A Study of Alternative TrkA Splicing Identifies TrkAIII as a Novel Potentially Targetable Participant in PitNET Progression

Simple Summary Pituitary neuroendocrine tumors (PitNETs) develop from anterior pituitary cells and, although generally benign, comprise a small subset of therapy-resistant aggressive or metastatic tumors. This highlights the need to identify novel potential therapeutic targets. PitNETs have low rates of somatic mutation and their pathogenesis is poorly understood. PitNETs are associated with conditions linked to alternative splicing, which may activate oncogenic pathways, and express the neurotrophin receptor tropomyosin receptor kinase A (TrkA), which exhibits oncogenic alternative TrkAIII splicing in other neuroendocrine tumors. In this study, we report for the first time that alternative TrkAIII mRNA splicing is common in PitNETs and can associate with intracellular TrkAIII activation, identifying TrkAIII as a novel potential targetable oncogenic participant in PitNET pathogenesis and progression. Abstract Pituitary neuroendocrine tumors (PitNETs) are generally benign but comprise an aggressive, invasive, therapy-resistant, metastatic subset, underpinning a need for novel therapeutic targets. PitNETs exhibit low mutation rates but are associated with conditions linked to alternative splicing, an alternative oncogene pathway activation mechanism. PitNETs express the neurotrophin receptor TrkA, which exhibits oncogenic alternative TrkAIII splicing in other neuroendocrine tumors. We, therefore, assessed whether TrkAIII splicing represents a potential oncogenic participant in PitNETs. TrkAIII splicing was RT-PCR assessed in 53 PitNETs and TrkA isoform(s) expression and activation were assessed by confocal immunofluorescence. TrkAIII splicing was also compared to HIF1α, HIF2α, SF3B1, SRSF2, U2AF1, and JCPyV large T antigen mRNA expression, Xbp1 splicing, and SF3B1 mutation. TrkAIII splicing was detected in all invasive and most non-invasive PitNETs and was significantly elevated in invasive cases. In PitNET lineages, TrkAIII splicing was significantly elevated in invasive PIT1 PitNETs and high in invasive and non-invasive SF1 and TPIT lineages. Immunoreactivity consistent with TrkAIII activation characterized PitNET expressing TrkAIII mRNA, and invasive Pit1 PitNETs exhibited elevated HIF2α expression. TrkAIII splicing did not associate with SF3B1 mutations, altered SF3B1, SRSF2, and U2AF1 or JCPyV large T antigen expression, or Xbp1 splicing. Therefore, TrkAIII splicing is common in PitNETs, is elevated in invasive, especially PIT1 tumors, can result in intracellular TrkAIII activation, and may involve hypoxia. The data support a role for TrkAIII splicing in PitNET pathogenesis and progression and identify TrkAIII as a novel potential target in refractory PitNETs.

Despite being typically benign, approximately 40% of PitNETs invade surrounding structures, and a small proportion of mainly lactotroph and corticotroph PitNETs develop into aggressive, therapy-resistant/refractory, sporadically metastatic tumors [2,6,7].Aggressive and metastatic PitNETs share a number of bio-clinical features, despite a lack of specific molecular markers [8], and current guidelines recommend treatment with temozolomide [6,7].However, primary and secondary therapeutic resistance to temozolomide is frequent [7], emphasizing the need to identify novel therapeutic targets for this specific subgroup, which continues to pose a significant therapeutic challenge.
Alternative TrkAIII splicing in NB cells is promoted by hypoxia, agents that cause endoplasmic reticulum (ER), Ca 2+ , redox, and nutrient stress, and by the simian vacuolating polyomavirus virus 40 (SV40) large T-antigen [25,30,36].Considering that PitNETs exhibit low mutation rates, express TrkA, and associate with conditions linked to alternative splicing and polyomavirus infection, we investigated alternative TrkAIII splicing as a potentially targetable participant in PitNET pathogenesis and progression.Overall, the data support a role for alternative TrkAIII splicing in PitNET pathogenesis and progression, potentially involving hypoxia, and identify TrkAIII as a novel potential therapeutic target in refractory PitNETs.

