Pathophysiological and translational implications of the dysregulation of molecular machineries controlling RNA metabolism in thyrotropinomas

Summary

The contribution of post-transcriptional regulation remains largely unexplored in thyrotropin-secreting pituitary tumors (TSHomas), a rare endocrine pathology responsible for inappropriate TSH secretion and central hyperthyroidism. We investigated the TSHoma post-transcriptional regulatory landscape by analyzing components of molecular machineries controlling RNA metabolism [spliceosome/RNA-exosome/nonsense-mediated decay (NMD)], their associations with clinical parameters, and the impact of their pharmacological inhibition in TSHoma cells. A drastic dysregulation of multiple components of spliceosome [spliceosome (SF3B2/U2AF35/PRPF40A) and splicing factors (e.g., SRP30C/EIF4A3/DDX1/TRA2B/HNRNNF…)], RNA-Exosome (e.g., DIS3/DIS3L/PABPN1), and NMD [e.g., PABPC1, and NMD-canonical targets (DDIT3/GADD45B/MAFF/PDRG1/ATF4/ATF3)] was found in TSHomas vs. non-tumor pituitaries with some of these alterations associated with relevant clinical features (e.g., tumor size, TSHB/fT4-levels). Moreover, we demonstrate a clear antiproliferative action of spliceosome/RNA-Exosome/NMD inhibitors in primary patient-derived TSHoma cells. Overall, a clinically relevant spliceosome/RNA-Exosome/NMD-associated molecular dysregulation and antiproliferative actions of these machineries’ inhibition are demonstrated in TSHomas, highlighting a potential source of novel diagnostic/prognostic biomarkers and therapeutic tools in TSHomas.

Graphical abstract

Highlights

• •The molecular machineries controlling RNA metabolism are dysregulated in TSHomas
• •Critical clinical features are correlated with alterations in RNA metabolism factors
• •Spliceosome/RNA-Exosome/NMD activity inhibition exerts antitumor actions in TSHomas
• •RNA metabolism machineries components represent a source of clinical tools in TSHomas

Cell biology; Cancer

Introduction

Thyrotropin-secreting pituitary tumors (TSHomas) are rare functional pituitary tumors (PiTs) characterized by inappropriate secretion of thyroid-stimulating hormone (TSH), leading to central hyperthyroidism. Representing less than 1% of all PiTs, TSHomas are often misdiagnosed due to their clinical overlap with resistance to thyroid hormone syndromes and other causes of altered thyroid function.^1,2 While advances in imaging, immunohistochemistry, and functional assays have improved diagnostic accuracy, the molecular mechanisms underlying TSHoma pathogenesis remain poorly understood.^2,3,4 In this regard, in contrast to the extensive literature on other functioning PiTs such as GH-/ACTH-/PRL-secreting tumors,^5,6,7,8 the biological drivers of TSHomas, including their transcriptomic and post-transcriptional landscapes, are largely unknown.^3,4

Growing evidence indicates that post-transcriptional regulation is a key determinant of gene expression control in both physiological and pathological contexts.^9,10 In fact, regulatory processes beyond transcription, such as alternative splicing (AS), RNA transport, and RNA stability processes, have emerged as critical modulators of tumor behavior in different endocrine-related and central nervous system (CNS) cancers, influencing proliferation, hormone secretion, and therapy resistance.^{10,11,12,13,14,15,16,17,18} Thus, given the secretory nature of PiTs, the involvement of post-transcriptional pathways in their development and functional phenotype is plausible, but the information reported so far is quite limited and fragmentary.^19,20,21,22

In this context, one of the most significant post-transcriptional mechanisms implicated in cancer biology is the regulation of RNA metabolism.^9,23,24 This cellular process is controlled by different cellular machineries involved in RNA processing, versatility, and degradation [i.e., splicing machinery (spliceosome),^15,25 the RNA-Exosome,^26,27,28 and the nonsense-mediated decay (NMD) complex^29,30,31 wherein many of their molecular components have been reported to hold potential as diagnostic and prognostic biomarkers and serve as a source of clinical tools for cancer. In fact, dysregulation of the spliceosome machinery itself has been linked to the aggressiveness of some PiTs^19,22; however, the relevance of the different cellular machineries controlling RNA metabolism (spliceosome, RNA-Exosome, and NMD) in TSHomas has not been systematically examined. Therefore, the present study was aimed at determining the systematic expression levels of the molecular components of the spliceosome, RNA-Exosome, and NMD machineries in TSHomas and their relationship with clinical features, as well as to evaluate the therapeutic potential of inhibiting the activity of these molecular machineries in primary patient-derived TSHoma cell cultures.

Results

Spliceosome dysregulation in TSHomas compared to NPs

A marked dysregulation in the expression levels of multiple components of the splicing machinery (38% of the spliceosome components and splicing factors analyzed) was found in TSHomas compared to the NP tissues. Specifically, there was a significant downregulation of the 3 major spliceosome components (PRPF40A, U2AF35, and SF3B2) and 24 splicing factors (EIF4A3, DDX1, SNRPC, HNRNPAB, SNRPB2, SNRPD1, SNRPD3, SART1, HNRNPA3, HNRNPG, SRPK1, RBM6, RBM4, PRP19, HNRNNPF, HNRNPA1, TRA2B, SRP30C, SRP20, SRSF10, SND1, SAM68, CELF4, and CUGBP1), whereas a significant upregulation of 3 splicing factors (ELAVL1, RBM5, and NSR100) was also found (Figure 1A). Interestingly, non-supervised hierarchical analysis based on the expression pattern of all spliceosome components and splicing factors was able to perfectly discriminate TSHomas and NP tissues into two independent clusters (Figure 1B), which was also confirmed by partial least squares-discriminant analysis (PLS-DA; Figure 1C). In fact, application of variable importance in projection (VIP) score of PLS-DA (Figure 1D) revealed that SRP30C, DDX1, TRA2B, SNRPB2, SF3B2, HNRNNF, and U2AF35 were the highest components (VIP score >1.5), capable of discriminating between TSHomas and control tissues. Indeed, receiver operating characteristic (ROC) curve analyses corroborated the capacity of these components to finely discriminate between TSHomas and control samples, showing an area under the curve (AUC) higher than 0.9 (Figure 1E).

Figure 1.

Dysregulation of splicing machinery in patients with TSHoma (n = 4) vs. non-tumor pituitary (n=11) samples

(A) Fold change plot shows the ratio of mean expression in patients with TSHoma and the mean of non-tumor samples (left), and a volcano plot displays downregulated (blue) and upregulated (red) genes in tumor vs. non-tumor samples with a p-value <0.05, and non-significantly altered (−1.2 > FC < 1.2) genes (gray) (right).

(B) Heatmap represents expression levels of TSHoma samples (green) and non-tumor samples (gray).

(C) PLS-DA considering mRNA expression levels of major/minor spliceosome and splicing factors.

(D) Ranked variable importance projection (VIP) scores of splicing machinery comparing TSHomas and non-tumor samples.

(E) ROC curve analyses regarding TSHoma and NP expression of the most discriminating factors. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 significantly different from control conditions. AUC: area under the curve.

RNA-exosome dysregulation in TSHomas compared to NPs

A dysregulation in the expression levels of 5 components of the RNA-Exosome machinery (19% of the analyzed) was found in TSHomas compared to the NP tissues. Specifically, there was a significant upregulation of two catalytic core components (endonucleases DIS3 and DIS3L) and two cofactors [one from the SKI complex (TTC37) and one from the TRAMP complex (MTREX)], whereas a significant downregulation of one cofactor of the SKI complex (WDR61) was also found (Figure 2A). Although the heatmap generated with the expression levels of the RNA-Exosome components did not completely segregate TSHomas and NP tissues (Figure 2B), PLS-DA showed a clear separation of these tissues based on this RNA-exosome expression profile (Figure 2C). Application of VIP score of PLS-DA revealed that WDR61, DIS3, TTC37, MTREX and DIS3L (all the significantly altered RNA-Exosome components) were the top components (VIP score >1.5) capable of discriminating between TSHomas and control tissues (Figure 2D). Indeed, ROC curve analyses corroborated the capacity of four of them to finely discriminate between TSHomas and control samples, showing an AUC superior to 0.8 (Figure 2E). Moreover, as depicted in Figure 2F, the heatmap generated with only these 5 components perfectly separated TSHomas from control-NPs, segregating them into two clusters.

Figure 2.

Expression profile of RNA-Exosome in patients with TSHoma (n = 4) vs. non-tumor pituitary (n = 11) samples

(A) Fold change plot shows the ratio of mean expression in patients with TSHoma and the mean of non-tumor samples (left), and the volcano plot displays downregulated (blue) and upregulated (red) genes in tumor vs. non-tumor samples with a p value <0.05, and non-significantly altered (−1.2 > FC < 1.2) genes (gray) (right).

