Activating PIK3CA mutations are known ‘drivers’ of human cancer and developmental overgrowth syndromes. We recently demonstrated that the ‘hotspot’ PIK3CAH1047R variant exerts unexpected allele dose-dependent effects on stemness in human pluripotent stem cells (hPSCs). In this study, we combine high-depth transcriptomics, total proteomics and reverse-phase protein arrays to reveal potentially disease-related alterations in heterozygous cells, and to assess the contribution of activated TGFβ signalling to the stemness phenotype of homozygous PIK3CAH1047R cells. We demonstrate signalling rewiring as a function of oncogenic PI3K signalling strength, and provide experimental evidence that self-sustained stemness is causally related to enhanced autocrine NODAL/TGFβ signalling. A significant transcriptomic signature of TGFβ pathway activation in heterozygous PIK3CAH1047R was observed but was modest and was not associated with the stemness phenotype seen in homozygous mutants. Notably, the stemness gene expression in homozygous PIK3CAH1047R hPSCs was reversed by pharmacological inhibition of NODAL/TGFβ signalling, but not by pharmacological PI3Kα pathway inhibition. Altogether, this provides the first in-depth analysis of PI3K signalling in hPSCs and directly links strong PI3K activation to developmental NODAL/TGFβ signalling. This work illustrates the importance of allele dosage and expression when artificial systems are used to model human genetic disease caused by activating PIK3CA mutations.
Class IA phosphoinositide 3-kinases (PI3Ks) are evolutionarily conserved enzymes that catalyse formation of the membrane-bound second messenger phosphatidylinositol-3,4,5-trisphosphate (PIP3). PI3Ks are activated downstream of receptor tyrosine kinases, with the ensuing increase in PIP3 and its derivative PI(3,4)P2 triggering a widespread signalling network, best known for the activation of the serine/threonine kinases AKT and mTORC1. PI3K activation promotes cell survival, glucose uptake, anabolic metabolism, cell proliferation and cell migration (Fruman et al., 2017). Among the class IA PI3K isoforms (PI3Kα, PI3Kβ, PI3Kδ), the ubiquitously expressed PI3Kα (encoded by the PIK3CA gene in humans) is the main regulator of organismal growth, development and survival (Bilanges et al., 2019).
Activating mutations in PIK3CA are among the most common somatic point mutations in cancer, together with inactivation or loss of the tumour suppressor PTEN (a negative regulator of PI3K) (Chang et al., 2015; Sanchez-Vega et al., 2018; Campbell et al., 2020). The same mutations in PIK3CA, when acquired postzygotically during development, also cause a range of largely benign overgrowth disorders (Madsen et al., 2018), for which the term PIK3CA-related overgrowth spectrum (PROS) has been coined. Motivated by the need to understand the role of PI3K signalling in a human developmental context, we previously generated an allelic series of human induced pluripotent stem cells (iPSCs) with heterozygous or homozygous expression of the activating mutation PIK3CAH1047R, the most commonly observed PIK3CA mutation in both cancer and PROS (Madsen et al., 2019). Despite the severe developmental disorders caused by heterozygosity for PIK3CAH1047R in humans in vivo, we found little discernible effect on germ layer specification from heterozygous iPSCs. In sharp contrast, homozygosity for PIK3CAH1047R led to self-sustained stemness and resistance to spontaneous differentiation in vitro and in vivo (Madsen et al., 2019). This suggested a previously unappreciated quantitative relationship between the strength of PI3K signalling and the gene regulatory network (GRN) in pluripotent stem cells.
The core pluripotency GRN features a feedforward autoregulatory circuit comprising three transcription factors, namely SRY box 2 (SOX2), Octamer-binding transcription factor 3/4 (OCT3/4; encoded by POU5F1), and the homeobox transcription factor NANOG (Boyer et al., 2005; Loh et al., 2006; Li and Belmonte, 2017). SOX2 helps sustain OCT3/4 expression, which is required for the establishment and maintenance of the pluripotent state (Nichols et al., 1998). However, even modest overexpression of OCT3/4 destabilises the pluripotency network and triggers differentiation (Niwa et al., 2000; Radzisheuskaya et al., 2013). In contrast, NANOG, although dispensable for the maintenance of pluripotency (Chambers et al., 2007), stabilises the pluripotency GRN. Overexpression of NANOG by as little as 1.5-fold leads to sustained self-renewal (or ‘stemness’) of murine and human PSCs (hPSCs) (Chambers et al., 2003; Mitsui et al., 2003; Darr, 2006; Ivanova et al., 2006). In hPSCs, NANOG expression is activated by the transcription factors SMAD2/3 (Xu et al., 2008), which in turn are activated by receptors binding TGFβ, Activin or NODAL (Pauklin and Vallier, 2015). Overexpression of NODAL thus results in self-sustained stemness of hPSCs even in differentiation-promoting conditions (Vallier et al., 2004, 2005).