Patients and Tumors
PitNETs from 53 patients were surgically removed at the Neuromed Institute (Pozzili, Italy).Prior to surgery, all patients were characterized for bio-clinical evidence of hormone hypersecretion and, by Magnetic Resonance Imaging (MRI), for macroscopic tumor characteristics.In total, 24 patients were clinically diagnosed with functioning PitNETs (6 prolactinomas, 13 acromegaly, 2 central hyperthyroidism, and 4 Cushing's disease), and the remaining 29 were diagnosed with clinically non-functioning tumors.With the exception of a young female with a micro-prolactinoma, all other patients had macro-tumors (maximal diameter > 10 mm).Tumor invasion of surrounding structures (cavernous sinus/sphenoid sinus/bone/dura) was identified by pre-operative MRI and surgical findings.Overall, 26 of 53 PitNETs and 4 of 6 recurrent PitNETs were invasive (49%), of which 2 were aggressive and 1 was metastatic.Routine immunohistochemical (IHC) pathological tumor classification and diagnosis were performed in accordance with European Pituitary Pathology Group proposals [39], using primary antibodies directed against pituitary hormone, transcription factors, and Ki67 (MIB1clone).Analyses were performed using an Ultraview DAB detection kit (Roche Diagnostics Int.; Rotkeuz, Switzerland) in an automatic VENTANA Benchmark ultra XT IHC/ISH System, as directed (Roche Diagnostics Int.; Rotkeuz, Switzerland).
The PitNETs examined in this study were classified according to their lineage of origin as follows: 24 PIT1, 24 SF1, and 5 TPIT-positive tumors, the details for which are provided in Table 1.For molecular studies, surgical tumor fragments were immediately placed in RNAlater TM nucleic acid stabilizing solution, as directed (Ambion ® , Life Technologies, Monza, Italy), and frozen at −80 • C prior to nucleic acid purification.In some cases, slide-mounted 4 µm FFPE PitNET tissue sections were also provided for confocal immunofluorescence analysis.This study was approved by the Neuromed Institute Internal Review Board, as a part of the Biopit study (Biopit 270423), and performed according to Helsinki declarations.Written informed consent was obtained from patients, with the exception of a minority of archived RNAs from patients lost to follow-up.

RNA Extraction and Reverse Transcriptase Polymerase Chain Reaction
Total RNAs were extracted from tissues using Trizol, according to the manufacturer's instructions (Life Technologies, Monza, Italy).Briefly, tumor tissues were homogenized in 1 mL of Trizol, and resulting supernatants were mixed with chloroform and centrifuged to obtain phase separation.The upper phase was recovered and washed in isopropanol, RNAs were then precipitated in 75% ethanol and centrifuged at 14,000× g in an Eppendorf microfuge at 4 • C, and RNA pellets were resuspended in 20 µL of RNase/DNAse-free water.RNA purity and concentrations were evaluated in a nanodrop spectrophotometer, as directed (Thermo Fisher Scientific, Carlsbad, CA, USA).Purified RNAs were reverse-transcribed using a Wonder RT transcription kit, as directed (Euroclone, Pero, Italy), and reverse transcription reactions, at various dilutions, were subjected to RT-PCR, using the primers and conditions detailed in Table 2.All RT-PCRs were performed in duplicate and repeated at least 2 times.For densitometric analysis, 1.5% agarose gels were digitally photographed and images analyzed by Image J software (ImageJ bundled with Java 1.8.0_172), with inter-gel comparisons performed using common 18S rRNA RT-PCR product and DNA ladder standards, where appropriate.

Tumor DNA Purification
Tumor DNA (tDNA) was extracted from 8 PRL PitNETs using Quick-DNA Miniprep Plus Kit, as directed (Zymo Research, Irvine, CA, USA).DNA quality was checked by 0.8% agarose gel electrophoresis and PCR amplification for the housekeeping gene GAPDH [40].

