(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



A scalable Drosophila assay for clinical interpretation of human PTEN variants in suppression of PI3K

['Payel Ganguly', 'Department Of Cellular', 'Physiological Sciences', 'Life Sciences Institute', 'Djavad Mowafaghian Centre For Brain Health', 'University Of British Columbia', 'Vancouver', 'British Columbia', 'Landiso Madonsela', 'Department Of Molecular Biology']

Date: 2021-11

Gene variant discovery is becoming routine, but it remains difficult to usefully interpret the functional consequence or disease relevance of most variants. To fill this interpretation gap, experimental assays of variant function are becoming common place. Yet, it remains challenging to make these assays reproducible, scalable to high numbers of variants, and capable of assessing defined gene-disease mechanism for clinical interpretation aligned to the ClinGen Sequence Variant Interpretation (SVI) Working Group guidelines for ‘well-established assays’. Drosophila melanogaster offers great potential as an assay platform, but was untested for high numbers of human variants adherent to these guidelines. Here, we wished to test the utility of Drosophila as a platform for scalable well-established assays. We took a genetic interaction approach to test the function of ~100 human PTEN variants in cancer-relevant suppression of PI3K/AKT signaling in cellular growth and proliferation. We validated the assay using biochemically characterized PTEN mutants as well as 23 total known pathogenic and benign PTEN variants, all of which the assay correctly assigned into predicted functional categories. Additionally, function calls for these variants correlated very well with our recent published data from a human cell line. Finally, using these pathogenic and benign variants to calibrate the assay, we could set readout thresholds for clinical interpretation of the pathogenicity of 70 other PTEN variants. Overall, we demonstrate that Drosophila offers a powerful assay platform for clinical variant interpretation, that can be used in conjunction with other well-established assays, to increase confidence in the accurate assessment of variant function and pathogenicity.

DNA sequencing is becoming commonplace in the clinic, as physicians read your DNA to determine if you have variations in gene sequence that may help with diagnosis and therapy planning. The assumption is that we would know a damaging sequence variation if we saw it. However, this is in fact extremely difficult, and to this day most sequence variations observed in people have unknown disease implication. We must turn to experimental assessments of the variant’s impact on gene function. This current report shows that an organism widely used to understand human disease mechanisms, Drosophila melanogaster, is a valid option for testing the function of hundreds of variants in a human gene of interest. In this case, we tested the function of a hundred variants in the human PTEN gene, found in patients with cancer or autism spectrum disorder, and we were able to pinpoint which ones are likely to contribute to disease, and which were not. Our work provides evidence that Drosophila offers a powerful experimental platform for establishing assay to easily test the function of high numbers of gene variations, that can be used to complement and extend other similar assays.

Funding: Work in the Allan and O’Connor labs was supported by a Simons Foundation for Autism Research Initiative, Award #573845 (2018-2021): “A multi-model screening approach for the functional characterization of large numbers of ASD variants” to D.A., T.O. and C.L. Work in the Verheyen lab was supported by a research grant from the Natural Sciences and Engineering Council of Canada (NSERC), grant number RGPIN/2014-05749 to E.V. Work in the Loewen lab was supported by Canadian Institutes for Health Research award # PJT-152967 to C.L. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2021 Ganguly et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The assay we developed exploits a genetic interaction approach which tests the established role of human PTEN as a suppressor of PI3K-induced increases in PIP3 and pAKT levels, as well as cellular proliferation and tissue growth. In this context, we tested the relative function of ~100 human PTEN variants. We successfully calibrated the assay against 23 total known pathogenic and benign variants, and 4 biochemically characterized variants. This shows the assay provides high positive predictive value for pathogenic variants and high negative predictive value for benign variants. Also, the assay performed very well when benchmarked against variant function data acquired from our previous report using a human cell line [ 50 ]. Overall, we validate the utility of Drosophila in providing reliable and reproducible functional data for high numbers of human variants, in assays conforming to guidelines for clinical interpretation. Notably, as an in vivo genetic model with efficient integrase-based transgenesis for locus-specific integration of human variants [ 23 – 25 ], the model offers tremendous reproducibility. Additionally, the ease of establishing a wide variety of sensitized genetic backgrounds in Drosophila tissues of interest provides a solid foundation for developing assays tuned to defined gene-disease mechanisms. Therefore, as confident assessments of variant function are best achieved through consensus across numerous well-established assays [ 14 , 50 ], we propose that Drosophila assays offer a powerful and flexible option for complementing and extending other well-established assays for clinical variant interpretation.