Given the unexpected and surprisingly mild phenotype caused by heterozygous PIK3CAH1047R expression in iPSCs, we reasoned that more sensitive assays would allow us to discern small but disease-relevant alterations in these cells. Thus, in this study, we first applied high-depth transcriptomics and proteomics to seek evidence of disease-related phenotypes in heterozygous cells, and to investigate how high-dose PI3K signalling leads to self-sustained stemness in homozygous PIK3CAH1047R iPSCs. We demonstrate that heterozygous cells do exhibit significant transcriptomic changes, although these are a weak echo of the widespread changes seen in homozygous cells. The mild transcriptional consequences of heterozygous expression of disease-relevant PIK3CA mutations were also validated in additional model systems and contrast with previous findings of major transcriptional rewiring in immortalised non-transformed breast epithelial cells (Hart et al., 2015; Kiselev et al., 2015). We demonstrate that the stemness phenotype of PIK3CAH1047R/H1047R iPSCs is maintained by self-sustained NODAL/TGFβ signalling, in line with increased PIK3CA-mediated NODAL expression, and that it is not reversible by PI3Kα-specific inhibition. This work provides in-depth characterisation of the near-binary PI3K signalling effects seen in hPSCs, and provides evidence for PI3Kα-induced NODAL/TGFβ signalling as the mechanism for self-sustained stemness in homozygous PIK3CAH1047R iPSCs. We discuss the implications of our findings for understanding and modelling developmental disorders and cancers driven by genetic PI3K activation.
A sharp PI3K activity threshold determines gene expression changes in PIK3CAH1047R iPSCs
We previously generated isogenic human iPSCs with heterozygous or homozygous knock-in of the ‘hotspot’ PIK3CAH1047R mutation. Surprisingly, heterozygous cells showed few phenotypic changes and differentially expressed protein-coding transcripts. In contrast, homozygous PIK3CAH1047R/H1047R cells exhibited marked morphological changes and altered gene expression, with strong enrichment for cancer-associated pathways (Madsen et al., 2019).
To substantiate the apparent PI3K activity threshold manifest in PIK3CAH1047R-driven gene expression changes, and to look for further disease-related changes in heterozygous cells, we undertook RNA sequencing at substantially greater depth, also increasing the sample size to four and including previously unstudied iPSC clones for the wild-type and homozygous PIK3CA genotype. All clones were obtained from CRISPR/Cas9 editing of the same parental iPSC line (WTC11), which has been extensively validated and used for genetic engineering by the Allen Cell Collection (http://www.allencell.org/). As before, PIK3CAH1047R homozygous mutant cells clearly separated from heterozygous and wild-type cells, which overlapped on multidimensional scaling (MDS) (Fig. 1A), but we now detected a reduction in the levels of 451 transcripts and an increase in the levels of 710 transcripts in PIK3CAWT/H1047R iPSCs (Fig. 1B). This dropped to 149 and 343 transcripts, respectively, after applying a fold-change cutoff of 1.3 (Fig. 1B; Table S5), indicative of the small magnitude of many expression changes in heterozygous mutants (Fig. S1A). Use of the same cutoff of 1.3, in sharp distinction, yielded 2873 and 2771 transcripts of decreased or increased abundance, respectively, in homozygous iPSC mutants (Fig. 1B; Table S6). Not only was the number of gene expression changes higher by an order of magnitude in homozygous cells, but many expression changes were large compared to wild-type controls (Fig. S1A). The magnitudes of gene expression changes in PIK3CAH1047R/H1047R cells correlated strongly with our previous findings (Spearman's ρ=0.74, P<2e-16) (Fig. S1B), whereas correlation was low (Spearman's ρ=0.1, P<2e-16) for PIK3CAWT/H1047R iPSCs (Fig. S1C), as expected given the smaller number and lower magnitude of observed gene expression changes in heterozygous cells, and the lower depth of previous transcriptomic studies.
Given previous reports that PIK3CAH1047R heterozygosity in breast epithelial cells extensively remodels gene expression (Hart et al., 2015; Kiselev et al., 2015), we undertook further transcriptional profiling in two unrelated cellular models of genetic PIK3CA activation. First, we examined iPSCs derived from a woman with clinically obvious but mild PROS due to mosaicism for PIK3CAE418K (Fig. 1C) (Parker et al., 2019). Heterozygous iPSCs were compared to wild-type lines established simultaneously from dermal fibroblasts from the same skin biopsy, which is possible due to genetic mosaicism of the sampled skin. Like PIK3CAWT/H1047R iPSCs, PIK3CAWT/E418K iPSCs closely clustered with isogenic wild-type controls on MDS plotting (Fig. 1D), with only 30 differentially expressed genes (Table S7). We also studied previously reported Pik3caWT/H1047R mouse embryonic fibroblasts (MEFs) 48 h after Cre-mediated Pik3caH1047R induction (Moniz et al., 2017). Wild-type and Pik3caWT/H1047R MEFs were superimposable on an MDS plot (Fig. 1E), with only 192 downregulated and 77 upregulated genes (Table S8). Our findings suggest that there are bona fide transcriptional changes induced by heterozygosity for PIK3CAH1047R, but these are dramatically smaller in number and magnitude than changes induced by homozygosity for PIK3CAH1047R.