Statistical Analysis
Data are expressed as median (range) and were statistically analyzed using the following non-parametric tests: Mann-Whitney U and Kruskal-Wallis tests for comparisons of continuous variables between 2 or 3 groups, respectively, and Spearman's correlation test.p values < 0.05 were considered significant.

TrkAIII Was the Only in-Frame Alternative TrkA Splice Variant Expressed in PitNETs
RT-PCR, using primers spanning NTRK1/TrkA exons 1 through 8, detected three products in PitNET cDNAs that were sequence characterized as the fully spliced TrkA transcript fs-TrkA; the exons 6 and 7 skipped transcript TrkAIII, and the exons 2-7 skipped transcript ∆2-7TrkA (Figure 1a-c).PitNETs were also analyzed using primers spanning NTRK1/TrkA exons 8 through 17, which produced single products (Figure 1b, middle panel) that were sequence characterized as containing fully spliced TrkA exons 8 through 17 (not shown).Fs-TrkA and TrkAIII were the only in-frame splice variant mRNAs expressed in PitNETs.The ∆2-7TrkA splice variant was sequence characterized as a nonsense mRNA (Figure 1c).This variant exhibits a frameshift at the novel exon 1/8 splice junction that results in a premature TGA stop codon at position 1039-1041 (fs-TrkA numeration) (this study) and in [26,27].
RT-PCR, using primers spanning NTRK1/TrkA exons 1 through 8, detected three products in PitNET cDNAs that were sequence characterized as the fully spliced TrkA transcript fs-TrkA; the exons 6 and 7 skipped transcript TrkAIII, and the exons 2-7 skipped transcript Δ2-7TrkA (Figure 1a-c).PitNETs were also analyzed using primers spanning NTRK1/TrkA exons 8 through 17, which produced single products (Figure 1b, middle panel) that were sequence characterized as containing fully spliced TrkA exons 8 through 17 (not shown).Fs-TrkA and TrkAIII were the only in-frame splice variant mRNAs expressed in PitNETs.The Δ2-7TrkA splice variant was sequence characterized as a nonsense mRNA (Figure 1c).This variant exhibits a frameshift at the novel exon 1/8 splice junction that results in a premature TGA stop codon at position 1039-1041 (fs-TrkA numeration) (this study) and in [26,27].RT-PCR using this primer set detected TrkAIII mRNA expression in all invasive Pit-NETs regardless of lineage, and also detected TrkAIII mRNA in ≈86% of non-invasive PitNETs, comprising 10 PIT1, 12 SF1, and 2 TPIT tumors.This primer set also confirmed exclusive TrkAIII mRNA expression in two invasive PIT1 and one invasive SF1 PitNETs, as well as in three invasive and one non-invasive TPIT PitNETs.In contrast, exclusive fs-TrkA mRNA expression was only detected in three non-invasive PIT1 and one non-invasive SF1 PitNETs but not in any invasive PitNET.

Enhanced Alternative TrkAIII Splicing in Invasive PIT1 PitNETs Associates with Increased HIF2α mRNA Expression
Potential hypoxia involvement in PitNET alternative TrkAIII mRNA splicing was assessed by RT-PCR analysis of HIF2α and HIF1α expression in cDNAs from 50 TrkAIII mRNA expressing PitNETs, for which RNAs were available.Densitometric RT-PCR analysis revealed that invasive PitNETs expressed significantly higher HIF2α levels than non-invasive PitNETs (p = 0.0028) (Figure 4).HIF-2α levels were also significantly higher in invasive compared to non-invasive PIT1 PitNETs (p = 0.0198) but did not distinguish between invasive and non-invasive SF1 PitNETs (p = 0.238).Invasive TPIT PitNETs, although too few for lineage-restricted statistical comparisons, also exhibited high levels of HIF2a expression compared to non-invasive counterparts.Kruskal-Wallis analysis did not detect a significant difference between HIF2α expression in combined invasive and non-invasive PIT1, SF1, and TPIT lineages (p = 0.849).Although alternative TrkAIII splicing and HIF2α were significantly elevated in invasive PIT1 PitNETs, Spearman's correlation coefficient analysis failed to confirm a direct correlation between HIF2α levels and TrkAIII to fs-TrkA RT-PCR ratios in individual invasive (p = 0.8) or non-invasive (p = 0.076) PIT1 PitNETs.