Identifying PTEN variants with reduced function is considered clinically actionable (ClinGen), and is an important contributory criterion for a PHTS diagnosis, typically resulting in a recommendation of surveillance and screening of related individuals [ 36 , 45 , 46 ]. In both PHTS and ASD, early detection and intervention are viewed as the most effective management strategies for both ASD or PTEN-associated cancers [ 34 , 45 – 49 ]. However, as of last database access (November 22, 2020), ClinVar classified 412 of the annotated 575 missense PTEN variants as VUS, and the COSMIC catalog of somatic mutations records 2492 missense PTEN variants and the gnomAD records 84 missense PTEN variants for which there is no easy way to assess their function. Thus, there is a pressing need to create robust, inexpensive, efficient assays for PTEN variant functionalization.

The goal of our study was to assess the efficacy of Drosophila in well-established assays for scalable variant functionalization and clinical variant interpretation of human coding variants. We use PTEN as an example. PTEN is a highly penetrant autosomal dominant cancer predisposition gene; haploinsufficiency or partial loss of PTEN tumor suppressor activity is observed in sporadic and heritable cancers [ 27 – 30 ]. PTEN hamartoma syndrome (PHTS), which can arise in childhood, and encompasses a variety of germline disorders such as Cowden and Bannayan-Riley-Ruvalcaba (BRR) syndromes [ 31 – 33 ], as well as macrocephaly, epilepsy, mental retardation/developmental delay and autism spectrum disorder (ASD) [ 34 – 36 ]. PTEN is a dual lipid and protein phosphatase that acts as a crucial repressor of the Phosphoinositide 3-kinase (PI3K) /Protein kinase B (PKB/AKT) pathway [ 37 – 39 ]. Increased PI3K/AKT pathway activity is a major contributor to human cancer [ 40 ]. PI3K catalyzes the conversion of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) to phosphatidylinositol-3, 4, 5-triphosphate (PIP 3 ), leading to AKT phosphorylation by phosphoinositide-dependent kinase-1 (PDK1) [ 41 – 44 ]. This promotes cellular survival, proliferation and growth. PTEN’s lipid phosphatase activity converts PIP 3 to PIP 2 , thus suppressing PI3K/AKT signaling.

Drosophila has tremendous potential as a complementary platform for clinical variant interpretation at the scale of hundreds of variants. Yet, to our knowledge, no more than 12 human variants for one gene have ever been tested in a Drosophila study [ 16 , 17 ], and none have been calibrated against enough pathogenic and benign variants. Key challenges include testing defined gene-disease mechanisms, and making the assay scalable which preferentially requires an easily scored phenotype in the F1 progeny of a single cross. Drosophila offers tremendous flexibility in establishing these conditions. While different at a gross anatomical level, most human organ systems and molecular pathways are well conserved [ 18 , 19 ]. Moreover, Drosophila offers an experimentally tractable platform, including a versatile and cost-effective suite of molecular genetic tools, and a wealth of annotated genetic and protein interactions for thousands of genes [ 20 – 22 ]. These establish a solid foundation for developing genetic interaction strategies for explicit testing of defined gene-disease mechanisms. Additionally, integrase-based transgenesis assures that all human variants can be efficiently generated for expression at the same level in specific tissues of interest, or in an orthologous gene replacement strategy [ 23 – 26 ].

Recently, the Clinical Genome Resource (ClinGen) Sequence Variant Interpretation (SVI) Working Group provided guidelines for setting up ‘well-established’ assays for clinical variant interpretation [ 12 ]. A critical consideration is how to translate a variant’s relative function into a clinical interpretation in an assay, reliably and reproducibly. Key criteria include: (i) Assessing variant function in an assay modeling a defined gene-disease mechanism. (ii) Calibrating the assay with a minimum of 11 known pathogenic and benign variants, preferably established on clinical or approved grounds. (iii) Assays must be optimized for rigor and reproducibility.

Discovery of disease-causing genetic variants has long served as a powerful tool in understanding and tackling disease. Exome and genome sequencing is becoming increasingly routine in clinical practice, and is finding variants at an increasing pace [ 1 , 2 ]. The underlying assumption is that the discovery of pathogenic variants will provide insight into disease aetiology and can guide precision therapy [ 3 – 5 ]. However, in spite of the many computational tools for variant effect prediction [ 6 – 11 ], making confident predictions of pathogenicity remains a challenge for many variants, and most remain ‘variants of uncertain significance’ (VUS) [ 12 , 13 ]. To fill this interpretation gap, a wide variety of functional assays are being developed [ 14 , 15 ].