(B) Heatmap represents expression levels of TSHoma samples (yellow) and non-tumor samples (gray).

(C) PLS-DA considering mRNA expression levels of different compartments that form RNA-Exosome.

(D) Ranked variable importance projection (VIP) Scores of RNA-Exosome compares TSHomas and non-tumor samples.

(E) ROC curve analyses regarding TSHoma and NP expression of the most discriminating factors.

(F) Heatmap of expression levels of the most discriminating components in TSHoma (yellow) and NP (gray) samples. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗∗p < 0.0001 significantly different from control conditions. AUC: area under the curve.

NMD dysregulation in TSHomas compared to NPs

A dysregulation in the expression levels of several components of the NMD complex (20% of the analyzed) was found in TSHomas compared to the NP tissues. Specifically, there was an overall significant upregulation of 9 NMD components [including components from the CCR4-NOT deadenylation complex (CNOT3, CNOT4, and CNOT6L) and key components in NMD activity (MAGOHB, EIF4G1, AQR, SMG9, SMG7, and PABPC1)], whereas a significant downregulation of 3 components (RUVBL2, ALYREF, and EIF4A3) was also found (Figure 3A). It should be mentioned that 2 of these 3 downregulated genes (ALYREF and EIF4A3) share functionality with the spliceosome, which were previously observed to be generally downregulated (Figure 1A). Like our previous observation with the RNA-Exosome, although the heatmap generated with the expression levels of the NMD components did not completely segregate TSHomas and NP-tissues (Figure 3B), PLS-DA showed a clear separation of these tissues (Figure 3C). Application of VIP score of PLS-DA revealed that RUVBL2, MAGOHB, CNOT3, PABPC1, CNOT4, ALYREF, AQR, and EIF4A3 were the highest components capable of discriminating between TSHomas and control tissues (VIP score >1.5; Figure 3D). Indeed, ROC curve analyses corroborated their capacity to finely discriminate between TSHomas and control samples, showing an AUC higher than 0.7 (Figure 3E). Moreover, as depicted in Figure 3F, the heatmap generated with these 8 components perfectly separated TSHomas from control NPs, segregating them into two clusters.

Figure 3.

Expression pattern of NMD in patients with TSHoma (n = 4) vs. non-tumor pituitary (n = 11) samples

(A) Fold change plot shows the ratio of mean expression in patients with TSHoma and mean of non-tumor samples (left), and the volcano plot displays downregulated (blue) and upregulated (red) genes in tumor vs. non-tumor samples with a p value <0.05, and non-significantly altered (−1.2 > FC < 1.2) genes (gray) (right).

(B) Heatmap represents expression levels of TSHoma samples (orange) and non-tumor samples (gray).

(C) PLS-DA considering mRNA expression levels of NMD machinery.

(D) Ranked variable importance projection (VIP) Scores of NMD comparing TSHomas and non-tumor samples.

(E) ROC curve analyses regarding TSHoma and NP expression of the most discriminating factors.

(F) Heatmap of expression levels of the most discriminating NMD components in TSHoma (orange) and NP (gray) samples.

(G) Log₂ fold change of NMD targets in TSHomas vs. non-tumor individuals. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 significantly different from control conditions. AUC: area under the curve.

It is important to mention at this point that the expression profile of some NMD-canonical targets might be directly associated with the NMD activity.^{29,32,33,34,35,36} Thus, we also analyzed the expression levels of these targets associated with NMD activity and found that 6 NMD targets were significantly downregulated (DDIT3, GADD45B, MAFF, PDRG1, ATF4, and ATF3) and one upregulated (TSTD2) in TSHomas (Figure 3G), which might suggest the existence of a putative increase in NMD activity in TSHomas.

Dysregulation of specific components of the RNA metabolism machineries correlates with important clinical features in TSHomas

We found that the patient with the highest expression levels of TSHB (the gene that codifies TSH hormone, hyper-produced/secreted in patients with TSHomas) was the only one without pre-surgery treatment (Figure S1A) and with no cure nowadays (Figure S1B). Moreover, a positive correlation between TSHB transcript levels and TSHB serum levels at diagnosis was observed (Figure S1C). No significant correlation was observed between TSHB transcript levels and fT4 serum levels or tumor size (Figures S1D and S1E, respectively), which could probably be accounted for by the limited size of TSHoma samples (n = 4). However, we found the following clinical correlations between different key components of the machineries controlling RNA metabolism:

• •Spliceosome components. Significant negative correlations between 1) RBM17, U2AF65, NOVA1 with TSHB expression (Figure 4A; green columns); 2) SRP30C with TSH serum levels (Figure 4B; green column); 3) EIF4A3, SRRM1, and DHX8 with fT4 serum levels (Figure 4C; green columns); and 4) DHX8, HNRNPL, SRRM1, and EIF4A3 with tumor size (Figure 4D; green columns).
• •RNA-Exosome components. Significant negative correlations between 1) DIS3 and MPHOSPH6 with TSHB expression (Figure 4A; yellow columns); 2) PABPN1 and EXOSC1 with TSH serum levels (Figure 4B; yellow columns); 3) EXOSC7, EXOSC2, ZCCHC8, and EXOSC1 with fT4 serum levels (Figure 4C; yellow columns); and 4) EXOSC7, DIS3L, EXOSC3, EXOSC2, and ZCCHC8 with tumor size (Figure 4D; yellow columns).
• •NMD components. We found: 1) significant negative correlations between PPP2R2A, CNOT7, and UPF3B with TSHB expression (Figure 4A; orange columns); 2) MAGOH and PABPC1 negatively while CNOT8 and ICE1 positively correlate with TSH serum levels (Figure 4B; orange columns); 3) significant positive correlations between RNPS1 and CWC22 with fT4 serum levels (Figure 4C; orange columns); and, 4) RNPS1 positively correlates with tumor size (Figure 4D; orange column).

Figure 4.

Correlations between expression of components of the RNA metabolism machineries and key clinical/molecular features in TSHomas

Ranking of Pearson’s coefficients of RNA metabolism machineries (i.e., spliceosome, RNA-Exosome, and NMD) indicating negative correlation (Pearson’s r <0) and positive correlation (Pearson’s r >0) with TSHB transcript levels (A), TSH hormone levels at diagnosis (B), fT4 serum levels (C), and tumor size (D). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001 significantly different from control conditions.

Among all the clinical correlations observed with the components of these three molecular machineries, SRP30C (from the spliceosome), DIS3 and DIS3L (from the RNA-Exosome), PABPC1 (from the NMD) and EIF4A3 (a common shared element between the spliceosome and the NMD) were the elements whose expression level was found to be statistically altered (Figures 1A, 2A, and 3A, respectively), suggesting a potential pathophysiological value as diagnostic and prognostic biomarkers in TSHomas. Therefore, these elements (as well as PABPN1 from the RNA-Exosome, which tends to be significantly altered (p = 0.053)] were selected for further discussion (see discussion section).

Other non-significant correlations regarding expression levels of TSHB, TSHB serum levels at diagnosis, fT4 levels, and tumor size with splicing genes (Figures S2A–S2D), RNA-Exosome genes (Figures S3A–S3D), and NMD genes (Figures S4A–S4D) were also obtained.

Finally, given that an overall difference in age range was found between the control subjects and patients with TSHomas [mean (range): 57 (44–79) vs. 40 (23–48), respectively], we performed two comparisons to determine whether age could be a potential confounding variable for the differentially expressed genes found in our study (spliceosome, RNA-exosome, or NMD molecular machineries). Specifically, we compared: 1) younger control subjects (<50 years at diagnosis) vs. older control subjects (>50 years at diagnosis); and 2) patients with TSHoma vs. younger control subjects (<50 years at diagnosis). These results (Figure S5) revealed that age should not be considered a potential confounding variable in our analyses of the spliceosome, RNA-exosome, or NMD molecular machineries since there were no relevant altered genes when comparing young and old control subjects (Figures S5A, S5C, and S5E), and most of the genes found to be dysregulated when comparing patients with TSHoma vs. control subjects were also validated when comparing patients with TSHoma vs. only young control subjects (Figures S5B, S5D, and S5F).

