random peri-tumoral fields). In each field, the density of MGMT positive cells was scored as No (no MGMT positive cells), Low (MGMT positivity in 1–33% of nucleated cells), Intermediate (MGMT positivity in 34–66% of nucleated cells), High (MGMT positivity in 67–100% of nucleated cells).

Bioinformatic analysis of TCGA data

MGMT mRNA and miR-181d expression analysis from TCGA datasets were performed as previously described.⁴³ In brief, preprocessed level 3 data were obtained from TCGA⁸³ for 582 glioblastoma specimens. Specimens were categorized based on whether they were newly diagnosed or recurrent. A normalized expression value for MGMT and miR-181d were calculated by subtracting the gene’s mean expression value across the dataset and then dividing by its standard deviation (SD).

Limiting dilution neurosphere forming assays

CMK3 cells were seeded in a 12-well plate at 10,000 cells per well. The cells were transfected with siPNPT1 or siNT. Twenty-four hours post-transfection, the cells were treated with or without 500 μM TMZ for 12 h. The cells were collected, single-cell suspensions were prepared, serially diluted, and inoculated into 96-well ultra-low attachment plates at specific densities. Cells were allowed to grow for 3 weeks. Each well was then examined for the absence or presence of a neurosphere (at least one aggregate of 10 or more cells was counted as one sphere). The frequency of sphere-forming cells was then calculated using Extreme Limiting Dilution Analysis (ELDA, http://bioinf.wehi.edu.au/software/elda/).

Immunoprecipitation, affinity purification, and western blot analysis

Dynabeads Protein G (30 μL) (Invitrogen) were washed two times in 10-bed volume of IP lysis buffer (20 mM Tris-HCl [pH 7.4], 150 mM NaCl, 2 mM EDTA, 1% NP-40). The beads were incubated with anti-rabbit PNPT1 antibody (Proteintech, 14487-1-AP or Abcam, ab157109) or anti-mouse turbo GFP antiserum (OriGene, TA150039) or anti-mouse IgG (Santa Cruz, sc-2025) in 10-bed volume of IP lysis buffer containing 1 mM BSA (Thermo Fisher) for 45 min at room temperature. The bead-antibody complexes were washed once in a 10-bed volume of IP wash buffer (20 mM Tris-HCl [pH 7.4], 300 mM NaCl, 0.5% NP-40) and incubated with 300 μg of total cell lysate or C-terminal GFP tagged WT-PNPT1 or point mutant PNPT1 expressing cell lysate at 4°C for 2 h with rotation. The enriched immune complexes were washed four times in IP wash buffer and collected by boiling with 1x SDS sample buffer.

Dynabeads M-280 Streptavidin (25 μL) (Invitrogen, 11206D) were equilibrated with a 10-bed volume of 1x PBS containing 0.5% BSA. The beads were incubated with 300 μg lysate of the cells transfected with the biotinylated miRNA at 4°C overnight with rotation. The beads were washed four times in IP wash buffer and collected by boiling them with 1x SDS sample buffer.

Cell lysates or affinity-purified proteins collected in SDS-sample buffer were resolved in SDS/PAGE and transferred onto the nitrocellulose membranes (Bio-Rad). Membranes were blocked using 5% BSA for 1 h at room temperature, followed by primary antibody incubation overnight at 4°C. After washing with 1x TBST buffer, membranes were incubated with species-specific secondary antibodies for 1 h at room temperature and developed in chemiluminescence reagent (GE biosciences) using Bio-Rad ChemiDoc MP Imaging System.

MGMT promotor methylation analysis

Genomic DNA was extracted from CMK3 cells transfected with miR-181d or miR-NT according to the manufacturer’s protocol using DNeasy blood and tissue kit (Qiagen). 1 μg of DNA was used for bisulfite treatment using Epitect bisulfite kit, (Qiagen). The bisulfite treated DNA was amplified by methylation specific PCR using GoTaq green PCR mix (Promega) using un-methylated and methylated MGMT specific primers.^28,84 The primer sets target unmethlyated and methylated alleles covering the CpG island of MGMT promoter. The amplified PCR product was resolved on a 2.5% agarose gel.

