Background
The ability of cancer cells to self-renew and undergo phenotypic changes has led to the postulate that some have similarities to germline cells and/or stem cells [
1,
2], leading to the suggestion that a key feature of oncogenesis is a cellular soma-to-germline transition [
1,
3‐
6]. This is supported by the finding that tumours in
Drosophila melanogaster activate a large cohort of germline genes during oncogenesis and that some of these are essential for tumour progression [
7‐
9]. Analysis of gene expression in human tumours indicates that a similar pattern of germline gene activation is also apparent, inferring a possible functional requirement [
3].
One major group of germline genes is termed the cancer/testis (CT) genes. These encode cancer/testis antigens (CTAs), proteins that are normally only present in healthy adult testis, but are also found in a wide range of cancers [
10‐
12]. Little is currently known about the function in the testis for most of these proteins, but evidence is emerging to indicate that CTAs function in oncogenic processes, supporting the idea of a functional soma-to-germline transition [
12,
13]. Examples include regulation of cellular mitotic fidelity [
14‐
20] and invasiveness [
21‐
25]. These findings offer attractive new avenues for cancer-specific therapeutic targeting via inhibition of oncogenic CTA functions [
2,
11‐
13].
Recently, a pipeline for the identification of new CT genes was developed [
26,
27]. One of the genes identified was
Testis
Expressed
19 (
TEX19), a mammalian specific gene with a poorly defined function [
28,
29]; subsequently, this expression profile was verified [
30] and TEX19 protein was shown to be a CTA [
31].
In rodents, the
TEX19 orthologue has undergone duplication to generate a paralogue pair of genes,
Tex19.1 and
Tex19.2 [
29]. Both murine genes are differentially expressed, with
Tex19.2 expression restricted to the developing gonadal ridge and adult testis and
Tex19.1 being expressed in the adult testis, the placenta and in the early embryo, in a pattern matching the pluripotency marker gene
Oct4 [
29], although expression control mechanisms of the two genes is distinct [
32].
Tex19.1 is expressed in embryonic stem cells (ESCs) and whilst this might infer a functional role in stemness [
29],
Tex19.1
-/- ESCs have no overt stemness/ proliferative defects [
33], nor are there any overt phenotypic defects in spermatagonial germ cells [
34]. Preliminary analysis of human
TEX19 expression indicates it is orthologous to
Tex19.1 as it is expressed in human ESCs [
29]. Tex19.1 is largely cytoplasmic and appears to be located in spermatagonial germline cells of testis seminiferous tubules, with levels diminishing as cells differentiate during spermatogenesis, suggesting a germline-specific function [
29,
33,
34].
Tex19.1
-/- mice are viable with no apparent behavioural defects [
33‐
35]. There is a slight increase in mortality of pups older than 5 days post-partum, but this has been attributed to in utero developmental defects linked to placental dysfunction [
33,
35]. Female fertility of
Tex19.1
-/- mice has been independently reported to be reduced [
34] and normal [
33], with the discrepancy being attributed to distinct genetic backgrounds [
33]. Males exhibit sub-normal levels of fertility with considerable inter-individual differences in spermatogenesis indicating a phenotypic variability, the cause of which is unknown [
33,
34]. Meiosis in
Tex19.1
-/- males has defects, which include impaired meiotic chromosome synapsis, the persistence of unprocessed DNA double-strand breaks, increased apoptosis and post pachytene meiosis I chromosome segregation defects, although these were not uniformly apparent [
33,
34]. Analysis of gene expression during early spermatogenesis did not reveal any notable changes to genes that could directly influence meiosis, but there was significant elevation in the expression of the class II long terminal repeat (LTR)-retrotransposon
MMERVK10C [
33,
34]. Expression of other transposable elements (TEs), such as
LINEs,
SINEs and
IAP retrotransposons did not appear to be altered, indicating TE specific suppressor activity for Tex19.1 in testis, which is proposed to be distinct from the Piwi-mediated pathway for TE regulation [
34].