Pharmacological inhibition of the spliceosome, RNA-Exosome, and NMD reduces the survival rate of TSHoma-patient-derived cells

As a proof of concept of the previously observed data, we evaluated the effect of the pharmacological inhibition of the spliceosome, RNA-Exosome, and NMD machineries (using pladienolide B, isoginkgetin, and NMDI14, respectively) as well as octreotide treatment in the survival rate of two different TSHoma-patient derived primary cells [Figure 5; a pure TSHoma (TSHoma-1: left panels), and a TSHoma that also overproduced GH (mix TSHoma/GHoma; TSHoma-2: right panels)]. In general, we observed similar results in both TSHoma primary cell cultures. Specifically, octreotide decreased the proliferation rate (40% reduction) at 24 h compared to control-treated cells in TSHoma-1, while TSHoma-2 showed a 20% reduction at 120 h (Figure 5A). Pladienolide B decreased the proliferation rate at 24–48 h in both TSHomas, especially at 96–120 h (more than 90% and 50% reduction, respectively; Figure 5B). Isoginkgetin exerted a similar effect as octreotide with only short-term antiproliferative capacity at 24 h in TSHoma-1 (∼40% reduction), while in TSHoma-2 it was mainly affected at 96 h (40% reduction) (Figure 5C). Finally, NMDI14 also exerted antiproliferative actions but after 96 h and 120 h of incubation in both TSHomas (40%–90% reduction in TSHoma-1, and 40–30% reduction in TSHoma-2; Figure 5D).

Figure 5.

Cell survival measurement of TSHoma patient-derived cell cultures in response to different treatments (n = 2)

Octreotide (10⁻⁷ M; A), pladienolide B (10⁻⁷ M; B), isoginkgetin (10⁻⁵ M; C), and NMDI14 (10⁻⁵ M; D) treatments compared to control DMSO-treated cells at 24, 48, 96, and 120 h.

Discussion

TSHomas are very infrequent, accounting for less than 1% of all PiTs, thus representing a very rare cause of hyperthyroidism.^2,3,4 Despite their low prevalence, translational studies in TSHomas are mandatory in order to improve the clinical knowledge of their pathophysiology that could be translated into better clinical management of this devastating pathology. To the best of our knowledge, this is the first study to provide a comprehensive analysis of post-transcriptional regulatory pathways, including splicing, RNA-Exosome, and NMD in TSHomas, integrated with clinically relevant parameters such as TSHB expression, TSH hormone levels, fT4 concentration, and tumor size. Specifically, we identified significant dysregulation of multiple components of the spliceosome, RNA-Exosome, and NMD molecular machineries in an available, well-characterized, set of TSHoma samples vs. NP tissues. Moreover, correlations between these molecular alterations and tumor behavior, along with a functional proof-of-concept in vitro, suggest a potential role for RNA metabolism in TSHoma pathophysiology.

The splicing process is a highly coordinated mechanism, regulated and carried out by the spliceosome, that relies on a combination of multiple spliceosome components and splicing factors to adequately control gene expression,²⁵ wherein this molecular machinery is often altered in cancer, increasing pathobiological versatility through the generation of distinct/novel AS variants.^15,37,38 Here, we demonstrate a drastic dysregulation of the expression profile of the splicing machinery in a well-characterized set of TSHomas, including significant alterations of major spliceosome components (SF3B2, U2AF35, PRPF40A) as well as multiple splicing factors [i.e., members of the hnRNP family (e.g., HNRNPG, HNRNPA3), SR proteins (e.g., SRP30C, SRSF10, TRA2B), a core component of the exon junction complex (EIF4A3), and neuron-specific regulators (NOVA1), among others]. In fact, bioinformatics analyses demonstrated an expression-based molecular fingerprint of the spliceosome components able to perfectly discriminate between TSHomas vs. control tissues. Moreover, these results indicate that, as it has been previously reported in multiple cancer pathologies,^16,17,19,22 TSHomas have a global spliceosome dysregulation which further reinforces our notion suggesting that spliceosome alteration might be considered a hallmark of cancer itself.

Hierarchical bioinformatics analyses revealed that SRP30C (also known as SRSF9) was the spliceosome component with higher capacity to discriminate between TSHomas and control tissues, being significantly downregulated in TSHomas and exhibiting a negative correlation with TSH levels at diagnosis, which might suggest that reduced SRP30C levels may contribute to dysregulated TSH secretion mechanisms, an idea that has been previously confirmed for other splicing events and hormonal systems.^39,40 Notably, SRP30C is a component of the serine/arginine-rich splicing factor family, known to influence splice site selection and RNA export in endocrine tissues, and its alteration has been linked to pathophysiological events in different tumor pathologies (including brain, breast, prostate, and liver cancers).^41,42,43,44 Moreover, in support of our findings in TSHomas, it has been recently reported through gene set enrichment analysis (GSEA) in different cancer types that SRSF9 expression is correlated with multiple functions and signaling pathways, representing a new biomarker for the prognosis and immunotherapy in various cancers.⁴⁵ In addition, we also found that alteration in EIF4A3 expression (a core component of the exon junction complex that plays a pivotal role in both splicing and RNA surveillance⁴⁶) was significantly decreased in TSHomas and negatively correlated with both serum fT4 and tumor size, which might suggest that reduced EIF4A3 expression may contribute to tumor growth via impaired RNA cleanup and increased oncogenic transcript stability. Interestingly, discordant results have been reported on the pathophysiological role of EIF4A3 in cancer, as we have demonstrated that overexpression of EIF4A3 plays an oncogenic role in pancreatic ductal adenocarcinoma and hepatocellular carcinoma,^47,48 while recent reports indicated a tumor-suppressor role associated to EIF4A3 in lung and gastric cancers,^49,50 suggesting a different tumor-specific regulation and tumor-dependent role of EIF4A3. Nonetheless, to the best of our knowledge, this is the first report identifying a relevant functional role of these splicing machinery elements in human TSHomas, remarking on a potential clinical value and pathophysiological relevance of the results reported herein.

Likewise, the RNA-Exosome is a multi-subunit ribonucleolytic complex combining nucleases, cofactors, and core elements that carries out the 3′-5′processing, quality control, and degradation of virtually all classes of nuclear and cytoplasmic RNAs, being a potential source of clinical tools for cancer.^26,27 In TSHomas, we found a molecularly relevant upregulation of two key catalytic core elements, DIS3 and DIS3L, compared with NP tissues, exhibiting a negative correlation with TSHB expression and tumor size, respectively. In support of the pathophysiological relevance of our findings are previous studies indicating that DIS3 mutations found in multiple myeloma (occurring in approximately 10% of the patients) impact their clinical outcome.⁵¹ Moreover, alterations in DIS3 elements expression/activity have been directly associated with the development and progression of different cancer types.^52,53 It is also noteworthy that PABPN1, a component of the PAXT complex essential for the RNA-Exosome activity regulating RNA decay and lncRNA turnover,^54,55 was slightly upregulated in TSHomas (p = 0.053) and negatively correlated with TSH hormone. Dysregulations of this factor have also been reported in different tumor pathologies and have been proposed as a prognostic biomarker.^{56,57,58,59,60} In this context, the negative correlations between DIS3/DIS3L/PABPN1 and clinical outcomes observed in our study could suggest that in TSHomas, increased RNA-Exosome activity may act as a protective feedback mechanism, reducing TSHB mRNA levels to counterbalance hormonal hypersecretion and limit tumor expansion. Although further studies would be necessary to clarify these unforeseen findings, we might speculate that the dysregulation of the RNA-Exosome observed in TSHomas may directly affect their secretory function, since this cellular machinery is necessary to maintain rRNA and snRNA processing fidelity, and its imbalance could affect the production of the elements required for active hormone synthesis and release.²⁶

Similarly, NMD molecular machinery is responsible for degrading transcripts that contain premature stop codons (and other specific features), therefore preventing the production of aberrant RNA variants with putative pathological cellular consequences.^31,61,62,63 In fact, NMD machinery is tightly associated with the spliceosome, by avoiding the production of aberrant variants derived from AS dysregulation and even sharing some of their regulatory proteins [e.g., EIF4A3 (previously discussed), MAGOH and SRRM1].⁶⁴ In TSHomas, our analysis revealed a dual alteration (upregulation and downregulation as compared with NP tissues) of pivotal elements of the NMD machinery, in accordance with previous works showing a bipolar role in other tumor pathologies, depending on multiple intrinsic and extrinsic factors.^62,65 Among all the elements altered, upregulation of PABPC1 stands out, as its levels were negatively correlated with TSH serum levels, suggesting a potential involvement of this NMD element in the dysregulated TSH secretion in TSHomas. This potential prognostic value of PABPC1 observed in TSHomas compared well with previous studies indicating that targeting PABPC1 might be a useful tool as a prognostic biomarker and therapeutic strategy for different endocrine cancer types.^{66,67,68,69,70} Moreover, we found the expression levels of several NMD targets associated with a global NMD activity were altered in TSHomas compared with NP tissues, suggesting the existence of a putative increase in the NMD activity in TSHomas that might be linked to the accumulation of aberrant transcripts that promote tumor maintenance (e.g., oncogenic splicing variants of the thyroid hormone receptor or the thyrotropin-releasing hormone receptor^71,72,73). Likewise, these results might be pathophysiologically relevant in TSHoma as NMD activation has been associated with cancer progression targeting tumor suppressor genes, and several inhibiting strategies have already been proposed using small molecules and antisense oligonucleotides (ASOs).⁶² Obviously, further work will be required to complete our understanding of this regulatory process in TSHomas and to fully elucidate the translational potential behind these interesting and potentially relevant observations.