Datasets were acquired as follows: PolyPhen-2 and CADD prediction results were parsed from the Ensembl Genome Browser [ 56 , 57 ]. HEK [ 50 ] and yeast [ 58 ] data were obtained from their publications. ClinVar classifications, gnomAD and COSMIC data were downloaded from their respective database [ 59 – 61 ]. The last date of access was November 22, 2020.

Bioinformatic analyses were done using Python 3.6 and the following packages: numpy 1.18.1, pandas 1.0.3 and matplotlib 3.1.3. Visit https://github.com/jessecanada/PTEN_Fly to view all datasets used in this study and our Jupyter notebooks for codes, instructions, and detailed analytics.

Quantification of pAkt immunoreactivity was performed on each L3 wing disc. In all cases, six or more wing discs were dissected and imaged for each genotype. Images were acquired with a Zeiss LSM 880 with a 34-channel spectral Quasar detector. Representative images of wing discs were processed in Adobe Photoshop CS5 (Adobe Systems, San Jose, CA). To quantify pAkt levels, all tissues were processed with the same reagents, and then imaged and analyzed in identical ways. For each image, 3 z-stacks were selected immediately under the peripodial cell layer from the apical region of each wing disc, to create a maximum projection image that was quantified using Image J. The pAkt intensity levels were reported as a posterior/anterior intensity ratio, where an identical square was drawn within each posterior and anterior compartment of the wing disc to measure the intensity in the same region. Statistical analysis and graphs of datasets were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA) and data within graphs were compared using one-way ANOVA followed by Tukey post hoc analysis. Differences between groups were considered statistically significant when p<0.05, since ANOVA corrects for multiple testing. Data are presented as mean ± Standard Deviation (SD).

Images of L3 wing discs were taken on a Nikon Air laser-scanning confocal microscope (Nikon, Tokyo, Japan). To quantify GFP area and PH3 staining, all tissues were processed with the same reagents, imaged, and analyzed in identical ways. The GFP area was calculated using ImageJ and reported as a ratio of the whole disc area to account for differences in wing disc size. PH3 cell counting was performed on max-projection images using ImageJ software. A minimum of 10 discs were quantified per genotype. Statistical analyses and graphing of datasets were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA) and data within graphs were compared using one-way ANOVA followed by Tukey post hoc analysis. Differences between groups were considered statistically significant when p<0.05. Data are presented as mean ± Standard Deviation (SD).

Wing hair counts (number) were calculated using ImageJ. Wing hairs were counted in a square of fixed size just above the posterior cross vein in the interval between longitudinal veins 3 and 4. The hairs were counted in each genotype in the same position and between 9–11 wings were measured, per genotype.

The relative function of PTEN variants [indicated by the amino acid number and substitution] was assessed using wing size as an assay across seven independent experimental batches ( S1 Table ). Each batch of crosses comprised flies in which a different random subset of variants was tested; PTEN-WT and attP2 were also included in every group of crosses, to control for any batch effects. Stocks were assigned numbers that did not reveal variant identity to assure blinding in scoring crosses. For every cross, the same number of males and females were utilized, under the same culture conditions. A single wing was dissected from each adult progeny at day 13 or 14, then stored in 70% ethanol until slide-mounting in Aquatex mounting solution (EMD Chemicals, USA). A minimum of 10 wings were mounted per genotype. Adult wings were imaged with an Axioplan-2 microscope (Zeiss, Oberkochen, Germany) at 5X and 20X magnification. Wing size areas were calculated using Adobe Photoshop CS3 (Adobe Systems, San Jose, CA).