Enhanced Alternative TrkAIII Splicing in Invasive PIT1 PitNETs Associates with Increased HIF2α mRNA Expression
Potential hypoxia involvement in PitNET alternative TrkAIII mRNA splicing was assessed by RT-PCR analysis of HIF2α and HIF1α expression in cDNAs from 50 TrkAIII mRNA expressing PitNETs, for which RNAs were available.Densitometric RT-PCR analysis revealed that invasive PitNETs expressed significantly higher HIF2α levels than non-invasive PitNETs (p = 0.0028) (Figure 4).HIF-2α levels were also significantly higher in invasive compared to non-invasive PIT1 PitNETs (p = 0.0198) but did not distinguish between invasive and non-invasive SF1 PitNETs (p = 0.238).Invasive TPIT PitNETs, although too few for lineage-restricted statistical comparisons, also exhibited high levels of HIF2a expression compared to non-invasive counterparts.Kruskal-Wallis analysis did not detect a significant difference between HIF2α expression in combined invasive and non-invasive PIT1, SF1, and TPIT lineages (p = 0.849).Although alternative TrkAIII splicing and HIF2α were significantly elevated in invasive PIT1 PitNETs, Spearman's correlation coefficient analysis failed to confirm a direct correlation between HIF2α levels and TrkAIII to fs-TrkA RT-PCR ratios in individual invasive (p = 0.8) or non-invasive (p = 0.076) PIT1 PitNETs.
These data confirm an association between elevated HIF2α expression and invasive PitNETs, especially invasive PIT1 PitNETs, and they also confirm an association but not a direct correlation between elevated HIF2α levels and elevated alternative TrkAIII splicing in PIT1 PitNETs.

Alternative TrkAIII Splicing in PitNETs Does Not Associate with Hotspot SF3B1 Mutations or De-Regulated SF3B1, SRSF2, U2AF1 Expression
Considering that lactotroph PitNETs associate with somatic hotspot SF3B1 mutations [14][15][16] In light of a report linking dysregulated splice factor expression to PitNET pathogenesis and aggressive behavior [17], PitNETs were also examined for alterations in SF3B1, U2AF1, and SRSF2 splice factor mRNA expression, by densitometric RT-PCR.No significant variations in SF3B1, U2AF1, or SRSF2 expression levels were detected between invasive and non-invasive in PitNETs, or individual PIT1 and SF1 lineages.This implies that variations in alternative TrkAIII splicing in PitNETs are unlikely to depend upon altered SF3B1, U2AF1, or SRSF2 mRNA expression (Figure 6).HIF1α levels were not significantly different in invasive compared to non-invasive PitNETs (p = 0.168), nor in invasive and non-invasive PIT1 (p = 0.78) or SF1 (p = 0.15) PitNETs, and were also similar in invasive and non-invasive TPIT PitNETs (Figure 4a,b).
These data confirm an association between elevated HIF2α expression and invasive PitNETs, especially invasive PIT1 PitNETs, and they also confirm an association but not a direct correlation between elevated HIF2α levels and elevated alternative TrkAIII splicing in PIT1 PitNETs.In light of a report linking dysregulated splice factor expression to PitNET pathogenesis and aggressive behavior [17], PitNETs were also examined for alterations in SF3B1, U2AF1, and SRSF2 splice factor mRNA expression, by densitometric RT-PCR.No significant variations in SF3B1, U2AF1, or SRSF2 expression levels were detected between invasive and non-invasive in PitNETs, or individual PIT1 and SF1 lineages.This implies that variations in alternative TrkAIII splicing in PitNETs are unlikely to depend upon altered SF3B1, U2AF1, or SRSF2 mRNA expression (Figure 6).