Mitochondrial isolation

Mitochondria were isolated from CMK3 cells (2 ×10⁷) using a differential centrifugation protocol (Mitochondria Isolation Kit, Cat#89874, Thermo Fisher). Briefly, cells were harvested by centrifugation at 850 ×g for 2 min at 4°C, resuspended in Reagent A supplemented with protease inhibitors, vortexed, and incubated on ice for 2 min. Reagent B was added, followed by brief vortexing, and the suspension was incubated on ice for 5 min with intermittent mixing. Subsequently, Reagent C was added, and the mixture was centrifuged at 700 ×g for 10 min at 4°C. The supernatant was collected and centrifuged at 12,000 ×g for 15 min at 4°C to pellet mitochondria; an intermediate centrifugation at 3,000 ×g was included to reduce lysosomal and peroxisomal contamination. The mitochondrial pellet was washed once with Reagent C at 12,000 ×g for 5 min and maintained on ice until downstream applications.

Luciferase reporter assay

Luciferase reporters bearing the truncated sections of PNPT1 3′ UTR containing predicted miR-181d miRNA Response Elements (MREs) or mutated miR-181d MREs were constructed by subcloning into pSiCheck-2 dual luciferase vector (Promega). CMK3 cells were co-transfected with miR-181d mimic and MRE containing luciferase reporter using Lipofectamine 2000 (Invitrogen) following the manufacturer’s protocol. Luciferase activity was measured 48 h post-transfection.

Clonogenic assay

The cells were washed with 1x phosphate buffered saline (PBS) and incubated in methanol (100%) for 30 min at room temperature. Methanol was removed, and cells were then allowed to dry at RT for 10 min and stained with methanol containing 0.5% crystal violet solution for 30 min. The cells were washed multiple times in tap water, allowed to dry at RT and the images were captured using the LASX-application suit in Lica-DMi8 microscope (Deerfield, IL).

Xenografts development and survival studies

Animal studies were performed following the Guide for the Care and Use of Laboratory Animals (Guide for the Care and Use of Laboratory Animals, 8th edition National Research Council (US) Committee for the Update of the Guide for the Care and Use of Laboratory Animals. Washington (DC): National Academies Press (US); 2011. ISBN-13: 978-0-309-15400-0ISBN-10: 0-309-15400-6). The animal study protocol was approved by the IACUC Protocol 2206-40091A.

For intracranial xenograft experiments, patient-derived BT-83 glioblastoma cells stably expressing miR-181d or miR-empty controls were dissociated into single-cell suspensions. Mice were anesthetized via intraperitoneal injection of ketamine (100 mg/kg, Dechra) and xylazine (10 mg/kg, Dechra), and anesthesia was confirmed by loss of reflexes. The cells were stereotactically injected into the brains of an athymic nude mice at 6 weeks of age (The Jackson Laboratory) using murine stereotaxic system (Stoelting Co.). The coordinates used were: 1.8 mm to the right of the bregma and 3 mm deep from the dura. Following tumor implantation, mice were maintained until the onset of overt neurological symptoms, including weight loss, lethargy, and hunched posture. After 7-day of tumor implant, the mice were randomly assigned to treatment groups (10 mice/group), and 50 mg/kg/day TMZ was administrated intraperitoneally for 5-day followed by 23-day treatment interruption. The mice were maintained for up to 90-day and the Kaplan-Meier survival curve was calculated. p values were determined using the log rank test.