Larval wing imaginal discs were fixed in 4% Paraformaldehyde (15 or 30 min, RT), followed by 2 rinses in 1X Phosphate buffer saline with 0.1% Tween 20 (0.1% PBT) and then washed in 0.1% PBT three times (5 min, 10 min, 15 min, RT), blocked in 2% bovine serum albumin or 5% donkey serum in 0.1% PBT (1 hr, RT), incubated with primary antibodies overnight (4°C), washed three times in 0.1% PBT, and incubated with secondary antibodies (1 hr 40 min, RT) and washed 3 times. Primary antibodies used: chicken α-GFP (1:1000; ab13970, Abcam, Ontario, Canada), rabbit α-Drosophila phospho-Ser505 Akt (1:200; #4054, Cell Signaling Technology, Danvers, MA), mouse α-human PTEN (1:50; #A487, R&D systems, Minneapolis, MN), rabbit α-Drosophila phospho Histone 3 (1:1000; #9701s, Cell Signaling Technology #9701S, Danvers, MA). Secondary antibodies used were: donkey anti-chicken 488 (1:700), donkey anti-rabbit Cy3 (1:500), donkey anti-mouse Cy5 (1:500) and donkey anti-rabbit 647 (1:500, Jackson ImmunoResearch, West Grove, PA).

Fly crosses were maintained on standard cornmeal food at room temperature. For every cross, the same number of parental flies were crossed and maintained for 36 hours prior to being transferred to grape juice/agar plates, for serial 24-hour egg collection windows. To measure pupal volume, larvae from different genotypes were synchronized such that they were collected in the early L1 stage and cultured under the same controlled conditions (50 larvae/vial) to avoid crowding [ 55 ]. Larvae were genotyped against a GFP-expressing balancer. The length and diameter of Pten heterozygote controls (Pten 100 /+;da-GAL4/attP2), Pten heteroallelic mutants (Pten 100 /Pten 117 ;da-GAL4/attP2) and Pten rescue (Pten 100 /Pten 117 ;da-GAL4/UAS-PTEN-WT) was measured using ImageJ and the pupal volume was calculated by using the formula 4/3π(L/2) (l/2) 2 (L, length; l, diameter) [ 55 ].

Flies were maintained on standard cornmeal food at room temperature. Crosses were raised at 70% humidity and 25°C, unless otherwise noted. The following lines were obtained from the Bloomington Drosophila Stock Center. P{w[+mC] = Dp110-CAAX}1, y[1] w[*] (BL25908) (referred to as PI3K act ). P{UAS-GFP.U}1, y[1] w* P{GawB}bi omb-Gal4 [omb-GAL4] (BL58815) (referred to as omb-GAL4, UAS-GFP, or simply omb-GAL4). y[1] w[*]; P{w[+mW.hs] = en2.4-Gal4}e22c; P{w[+mC] = tGPH}4/TM3, Ser[1] (BL8165). Pten 117 was a kind gift from Hugo Stocker, ETH Zurich, Switzerland [ 51 ]. Pten 100 was a kind gift from Elizabeth Rideout, UBC, Vancouver, Canada [ 52 ]. All UAS-PTEN stocks were previously integrated into the attP2 locus of the Drosophila genome [ 23 , 50 , 53 ]. Integration into the same site ensures reproducible expression of all variants to allow for direct comparison of PTEN variant function in Drosophila tissues. For all crosses, the parental attP2 containing stock was used as a control. Constitutively active Drosophila PI3K (PI3K92E-CAAX, referred to as PI3K act ) [ 54 ], was recombined onto the same chromosome as omb-GAL4 and UAS-GFP. The recombinant line is referred to as omb-GAL4>UAS-PI3K act /FM7, or omb>PI3K act .