PitNET Alternative TrkAIII Splicing Does Not Associate with Unconventional Xbp1 Splicing or JCPyV Large T Antigen mRNA Expression
Agents that activate the UPR also promote alternative TrkAIII splicing in NB cells [25,30,33,36].A potential role for UPR activation in PitNET alternative TrkAIII splicing was examined by RT-PCR analysis of unconventional Xbp1 splicing, which serves as an index of UPR activation [42].Unconventional Xbp1 splicing, detected in DTT-treated (positive control) but not untreated (negative control) SH-SY5Y cDNAs, was not detected in any of the 50 PitNET cDNAs analyzed (Figure 7).

PitNET Alternative TrkAIII Splicing Does Not Associate with Unconventional Xbp1 Splicing or JCPyV Large T Antigen mRNA Expression
Agents that activate the UPR also promote alternative TrkAIII splicing in NB cells [25,30,33,36].A potential role for UPR activation in PitNET alternative TrkAIII splicing was examined by RT-PCR analysis of unconventional Xbp1 splicing, which serves as an index of UPR activation [42].Unconventional Xbp1 splicing, detected in DTT-treated (positive control) but not untreated (negative control) SH-SY5Y cDNAs, was not detected in any of the 50 PitNET cDNAs analyzed (Figure 7).JCPyV polyomavirus infection has been implicated in PitNET pathogenesis [18,19].JCPyV large T-antigen mRNA expression, as a potential indicator of JCPyV infection, was assessed in PitNETs, by RT-PCR.JCPyV large T-antigen mRNA expression was not detected in 45 PitNETs exhibiting alternative TrkAIII splicing, suggesting that JCPyV infection is unlikely to be responsible for alternative TrkAIII splicing in this PitNET cohort.

Discussion
In this study, we report that alternative TrkA mRNA splicing, limited to NTRK1/TrkA exons 1 through 8, is highly prevalent in PitNETs, regardless of their lineage of origin.We validate that PitNETs express three alternative splice variants fs-TrkA, TrkAIII, and Δ2-8TrkA and that TrkAIII is the only in-frame, tyrosine kinase-domain encoding, potentially oncogenic alternative to fs-TrkA.Although alternative TrkAIII mRNA splicing was detected in both invasive and non-invasive PitNETs, it was significantly elevated in invasive tumors, particularly in the invasive PIT1 PitNET group.
The detection of TrkAIII mRNA expression in invasive and non-invasive PitNETs and immunoreactivity is consistent with intracellular TrkAIII activation in PitNETs exhibiting exclusive or near-exclusive TrkAIII mRNA expression, suggesting that TrkAIII participates in different stages of PitNET pathogenesis and progression.Interestingly, exclusive and predominant TrkAIII mRNA expression was more common in invasive PitNETs, whereas exclusive fs-TrkA expression was only detected in non-invasive PitNETs.In accordance with this, alternative TrkAIII splicing was significantly higher in invasive compared to non-invasive PitNETs.However, when grouped according to lineage, it was only significantly higher in invasive compared to non-invasive PIT1 PitNETs.In contrast, invasive and non-invasive SF1 PitNETs exhibited similar levels of alternative TrkAIII splicing.Although TPIT PitNETs were too few for statistical comparisons within the group, it is remarkable that all three invasive cases exhibited exclusive TrkAIII mRNA expression.Overall, these findings indicate that divergent factors may influence alternative TrkAIII splicing in different PitNET lineages.Furthermore, they suggest enhanced potential for TrkAIII involvement in invasive PIT1 PitNET behavior and similar potential for involvement in both invasive and non-invasive SF1 PitNET behavior.Alternative TrkAIII splicing should, therefore, be added to the growing network of molecular changes associated with PitNET pathogenesis and progression [5,10,43].
The strongest evidence for TrkAIII involvement in PitNET pathogenesis and progression can be detected in high-level over-lapping non-phosphorylated and phosphorylated TrkA isoform(s) immunoreactivity in PitNETs exhibiting exclusive or JCPyV polyomavirus infection has been implicated in PitNET pathogenesis [18,19].JCPyV large T-antigen mRNA expression, as a potential indicator of JCPyV infection, was assessed in PitNETs, by RT-PCR.JCPyV large T-antigen mRNA expression was not detected in 45 PitNETs exhibiting alternative TrkAIII splicing, suggesting that JCPyV infection is unlikely to be responsible for alternative TrkAIII splicing in this PitNET cohort.