The proposal that Tex19.1 functions in an independent TE regulatory pathway is further supported by the finding that in murine placenta, where there are alterations to expression levels of some TEs,
Tex19.1 is the only known methylation-sensitive genome defense gene that is highly expressed [
32], suggesting it may independently serve to protect placental cells from elevated TE expression [
35]. Female
Tex19.1
-/-
mutant mice also exhibit impaired placental function [
33,
35]. Unlike
Tex19.1
-/- male testis tissue,
Tex19.1
-/- placental tissue exhibits elevated
LINE expression and also exhibits some differential expression of protein coding genes [
35]. Collectively, these findings suggest that Tex19.1 controls transcription/transcript related mechanisms to protect the germline and placental genomes [
29,
33‐
35].
The finding that human
TEX19 is a CT gene opens the question of whether
TEX19 expression is oncogenic or provides a functional advantage to cancer cells [
26,
31]. Expression of germline genes has been linked to poor patient prognosis in cancers, such as lung cancer (for example, see [
36]), so revealing functional roles of these genes, if any, is important to understand the mechanisms of cancer development/progression. In this study we identify a requirement for TEX19 in human cancer cells to drive proliferation that reveal it to be a potential cancer-specific drug target and prognostic indicator.
Methods
Cell culture and proliferation/self-renewal assays
Human cacner cell lines used in this study are provided in Additional file
1; Table S1. Cells were cultured in McCoy’s 5A medium (Thermo Fisher Scientific, Runcorn, UK) supplemented with 10% fetal bovine serum (FBS; Life Technologies) or in RPMI medium supplemented with 10% FBS and 2 mM sodium pyruvate (Thermo Fisher Scientific, Runcorn, UK); SW480 cells were cultured in Dulbecco’s modified Eagle’s Medium (Thermo Fisher Scientific, Runcorn, UK).. All cells were cultured at 37 °C in 5% CO
2 in a humidified incubator. All cells were authenticated once every 12 months using LGC Standards Cell Line Authentication service (last report number: 710236782; Teddington, UK). Cells were regularly checked for mycoplasma using the LookOut Mycoplama Detection Kit (Sigma, Irvine, UK).
For leptomycin B (LMB) treatment cells were seeded into 40 mm tissue culture dishes and grown to the required density. Cells were then treated with 10 ng/ml LMB (L2913; Sigma, Irvine, UK) and incubated for a further 16 h.
Extreme limiting dilution analysis (ELDA) was performed as previously described [
37,
38]. Briefly, sphere-derived cell were collected from 10 cm dishes and diluted into single cell suspension and plated at concentrations of 1000 to 1 cells per 100 μl SCM using repeats of defined experimental conditions in 96 well ultra-low attachment plates (Costar Corning; Sigma, Irvine, UK). Cells were incubated at 37 °C in a 5% CO
2 atm for 10 days. Cells were supplemented with 50 μl of stem cell media (SCM) and transfection complexes re-applied after 4 and 8 days of incubation. At the end of 10 days the number of wells showing spheres with more than 20 cells were counted by light microscope. ELDA web tool (hrrp://bioinf.wehi.edu.au/software/elda) was used to determine frequencies of sphere forming cells.
Staining for senescence was carried out using β-galactosidase Staining Kit (Cell Signaling, Leiden, Holland) following the manufacturer’s instructions.
siRNA transfection
siRNAs used in this study are TEX19 siRNA A (5′-AGGATTCACCATAGTCTCTTA-3′), TEX19 siRNA B (5′-TTCAACATGGAGATCAGCTAA-3′) and a negative control (Qiagen, Manchester, UK, Allstars Negative Control siRNA). Transfection was carried out with HiPerFect (Qiagen, Manchester, UK) following the manufacturer’s instructions. Briefly, 150 ng of siRNA was mixed with 6 μl of HiPerFect and 100 μl of cell specific medium. This mix was incubated at room temperature for 15 min to permit transfection complexes to form and was then added in a dropwise fashion to approximately 1.5 × 105 cells. The number of siRNA treatments per cell culture was dependent upon the specific experiment and siRNA was added to cells at least once every 24 h for proliferation assays over extended periods. Depletion was verified by RT-qPCR and/or western blotting.