To assess whether transcriptional changes observed in iPSCs were mirrored in the proteome, we applied label-free proteomics to the iPSC lines used in our previous study (Madsen et al., 2019). Approximately 4600 protein ratios were obtained for both heterozygous versus wild-type and homozygous versus wild-type iPSC comparisons, as estimated using a novel Bayesian approach based on the Markov Chain Monte Carlo (MCMC) method (Robin et al., 2019 preprint). In contrast to other algorithms, the MCMC method generates an error estimate alongside each protein concentration that permits a more confident determination of proteins with the most robust differential expression. The number of differentially expressed proteins correlated with PIK3CAH1047R allele dosage, with 54 and 258 differentially expressed proteins in PIK3CAWT/H1047R and PIK3CAH1047R/H1047R cells, respectively (Fig. 1F, Table S9,S10). Of these, 27 proteins were differentially expressed in both heterozygous and homozygous PIK3CAH1047R iPSCs (Table S11), with 16 changing in opposite directions (Fig. 1F). There was a strong correlation between differentially expressed proteins and corresponding transcripts in PIK3CAH1047R/H1047R iPSCs (Fig. S2A,B), but not in heterozygous mutants (Fig. S2C,D). As for the relatively weak correlation seen between transcriptomic experiments for heterozygous cells, this likely reflects the small magnitude of gene expression changes induced by heterozygous PIK3CAH1047R (Fig. 1B; Fig. S1C).
Collectively, these findings corroborate the existence of a threshold of PI3K pathway activity that determines the large majority of gene expression changes in PIK3CAH1047R/H1047R iPSCs in a near-binary manner. Although deeper sequencing did reveal statistically significant gene expression changes in heterozygous iPSCs, and although these changes may contribute to growth-related phenotypes in PROS when sustained across development, effect sizes were modest and more variable. Similar findings in heterozygous MEFs suggest that this may be generalisable to differentiated cell types, irrespective of species. This consolidates the view that only homozygosity for PIK3CAH1047R results in robust and widespread transcriptional changes in otherwise normal diploid cells, arguing against a universal ‘butterfly’ effect of heterozygosity based on studies of a genetically abnormal breast epithelial cell line (Hart et al., 2015; Kiselev et al., 2015).
PIK3CAH1047R/H1047R iPSCs show evidence of signalling ‘rewiring’
We previously demonstrated a graded increase in AKT (S473) phosphorylation across heterozygous and homozygous PIK3CAH1047R iPSCs (Madsen et al., 2019). To assess in more detail whether the near-binary gene expression difference between heterozygous and homozygous PIK3CAH1047R cells is underpinned by corresponding differences in indices of PI3K pathway activation, we profiled phosphorylation of a wider repertoire of pathway components using reverse phase phosphoprotein array (RPPA) technology.
Changes in protein phosphorylation were surprisingly modest, with the largest change a twofold increase in AKT phosphorylation (on S473 and T308) in PIK3CAH1047R/H1047R cells. Contrasting with the near-binary response seen at the transcriptional level, heterozygous and homozygous PIK3CAH1047R expression generally produced graded phosphorylation of PI3K pathway components, with slightly higher levels in homozygous iPSCs (Fig. 2A). None of the mutant genotypes showed consistently increased phosphorylation of the mTORC1 target P70S6K or its downstream substrate S6 (Fig. S3A), perhaps reflecting saturation at this level of the pathway due to other stimuli for mTORC1 in the complete culture medium (e.g. amino acids) (Valvezan and Manning, 2019). When deprived of growth factors for 1 h before RPPA profiling, both heterozygous and homozygous mutants exhibited increased P70S6K phosphorylation, whereas S6 phosphorylation remained similar to wild-type cells (Fig. 2B).
Inhibition of PI3Kα activity with the PI3Kα-selective inhibitor BYL719 for 24 h fully reversed canonical PI3K signalling-related changes in the phosphorylation of downstream proteins, including AKT, GSK3, FOXO1, TSC2 and P70S6K (Fig. 2B). Consistent with these signalling changes, we previously showed that the same dose of BYL719 (100 nM) abolishes the increased tolerance to growth factor deprivation-induced death conferred by heterozygous or homozygous PIK3CAH1047R in iPSCs (Madsen et al., 2019). Despite its effects on the primary PI3K signalling cascade, PI3Kα inhibition failed to reverse other changes observed in PIK3CAH1047R/H1047R iPSCs, including increased phosphorylation of SMAD2 and ERK1/2, and increased expression of c-MYC and IGF1R (Fig. 2B; Fig. S3B). This suggests signalling rewiring in PIK3CAH1047R/H1047R iPSCs that is partially resistant to relatively short-term inhibition of the inducing stimulus.
Pathway and network analyses implicate NODAL/TGFβ signalling in PIK3CAH1047R allele dose-dependent stemness
Pathway and network analyses were next applied to proteomic and transcriptomic data to identify candidate mechanism(s) mediating PIK3CAH1047R allele dose-dependent stemness. Consistent with our previous study (Madsen et al., 2019), TGFβ1 was again the most significant predicted upstream activator according to Ingenuity Pathway Analysis (IPA) of the top 2000 upregulated and top 2000 downregulated transcripts in PIK3CAH1047R/H1047R iPSCs (Fig. 3A). TGFβ1 was also the most significant upstream activator predicted by analysis of PIK3CAH1047R/H1047R proteomic data (Fig. 3B). This is consistent with strong NODAL mRNA upregulation and increased pSMAD2 (S465/S467) in PIK3CAH1047R/H1047R iPSCs in this study (Table S6 and RPPA data in Fig. 2, respectively).