Discussion
In this study, we report that alternative TrkA mRNA splicing, limited to NTRK1/TrkA exons 1 through 8, is highly prevalent in PitNETs, regardless of their lineage of origin.We validate that PitNETs express three alternative splice variants fs-TrkA, TrkAIII, and ∆2-8TrkA and that TrkAIII is the only in-frame, tyrosine kinase-domain encoding, potentially oncogenic alternative to fs-TrkA.Although alternative TrkAIII mRNA splicing was detected in both invasive and non-invasive PitNETs, it was significantly elevated in invasive tumors, particularly in the invasive PIT1 PitNET group.
The detection of TrkAIII mRNA expression in invasive and non-invasive PitNETs and immunoreactivity is consistent with intracellular TrkAIII activation in PitNETs exhibiting exclusive or near-exclusive TrkAIII mRNA expression, suggesting that TrkAIII participates in different stages of PitNET pathogenesis and progression.Interestingly, exclusive and predominant TrkAIII mRNA expression was more common in invasive PitNETs, whereas exclusive fs-TrkA expression was only detected in non-invasive PitNETs.In accordance with this, alternative TrkAIII splicing was significantly higher in invasive compared to noninvasive PitNETs.However, when grouped according to lineage, it was only significantly higher in invasive compared to non-invasive PIT1 PitNETs.In contrast, invasive and noninvasive SF1 PitNETs exhibited similar levels of alternative TrkAIII splicing.Although TPIT PitNETs were too few for statistical comparisons within the group, it is remarkable that all three invasive cases exhibited exclusive TrkAIII mRNA expression.Overall, these findings indicate that divergent factors may influence alternative TrkAIII splicing in different PitNET lineages.Furthermore, they suggest enhanced potential for TrkAIII involvement in invasive PIT1 PitNET behavior and similar potential for involvement in both invasive and noninvasive SF1 PitNET behavior.Alternative TrkAIII splicing should, therefore, be added to the growing network of molecular changes associated with PitNET pathogenesis and progression [5,10,43].
The strongest evidence for TrkAIII involvement in PitNET pathogenesis and progression can be detected in high-level over-lapping non-phosphorylated and phosphorylated TrkA isoform(s) immunoreactivity in PitNETs exhibiting exclusive or near-exclusive TrkAIII mRNA expression.Lower levels of overlapping immunoreactivity were also detected in three invasive SF1 PitNETs exhibiting variable levels of alternative TrkAIII splicing.In contrast, immunoreactivity was barely detectable in two non-invasive SF1 (n.41) and PIT1(n.17)PitNETs exhibiting predominant fs-TrkA to TrkAIII RT-PCR ratios.Overall, these findings support a functional relationship between TrkAIII mRNA expression and intracellular TrkA isoform(s) expression and activation, including the TrkAIII oncoprotein.This is in line with reports that PitNETs and pituitary cell types exhibit heterogeneous TrkA expression, which has previously limited research interest in the potential significance of TrkA in these tumors [24,44].
Hypoxia promotes alternative TrkAIII splicing in neural crest progenitors, neural stem cells, and NB cells [25,36].PitNETs also show activated hypoxia responses, including HIF1α-RSUME-VEGF pathway activation, which is involved in PitNET progression and represents a current therapeutic target in refractory disease [6,7,10,[43][44][45].Because PitNET protein extracts were not available, the investigation into potential hypoxia participation in PitNET alternative TrkAIII splicing was restricted to RT-PCR comparisons with HIF1α and HIF2α expression.Elevated alternative TrkAIII splicing was linked to significantly higher levels of HIF2α but not HIF1α mRNA expression in invasive PIT1 PitNETs, suggesting a potential role for hypoxia in PitNET alternative TrkAIII splicing.This finding also identifies HIF2α as a novel potential marker of invasive PIT1 PitNET behavior.However, no significant correlation could be found between HIF2α expression and alternative TrkAIII mRNA splicing in individual PIT1 PitNETs.This does not rule out a role for hypoxia in alternative TrkAIII splicing, since all PitNETs expressed HIF1α mRNA, and HIF1α is involved in the PitNET hypoxia response [6,[45][46][47].Hypoxia also stimulates HIF1α and HIF2α protein expression at the post-transcriptional level [48].Notably, NB cells are one of the few cell types that show HIF2α transcriptional sensitivity to hypoxia [49] and also exhibit hypoxia-regulated alternative TrkAIII splicing [25,36], revealing a similarity between NBs and PIT1 PitNETs, potentially based on a common neural crest cell origin [50][51][52][53][54][55].