Bioluminescence imaging

Mice were anesthetized by inhalation of 2% isoflurane and the dose was maintained throughout the imaging procedures. 150 mg/kg D-luciferin, dissolved in 1x PBS, was injected into the peritoneum, and the mice underwent sequential exposures in auto mode (IVIS50 imaging system) according to the manufacturer’s instructions. The images were acquired on days 7, 14, and 21 post-tumor implantations. Bioluminescence intensity was assessed using the Living Image 3.2 software (Caliper Life Sciences, Hopkinton, MA). Total flux values (photons (p)/second (sec)) were determined by delineating the regions of interest (ROIs) of uniform size on each mouse.

QUANTIFICATION AND STATISTICAL ANALYSIS

Three or more independent experiments were performed for each assay and results were combined to define as mean ± SD. Statistical analyses were conducted using GraphPad Prism software 10. The statistical significance was evaluated using an unpaired two-tailed Student’s t-test or one-way ANOVA. Following ANOVA, post hoc multiple comparisons were conducted using Tukey’s or Dunnett’s test to identify statistically significant differences between specific group pairs. p value (*) of ≤0.05 was considered statistically significant.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116516.

Highlights.

• Feedforward microRNA degradation drives cell-to-cell gene variance and therapy resistance

• TMZ triggers feedforward loop between PNPT1 and miR-181d, driving rapid miR-181d degradation

• Degradation of miR-181d increases the cell-to-cell variability in MGMT expression

• Cell-to-cell variability in MGMT expression contributes to acquired TMZ resistance