Finally, because of their low prevalence, to the best of our knowledge, there are no currently active clinical trials specific for TSHomas. The use of SSA has been proven to be effective in patients with TSHoma by managing the clinical symptoms of hypersecretion and decreasing tumor size^3,74; however, the level of response to SSA is varied, likely due to the heterogeneity of somatostatin receptor types expressed by the tumor. Therefore, more effective therapeutic tools are deemed necessary in TSHomas. Interestingly, a proof-of-concept was performed in a TSHoma-derived cell culture wherein we could compare the antiproliferative capacity of octreotide and three pharmacological inhibitors of the spliceosome, RNA-Exosome, and NMD activity (pladienolide B, isoginkgetin, and NMDI14, respectively). Our results demonstrated that inhibition of spliceosome, RNA-Exosome, and NMD activity might be a potential therapeutic avenue for TSHoma treatment by demonstrating that the pharmacological impact of inhibiting these machineries has beneficial antiproliferative actions in TSHoma cells [especially pladienolide B and NMDI14 (up to ∼90% reduction) compared with octreotide and isoginkgetin (up to ∼40% reduction) treatments]. Some of these results compare well with recent data from our and other groups showing that these compounds reduced proliferation rates in different endocrine-related tumors.^{16,22,56,75,76,77,78} Additionally, although these specific inhibitors have not been used in clinical trials yet, some approaches targeting these machineries, especially spliceosome, have been documented in both tumor pathologies and non-tumor diseases (e.g., NCT00459823, NCT02841540, RO6885247, NCT05024994, NCT06501196, NCT04676516, NCT03032172, NCT06873334, NCT05100823, NCT02240355, and NCT03670472: information according to ClinicalTrials.gov).

Limitations of the study

This study has some limitations, especially the number of TSHoma samples analyzed, the lack of an additional validation because of the absence of external publicly available cohorts or proteomic datasets, and a profound functional validation. However, when viewed together, our data add compelling evidence, with potential clinical/therapeutic implications, demonstrating that the spliceosome, RNA-Exosome, and NMD machineries are dysregulated in TSHomas, and that some elements of these machineries are associated with relevant clinical parameters, supporting not only a diagnostic but also a prognostic value. Additionally, three of four TSHoma samples used in this study for the molecular characterization of the spliceosome, RNA-Exosome, and NMD were pre-surgery treated with first-generation SSA, which could potentially confound these molecular characterization results. However, the naive patient showed the same direction of dysregulation compared to NP samples in all the significantly dysregulated factors (except for DIS3L and CNOT6L; data not shown), suggesting that pre-surgery treatment is likely not a confounding variable in our initial characterization of the components of the spliceosome, RNA-Exosome, and NMD machineries in TSHomas compared to NP samples. Moreover, our study also suggests that targeting the activity of the spliceosome, RNA-Exosome, and NMD machineries might translate into a beneficial effect on patients with TSHoma, an observation that certainly warrants further investigation.

Resource availability

Lead contact

Further information and requests for resources should be directed to the lead contact, Raúl M. Luque (raul.luque@uco.es).

Materials availability

All materials reported in this paper will be shared by the lead contact upon request.

Data and code availability

All data generated or analyzed during this study are included in this published article. No datasets requiring deposition in a public repository were generated as part of this work; therefore, no accession numbers or DOIs are applicable. Further inquiries can be directed to the lead contact upon reasonable request. This paper does not report original code.

Acknowledgments

We deeply thank all the patients and their families for generously donating the samples and clinical data for research purposes. Special thanks to the staff of the Biobank of IMIBIC. This work was funded by Junta de Andalucía (P20_00442, PI-0117-2025, DGP_PIDI_2024_00134, DGP_POST_2024_00272, BIO-0139), Foundation of the Spanish Society of Endocrinology and Nutrition (FSEEN), Spanish Ministry of Science, Innovation and Universities (PID2022-1381850B-I00, FPU20/03954), and CIBEROBN. CIBER is an initiative of Instituto de Salud Carlos III, Ministerio de Sanidad, Servicios Sociales e Igualdad, Spain. This retrospective multicenter study was approved by the Reina Sofía University Hospital/IMIBIC/Junta de Andalucia Ethics Committee.

Author contributions

M.E.G.G.: conceptualization, data curation, formal analysis, investigation, methodology, validation, visualization, writing – original draft, and writing – review and editing; Á.F.M.: conceptualization, investigation, and methodology; L.A.M.: investigation and methodology; E.V.: resources; C.M.D.P.: resources; S.M.: resources; D.C.: conceptualization and resources; C.A.: resources; M.A.A.: resources; M.A.J.: methodology; A.J.M.F.: resources and supervision; A.S.M.: resources; A.C.F.F.: conceptualization, data curation, investigation, supervision, writing – original draft, and writing – review and editing; A.D.H.M.: conceptualization, data curation, investigation, project administration, resources, supervision, writing – original draft, and writing – review and editing; R.M.L.: conceptualization, data curation, investigation, project administration, supervision, writing – original draft, and writing – review and editing

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
TSHoma tissue samples	Virgen del Rocio University Hospital and Virgen de la Victoria University Hospital	N/A
Non-tumor pituitary tissue samples	Reina Sofia University Hospital	N/A
Chemicals, peptides, and recombinant proteins
S-MEM	Gibco	Cat#11380037
DMEM-D-valine	Sartorius	Cat# 06-1055-09-1A
BSA	Sigma Aldrich	Cat#A9418
HEPES	Sigma Aldrich	Cat#H3537
Trypsin	BD Biosciences	Cat#215250
DNAse I	Sigma Aldrich	Cat# 10104159001
Red Blood Cell lysis buffer	BIOLEGEN	Cat# 420301
Poly-L-Lysine	Sigma Aldrich	Cat# P1524-100 MG
Critical commercial assays
AllPrep DNA/RNA/Protein Mini kit	Qiagen	Cat#80004
RevertAid RT Reverse Transcription kit	Thermo Fisher	Cat#K1691
Preamp Master Mix	Standard BioTools	Cat#100-5581
Exonuclease I	New England Biolabs	Cat#M0293L
GE 96.96 Dynamic Array	Standard BioTools	Cat#BMK-M-96.96
Resazurin	Canvax	Cat#CAO35S
Octreotide	GP-Pharm	N/A
Pladienolide B	Santa Cruz Biotechnology	Cat# STC-SC-391691
Isoginkgetin	Sigma Aldrich	Cat#416154
NMDI14	MedChemExpress	Cat#HY-111374
Experimental models: Cell lines
Tumor patient-derived cell cultures	This paper	N/A
Oligonucleotides
See Table 2 with designed primers	This paper	N/A
Software and algorithms
Fluidigm Real-Time PCR Analysis Software v.4.7	Standard BioTools	https://www.standardbio.com/products/software
GeNorm v3.3	Vandesompele et al.79	https://genorm.cmgg.be
MetaboAnalyst Software v.6.0	McGill University	https://www.metaboanalyst.ca
GraphPad Prism v.8	GraphPad	https://www.graphpad.com/
Other
UV-visible spectrophotometer	Thermo Fisher	NanoDrop2000 spectrophotometer
BioMark HD System	Standard BioTools	N/A
VANTAstar System	BMG LABTECH	N/A