Whole-cell lysates were prepared using M-PER lysis buffer (Thermo Fisher Scientific, Runcorn, UK #78503), Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific, Runcorn, UK) and Halt Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific, Runcorn, UK). Approximately 30 μg of protein extract were used for western blotting (WB). Samples were mixed with 2X Laemmli Buffer (1:1) (Sigma, Irvine, UK; S3401) and boiled at 100 °C for 5 min prior to electrophoresis. Precision Plus Protein Dual Color Standards (BioRad, Watford, UK) was used as a protein ladder. NuPAGE Novex 4–12% Bis-Tris gels (Thermo Fisher Scientific, Runcorn, UK) were used and electrophoresis was carried out in NUPAGE MOPS SDS buffer (Thermo Fisher Scientific, Runcorn, UK) for 90 min at 120 v. Fast Western Blot Kit, ECL substrate (Thermo Fisher Scientific, Runcorn, UK) was used according to manufacturer’s instructions to detect the primary antibodies. Membranes were probed with primary antibodies in 10% dry milk/PBS/0.5% Tween 20. Incubation with secondary antibodies was performed at room temperature for 1 h, followed by a 10 min wash in milk solution and 3 additional 10 min washes in PBS/0.5% Tween 20 at room temperature. Antibody detection was performed using Pierce ECL Plus Western Blotting Substrate (Thermo Fisher Scientific, Runcorn, UK).
Subcellular fractionation was carried out as follows. Following harvesting cells were resuspended in hypotonic buffer [50 mM Tris-HCl (pH 7.4), 0.1 M sucrose, 1 mM AEBSF] and lysis buffer C (1% Triton, 10 mM MgCl2, 1 mM AEBSF) at 1:1 ratio. Following incubation on ice for 30 min tubes were spun at 6000 g for 2 min. Supernatant contained cytoplasmic proteins and the pellet was resuspended in lysis buffer N [50 mM Tris-HCl (pH 7.4), 100 mM potassium acetate, 1 mM AEBSF] to extract nuclear protein.
The following antibodies were used in this study: anti-TEX19 (R & D Systems, AF6319), 1:200 dilution for WB; anti-Lamin B1 (Abcam, AB16048), 1:1000 dilution for WB; anti-tubulin (Sigma, T6074), 1:8000 dilution for WB; anti-cleaved caspase-3 (Cell Signaling, Leiden, Holland; 9664), 1:1000 dilution for WB; cell cycle cocktail (anti-pCDK2, anti-Actin, anti-pH3) (Abcam, Cambridge, UK; AB136810), 1:250 dilution for WB; anti-rabbit secondary antibody (Cell Signaling, Leiden, Holland; 7074), 1:3000 dilution for WB; anti-mouse secondary antibody (Cell Signaling, Leiden, Holland; 7076).
For the chromatin association assay (Ch) protein lysates were prepared consecutively with increased concentrations of NaCl. Protein extracts were subjected to western blotting as described using anti-a-tubulin and anti-histone H3 antibodies in addition to anti-TEX19 antibodies. 10 ng/ml KaryoMAX colcemid (Gibco, Runcorn, UK; 15212-012) was added to the growth medium to synchronize cells in metaphase prior to chromatin extraction.
Reverse transcription quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted from appropriate cell cultures using the RNeasy Plus Mini Kit (Qiagen, Manchester, UK) following the manufacturer’s instructions. First-strand cDNA synthesis was carried out using SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific, Runcorn, UK) following the manufacturer’s instructions.
qPCR reactions were carried out using GoTaq qPCR Master Mix (Promega, Southampton, UK) in a CFX96 Real-Time PCR Detection System C100 thermal cycler (BioRad, Watford, UK). All RT-qPCR primers were obtained from Qiagen (Manchester, UK). The exceptions were the HERV primers and primers for
PIWIL3 and
PIWIL4. PIWIL3 primers were PIWIL3F (5′-TGGCATTGCATTAAGTAAGGG-3′) and PIWIL3R (5′-TTTGAAAAACGCAAACATCG-3′).
PIWIL4 primers were PIWIL4F (5′-CTGAAGGATACAGCGGGAAA-3′) and PIWIL4R (5′-AAAGATGCACTCAGCAAGGAC-3′). HERV primers are listed in Additional file
2; Table S2; other RT-qPCR primers are shown in Additional file
3; Table S3. Reactions were carried out in triplicate with PCR primers at a final concentration of 0.2 μM in a final volume of 25 μl. All PCR primers are available upon request. BioRad CFX Manager 2.0 software was used to determine primer efficiency/specificity, threshold cycle values (Ct values) and expression values using default parameters. Results were normalized using two or three reference genes and fold-change values were calculated based on the ΔΔCT method.