Data and code availability

RNA sequencing data of two patient-derived glioblastoma cell lines, BT-83 and CMK3, in response to TMZ therapy have been deposited at the “Mendeley Data” (https://data.mendeley.com/datasets/pdvttwws8m/1) and are publicly available as of the date of publication. “DOI:https://doi.org/10.17632/pdvttwws8m.1” is listed in the key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-PNPT1 antibody	Proteintech	Cat# 14487-1-AP; RRID: AB_2165820
Anti-PNPT1 monoclonal antibody	Invitrogen	Cat# MA5-27859; RRID: AB_2735325
Anti-PNPT1 antibody	Abcam	Cat# ab157109; RRID: AB_2910229
Anti-MGMT antibody	Proteintech	Cat# 17195-1-AP; RRID: AB_2143221
Anti-MGMT monoclonal antibody	Invitrogen	Cat# 35-7000; RRID: AB_87574
Anti-MGMT antibody	Abcam	Cat# ab39253; RRID: AB_776338
Anti-Alpha (α) tubulin antibody	Santa Cruz	Cat# sc-8035; RRID: AB_628408
Anti-GFP monoclonal antibody	Santa Cruz	Cat# sc-9996; RRID: AB_627695
Anti-GFP antibody	Proteintech	Cat# 50430-2-AP; RRID: AB_11042881
Anti-GFP antibody	OriGene	Cat# TA150039; RRID: AB_2622255
Anti-GAPDH antibody	Proteintech	Cat# 10494-1-AP; RRID: AB_2263076
Anti-Succinate dehydrogenase antibody	Proteintech	Cat# 14865-1-AP; RRID: AB_11182164
Anti-FLAG monoclonal antibody	Millipore-Sigma	Cat# F3165; RRID: AB_259529
Anti-IgG mouse	Cell Signaling	Cat# 7074; RRID: AB_2099233
Anti-IgG mouse	Santa Cruz	Cat# sc-2025; RRID: AB_737182
Anti-IgG rabbit	Cell Signaling	Cat# 7076; RRID: AB_330924
Bacterial and virus strains
DH5α	Invitrogen	18265017
XL1-Blue	Agilent	200150
Stbl3	Invitrogen	C737303
Chemicals, peptides, and recombinant proteins
Temozolomide	Sigma-Aldrich	T2577
Actinomycin D1	Gibco	11805017
VE821	MCE	HY-14731
TRIzol reagent	Invitrogen	15596026
DMSO	Sigma-Aldrich	D2438
Dynabeads M-280 Streptavidin	Invitrogen	11206D
Dynabeads Protein G	Invitrogen	10009D
Dynabeads Protein A	Invitrogen	10008D
Ketamine	Dechra	200–073
Xylazine	Dechra	047–956
Heparin Solution	STEMCELL Technologies	07980
Human Epidermal Growth Factor	STEMCELL Technologies	78006
Human Fibroblast Growth Factor	STEMCELL Technologies	78134
Puromycin	Sigma-Aldrich	P9620
Neomycin	Sigma-Aldrich	N1142
Doxycycline	Sigma-Aldrich	D3447
D-Luciferin	Thermo Fisher	L2916
Critical chemical assays
Dual-Glo® Luciferase Assay System	Promega	E2920
miRNeasy Kit	Qiagen	217084
OmniScript RT Kit	Qiagen	205113
miRCURY LNA RT Kit	Qiagen	339340
QuikChange Lightning Site-Directed mutagenesis Kit	Agilent	210519
Bioanalyzer RNA 6000 Nano Kit	Agilent	5067–1511
Rabbit EnVision™ Kit	Dako	K4003
DNeasy blood and tissue Kit	Qiagen	69504
EpiTect bisulfite Kit	Qiagen	59104
Mitochondria isolation Kit	Thermo Fisher	89874
Universal Mycoplasma Detection Kit	ATCC	30-1012K
Deposited data
RNA Sequencing of BT-83 and CMK3 cell lines	This paper	DOI: https://doi.org/10.17632/pdvttwws8m.1
Experimental models: Cell lines
CMK3	This paper	N/A
BT-83	This paper	N/A
Experimental models: Organisms/strains
Wild-type nude mice	The Jackson Laboratory	Strain #002019
Oligonucleotides