Experimental model and study participant details

All techniques carried out in this study were conducted in accordance with the ethical standards of the Helsinki Declaration, of the World Medical Association, and Ethics Committee approval was obtained (Code: S2400287). Written informed consent was obtained from all individuals included in the study or by a family member. Specifically, four human pituitary TSHoma samples (mean age: 39.5 (23–48); 50% men and 50% women] were collected during transsphenoidal surgery from two different hospitals of Spain [Virgen del Rocio University Hospital (Seville) and Virgen de la Victoria University Hospital (Malaga)]. Additionally, 11 Non-tumor Pituitary samples [NPs; mean age: 57 (44–79); 37% men and 63% women] were obtained from autopsies by an expert pathologist following standard autopsy procedures. Briefly, after scalp incision and removal of the cranial vault, the brain was extracted, allowing access to the pituitary gland. The pituitary was then carefully dissected, and representative tissue fragments were immediately transferred to the biobank staff for snap-freezing and storage at −80°C until further use. Tumor samples were immediately included in sterile cold S-MEM medium (Gibco) after surgery in hospitals participating in this study for immediate exam by an experienced pathologist following standardized diagnosis criteria. Then, the pathologist preserved a fragment for further anatomopathological analyses, and the remaining fragment was rapidly stored in RNA-later reagent (a solution used for RNA stabilization and storage that protects the integrity of RNA in unfrozen tissue samples) and immediately transported to the laboratory in Cordoba wherein RNA extraction was carried out using the AllPrep DNA/RNA/Protein Mini Kit (Qiagen, Cat#80004). Purity of RNA was analyzed using a Nanodrop system (Thermo Fisher Scientific) by assessing the A_260/280 and A_260/230 ratios, with values close to 2.0 considered indicative of minimal protein, DNA, and phenol contamination. Additionally, RNA integrity was assessed in available TSHoma samples by determining the RNA Integrity Number (RIN). Additionally, when tumor sample size was enough (two TSHoma cases), another tumor fragment was maintained in sterile cold medium and used to perform single-cell dissociation in order to obtain primary cell cultures (see section below). The diagnosis of TSHoma was confirmed by expert pathologists and by molecular screening using qPCR to measure key pituitary hormones and respective receptors. Additionally, immunohistochemical analysis of the transcription factor PIT-1 (Thermo Fisher Scientific, catalog number PA5-84020) was performed on the available formalin-fixed, paraffin-embedded TSHoma tissue sections. Staining was performed on an automated platform (VENTANA, Roche, Basel, Switzerland) according to the manufacturer’s instructions. This analysis was used to confirm tumor lineage assignment in accordance with the 5th WHO classification of pituitary tumors.⁸⁰ Representative images are shown in Figure S6.

Demographic and clinical characteristics from all patients/donors were collected to perform clinical correlations and are summarized in Table 1. Measurement of TSH and free T4 (fT4) levels was performed in the laboratory services of the different hospitals involved using specific assays (following the manufacturers’ instructions) without bias attributable to inter-assay variability (intra-laboratory/inter-laboratory) inside/between the different hospitals since all the assays were performed under the same quality control program [i.e., an internal quality control program and an external quality control program according to the indications of the International Federation of Clinical Chemistry (IFCC) and the American Association for Clinical Chemistry (AACC)]. Pituitary resonance imaging was conducted at each hospital using state-of-the-art resonance imaging equipment and patients were treated according to the available clinical guidelines, being surgery the treatment of choice in all patients. Moreover, three patients with TSHomas were pretreated with first generation somatostatin analogues (SSA; octreotide, n = 1; lanreotide, n = 2).⁷⁴

Table 1.

Demographic, clinical, and molecular information available from the internal cohort of patients with TSHoma (n = 4) and from individuals without pituitary tumor (n = 11)

	TSHomas	Normal pituitaries
Patient/sample’s number	4	11
Age (mean; range)	40 (23–48)	57 (44–79)
Sex (men/women; %)	50/50	37/63
Complete resection (yes/no)	3/1	–
Curation (yes/no)	3/1	–
Lesion size (mm; mean ± SD)	15.75 ± 8.1	–
Presurgery treatment (yes/no; with first-generation SSA)	3/1	–
TSHB mRNA level (normalized copy number)	6.6·106 ± 7.6·106	–
TSH level (μUI/mL)	10.3 ± 9.0	–
fT4 level (ng/mL)	3.6 ± 1.2	–

Cohort size, age, sex, clinical features (i.e., tumor size, pre-surgery treatment, curation), and molecular features (hormonal levels) are described.

Method details

RT-qPCR custom dynamic array

RNA (1 μg) was retrotranscribed to cDNA using random hexamer primers. qPCR dynamic arrays based on microfluidic technology was implemented to determine the simultaneous expression of 184 transcripts: i) the molecular components belonging to different cellular machineries controlling RNA metabolism [Spliceosome (12 factors from major spliceosome, 3 from minor spliceosome, and 64 splicing factors); RNA-Exosome (9 structural core elements, 3 catalytic core nucleases, and 14 cofactors), NMD (60 components and 15 NMD targets)]; ii) TSHB transcript levels; and, iii) 3 housekeeping genes [β-actin (ACTB), hypoxanthine-guanine phosphoribosyl-transferase (HPRT), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH)]. The thermocycling profile consisted of 1) 95 °C for 1 min; 2) 35 cycles of denaturing (95 °C for 5 s) and annealing/extension (60 °C for 20s); and 3) a last cycle where final PCR products were subjected to graded temperature-dependent dissociation (60 °C–95 °C, increasing 1 °C/3 s). Housekeeping genes were measured in order to normalize variations in RNA quantification and reverse transcription reaction efficiency using a Normalization Factor (NF) calculated with the GeNorm v3.3 software. All these primers were collected along with their relevant features in Table 2. Microfluidic arrays were measured using the BioMark HD System and the Fluidigm Real-Time PCR Analysis Software v.4.72 (Standard BioTools). Gene selection was based on previous bibliography reporting functional relevance in the respective biological process (splicing and decay by RNA-Exosome or NMD) and significant alterations in expression levels comparing tumor vs. non-tumor tissues.^{17,18,22,27,29,56,75,76,81,82,83}

Table 2.

Specific primers for human transcripts used in this study, including spliceosome, splicing factors, RNA-exosome, nonsense-mediated decay, TSHB, and housekeeping genes