Results

Human PTEN rescues a Drosophila Pten hypomorph overgrowth phenotype As a first step towards characterizing PTEN variants, we sought to establish that the annotated wild type PTEN (PTEN-WT) could functionally replace Pten in Drosophila tissues, when integrated into the attP2 locus. Two previous studies show differing phenotypic results for PTEN overexpression in the eye disc, but neither tested if PTEN could rescue loss of Drosophila Pten [62, 63]. Pten nulls are lethal [64], therefore we chose to use a strong hypomorphic heteroallelic combination, Pten100/Pten117, that exhibits increased larval growth leading to increased pupal volume and adult weight [52] (Fig 1A and 1D). Thus, we could be sure we were testing a growth phenotype, rather than lethality which is reported to have contributions from Pten regulation of the actin cytoskeleton as well as cellular growth mechanisms [64]. We examined the capacity for PTEN to rescue the overgrowth phenotype, examining males and females separately due to their difference in size. UAS-PTEN-WT was expressed using the ubiquitous daughterless (da)-GAL4 driver in developing embryos and larvae. We confirmed that Pten100/Pten117; da-GAL4/attP2 mutants had significantly greater pupal volume and adult fly weight than did Pten100/+; da-GAL4/attP2 heterozygous controls (Fig 1A and 1D). In Pten100/Pten117; da-GAL4/UAS-PTEN-WT animals, we observed a rescue of mutant pupal volume and adult fly weight to control levels, in both sexes (Fig 1A–1F). Thus, PTEN-WT integrated into the attP2 locus rescued the overgrowth phenotype found in Pten hypomorphic mutants. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Ubiquitous expression of human PTEN rescues overgrowth of Drosophila Pten hypomorphic mutants. Representative images of female pupae (A) and male pupae (D) of Pten heterozygotes (Pten100/+; da-GAL4/attp2), Pten mutants (Pten100/Pten117; da-GAL4/attp2), and Pten mutants rescued by PTEN (Pten100/Pten117; da-GAL4/UAS-PTEN-WT). Graphs showing female and male pupal volume (B, E), as well as female and male body weight (C, F) for each genotype shown. Sample size (n) is indicated for each genotype. Each datum point within scatter plots represents a single pupa (B, E) or the mean of 10 adult flies (C, D). Expression of PTEN-WT in Pten strong hypomorphs rescued pupal volume and adult body weight in both sexes. Data expressed as mean ± SD and analyzed using one-way ANOVA with Tukey HSD post-hoc; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, ns = not significant. https://doi.org/10.1371/journal.pgen.1009774.g001

Human PTEN suppresses PI3K induced proliferation in developing Drosophila wing discs We further characterized the effects of UAS-PI3Kact expression by examining growth of the developing wing imaginal disc. We marked cells in which transgenes were expressed with UAS-GFP. This served as a proxy for cellular growth or proliferation, since changes in the size of the GFP marked region could reflect increases in either parameter. The omb-GAL4 driver is expressed in a wide domain that encompasses the wing pouch as well as a portion of the presumptive hinge region (Fig 3A). We examined discs from female omb>PI3Kact larvae. Compared to omb>attP2 controls, PI3Kact expression caused a significant increase in the size of the GFP expression domain (Fig 3B, 3C and 3E). This increased size was suppressed to control (omb>GFP) size by co-expression of PTEN-WT (Fig 3D and 3E). Biochemical variants C124S and G129E failed to suppress the PI3Kact-induced increase in the GFP domain, whereas Y138L was partially but not fully suppressive, as was expected since it has an intact lipid phosphatase activity. In contrast, 4A induced a significant reduction in GFP expressing domain beyond control size in the presence of PI3Kact expression. These data are consistent with the impact of these variants on adult wing size, shown in Fig 2D. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Human PTEN suppress PI3K-induced proliferation in Drosophila wing imaginal discs. Wing pouch specific driver omb-GAL4 expressed UAS-PI3Kact and UAS-PTEN variants in the developing wing disc. Illustration of a Drosophila imaginal wing disc showing regions that develop into specific adult tissues after metamorphosis. The anterior to posterior axis (A>P) and the omb-GAL4 expression domain (omb>GFP) are shown (A). Representative images of wing discs of 3rd instar female larva, with the wing pouch area marked by GFP (in grey) and PH3+ cells (in green) stained with anti-PH3 antibody for genotypes as shown (B–D). Graph showing ratio of GFP area over total area of the disc (E) and PH3 positive cells (F) in wing imaginal discs of each genotype. Expressing PTEN-WT reduced PI3Kact induced proliferation. Each datum point in scatter plots represents a single wing imaginal disc. Data are expressed as mean ± SD and analyzed using one-way ANOVA with Tukey HSD post-hoc; p > 0.05, * p < 0.05, ** p < 0.01, **** p < 0.000.1. ns = not significant. Scale bar is 100 μm. https://doi.org/10.1371/journal.pgen.1009774.g003 PI3K/AKT signalling had been shown to promote wing growth by increasing both cell size and proliferation [54, 70]. To determine the relative contribution of these in our model, we first examined cellular proliferation using immunoreactivity to phospho Histone 3 (PH3), a marker of cell mitosis. Expression of PI3Kact increased proliferation significantly (Fig 3C and 3F), and this was suppressed to control levels by co-expression of PTEN-WT (Fig 3D and 3F). A significant failure to suppress this increase in the proliferative marker was observed for the C124S variant only, while 4A resulted in a significant reduction in proliferation. However, the G129E and Y138L variants failed to exhibit any substantive difference from attP2 controls and PTEN-WT. We reason that this is due to the high variability in the number of PH3+ cells observed within each genotype, leading to a substantial overlap of the data, even between attP2 and PTEN-WT. This may be due to variability in the level of active mitosis captured at the time of fixation, in a tissue that proliferates over multiple days. We conclude that enhanced proliferation is a contributor to PI3Kact-induced increase in wing disc size, but that this assay is not, by itself, sufficiently robust to assess differences in variant function. To determine whether increased size of the imaginal wing disc and adult wing results from increased cell size, we counted individual wing hairs within the same fixed area of adult wings. As each cell in the wing blade produces a single hair, the density of hairs in a fixed area can be used as a proxy to determine cell size. When PTEN-WT was co-expressed with PI3Kact, the number of hairs within a fixed area was not significantly different, even though the wing size was significantly smaller, which suggested no change in cell size, but a change in cell number (S2 Fig). Thus, in this context, expression of PI3Kact enhanced proliferation, and this effect was suppressed by co-expression of PTEN-WT. Three of the biochemical variants also showed no change in cell size in the omb>PI3Kact background (S2 Fig). In contrast, the 4A hypermorphic mutant significantly reduced cell size and increased hair count per unit area, showing that cell size can also be impacted by PTEN, but this is only clearly evident for variants with potently abnormal function.