Although PIK3CAWT/H1047R iPSCs showed ∼tenfold fewer differentially expressed genes than homozygous iPSC cells, IPA in heterozygous iPSCs also revealed multiple TGFβ pathway-related stimuli among predicted upstream activators (Fig. 3C). Moreover, TGFβ1 was predicted to be one of only two significant upstream activators when analysis was performed on genes concordantly differentially expressed (N=180) in PIK3CAH1047R mutant iPSCs versus wild-type controls (Fig. 3C; Table S12).
The other significant upstream regulator common to heterozygous and homozygous PIK3CAH1047R iPSCs was MAPK1 (also known as ERK2), consistent with RPPA findings and immunoblot evidence of increased ERK kinase phosphorylation in PIK3CAH1047R mutant iPSCs (Madsen et al., 2019; Fig. 2A; Fig. S3A). The significance of predicted TGFβ activation in heterozygous PIK3CAH1047R iPSCs (overlap P-value=1.7e-05) was much lower than in homozygous (overlap P=4.3e-21) mutants. This is in keeping with the much lower effect size in heterozygous cells, and consistent with a critical role for the NODAL/TGFβ pathway in mediating the allele dose-dependent effect of PIK3CAH1047R in human iPSCs.
To complement IPA analysis, which is based on highly curated proprietary datasets, we undertook non-hypothesis-based weighted gene correlation network analysis (WGCNA) – a network-based data reduction method that seeks to determine gene correlation patterns across multiple samples, irrespective of the function of individual genes (Langfelder and Horvath, 2008). Using all transcripts expressed in wild-type, heterozygous and homozygous PIK3CAH1047R iPSCs (Fig. 4A), this analysis returned 43 modules (or clusters) of highly interconnected genes (Fig. 4B). Of the two modules with the highest correlation with the homozygous trait, one showed enrichment for several Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway terms relevant to stemness of PIK3CAH1047R/H1047R iPSCs, notably including ‘signalling pathways regulating pluripotency in stem cells’ (Fig. 4C).
Given previous evidence of strong activation of NODAL/TGFβ signalling in homozygous mutant cells, we next constructed the minimal network of differentially expressed genes in PIK3CAH1047R/H1047R iPSCs that linked pluripotency, PI3K and TGFβ signalling pathways (Fig. 4D). This approach allowed us to navigate the signalling rewiring and to link strong PI3K pathway activation, stemness and NODAL/TGFβ signalling in an unbiased manner. Indeed, the resulting network exhibited high interconnectivity, with multiple shared nodes across all three pathways, suggesting close crosstalk between PI3K and NODAL/TGFβ signalling in stemness regulation. That most nodes represented genes with increased expression in homozygous mutants strengthens the notion that strong oncogenic PI3Kα activation stabilises the pluripotency network in human iPSCs. The MYC oncogene stood out as the only network node intersecting with all three signalling pathways, suggesting it may comprise a key mechanistic link in the observed phenotype.
Inhibition of NODAL/TGFβ signalling destabilises the pluripotency gene network in PIK3CAH1047R/H1047R iPSCs
NODAL/TGFβ signalling plays a critical role in pluripotency regulation (Vallier et al., 2005; Mesnard et al., 2006; Xu et al., 2008), and a differentiation-resistant phenotype has been reported in NODAL-overexpressing iPSCs (Vallier et al., 2004). Together with increased NODAL expression in homozygous PIK3CAH1047R iPSCs and computational identification of enhanced NODAL/TGFβ pathway activity in PI3K-driven ‘constitutive’ stemness (Madsen et al., 2019 and current study), this led us to hypothesise that strong PI3Kα-dependent induction of NODAL underlies establishment of the differentiation-resistant phenotype of these cells. Specifically, we hypothesised that autocrine NODAL enhances NODAL/TGFβ signalling in PIK3CAH1047R/H1047R iPSCs, with resulting increased NANOG expression ‘locking’ the cells in perpetual stemness (Xu et al., 2008).
Testing this hypothesis in iPSCs is challenging for biological and technical reasons, including lack of specific pharmacological inhibitors of NODAL, and difficulty in detecting subtle early phenotypic consequences of partial destabilisation of the iPSC pluripotency gene regulatory network. Moreover, the widely adopted maintenance medium and coating substrate we used for cell culture both contain TGFβ ligands (Vukicevic et al., 1992; Chen et al., 2011), which may mask effects of NODAL repression by PI3Kα-specific inhibition. We previously found that treatment of PIK3CAH1047R/H1047R iPSCs in this ‘complete’ maintenance medium with 500 nM BYL719 reduces NODAL mRNA expression within 24 h, but has no discernible effect on increased NANOG mRNA levels (Madsen et al., 2019).