RNA sequencing and data analysis
Total RNA was extracted using RNeasy Plus mini kit (Qiagen, Manchester, UK) according to the manufacturer’s protocol. RNA quality was checked on an Agilent Bioanalyzer RNA 6000 nano chip and was assessed to be of high quality (RIN > 9.8). Indexed sequencing libraries were then prepared using the Illumina TruSeq v2 protocol. Briefly, polyA-tailed RNA enriched on oligo-dT beads before fragmentation and random priming. Reverse transcription was carried out with second strand synthesis and the resultant double-stranded cDNA was end repaired, A-tailed and Illumina TruSeq adapters were ligated. Correctly ligated fragments were enriched by performing 12 cycles of PCR with primers complementary to the Illumina adapters. The final libraries were checked and quantified on the Agilent Bioanalyzer DNA 1000 chip and the Life Technologies Qubit High Sensitivity DNA assay system before being pooled to an equimolar concentration of approximately 10 nM. qPCR was performed on a 105 dilution of the multiplex pool (Kapa Biosystems Library Quantification Kit; Sigma, Irvine, UK) before 12 pM of multiplex library was sequenced on one lane of an Illumina HiSeq (TruSeq v3 chemistry) generating 190 million reads passing filter. Reads were demultiplexed and fastq files generated using Illumina CASAVA v1.8.2 software.
Fastq data underwent guided alignment to the human genome (NCBI Build 37.2) using Tophat v2.0.6 [
39] with default parameters. Read duplicates were removed using Picard (
http://broadinstitute.github.io/picard) and counts per gene generated using HTSeq [
40]. Differential expression at both the gene and exon level was carried out in R (
https://www.r-project.org/) using the ‘DESeq’ and DEXSeq’ R packages [
41]. The resulting
P values were adjusted for multiple testing with Benjamini and Hochberg’s [
42] method to control the false discovery rate. Genes with a
P value <0.05 and a log
2-fold change > 1.0 have been selected as significant. Pathway and gene ontology (GO) analysis was carried out in R v3.2.3 (
https://www.r-project.org/) using the ‘GOstats’ R package [
43].
Fluorescence activated cell sorting cell cycle analysis
Following trypsinization cells were fixed in 70% ethanol at 4 °C overnight. Fixed cells were treated with 0.5 mg/ml RNase A (Sigma, Irvine, UK) and stained with propidium iodide (500 nM, Sigma, Irvine, UK). Stained cells were analyzed using a Partec CyFlow Cube 8 and cell cycle analysis was carried out using FCS Express 4 software.
Immunohistochemistry (IHC) and human tissue
Human tissue was obtained from patients following the guidelines of the North Wales Research Ethics Committee – West. All tissues were fixed in formalin, embedded in paraffin and prepared as 4 mm slices. Tumour/normal tissue arrays were obtained from the Cooperative Human Tissue Network (University of Virginia, USA). Staining was automated using the Ventana Benchmark XT instrument. Chromogenic reactions were carried out using 3,3′-diaminobenzidine and slides were counter stained with haematoxylin. The rabbit polyclonal anti-TEX19 antibody (Abcam, Cambridge, UK; 185507) was used for TEX19 staining. Secondary antibody only staining was used as a control. Antigen retrieval consisted of 4 min wash with protease I. All slides were scanned using and Axio Scanner.Z1 Scanner (Zeiss Cambridge, UK).
Tissue/cell immunofluorescence imaging
For staining of cultured cells 105 cells were seeded on a cover slip in a 24-well plate with appropriate medium and grown to the required density. Cells were fixed in 4% paraformaldehyde in PBS for 10 min at room temperature and then washed three times with PBS at room temperature. Cells were incubated for 1 h in 5% FBA/0.3% Triton in PBS at room temperature. Cells were incubated with primary antibodies in 1% BSA/0.3% Triton in PBS at appropriate concentrations overnight at 4 °C. Following three 5 min washes in PBS appropriate concentrations of the required secondary antibody were incubated with the cells in the same buffer for 1 h at room temperature in the dark. Following final washing cover slips were mounted on slides with Vectashiled Hard Set Antifade Mounting Medium (Vector Laboratories, Peterborough, UK) and counter stained with DAPI (Sigma) as required. Images were acquired using a Zeiss LSM 710 confocal microscope and analyzed using ZEN software (Zeiss, Cambridge, UK).