Functionalization of ~100 PTEN variants in suppression of PI3K-induced wing overgrowth Having established that PTEN can suppress PI3K/AKT pathway in a wing growth assay, we proceeded to screen the relative function of PTEN variants from individuals with PHTS, somatic cancer, or ASD (S1 and S2 Tables), using the omb>PI3Kact wing size assay, in addition to PTEN-WT, attP2 (no PTEN) control, and the panel of biochemical variants (C124S, G129E, Y138L, 4A). We crossed homozygous male UAS-PTEN variant flies to omb>PI3Kact females and assessed the wing size of adult female offspring (Fig 5). In brief, a minimum of 10 adult wings were quantified for each variant, and each wing was taken from a different adult female. Variants were tested in seven batches. To normalize for batch effects, we included PTEN-WT and attP2 controls within each batch of crosses (S4 and S5 Figs). The results were plotted to show normalized wing size with respect to PTEN-WT wing size (at 1) and attP2 wing size (at 0) (Fig 5). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Functionalization of human PTEN variants using the wing size assay. 97 UAS-PTEN variant flies were crossed to omb>PI3Kact females and progeny were assessed for wing size (see S1 Table). Graph showing normalized variant activity of PTEN variants where attP2 (no PTEN) = 0 and PTEN-WT = 1. Each datum point indicates a normalized variant mean. Data are expressed as normalized mean. Error bars indicate propagated error (relative standard deviation). https://doi.org/10.1371/journal.pgen.1009774.g005 In these tests, PTEN-WT and the biochemical controls reproduced their relative abilities to suppress omb>PI3Kact wing size. Importantly, C124S and G129E demonstrated no PTEN activity, while Y138L showed around 27% of PTEN-WT activity. The hypermorphic 4A variant acted as a significant gain of function variant causing 374% of PTEN-WT function, and was significantly different from PTEN-WT when performing in-batch statistical comparisons. All other variants showed a range of function across the spectrum of PTEN activity. To test the significant difference between variants, we employed one way ANOVA and a post-hoc Tukey test within each of the seven in-batch datasets (S4 and S5 Figs). Notably, these data showed that only G127R exhibited significantly less activity than attP2, suggestive of interference with endogenous PI3K/AKT signaling potentially through suppression of endogenous Pten activity. Additionally, both E256K and K330E variants were found to be gain of function in this assay. There are currently no studies to corroborate that these variants cause an increase in PTEN function in PI3K/AKT signaling, as seen in our assay.