To minimise confounding effects of exogenous TGFβ ligands, we prepared medium with and without recombinant NODAL supplementation, and assessed expression of NODAL and NANOG as surrogate markers of stemness over 72 h of culture. We also reduced the BYL719 concentration to 250 nM given the increased iPSC toxicity observed with 500 nM BYL719 (Madsen et al., 2019); and pilot experiments (not shown) in which 24 h treatment with 250 nM but not 100 nM BYL719 in complete medium reduced NODAL mRNA expression in PIK3CAH1047R/H1047R iPSC clones. Within 48 h, exclusion of NODAL from the medium resulted in the expected downregulation of NODAL and NANOG expression in wild-type iPSCs, and this was greater still at 72 h (Fig. 5; Fig. S4A). However, in PIK3CAH1047R/H1047R iPSCs, NODAL removal had no effect on the increased NODAL and NANOG expression (Fig. 5; Fig. S4A), in line with a self-sustained stemness phenotype. Exposure of NODAL-free PIK3CAH1047R/H1047R cultures to 250 nM BYL719 had a visible colony growth-inhibitory effect (Fig. S5) and decreased NODAL expression within 24 h, and this continued to decrease subsequently (Fig. 5). This is consistent with the known ability of NODAL to control its own expression through a feed-forward loop (Hill, 2018). However, despite a 55% reduction in NODAL mRNA after 72 h, little effect on NANOG expression was seen (Fig. 5). This may reflect the short time course studied (to avoid confounding effect of passaging), or the exquisite sensitivity of iPSCs to residual upregulation of NODAL in homozygous PIK3CAH1047R iPSCs. This may be compounded by residual low levels of TGFβ-like ligands in the coating substrate, or possibly by increased expression of two other TGFβ superfamily ligands, GDF3 and TGFB2, observed in homozygous mutant cells (Table S6).
To confirm that NODAL/TGFβ signalling is required for the maintenance of stemness in PIK3CAH1047R/H1047R iPSCs, the cells were treated with SB431542 – a specific inhibitor of TGFβ and NODAL type I receptors (Inman et al., 2002). This completely repressed NODAL expression within 24 h, accompanied by downregulation of NANOG and POU5F1 expression (Fig. 5). Given that the pluripotent stem cell state is stabilised by an autoregulatory feedforward interaction between these transcription factors, their downregulation is expected to result in altered expression of multiple stemness and differentiation markers. Confirming this, we used a lineage-specific gene expression array to demonstrate a similar reduction in mRNA expression of several other well-established stemness markers (MYC, FGF4 and GDF3), which were upregulated in PIK3CAH1047R/H1047R iPSCs at baseline (Table S6), performing the analysis after 48 h of NODAL/TGFβ pathway inhibition (Fig. S4B). Despite the short treatment, we also found evidence for the expected neuroectoderm induction upon inhibition of the NODAL/TGFβ pathway (Vallier et al., 2004, 2009), reflected by increased expression of CDH9, MAP2 and PAPLN (Fig. S4B).
Collectively, these data suggest that the stemness phenotype of PIK3CAH1047R/H1047R iPSCs is mediated by self-sustained NODAL/TGFβ signalling, most likely through PI3K dose-dependent increase in NODAL expression, and that this is amenable to reversal through inhibition of the NODAL/TGFβ pathway but not of PI3Kα itself.
PIK3CAH1047R is the most common activating PIK3CA mutation in human cancers and in PROS (Madsen et al., 2018). We recently found that PIK3CA-associated cancers often harbour multiple mutated PIK3CA copies, and demonstrated that homozygosity but not heterozygosity for PIK3CAH1047R leads to self-sustained stemness in hPSCs (Madsen et al., 2019). High-depth transcriptomics in this study confirmed that heterozygosity for PIK3CAH1047R induces significant but very modest transcriptional changes; observed both in CRISPR-edited hPSCs with long-term PIK3CAH1047R expression and in mouse embryonic fibroblasts (MEFs) upon acute PIK3CAH1047R induction by Cre, with canonical PI3K pathway activation seen in both cases (this study and Berenjeno et al., 2017; Moniz et al., 2017; Madsen et al., 2019). Similarly, hPSCs with heterozygous expression of PIK3CAE418K, a ‘non-hotspot’ mutation, were transcriptionally indistinguishable from their isogenic wild-type controls. However, in contrast to the mild transcriptional consequences of these heterozygous variants, homozygosity for PIK3CAH1047R was associated with differential expression of nearly a third of the hPSC transcriptome, suggesting widespread epigenetic reprogramming. This near-binary response is not a consequence of a similar quantitative difference in PI3K pathway activation, as assessed by phosphoprotein profiling, which instead showed a relatively modest and graded increase in homozygous versus heterozygous PIK3CAH1047R hPSCs. This implies that the apparent PI3K signalling threshold that determines the cellular response in hPSCs is ‘decoded’ distally to the canonical pathway activation.
Using a combination of computational analyses and targeted experiments, this study further provides evidence for self-sustained NODAL/TGFβ pathway activation as the main mechanism through which PIK3CAH1047R homozygosity ‘locks’ hPSCs in a differentiation-resistant state that has become independent of the driver mutation and the associated PI3K pathway activation. We suggest that homozygosity but not heterozygosity for PIK3CAH1047R promotes sufficient NODAL/TGFβ pathway activity to induce increased NODAL and downstream NANOG expression to levels that stabilise the stem cell state, yet are not high enough to tip the balance towards mesendoderm differentiation (Madsen et al., 2019). Exactly how PI3K activation regulates NODAL expression remains unknown. A potential mechanism involves increased expression of the stem cell reprogramming factor MYC, which was observed at both mRNA and protein levels in homozygous but not heterozygous PIK3CAH1047R iPSCs. Furthermore, MYC was the only node in the WGCNA-based network of pluripotency, PI3K and TGFβ pathway components that was shared by all three pathways (Fig. 4D). MYC has previously been shown to exert oncogenic effects that depend on a sharp threshold of MYC expression, reminiscent of the effects we observe for allele dose-dependent PIK3CA activation (Murphy et al., 2008). Elevated MYC has also been shown to allow PIK3CAH1047R-induced murine breast cancers to become independent of continuous PIK3CAH1047R expression (Liu et al., 2011).