Gene	Accession number	Primer sequence (Sense, sn)	Primer sequence (antisense, As)	Product size (bp)
Splicing
CUGBP1	NM_001376437.1	AACAGAAGAGAATGGCCCAGC	TGCTGAAGGAGTGCTAAATACTGG	121
CELF4	NM_020180.3	CCCCAGCAGCAGAGAGAA	GAAGCCGAAAGGGAGGAA	108
ESRP1	NM_020180.3	TTTTGGGATCACTGCTGGGG	TGTCCCACCTTCTTGTTGGC	108
ESRP2	NM_024939.2	AGAGCCCAGCAGTCAATTGTT	GTCTCACTGTCCACCACATCAG	96
SAM68	NM_006559.3	GAGCGAGTGCTGATACCTGTC	CACCAGTCTCTTCCTGCAGTC	106
NOVA1	NM_002515.2	TACCCAGGTACTACTGAGCGAG	CTGGTTCTGTCTTGGCCACAT	124
PRPF40A	NM_001354431.4	GCTCGGAAGATGAAACGAAA	TGTCCTCAAATGCTGGCTCT	130
PTBP1	NM_002819.4	TGGGTCGGTTCCTGCTATT	CAGATCCCCGCTTTGTAC	111
RAVER1	NM_133452.2	GTAACCGCCGCAAGATACTG	CGAAGGCTGTCCCTTTGTATT	126
RBM17	NM_032905.4	CAAAGAGCCAAAGGACGAAA	TACATGCGGTGGAGTGTCC	107
RBM22	NM_018047.2	CTCTGGGTTCCAACACCTACA	GGCACAGATTTTGCATTCCT	137
RBM3	NM_006743.4	AAGCTCTTCGTGGGAGGG	TTGACAACGACCACCTCAGA	98
RBM45	NM_152945.3	CCCATCAAGGTTTTCATTGC	TTCCCGCAGATCTTCTTCTG	123
RNU1	NR_004430.2	ATCACGAAGGTGGTTTTCC	GCAGTCGAGTTTCCCACA	94
RNU11	NR_004407.1	AAGGGCTTCTGTCGTGAGTG	CCAGCTGCCCAAATACCA	108
RNU12	NR_029422.1	ATAACGATTCGGGGTGACG	CAGGCATCCCGCAAAGTA	106
RNU2	NR_002716.3	CTCGGCCTTTTGGCTAAGAT	TATTCCATCTCCCTGCTCCA	116
RNU4	NR_003925.1	TCGTAGCCAATGAGGTCTATCC	AAAATTGCCAGTGCCGACTA	103
RNU5	NR_002756.2	TGGTTTCTCTTCAGATCGCA	GTTGTTCCTCTCCACGGAAA	65
RNU6	NR_004394.1	CGCTTCGGCAGCACATATA	AAAATATGGAACGCTTCACGAA	101
RNU6ATAC	NR_023344.1	TGAAAGGAGAGAAGGTTAGCACTC	CGATGGTTAGATGCCACGA	112
SF3B1 TV1	NM_012433.3	GCAGACCGGGAAGATGAATA	TTTTCCCTCCATCTGCAAAA	88
SFPQ	NM_005066.2	TGGTAGGGGGTGAAAGTG	TTAAAAACAAGAAATGGGGAAATG	125
SND1	NM_014390.3	ACTACGGCAACAGAGAGGTCC	GAAGGCATACTCCGTGGCT	101
PRP45	NM_001318844.2	ATGCGTGCCCAAGTAGAGAG	TCCCCATCCTCTTTTTCCA	134
NSR100	NM_194286.4	CCTTCACCACCTCCTCACC	TTCGGCACATTCCAGACAC	113
SRSF10	NM_006625.5	CTACACTCGCCGTCCAAGAG	CCGTCCACAAATCCACTTTC	103
SC35	NM_001195427.2	TGTCCAAGAGGGAATCCAAA	GTTTACACTGCTTGCCGATACA	113
SRP20	NM_003017.5	TAACCCTAGATCTCGAAATGCATC	CATAGTAGCCAAAAGCCCGTT	117
SRP75	NM_005626.5	GGAACTGAAGTCAATGGGAGAA	CTTCGAGAGCGAGACCTTGA	110
SRP40	NM_001039465.2	GCAAAAGGCACAGTAGGTCAA	TTTGCGACTACGGGAACG	92
SRP55	NM_006275.6	AGACCTCAAAAATGGGTACGG	CTTGCCGTTCAGCTCGTAA	82
SRP30C	NM_003769.3	CCCTGCGTAAACTGGATGAC	AGCTGGTGCTTCTCTCAGGA	87
TCERG1	NM_006706.3	GAGGAGCCCAAAGAAGAGGA	CACCAGTCCAAACGACACAC	112
TIA1	NM_022037.2	TAAATCCCGTGCAACAGCAGA	TATGCAGGAACTTGCCAACCA	124
TRA2B	NM_004593.2	GATGATGCCAAGGAAGCTAAAG	AGGTAGGTCTCCCCATGTAAATTC	130
TRA2Α	NM_013293.4	TCAAAGGAGGCTATGGAAAGG	TGTGTGCGCTCTCTTGGTTA	90
U2AF35	NM_006758.2	GAAGTATGGGGAAGTAGAGGAGATG	TTCAAGTCAATCACAGCCTTTTC	120
U2AF65	NM_007279.2	CTTTGACCAGAGGCGCTAAA	TACTGCATTGGGGTGATGTG	130
DHX9	NM_001357.5	ACAGGTTCCCCAGTTCATTCT	TTTTTCCAGGCTCTTCTCCTC	146
ELAVL4	NM_001324216.1	AGCAAAACCAACCTCATCGT	TCCATACCCTAAACTCTGTCCTGT	141
HNRNPA1	NM_001396241.1	AAAGCCCTGTCAAAGCAAGA	AGTTGTCATTCCCACCGAAA	112
HNRNP A2/B1	NM_001438570.1	CAGAGTTCTAGGAGTGGAAGAGGA	CCATTATAGCCATCCCCAAA	149
HNRNPF	NM_001098205.2	AGTCCCACAGAACCGAGATG	CCAACCCTGAGAAGAACTGAAC	144
HNRNP K	NM_031263.4	CTGGGGTGTCAGTTGTTGG	TGGTTTCAGTGTTAGGGAAGG	141
MBNL1	NM_001387805.1	CGCAGTTGGAGATAAATGGA	GGGCACCAGGCATCA	102
MBNL2	NM_001382675.1	ACCACGCCTGTTATTGTTCC	TCCCTGCATACCTCCAGTTT	101
PRPF19	NM_014502.5	CCAAGTTCCCAACCAAGTGT	GGCACAGTCTTCCCTCTCTTC	146
RBM10	NM_005676.5	CAGCACTCCCTCAACATCCT	AGCACTTCTCTCGGCGTTT	127
RBM4	NM_001198843.2	GTCCCACCTGCACCAATAAG	CCGCTCCATGTGTACGAAG	104
RBM5	NM_005778.4	TCAGGCACCAGCAACTCTC	CGGTCTCGGTATTTCATCTCTC	124
RBM6	NM_005777.3	CCAGGATGGAGAGAGCAAAA	CAGTAGTAAGGCGGACATAGGG	104
RBM39	NM_001242600.2	AGTTGGATGGGATACCGAGA	TTGCCCTGAGCTGAATTTTT	102
RBM25	NM_021239.3	GCTAAATGCCCCCTCACAG	CTGGAAATCTGCGGAAAATG	86
RAVER2	NM_018211.4	TGGGAGAACCACCAAAAGAA	GCAGGGGATGATAAGCACAC	91
SRPK1	NM_003137.5	AAATGGAGACAGCAGCACATC	TGAGGAAGACTGGCACACC	91
SLU7	NM_001364521.2	TGACCAGAGAGGACTGGAGAA	GAGGAATATGGGGGTTGATG	111
SLM2 (KHDRBS3)	NM_006558.3	TGGTGCTGATTACTATGATTACGG	CTTTGCTGTCCTCGCTGAA	115
SF3B2	NM_006842.3	CTGCCAAACAGAAGCAAAAA	TGTGAGGGGACCTAAAACTTG	97
HNRNP G (RBMX)	NM_001162536.3	AGAGATTATGCACCACCACCA	CACGATCACGACCATATCCA	118
HNRNP A3	NM_001395170.1	ATGGGGCACACTCACAGATT	GCATCCACCTCTTCAACACA	102
SART1	NM_005146.5	GGGCAGAGAAAAATGTGGAG	TGGACAGGATAGAGCGAGGT	111
KHSRP	NM_001366299.1	GCTGCCACGACAGTGAATAA	TCTCCGGTTGATCTCCATCT	85
LSM2	NM_021177.5	AACAGTGATGGCTCCTCCTC	TATTGGGGGTTAGGGGTTCT	90
SNRPD3	NM_001278656.2	GTTGACGCAGATTGAAAACAAG	GTGGGGTAAAGGCAGGGTAG	80
DHX15	NM_001358.3	TCTACACTTCCACCTCAGCAGCA	CCAGGATCAATCACAAACACCACAC	152
SNRPD1	NM_006938.4	CCCGGCCTATACCTTTCATT	GCCTTTTGCCTTCTACTTGG	120
SNRPB2	NM_198220.3	ATGCGTGGAACTTTTGCTG	CAGGCTTTTTGTTTGTGGTTG	94
ELAVL1	NM_001419.3	GCTGGGACAGGTGAGAAGAA	CGGAGACCACATACAATGAGAA	97
HNRNPAB	NM_004499.4	GCACACATGCTTTGTTTGGA	CACATGGGACAGGCAATAGA	85
SNRPA1	NM_003090.4	TGCCGTATAGGTGAGGGACT	GCGATTTGAGAGATGCCAGA	115
SNRPC	NM_003093.3	TACCTCACCCATGACTCTCCA	CCTGCTCTTCCATCCATTTC	103
DDX1	NM_004939.3	AGGTTGAGCCGGATATAAAGG	TATAGCTTCCACCACCAGCA	90
DHX8	NM_001322217.2	AGCCTGAGCATGAAGGATGT	AAGGGACAAGTGAGTGGGTCT	132
EIF4A3	NM_001130678.4	ACAGAGCCCAAAAGTCTTCAAC	ACATCCCCTACAGAACAGATGG	119
HNRNPL	NM_001005335.2	GTGGTGGAAGCAGACCTTGT	CCCCCAACACATCTTCAAAC	112
MAGOH	NM_002370.3	GCCAACAACAGCAATTACAAGA	TTATTCTCTTCAGTTCCTCCATCAC	88
SRRM1	NM_001303448.1	GTAGCCCAAGAAGACGCAAA	TGGTTCTGTGACGGGGAG	108
SRSF1	NM_006924.4	TGTCTCTGGACTGCCTCCA	TGCCATCTCGGTAAACATCA	98
RNA-exosome
EXOSC1	NM_016046.5	TATAAGAGTTTCCGCCCAGGT	ACCACTCCCAGCTCGTTCT	107
EXOSC2	NM_014285.7	GGCCGCGACACTAAGAAA	ATGCAATGAGCTTCTCTTCTCC	109
EXOSC3	NM_016042.4	CGGTGTTTACTGGGTGGACT	TCACTCCCTCCAACATCAACT	120
EXOSC4	NM_019037.3	CACGCACACACAAAGTCTCTC	AGGTGCTACAGGCAGATGGT	103
EXOSC5	NM_020158.4	CCTACATCCAAGCAAGAAAAGG	AGTGTCTGAGTAGAGCCCCTTG	108
EXOSC6	NM_058219.3	GACCCAACGCGGCTAC	CGACACGGCACACAGC	109
EXOSC7	NM_015004.4	CCCCATTTCTGCTTTCACTC	TGAGGACTACCGATGTGTCG	115
EXOSC8	NM_181503.3	CAACTTCCTTTGCTGTGTTTGA	TTTGCCTTCCTCATCCATTACT	110
EXOSC9	NM_001034194.2	TTTGGAACAGATTACGGATGC	GTTGCCCGATTGAGTTTTG	101
EXOSC10	NM_001001998.3	CATCAAACCCAATGCTCAGA	GATGAAATCAGCCAGTGCAG	115
DIS3	NM_014953.5	CGAAAACATCCTGCTCCAC	TCCAAAGACTCAGCCAAAGAC	113
DIS3L	NM_001143688.3	GATGTGTGAGATGCCAGTAAACA	GGGTCAATGCTGAATACGAGA	111