For tissue staining 4 μM paraffin embedded section were deparaffinised and re-hydrated as follows: three times through xylene, two times through 100% ethanol, two times through 70% ethanol, two times through sterile distilled H2O. Antigen retrieval was performed by heating samples in boiling 10 nM sodium citrate buffer (pH 6.0) for 10 min and cooling slides to room temperature. Slides were washed in sterile distilled water and stained as for the cultured cells (see above).
Antibodies used for staining were as follows, anti-TEX19 (Abcam, Cambridge, UK; AB185507), dilution of 1:50; anti-MAGE-A1 (LSBio, Nottingham, UK; LS-C87868), dilution of 1:20; anti-vimentin (LSBio, Nottingham, UK; LS-B7191), dilution of 1:100; secondary goat anti-rabbit (Alexa Fluor 488; Thermo Fisher Scientific, Runcorn, UK; A11034), dilution of 1:1,1000; secondary goat anti-mouse (Alexa Fluor 568; Thermo Fisher Scientific, Runcorn, UK; A11031), dilution of 1:1000.
Mouse tumourgenicity assay
In vivo tumour growth capability of Tex19shRNA-SW480-c3 cells was assessed by sub-cutaneous xenograft into immune deficient NSG mice (Envigo, Derby, UK). Cells were harvested using 1 mM EDTA and re-suspended at a density of 5 × 107 cells/ml in serum free DMEM medium. A total of 5 × 106 cells were injected sub-cutaneously into the flank of each mouse. Cells were allowed to establish for 6 days prior to induction of shRNA expression with doxycycline. To induce shRNA expression, mice were injected with 10 mg/kg intra-peritoneally every 2 days. Mice were then monitored, and when palpable, tumour volume was measured twice weekly with a digital caliper. Relative tumour volume (RTV) was calculated using xy2/2; where x is the longest axis of the tumour and y is the shortest axis of the tumour.
Survival analysis
Cancer data sets available from The Cancer Genome Atlas (TCGA;
http://cancergenome.nih.gov) were analyzed using R v3.2.3 (
https://www.r-project.org/) to assess the association of
TEX19 expression and clinical data. Normalized gene RSEM values for all TCGA RNA-seq data sets as well as corresponding clinical data were downloaded from
http://firebrowse.org. The survival analysis was carried out on primary tumour samples apart from leukaemia, where the primary blood derived cancer samples from peripheral blood were used. Normalized RSEM values were transformed to log
2 counts per million prior to survival analysis using voom [
44] in the ‘limma’ R package [
45]. The patients were categorized in to two groups, low and high
TEX19 expression in cancer, split by the median value. Overall survival (in years) related to
TEX19 expression was computed with the Kaplan-Meier method and compared by the log rank test using the ‘survival’ R package [
46].
P values of <0.05 were considered statistically significant.
Discussion
The idea that cancer cells achieve self-renewal potential by re-activating programmes that regulate the germ/stem like state, and undergo a soma-to-germline transition is gaining traction (for example, see [
2,
3,
5]). Germline functions are known to contribute to distinct biological features of cancers [
11‐
13], including invasiveness/metastasis (for example, [
21‐
23,
25]) and maintenance of proliferative potential (for example, [
14‐
19]). Some activities are linked to poor outcomes; for example, elevated expression of
SPANX-A/C/D is linked to poor prognosis in breast cancers [
25]. Furthermore, germline gene expression in lung cancer has been linked with aggressive, metastases prone disease and can be used for stratification of patients to identify cohorts who might benefit from targeted therapies [
36]. Even meiotic chromosome regulators have been shown to contribute to oncogenesis by driving inter chromosomal associations required for oncogenic alternative lengthening of telomeres [
60]. So, germline/stem cell factors can make distinct contributions to cancer progression, maintenance and evolution.