Stabilisation of the stemness phenotype in hPSCs by strong genetic PI3K pathway activation may be generalisable beyond this model system. BYL719 (alpelisib; Novartis), the PI3Kα-selective inhibitor used in our cellular studies, was recently approved for use in combination with anti-oestrogen therapy in oestrogen receptor-positive breast cancers (André et al., 2019). In a separate study focusing on human breast cancer, we have described the use of computational analyses to demonstrate a strong positive relationship between a transcriptomically derived PI3K activity score, stemness gene expression and tumour grade in breast cancer (Madsen et al., 2020 preprint). Previous reports have suggested a role for NODAL in driving breast cancer stemness and aggressive disease (Bar-Eli, 2012; Margaryan et al., 2019), with potential links to mTORC1 activation (Katsuno et al., 2019; Jewer et al., 2020). Our findings that BYL719 fails to fully reverse the increased NODAL and stemness gene expression in homozygous PIK3CAH1047R iPSCs suggests that inhibition of TGFβ signalling as a pro-differentiation therapy warrants investigation as a co-therapy with PI3K inhibitors in breast tumours with strong PI3K pathway activation. The lack of widespread transcriptional changes upon heterozygous expression of mutant PIK3CA in otherwise genetically normal cell models may explain the low oncogenicity of this genotype in isolation in vivo.
Finally, our observations are important for future studies seeking to model human PIK3CA-related diseases. The modest changes observed in heterozygous PIK3CAH1047R cells, in sharp contrast to the radical transcriptional alterations in homozygous cells, emphasise the importance of careful allele dose titration when artificial overexpression systems are used to model disorders caused by genetic PIK3CA activation. Our findings in heterozygous cells are also a reminder that very small effect sizes in cellular systems may summate and result in major human phenotypes over a life course. That such minor changes are found in a cellular study of a rare and severe disorder emphasises the challenges of modelling much more subtle disease susceptibility conferred by GWAS-detected genetic associations, in which cellular effect sizes are likely to be smaller still.
MATERIALS AND METHODS
iPSC culture and treatments
The derivation of the iPSC lines, including associated ethics statements, has been described previously (Madsen et al., 2019). All lines were grown at 37°C and 5% CO2 in Essential 8 Flex (E8/F) medium on Geltrex-coated plates, in the absence of antibiotics. For maintenance, cells at 70-90% confluency were passaged as aggregates with ReLeSR, using E8 supplemented with RevitaCell (E8/F+R) during the first 24 h to promote survival. A detailed version of this protocol is available via protocols.io (doi: dx.doi.org/10.17504/protocols.io.4rtgv6n).
All cell lines were tested negative for mycoplasma and genotyped routinely to rule out cross-contamination during prolonged culture. Short tandem repeat profiling was not performed. All experiments were performed on cells within ten passages after thawing.
Collection for RNA sequencing and total proteomics
For RNA sequencing and total proteomics, subconfluent cells were fed fresh E8/F 3 h before snap-freezing on dry ice and subsequent RNA or protein extraction. Relative to the results in Madsen et al. (2019), the current transcriptomic data of PIK3CAH1047R were obtained more than 6 months following the first study, with cells at different passages, and were thus independent from one another. Moreover, sample collection for the second transcriptomics experiment was conducted over 3 days according to a block design, thus allowing us to determine transcriptional differences that are robust to biological variability.
Cell lysate collection for RPPA
For RPPA in growth factor-replete conditions, cells were fed fresh E8/F 3 h before collection. To assess variability due to differences in collection timing, clones from each iPSC genotype were collected on each one of 3 days according to a block design, giving rise to a total of 22 cultures. To test the effect of the PI3Kα-specific inhibitor BYL719, cells were treated with 100 nM drug (or DMSO only as control treatment) for 24 h and exposed to growth factor removal within the last hour before collection. All cells were washed in Dulbecco's PBS (without calcium and magnesium) before collection to rinse off residual proteins and cell debris.
NODAL/TGFβ signalling studies
Wild-type or homozygous PIK3CAH1047R iPSCs were seeded in 12-well plates all coated with Geltrex from the same lot (2052962; diluted in Dulbecco's modified Eagle's medium/F12, lot RNBH0692). Cells were processed for seeding at a ratio of 1:15 according to the standard maintenance protocol. One day after seeding, individual treatments were applied to triplicate wells. Briefly, cells were first washed twice with 2 ml and 1 ml of Dulbecco's PBS to remove residual growth factors. The base medium for individual treatments was Essential 6 supplemented with 10 ng ml−1 heat-stable FGF2. This was combined with one of the following reagents or their diluent equivalents: 100 ng ml−1 NODAL (diluent, 4 mM HCl); 250 nM BYL719 (diluent, DMSO); and 5 µM SB431542 (diluent, DMSO). Cells were snap-frozen on dry ice after 24, 48 and 72 h following a single Dulbecco's PBS wash. Individual treatments were replenished daily at the same time of day to limit temporal confounders.