ZCCHC7	NM_001289121.2	GCTAATAACCGAACACCTGGA	CGTGGTAAGGGGCAGTTTT	116
TENT4B	NM_001040284.3	GTCCTCTTCTGCCACACAGTC	CATCTTGCGACCCTACTCG	119
SKIV2L	NM_006929.4	ACGGCATGACTCTGTCTTTGT	TGATGGGCGAAGTGTAGATG	119
MTREX	NM_015360.5	AAGGGAAAAGCAGCGTGTAA	TCACCAGTCATCAAACCAACA	108
RBM7	NM_001286045.2	CAGGATTTTCACCATCAGTTCA	TTCATTCTGACTTTACGCTGTGA	106
ZCCHC8	NM_017612.5	CACGCAGAAGAAGTAGAAGAAAGAT	TCCGATAGATAAAAGGTGGAAGAC	115
TTC37	NM_014639.4	AGCCTATGAGAGAGCCTTGTCTATT	GCTACATCCGTTTTTCCTTGTT	105
WDR61	NM_001303247.1	ACTCACTTTGTTTCCAGTTCGTC	CATTGTATTTTACTCCCCAGACCT	118
ZFC3H1	NM_144982.5	TCATCGTCCTCTTCTCAGCA	GGTTCTTTGGGTCGGTAGC	104
BCL2L2-PABPN1	NM_004050.5	ATTAAGGGGTTTGCTGCTGA	GGGTGAAGAGTGTGCTCCAT	108
PABPN1	NM_004643.4	TAAAGGAGCTACAGAACGAGGTAGA	ATAGATGGAACGGGCATCAG	119
C1D	NM_006333.4	TGGCAACCCAAGGAGTTAAT	GCCTTTTTCTTGTCTGTTATTTCC	106
HBSL1	NM_001145207.2	GGCATCGGAATGTTCGAG	CTGTTGACGGCGAAATACAA	102
MPHOSPH6	NM_005792.2	AAACAGTAGAGCTTGATGTGTCAGA	CTTCATAATTGGCATGGTCTCTC	111
Nonsense-mediated decay
ALYREF	NM_005782.4	TTTGGAACGCTGAAGAAGGC	CGCCACGGTTTCTAGTCATG	232
AQR	NM_014691.3	CACAACTTCCCAGAACAGAGG	TGACCAAGACGCAGTAGGTG	119
CASC3	NM_007359.5	CAAAGTTGGAGATGCAGTCAAG	TCCCAGGTCCCAGTCTTAGTAG	120
CNOT1	NM_001265612.2	GAGGGGTTTGGGTATGGAA	GGGAATGTTGAATGAAGGAGAG	92
CNOT10	NM_001393368.1	AGAAATGCCTTGTTGCTGCT	TGCTTTCGCTGCTCTCTGT	106
CNOT11	NM_017546.5	TGGTGTGGAGATCAAACGAA	TGGGGTCTTTTTCCAACTCA	98
CNOT2	NM_001414662.1	TTATCAGGGCAGCAGAGACA	TTGGGGTAGAGATTTTCAGGAG	106
CNOT3	NM_014516.4	GAATACCATCGACACGCTCA	ATCCGGTCCTGCTTATCCTT	111
CNOT4	NM_001393375.1	CCAGAGGATGACTTGGGTTTT	CGAAAGGGAAGGTTGGTCTT	102
CNOT6	NM_001370474.1	GGCTGCTGAACTATTTGCTTG	GTTGGCCTTGTCCTATCTGGT	115
CNOT6L	NM_001387839.1	GGAACCCGAAAGCTACTGAAC	CTGACGGCAGAATTTGGTCT	112
CNOT7	NM_001322088.2	TTGTGAAGTTTGGGCTTGC	CGGTGTCCATAGCAACGTAAT	89
CNOT8	NM_001301080.2	CTGAACCTTTTCTTCCCATCC	TTCCAATCCTCTGCAAATCC	115
CNOT9	NM_005444.3	GCTGTGGCATTCATTTGGT	TAGACTGGTGTGCTGTCAAGGT	98
CWC22	NM_001376032.1	TATTAGCTGGGCGATTTTGC	TTCCAAGCGATGGATGGTAT	95
DCP1A	NM_001290207.2	AGACAGCAGCAGCAAGAGTG	AGGCTGCACAAATACTTCAGG	107
DCP1B	NM_001319292.2	CCCTATATCAACCGCATCGT	GGTTCCTTCCACATCAGTTTTC	99
DCP2	NM_001242377.2	ATGACCCCCAAATCCAAACT	TCGAGAAAGCCAGTCCCTTA	90
DDX17	NM_001098504.2	TTTCCGTTGGCTCTTAGTGG	TCCAAGTATGGCTGGTGGTT	116
DDX6	NM_001425148.1	AACCTGATGGAGGAACTAACTCTG	TGAAGCCTGGAGAAAAGTGTG	104
DHX34	NM_014681.6	AAACCGCCATCCTCTACCTC	CAGCATCTTCCCAATCACAA	116
DHX8	NM_001322217.2	AGCCTGAGCATGAAGGATGT	AAGGGACAAGTGAGTGGGTCT	132
DIS3L2	NM_001257281.2	CGAGGAGCATTGGAAGGTAG	ACTGTTGCGGGAGATGGT	110
EDC4	NM_001427345.1	ACAGAACAGGAGGCAGGAGA	ACATTGAGGGGCAAGATGAG	83
EIF4A3	NM_001130678.4	ACAGAGCCCAAAAGTCTTCAAC	ACATCCCCTACAGAACAGATGG	119
EIF4G1	NM_198244.3	CCGGGGAACAGAAGTATGAA	GAACTGAAAACCAAGCAGGAAC	101
GNL2	NM_001323623.2	GTCCAAAAGAATATGGGGTGAG	TCAATGTGAGGGGAACGAG	108
GSPT1	NM_002094.4	ACCTTTTCTTACTGGCATGGA	ACTGCGTATTTTCCCCTTGA	105
GSPT2	NM_018094.5	TCCAAATCTGTGATCGTACCC	TGTCCTCCGATGGTTGACTT	101
HNRNPL	NM_001005335.2	GTGGTGGAAGCAGACCTTGT	CCCCCAACACATCTTCAAAC	112
ICE1	NM_015325.3	TGTGGCCTGTGATGGATAAA	TGGCCTAAACGACCAATCA	112
MAGOH	NM_002370.3	GCCAACAACAGCAATTACAAGA	TTATTCTCTTCAGTTCCTCCATCAC	88
MAGOHB	NM_001319985.2	GTCCCCGTTCTCTAACTTCCTT	TGACCTAGTCTCCCATCTTGGT	100
MOV10	NM_020963.5	ACAGGTCTCATCCCACCATC	GTTCTCGATCCACGACATCA	90
NBAS	NM_015909.4	CAGACCAGAGCAGAGGAAATAGA	CACAGAGAACCAGCAAACCA	109
NCBP2	NM_001308036.2	CGTCTGGATGACCGAATCA	TCGTAGTCCTGCCGATACTCA	116
PABPC1	NM_002568.4	AGTTCGCAATCCTCAGCAAC	CCAACATGGAAGCAGTCAAA	104
PARN	NM_002582.4	AGATAGCTGGAAGGAGGCTGA	ACTGTGCTGGGAGCTGTAAAA	108
PNRC2	NM_017761.4	GTATAACATTCCAGCCCCTCAA	TCTGTTCCTTGGTCTTCTGTCTG	80
PPP2CA	NM_001355019.2	GGTTACACCTTTGGGCAAGA	GCACCAGTTATATCCCTCCATC	105
PPP2R1A	NM_014225.6	TGGGAGTGGAGTTCTTTGATG	CTAGCTTCTTCAGGTTGCTGGT	108
PPP2R2A	NM_002717.4	TGGTTACCCCAGAAAAATGC	CTGGTCTTTTGTCCCTTTCACT	94
RNPS1	NM_080594.4	TCTTCTGGCTCTCCAAGTCC	CGCCTTTTCCTCTCCTTTTC	104
RUVBL1	NM_003707.3	TCTATCGCTCCCATCGTCA	CATCACTCGGTCCAGAAGGT	114
RUVBL2	NM_001321190.2	ACAGAAGTGCAGGTGGATGAC	CTCGCCTTTGAGTTCGTTG	120
SEC13	NM_001136026.3	CCTCCCATGAGGACATGATT	CCATTGCGCACATCAAAG	106
SMG1	NM_015092.5	GAGCCAAAGAGCAAGTCAGG	AGTTCAAATAACGCCGCATC	111
SMG5	NM_001323615.2	CCCCTCATAGGATGCAAGAA	GGACAAATCCCCCAGATACA	102
SMG7	NM_001331007.2	ATGAAACCGAGCAGCACAC	GCACAGGATGCCAAGAAAA	92
SMG8	NM_018149.7	ACCAACAGGGCTTTATTCCA	AGCAGGCCAAGAGTTTGTGT	107
SMG9	NM_019108.4	ATTGCTGCCTTCCTTTTCAC	GGCTTCACCATCTCTGCTGT	107
SRRM1	NM_001303448.1	GTAGCCCAAGAAGACGCAAA	TGGTTCTGTGACGGGGAG	108
SRSF1	NM_006924.4	TGTCTCTGGACTGCCTCCA	TGCCATCTCGGTAAACATCA	98
TUT4	NM_001009881.3	GAACAACACAACAGGGAGCA	GAAGAGCCAAATAAGCACAACC	95
TUT7	NM_001185074.2	GGCTGATCTTGATGGAGAAAGT	CCTGTACTGAGTGGGTAAAGTGG	80
UPF1	NM_002911.4	CTGCACACCAAGCTCTACCA	CACACAGGACAGGATGATGAA	90
UPF2	NM_080599.3	CAATGAACGGCAAGAACAAG	CTCCCTTCGGATGTTGGTAG	116
UPF3A	NM_001353647.1	ACAGTCCAGCACCCAGAAAA	CTGTTTCCGCCACACTCTC	106
UPF3B	NM_080632.3	GATAAGCAGGATCGCAACAAG	TTGGTCAAAGTGGGAGGTAATC	80
XRN1	NM_001282857.2	AGATTTTGATCGGGAGCACTT	CTTCTTCTGCTGCGACACCT	110
NMD targets
ATF3	NM_001674.4	CCTCTGCGCTGGAATCAGTC	TTCTTTCTCGTCGCCTCTTTTT	111
ATF4	NM_001675.4	CCCCTTCACCTTCTTACAAC	CTTCACTGCCCAGCTCTAAA	92
CDKN1A	NM_000389.5	TGGAGACTCTCAGGGTCGAA	GGATTAGGGCTTCCTCTTGG	97
DDIT3	NM_001195053.1	TGGAAGCCTGGTATGAGGAC	CAGGGTCAAGAGTGGTGAAGA	120
TSTD2	NM_139246.5	CATGCTTTCCTTCCCATTGTT	ATGGGCACGATTTCTTCAAA	120
FSD1L	NM_001330739.2	CGTGTAAAGGATGAGCGATG	GGTTCAAATGGGATGAGGAG	208
TRAF2	NM_021138.4	AAGCCCTGAGTAGCAAGGTG	TGAAGACCCCATCGTAGGTG	120
GADD45A	NM_001924.4	GGAGGAATTCTCGGCTGGAG	CGTTATCGGGGTCGACGTT	152
GADD45B	NM_015675.4	TTGTCTCCTGGTCACGAACC	TGTGGCAGCAACTCAACAGA	170
MAFF	NM_012323.4	CGTAAGTCACAGACAGCCAA	CCCTCATCTTGGAAACTGCT	90
MAP3K14	NM_003954.5	CTGTCCCAGCCATTTTCTCT	GATGGGTTCTTCTCACTGTC	98
NAT9	NM_015654.5	ATTGTGCTGGATGCCGAGA	ACCTAGCGTGGTCACTCCGTA	210
PDRG1	NM_030815.3	GACCTGGACACTAAAAGGAA	CGAAGCAAACCATCACATCT	91
PANK2	NM_001324191.2	GGGATATAGACGGGAGCCAT	CCACCGATATCCAGTCCAAA	114
RASSF1	NM_007182.5	AAATGACTCTGGGGAGGTGAA	TTCTGGCGGCAATAGGAGTA	132
HK
ACTB	NM_001101.5	ACTCTTCCAGCCTTCCTTCCT	CAGTGATCTCCTTCTGCATCCT	176
GAPDH	NM_002046.7	AATCCCATCACCATCTTCCA	AAATGAGCCCCAGCCTTC	122
HPRT1	NM_000194.3	CTGAGGATTTGGAAAGGGTGT	TAATCCAGCAGGTCAGCAAAG	157