TEX19 as an oncogenic driver
Despite the growing evidence for the soma-to-germline transition of tumours, the reported roles of specific germline genes in cancers remains limited. Here we find that
TEX19 expression is required to maintain the proliferative potential of a range of different cancer cells. Given that
TEX19 is expressed in many cancer types, this might infer that it is an oncogenic factor, indeed, all cell lines / tumour samples we analysed showed evidence of
TEX19 expression. This is not the case for all database studies [for example, the Human Protein Atlas (
www.proteinatlas.org) reports only limited
TEX19 expression], but we have demonstrated that some areas of later stage tumours are more prone to
TEX19 expression, suggesting that expression can be regional within an advanced tumour. It is known that tumours can evolve and acquire a high degree of intra- and inter-tumour heterogeneity, and so database samples may also include information acquired from tumour regions that do not express
TEX19 [
49,
50]. The heterogeneous distribution of TEX19 indicates two possibilities; firstly,
TEX19 may be ‘on’ during the early stages of tumourigenesis and becomes deactivated during the evolution of the tumour, or, alternatively,
TEX19 only becomes activated in specific regions as the tumour develops and grows. The finding that germline genes are actually required for the early oncogenic process in
D. melanogaster l(3)mbt tumours might suggest the former is the case [
7]. Our analysis of colorectal tumour progression profiles (Fig.
5) supports this view, as TEX19 was detected in all early adenomas, but was not detected in some samples of later stages of tumour progression which could point to an early ‘on’ later ‘off’ model for
TEX19. This suggests that therapeutic targeting of TEX19 might not eliminate all tumour cells in later stage tumours. However, given the fact that TEX19 function is implicated in stemness [
29], it might be the case that the TEX19 positive cells are those that retain stem-like features and thus therapeutic targeting of these cells remains important, as these cells might be driving the poor prognosis/therapeutic resistance [
1].
How does TEX19 modulate proliferation?
We have demonstrated that
TEX19 expression is required to maintain proliferation/self-renewal in a range of cancer cell types. Loss of murine
Tex19.1 can result in spermatogenic cells entering apoptosis, however, analysis of TEX19-depleted human cancer cells indicates that they are likely to be in a quiescent-like state, which appears to be linked to a failure to proceed through S-phase with normal kinetics. This observation is in contrast to murine
Tex19.1
-/- ESCs, which, whilst being defective in self-renewal, do not exhibit any overt S-phase defects [
33]. Murine Tex19.1 has been shown to control levels of TE transcripts and, in the case of the female placenta, other protein coding transcripts [
35]. We have demonstrated here that TEX19 in SW480 cells can influence TE transcript levels. In SW480 cells this appears to be counteracted by a PIWIL1-dependent mechanism. A model in which TEX19 and PIWIL1 serve in opposing and independent mechanisms for TE transcript control is consistent with murine Tex19.1 acting on TEs in a PIWI-independent fashion [
34]. Indeed, in other cancer cells lines we tested PIWIL1 is not expressed and yet TE transcripts exhibit a measurable change upon TEX19 depletion, which supports a TEX19-dependent, PIWI-independent pathway.
The ovarian carcinoma cell line A2780 showed considerable activation of HERVK
gag transcript levels upon a relatively moderate reduction of TEX19. A2780 cells do not express measurable levels of
PIWIL1 and the
PIWI orthologues that are expressed are not altered upon TEX19 depletion. This indicates that in A2780 cells TEX19 appears to operate in a PIWI-independent mechanism for TE suppression, similar to that proposed for murine placental TE regulation [
35]. These findings also demonstrated that TEX19 can act in a repressive and activating fashion for some TE transcripts in a cell- and/or dose-dependent fashion.
Given this, we explored protein coding gene changes with an aim of identifying changes common to all cells. This showed that no gene transcripts were consistently altered upon TEX19 depletion. However, coding gene transcripts were altered (up and down) indicating that TEX19 could regulate the transcripts/transcription of a cohort of oncogenic protein coding genes to foster a proliferative state, possibly in a dose-dependent fashion.
SSX2, another CTA, is a chromatin regulator [
61] and it has been inferred that it may regulate cancer cell proliferation through transcriptional regulation [
13]. Also, the germline-specific chromatin regulator ATAD2 drives various cancer progression phenotypes via transcriptional regulation and is linked to poor prognoses in various cancers and offers an important potential therapeutic target [
62‐
72]. These findings indicate there are CTAs that can directly modulate transcription, and we postulate that TEX19 could function in a similar fashion controlling cellular transcript levels, either at the transcriptional and/or post-transcriptional levels. As for mouse Tex19.1, TEX19 in cancer cells can operate on a small sub-set of protein coding and/or TE transcripts, although the latter may be indirect.