Mouse embryonic fibroblast culture
The derivation and culture of the wild-type and PIK3CAWT/H1047R MEFs used in this study have been reported previously (Moniz et al., 2017). Cell pellets were collected on dry ice 48 h after induction of heterozygous PIK3CAH1047R expression, without prior starvation.
Induced pluripotent stem cell lysates were collected in QIAzol and processed for RNA extraction with the DirectZol Kit as per the manufacturer's instructions. The final RNA was subjected to quantification and quality assessment on an Agilent Bioanalyzer using the RNA 6000 Nano Kit, confirming that all samples had a RNA integrity number (RIN) score of 10. For PIK3CAH1047R iPSCs and corresponding wild types, an Illumina TruSeq Stranded mRNA Library Prep Kit was used to synthesise 150-bp-long paired-end mRNA libraries, followed by sequencing on an Illumina HiSeq 4000, with an average depth of 70 million reads per sample. PIK3CAWT/E418K and isogenic control iPSCs were subjected to 50-bp-long single-end RNA-seq at an average depth of 20 million reads per sample.
MEF RNA was extracted using Qiagen's RNAeasy miniprep (with QIAshredder). All samples had a confirmed Agilent Bioanalyzer RIN score of 10. An Illumina TruSeq Unstranded mRNA kit was used to prepare 100-bp-long paired-end libraries, followed by Illumina HiSeq 2000 sequencing. Details of the subsequent data analyses (raw read mapping, counting, statistical testing, pathway and network analyses) are provided in supplementary Materials and Methods.
Label-free total proteomics
Cells were cultured to subconfluence in Geltrex-coated T175 flasks, and protein was harvested by lysis in 3 ml modified RIPA buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% NP-40, 0.5% Na-deoxycholate and 1 mM EDTA] supplemented with phosphatase inhibitors (5 mM β-glycerophosphate, 5 mM NaF, 1 mM Na3VO4) and protease inhibitors (Roche cOmplete ULTRA Tablets, EDTA-free). The lysates were sonicated on ice (4×10 s bursts, amplitude, 60%; Bandelin Sonopuls HD2070 sonicator) and spun down for 20 min at 4300 g. Ice-cold acetone was added to the supernatant to achieve a final concentration of 80% acetone, and protein was left to precipitate overnight at −20°C. Precipitated protein was pelleted by centrifugation at 2000 g for 5 min and solubilised in 6 M urea, 2 M thiourea and 10 mM HEPES (pH 8.0). Protein was quantified using the Bradford assay and 8 mg of each sample was reduced with 1 mM dithiothreitol, alkylated with 5 mM chloroacetamide and digested with endopeptidase Lys-C (1:200 v/v) for 3 h. Samples were diluted to 1 mg ml−1 protein using 50 mM ammonium bicarbonate and incubated overnight with trypsin (1:200 v/v). Digested samples were acidified and urea removed using SepPak C18 cartridges. Peptides were eluted, and an aliquot of 100 μg set aside for total proteome analysis. The peptides were quantified using the Pierce quantitative colorimetric peptide assay. The equalised peptide amounts were lyophilised and resolubilised in 2% acetonitrile and 1% trifluoroacetic acid in order to achieve a final 2 μg on-column peptide load.
Mass spectrometry data acquisition
All spectra were acquired on an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific) operated in data-dependent mode coupled to an EASY-nLC 1200 liquid chromatography pump (Thermo Fisher Scientific) and separated on a 50 cm reversed phase column (Thermo Fisher Scientific, PepMap RSLC C18, 2 µM, 100A, 75 µm×50 cm). Proteome samples (non-enriched) were eluted over a linear gradient ranging from 0-11% acetonitrile over 70 min, 11-20% acetonitrile for 80 min, 21-30% acetonitrile for 50 min, 31-48% acetonitrile for 30 min, followed by 76% acetonitrile for the final 10 min with a flow rate of 250 nl/min.
Survey-full scan mass spectrometry spectra were acquired in the Orbitrap at a resolution of 120,000 from m/z 350-2000, automated gain control (AGC) target of 4×105 ions, and maximum injection time of 20 ms. Precursors were filtered based on charge state (≥2) and monoisotopic peak assignment, and dynamic exclusion was applied for 45 s. A decision tree method allowed fragmentation for ion trap MS2 via electron transfer dissociation (ETD) or higher-energy collision dissociation (HCD), depending on charge state and m/z. Precursor ions were isolated with the quadrupole set to an isolation width of 1.6 m/z. MS2 spectra fragmented by ETD and HCD (35% collision energy) were acquired in the ion trap with an AGC target of 1e4. Maximum injection time for HCD and ETD was 80 ms for proteome samples. Details of the subsequent data analyses (FASTA file generation and mass spectrometry searches) are provided in supplementary Materials and Methods.