PNPT1 For: 5′TGTCATGTTGGAAGCCTCTG3′	IDT	N/A
PNPT1 Rev: 5′ACTGCTGAATGCCCTGAATTA3′	IDT	N/A
PNPT1-Myc For: 5′CAGTCGCCAGCTACAACCGTGGT3′	IDT	N/A
PNPT1-Myc Rev: 5′CTCTTCTGAGATGAGTTTCTGCTC3′	IDT	N/A
MGMT For: 5′GCCTGGCTGAATGCCTATTT3′	IDT	N/A
MGMT Rev: 5′AACCTTCAGCTTCCATAAC3′	IDT	N/A
PNPT1-MRE2: 5′ CGCGTAGCCCAGAAGT TCGACACTGTAGTGGC3′	IDT	N/A
PNPT1-MRE2 Rev: 5′ GGCCGCGACTACAGT GTCGAACTTCTGGGCTA3′	IDT	N/A
PNPT1-MRE3: 5′ CGCGTGACCAGCCTGGC CAACATGGTGAAAGC3′	IDT	N/A
PNPT1-MRE3 Rev: 5′ GGCCGCTTTCACCA TGTTCGCCAGGCTGGTCA3′	IDT	N/A
PNPT1-MRE2-mut For: 5′ CGCGTAGAAAA GAAGTTCGACACTGTAGTGGC3′	IDT	N/A
PNPT1-MRE2-mut Rev: 5′ GGCCGCGACT ACAGTGTCGAACTTCTTTTCTA3′	IDT	N/A
PNPT1-MRE3-mut For: 5′ CGCGTGACCA TAATGGCCAACATGGTGAAAGC3′	IDT	N/A
PNPT1-MRE3-mut Rev: 5′ GGCCGCTTTC ACCATGTTCGCCATTATGGTCA3′	IDT	N/A
PNPT1 S135G For: 5′ATTGGTACTGGTGATAAAGAA3′	IDT	N/A
PNPT1 S135G Rev: 5′TTCTTTATCACCAGTACCAAT3′	IDT	N/A
PNPT1 Q387R For: 5′GCATTATTTCGAAGAGGACAA3′	IDT	N/A
PNPT1 Q387R Rev: 5′TTGTCCTCTTCGAAATAATGC3′	IDT	N/A
PNPT1 R445E For: 5′GGTTTAAATGAAAGAGAACTT3′	IDT	N/A
PNPT1 R445E Rev: 5′AAGTTCTCTTTCATTTAAACC3′	IDT	N/A
PNPT1 R446E For: 5′TTAAATAGAGAAGAACTTGGG3′	IDT	N/A
PNPT1 R446E Rev: 5′CCCAAGTTCTTCTCTATTTAA3′	IDT	N/A
PNPT1 S484A For: 5′GGGTCATCTGCTATGGCATCT3′	IDT	N/A
PNPT1 S484A Rev: 5′AGATGCCATAGCAGATGACCC3′	IDT	N/A
PNPT1 D538A For: 5′GGAATTGAAGCTTACAATGGT3′	IDT	N/A
PNPT1 D538A Rev: 5′ACCATTGTAAGCTTCAATTCC3′	IDT	N/A
PNPT1 D544G For: 5′GGTGACATGGGCTTCAAAATA3′	IDT	N/A
PNPT1 D544G Rev: 5′TATTTTGAAGCCCATGTCACC3′	IDT	N/A
Unmet MGMT For: 5′TTTGTGTTTTGATGTTTGTAGGTTTTTGT3′	IDT	N/A
Unmet MGMT Rev: 5′AACTCCACACTCT TCCAAAAACAAAACA3′	IDT	N/A
Met MGMT For: 5′TTTCGACGTTCGTAG GTTTTCGC3′	IDT	N/A
Met MGMT Rev: 5′GCACTCTTCCGAAA ACGAAACG3′	IDT	N/A
Recombinant DNA
pLenti-CRISPR v2-sgPNPT1-2	Addgene	234770
pCMV-Entry-miR-181d	This paper	N/A
pCMV-VSV-G	Addgene	8454
psPAX2	Addgene	12260
pCMV6-Entry-PNPT1	OriGene	RC208800
pCMV6-AC-GFP-PNPT1	OriGene	RG208800
pCMV6-AC-GFP-PNPT1 S135G	This paper	N/A
pCMV6-AC-GFP-PNPT1 Q387R	This paper	N/A
pCMV6-AC-GFP-PNPT1 R445E	This paper	N/A
pCMV6-AC-GFP-PNPT1 R446E	This paper	N/A
pCMV6-AC-GFP-PNPT1 S484A	This paper	N/A
pCMV6-AC-GFP-PNPT1 D538A	This paper	N/A
pCMV6-AC-GFP-PNPT1 D544G	This paper	N/A
psiCHECK-2	Promega	N/A
psiCHECK-2-PNPT1 MRE1	This paper	N/A
psiCHECK-2-PNPT1 MRE2	This paper	N/A
psiCHECK-2-PNPT1 MRE3	This paper	N/A
psiCHECK-2-PNPT1 mut-MRE2	This paper	N/A
psiCHECK-2-PNPT1 mut-MRE3	This paper	N/A
pCMV6-Entry-PNPT1-MRE2	This paper	N/A
pCMV6-Entry-PNPT1-MRE3	This paper	N/A
pCMV6-Entry-PNPT1-MRE2-mut	This paper	N/A
pCMV6-Entry-PNPT1-MRE3-mut	This paper	N/A
pINDUCER20_3xFlag	Addgene	190028
pINDUCER20_3xFlag-shATR	This paper	N/A
pLV-fLuc/puro	This paper	Gifted by Dr. Robert Galvin
Software and algorithms
Graphpad Prsim	Graphpad	Version 10.0
ImageJ	NIH	Version 1.50e
Adobe photoshop	Adobe	Version 21.0.2

• All data reported in this manuscript will be shared by the lead contact upon request.

• Full western blot image data will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.