TEX19 is both cytoplasmic and nuclear in cancer cells and testis cells
Consistent with a direct role in transcriptional regulation, we demonstrate that human TEX19 can locate to the nucleus, although there is no apparent nuclear localization/export signals on TEX19, which might suggest nuclear localization in response to cellular status is regulated by interacting partners. Murine Tex19.1 interacts with the E3 ubiquitin ligase Ubr2 which contains a nuclear localization signal, so a role for a UBR2-TEX19 interaction in human cancers would be worthy of further investigation [
73]. TEX19 nuclear localisation appears to be linked to cell density and may be related to proliferative state, with nuclear TEX19 being associated with proliferation. Whilst murine Tex19.1 is predominantly localized to the cytoplasm, nuclear Tex19.1 can be observed in placental tissues indicating commonalities between murine and human proteins [
33,
34]. Additionally, we note nuclear foci of TEX19 in testis, germ cell tumour line NTERA2 and LMB treated nuclei; the functional relevance, if any, is unknown although it is a noteworthy observation.
Recently, immunohistochemistry studies indicated that human TEX19 was localized to Sertoli cells [
31]. Our immunofluorescence analysis of co-staining with the Sertoli cell marker vimentin did not show direct co-localisation of the high intensity TEX19 staining regions (although there appears to be a low intensity TEX19 staining throughout the basal region, including Sertoli cells), rather a close association, suggesting that TEX19 may not be exclusively Sertoli cell specific. Furthermore, IHC staining of normal testis appears to show nuclear localization/speckling in some of the large nuclei [
31], consistent with our observation. Zhong and co-workers [
31] extended their analysis to demonstrate that TEX19 was present in bladder cancer samples. Whilst most of the TEX19 they observed in these samples appeared to be cytoplasmic, there are clearly cells within the tumours that exhibit some nuclear staining with the anti-TEX19 antibodies employed [
31].
TEX19 expression influences clinical outcomes
That
TEX19 expression helps drive proliferative potential of cancer cells suggests that it might influence disease progression / outcome in patients. Our analysis of overall survival indicates that for a number of cancers, including breast cancer and renal cancers, that there is a significant correlation between higher
TEX19 expression and poor prognosis, supporting a potential functional association. These analyses might be an underestimation of the influence of
TEX19 expression as our analysis was based on splitting cancer patient cohorts based on median
TEX19 expression. Use of the median split means that the two populations being compared both have high numbers of
TEX19 expressing cancers; if
TEX19 expression alone (irrespective of the levels) is sufficient to drive cancers, then in many of our analyses a split based on the median would not detect a correlative link between
TEX19 expression and a poor prognosis. Ideally, we would have split all cohorts into those expressing
TEX19 and those not. This approach, however, was untenable as many data sets had an imbalance of negative
vs. positive
TEX19 expressing cancers, negating statistical analysis. Moreover, this is further complicated by the difficulties of ascribing tumour biopsies as having no expression given the sensitivity/depth of modern deep RNA sequencing technologies, which can identify low abundance transcripts for almost all annotated genes [
74]. Despite these limitations, the analysis of breast cancers and renal cancers clearly indicate a correlative link between high
TEX19 expression and poor prognosis, consistent with a functional role for TEX19 as an oncogenic proliferative driver.
Surprisingly, we observed the converse relationship for gliomas where higher
TEX19 expression is linked to a better prognosis, suggesting it has favorable activity in neuronal cells. This inverse influence on disease outcome has also been observed for T-box transcription factors, which both positively/negatively regulate gene expression during embryonic development [
75]. For example, as for TEX19, TBX3 appears to have tumour suppressing activity in glioblastomas [
76] but oncogenic activity in a number of solid tumours (for examples, see [
77‐
79]; for review, see [
75]). This commonality between T-box transcription factors and TEX19 could infer a functional link in the distinct regulation of developmental/proliferative genes in distinct cancer types. The factors that set neuronal tumours apart from other tumour types are likely to be multifold, however, it is noteworthy that
LINE-1 retrotransposition is active in somatic neuronal cells [
80]; it is not unreasonable to postulate that TEX19 production in neuronal malignancies could limit
LINE-1 transposition events and thus limit the evolutionary capacity of the diseased genome and therefore the aggressiveness of the tumour.
Acknowledgements
The authors would like to extend their thanks to Dr. Ian Adams (Edinburgh University) for his invaluable and insightful comments on the manuscript.