They were specifically designed and used in qPCR-based microfluidic. The official name of the genes, species, NCBI accession number of the transcripts, primers sequences (Sense and Antisense), and product sizes of the amplification products are included.

Cell proliferation

Furthermore, two available fresh TSHoma samples were transferred into sterile cold medium after surgery [S-MEM (Gibco), supplemented with 0.1% bovine serum albumin (BSA), 0.01% L-glutamine, 1% antibiotic-antimycotic solution, and 2.5% 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)] and were dispersed into single cells. Briefly, a mechanical and enzymatic [trypsin (BD Biosciences) and DNase I (Sigma Aldrich)] protocol was performed until the tissue was completely dissociated. Afterward, red blood cells were removed using the Red Blood Cell lysis buffer (Biolegend). Then, single cells were cultured onto poly-L-Lysine-coated tissue culture plates in DMEM-D-valine (Sartorius) containing 10% Fetal Bovine Serum (FBS) and complemented as S-MEM medium, under sterile conditions in a humidified atmosphere at 37 °C and 5% CO₂.

Cell proliferation/viability of TSHoma cell cultures was evaluated in response to different experimental treatments every 24h (up to 120 h), seeding 15,000 cells/well in 96-well plates. Specifically, cells were treated with i) a pharmacological gold-standard SSA (octreotide; 10⁻⁷ M); ii) a splicing inhibitor (pladienolide B; 10⁻⁷ M); iii) an RNA-Exosome inhibitor (isoginkgetin; 10⁻⁵ M); and iv) an NMD inhibitor (NMDI14; 10⁻⁵ M). Since dimethyl sulfoxide (DMSO) was used to dissolve all these compounds, DMSO at 0.1% in control and treatment wells was present. Doses of each treatment were based on bibliography.^{22,56,75,84,85,86,87} Treatments were daily refreshed after each measurement and three technical replicates per condition were performed using the Resazurin reagent (Canvax) and the VANTAstar system (BMG LABTECH), blanking with only medium wells and measuring fluorescence intensity (excitation at 540 nm, emission at 590 nm). Demographic and clinical characteristics from the two patients whose samples were used on these in vitro measurements are summarized in Table 3.

Table 3.

Demographic, clinical, and molecular available information from the patients with TSHoma whose samples were used for in vitro experiments (n = 2)

	TSHoma-1	TSHoma-2
Tumor type	TSHoma	Mix GH/TSHoma
Age at diagnosis	22	67
Sex	male	female
Surgery	transsphenoidal	transsphenoidal
Complete resection	yes	yes
Curation	yes	NA
Lesion size (mm)	10	18
Presurgery treatment	yes (lanreotide)	naive
TSH level (μUI/mL)	2.1	4.82
fT4 level (ng/mL)	2.22	1.6

Type, age, sex, clinical features (i.e., tumor size, pre-surgery treatment, curation, surgery), and molecular features (hormonal levels) are provided.

Quantification and statistical analysis

Heatmaps, Partial Least Squares Discriminant Analysis (PLS-DA), Volcano plots, and VIP Scores analyses were performed with MetaboAnalyst Software v.6.0 (McGill University). All data are represented as mean ± SEM, except for boxplots which represent the median and quartiles of each gene expression. All statistical analyses, including fold change plots, clinical associations, and in vitro results were conducted using GraphPad Prism v.8.0 (GraphPad software). Normality was assessed using Shapiro-Wilk or Kolmogorov-Smirnov tests. Consequently, either parametric [Student’s t test (paired or unpaired)] or nonparametric (Mann-Whitney U) tests were applied for quantitative parameters. Pearson coefficient was determined to assess statistical correlation among gene expression and clinical features. Statistical significance was set at (∗) p ≤ 0.05, (∗∗) p ≤ 0.01, (∗∗∗) p ≤ 0.001, (∗∗∗∗) p ≤ 0.0001, compared to control conditions.