Reverse phase protein array
For RPPA, snap-frozen cells were lysed in ice-cold protein lysis buffer containing 50 mM HEPES, 150 mM NaCl, 1.5 mM MgCl2, 10% (v/v) glycerol, 1% (v/v) Triton X-100, 1 mM EGTA, 100 mM NaF, 10 mM Na4P2O7, 2 mM Na3VO4 (added fresh), a 1× EDTA-free protease inhibitor tablet and a 1× PhosStop tablet. Protein concentrations were measured using the Bio-Rad DC protein assay, and all concentrations were adjusted to 1 mg ml−1 with lysis buffer and 1× SDS sample buffer [10% glycerol, 2% SDS and 62.5 mM Tris-HCl (pH 6.8)] supplemented with 2.5% β-mercaptoethanol.
The protein lysates were processed for slide spotting and antibody incubations as described previously (Macleod et al., 2017). Briefly, a four-point dilution series was prepared for each sample and printed in triplicate on single pad Avid nitrocellulose slides (Grace Biolabs) consisting of eight arrays with 36×12 spots each. Next, slides were blocked and incubated in primary and secondary antibodies. The processed arrays were imaged using an Innopsys 710 slide scanner. Non-specific signals were determined for each slide by omitting the primary antibody incubation step. For normalisation, sample loading on each array was determined by staining with Fast Green dye and recording the corresponding signal at 800 nm. Details for all primary and secondary RPPA antibodies are included in Table S3. Details of all subsequent data analyses, including statistical testing and evaluation of antibody/assay performance, are provided in supplementary Materials and Methods.
Cellular RNA was extracted as described above for RNA-seq, and 200 ng was used for cDNA synthesis with High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Subsequent SYBR Green-based qPCRs were performed on 2.5 ng total cDNA. TaqMan hPSC Scorecards (384-well) were used according to the manufacturer's instructions with minor modifications. Further details on protocol modifications and all data analysis steps are provided in supplementary Materials and Methods. Details for all primers are included in Table S4.
Bespoke statistical analyses are specified in the relevant sections above and in supplementary Materials and Methods. No statistical methods were used to predetermine sample size.
We thank Dominique McCormick and Ineke Luijten for help with RNA extraction and cDNA synthesis; Cornelia Gewert for technical support; and Marcella Ma, Brian Lam and Michelle Dietzen for technical support with RNA-seq and genomics analyses. We thank Evelyn K. Lau for help with GEO upload of the MEF RNA-seq data. We are grateful to Prof. Siddhartan Chandran and his group for iPSC culturing facilities and to Pamela Brown (Shared University Research Facilities Biomolecular Core, University of Edinburgh, UK) for access to qPCR facilities.
Conceptualization: R.R.M., R.K.S.; Methodology: R.R.M., J.L., R.K.S.; Software: X.R.; Validation: R.R.M.; Formal analysis: R.R.M., J.L., R.G.K., X.R., F.V., K.G.M.; Investigation: R.R.M., J.L., R.G.K., F.V., K.G.M., L.S.M.; Resources: N.O.C., R.L., B.V., R.K.S.; Data curation: R.R.M., J.L., R.G.K., X.R., F.V., K.G.M., L.S.M.; Writing - original draft: R.R.M., R.K.S.; Writing - review & editing: R.R.M., J.L., R.L., B.V., R.K.S.; Visualization: R.R.M., J.L.; Supervision: R.R.M., R.L., B.V., R.K.S.; Project administration: R.R.M., R.K.S.; Funding acquisition: R.L., B.V., R.R.M., R.K.S.
This work was supported by the Wellcome Trust (105371/Z/14/Z and 220464/Z/20/Z to R.R.M., and 210752/Z/18/Z to R.K.S.); the UK National Institute for Health Research (NIHR) Cambridge Biomedical Research Centre; and a Boak Student Award from Clare Hall (to R.R.M.). Work in the laboratory of B.V. is supported by Cancer Research UK (C23338/A25722) and the UK NIHR University College London Hospitals Biomedical Research Centre. Metabolic Research Laboratories Core facilities are supported by the Medical Research Council Metabolic Diseases Unit (MC_UU_12012/5) and a Wellcome Major Award (208363/Z/17/Z). R.L. is funded by a Lundbeckfonden Fellowship. K.G.M., N.O.C. and the University of Edinburgh RPPA facility is supported by a Cancer Research UK Centre award.
Raw data and bespoke RNotebooks containing guided scripts and plots are available via the Open Science Framework (doi: 10.17605/OSF.IO/MUERY). The original RNA-seq data have been deposited in GEO under accession numbers GSE134076 (H1047R iPSC data), GSE138161 (E418K iPSC data) and GSE135046 (MEF data). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2019) partner repository with the dataset identifier PXD014719.
R.K.S. is a consultant for HotSpot Therapeutics (Boston, MA, USA). B.V. is a consultant for Karus Therapeutics (Oxford, UK), iOnctura (Geneva, Switzerland) and Venthera (Palo Alto, CA, USA), and has received speaker fees from Gilead Sciences (Foster City, CA, USA). N.O.C. is a director of Ther-IP Ltd (Edinburgh, UK) and founder, shareholder and advisor for PhenoTherapeutics Ltd (Edinburgh, UK), and a member of the advisory board and shareholder of Amplia Therapeutics Ltd (Melbourne, Australia).