In relation to potential molecular mechanisms that could promote TrkAIII splicing in PitNETs, hotspot mutations in splicing factor SF3B1 have been reported in lactotroph PitNETs and have been shown to induce aberrant splicing [14][15][16][17] PitNET pathogenesis and aggressive behavior have also been linked to dysregulated splice factor expression [17].PitNET RNA availability limited the examination of dysregulated splicing factors in this study to SF3B1, SRSF2, or U2AF1.These were selected for analysis based on observations that SF3B1 regulates splicing in PitNET cells, and both SRSF2 and U2AF1 are differentially expressed in different PitNET lineages [17].No significant variations in SF3B1, SRSF2, and U2AF1 expression were detected between invasive and non-invasive PitNETs, either as a whole or grouped according to PIT1 and SF1 lineages.Furthermore, altered SF3B1, SRSF2, and U2AF1 expression did not correlate with enhanced alternative TrkAIII splicing in invasive PIT1 PitNETs.However, since the splicing machinery is complicated, we do not rule out the potential involvement of other splicing factors dysregulated in PitNETs [17].
Agents that cause ER stress and activate the UPR also promote alternative TrkAIII mRNA splicing in NB cells [25,30,33,36].In this investigation, unconventional Xbp-1 splicing, which serves as an index of UPR activation [42], was assessed in order to evaluate the relationship between the UPR and PitNET alternative TrkAIII splicing.No PitNETs displaying TrkAIII expression exhibited unconventional Xbp-1 splicing, potentially ruling out a role for the UPR.This was surprising, considering that hypoxia triggers UPR activation [56] and PitNETs exhibit activated hypoxia responses [6,[45][46][47].It is unclear if this may reflect a malfunctioning IRE1/Xbp1 arm of the UPR, TrkAIII modification of the UPR [30], or some other mechanism.
Finally, an examination of JCPyV large T antigen expression in PitNETs was prompted by a possible role for JCPyV infection in PitNET pathogenesis [18,19], by SV40 large T antigen promotion of alternative TrkAIII splicing in NB cells, and by alternative TrkAIII splicing association with MCPyV large T antigen expression in Merkel cell carcinomas.JCPyV large T antigen expression, however, was not detected in any of the PitNETs examined, suggesting that JCPyV is not involved in PitNET alternative TrkAIII splicing.

Conclusions
In conclusion, this study reveals that alternative TrkAIII mRNA splicing is common in all PitNET lineages and is significantly more pronounced in invasive PitNETs, especially invasive PIT1 PitNETs.It also reveals that significant increases in alternative TrkAIII mRNA splicing are associated with significantly elevated HIF2α mRNA expression in invasive PIT1 PitNETs, linking alternative TrkAIII splicing to the hypoxia response.We also verify that exclusive TrkAIII mRNA expression is associated with immunoreactivity consistent with intracellular expression and activation of the TrkAIII oncoprotein and that TPIT PitNETs appear to be especially susceptible to exclusive TrkAIII mRNA expression and intracellular TrkAIII activation.We conclude, therefore, that alternative TrkAIII mRNA splicing, leading to intracellular expression and activation of the TrkAIII oncoprotein, is likely to participate in PitNET pathogenesis and progression.TrkAIII may, therefore, represent a novel potential oncogenic target for clinically approved Trk inhibitors in refractory PitNETs [57,58].

Table 2 .
RT-PCR primers and conditions used in this study.