The fly wing assay accurately assigns PTEN variant function We next assessed the relative accuracy of the wing size assay by comparing it to two other functional datasets from non-Drosophila models for the same set of PTEN variants. These included a HEK cell PTEN assay which is based on measuring relative pAKT/AKT immunofluorescence in the presence of endogenous PTEN [50], as well as a yeast assay that uses colony size as a readout of the lipid phosphatase activity of PTEN [58]. To determine how well the three datasets correlate, we performed hierarchical clustering of the normalized relative PTEN functional scores from the Drosophila wing, HEK and yeast assays (Fig 6A, raw and normalized data provided in S2 and S3 Tables). A heatmap comparing relative PTEN variant function for each assay demonstrated that all three assays generally produced similar functional readouts. However, we noted that the yeast assay assessed more variants as functionally normal than did the other two. Colour-coding of the hierarchical clustering branches was used to visualize groups of variants showing similar functional scores across the three assays (Fig 6B–6D). The branches denoted in yellow, dark orange and cornflower blue showed the greatest discrepancy as over-calling normal PTEN function for numerous variants in the yeast assay (Fig 6C and 6D), when compared to either wing or HEK assays. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. PTEN variant function in the wing assay compared to HEK and yeast assays. We tested the accuracy of the Drosophila wing size assay in predicting PTEN variant activity compared to available datasets assaying the ratio of pAKT/AKT immunoreactivity levels in HEK, or yeast growth dependent on PIP 3 to PlP 2 hydrolysis. (A) Hierarchical clustering to visualize functional similarities in normalized PTEN variant function from each assay, and a heatmap showing relative function. (B-D). Pairwise correlation scatter plots between the assays. Each datum point indicates an individual normalized variant mean (where attP2 = 0 and PTEN-WT = 1) and each cluster is colour coded as in (A). A high correlation of the wing size and HEK assays support the accuracy of wing size assay to accurately predict functions of PTEN variants in vivo across ~100 variants. (C, D) The over-calling of wildtype function for a set of variants (mostly those in the yellow, dark orange and cornflower blue clusters) is evident in the lower right quadrant of both plots. https://doi.org/10.1371/journal.pgen.1009774.g006 To better compare these datasets, we calculated pair-wise Pearson’s correlation coefficients (Fig 6B–6D). While the correlations were significant (p<0.0001) in each case, the Drosophila wing versus HEK assays had the highest (at r = 0.7318), compared to wing versus yeast (r = 0.4999) and HEK versus yeast (r = 0.6287). Notably, these correlations highlighted the variants overcalled as normal by the yeast assay, in the bottom right quadrant of Fig 6C and 6D. Finally, we compared the wing assay to the previous yeast SIM data for PTEN variant interaction with vac14 mutants (r = 0.6718), and to the Drosophila time to eclosion assay (r = 0.6429) [50] (S6 Fig). It is notable that the wing and HEK assays exhibited the highest correlation between any two assays. A small number of variants were quite differentially assessed between the wing and eclosion assays, in particular Y180H, K322E in which the wing assay seems to be the outlier of most assays, and I135T, N276S, K330E in which the eclosion assay seems to be the outlier (S7 Fig). The specific reasons for these differences are unknown, likely reflecting some idiosyncrasy in each assay, perhaps impacting protein stability, localization, protein/substrate interaction, protein translation or post-translational modifications, or any number of other processes.

Comparison of wing size scores to computational predictions Computational prediction tools such as CADD and Polyphen-2 are used to efficiently predict the functional consequence of variants, in lieu of experimental assays [56, 57]. Here, we compared function calls for PTEN variants by the wing assay with CADD Phred and PolyPhen-2 scores. The core of the CADD algorithm uses a logistic regression model which outputs a continuous range of values [56]. CADD does not recommend cut offs to predict the function of a human gene variant. However, other methods such as GAVIN suggest that, for PTEN, CADD Phred scores above 29.3 predict that the variant is damaging, and below 17.33 as benign. Using these criteria, only 13 of the 97 variants were assessed as damaging by this criteria, a significant under-representation of the variants found to be damaging according the wing and HEK assays [76] (Fig 7A). PolyPhen-2 uses a naïve Bayes model whose output is binary, and the scores are derived from the posterior probability. PolyPhen-2 classified many more variants as damaging than CADD (Fig 7B), however there were still 17 variants that disagreed with results from the fly assay; for example, ten loss of function (<50% of PTEN wildtype activity) variants were predicted to be benign (Fig 7B). Overall, we find that these computational methods disagree for many variants, and that the binary Polyphen-2 matches our experimental data better. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. Comparison of PTEN variant function in wings with computational predictions. Graph showing normalized wing size data compared to CADD phred (A) and PolyPhen-2 (B). A score above 0.45 indicates a likely damaging variant by PolyPhen-2 and is shaded dark grey. Each datum point indicates an individual normalized variant mean (where attP2 = 0 and PTEN-WT = 1). https://doi.org/10.1371/journal.pgen.1009774.g007

[END]

[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1009774

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.