frequently rearranged in advanced T-cell lymphomas FRAT regulator of WNT signaling pathway 1 (FRAT1) : Time behavioural study of 3rd order combinations in WNT3A stimulated HEK 293 cells

frequently rearranged in advanced T-cell lymphomas FRAT regulator of WNT signaling pathway 1 (FRAT1) : Time behavioural study of 3rd order combinations in WNT3A stimulated HEK 293 cells shriprakash sinha Independent Researcher; Orcid ID : orcid.org/0000-0001-7027-5788 Address : 104-Madhurisha Heights Phase 1, Risali, Bhilai-490006, India Corresponding author email : [email protected] Abstract FRAT1 is homologous to GSK3 binding protein (GBP), that is known to act as a positive regulator of WNT signaling by stabilizing β-catenin. Gujral and MacBeath [1] provides a quantitative, and dynamic study of WNT3A-mediated stimulation of HEK 293 cells, where they record time based expression profiles of several response genes which correlated significantly with proliferation and migration. By monitoring the dynamics of gene expression using self-organizing maps, they identified clusters of genes that exhibit similar expression dynamics and uncovered previously unrecognized positive and negative feedback loops. However, their study depicts/uses singular measurements of individual gene expression at different time snapshots/points to infer the system wide analysis of the pathway. At any particular time point, it is often the case that genes are working synergistically in combinations, even though their expression measurements are singular in nature. Here, I •enumerate and rank all 2415 FRAT1 related 3rd order combinations in a forest of 71C3combinations using four different sensitivity methods; •show the conserved rankings for FRAT1-X-X combinations, which point to existence of biological synergy of some of these combinations across the different sensitivity methods; and •study the behaviour of some of these combinations related to WNT3A response genes that are ranked by the machine learning search engine (Sinha [2]) in time. Patterns of combinations emerge, some of which have been tested in wet lab, while others require further wet lab analysis. Keywords: Sensitivity analysis, Support vector ranking, Hilbert Schmidt Independence Criterion indices (HSIC) and Sobol indicies, WNT3A ITime behavioural study of 3-odr FRAT1 comb. in WNT3A stimulated cells 1Aspects of unpublished work were presented in a poster session at Cell Symposia: Technology. Biology. Data Science, 9-11 October 2016, Berkeley, California, USA. Preprint submitted to Preprint March 16, 2025 1. Significance Sinha [2] recently demonstrated the use of machine learning based search engine to rank/reveal gene combinations at 2nd order for the time series data by Gujral and MacBeath [1] and showed how it is possible to locate combinations of priority that might be working synergistically, using sensitivity methods and powerful support vector ranking algorithm. However, the problem explodes combinatorially with even a small set of 71 recorded genes in the study by Gujral and MacBeath [1], when one steps to explore 3rd order combinations. With the total number of 71C3(= 57155) combinations, it becomes nearly impossible for any biologist to study the system wide dynamics of any pathway. Also, the amount of time usually needed to search for and test a combination is far more than the search down by the machine learning based search engine. Here, I extend the research work by Sinha [2] to conduct a behavioral study of 3rd order FRAT1 related combinations using individual gene expressions measured in time, in WNT3A stimulated HEK 293 cells. 2. Introduction The details of the machine learning based search engine has been recently published in Sinha [2] and deployed to explore the 2nd order combinations of genes in the data set provided by Gujral and MacBeath [1]. Nevertheless, here, I point to the fundamentals of the published work for completeness. 2.1. A combinatorial problem Sensitivity analysis plays a major role in computing the strength of the influence of involved factors in any phenomena under investigation. When applied to expression profiles of various intra/extracellular factors that form an integral part of a signaling pathway, the variance and density based analysis yields a range of sensitivity indices for individual as well as various combinations of factors. These combinations denote the higher order interactions among the involved factors. Computation of higher order interactions is often time consuming but it gives a chance to explore the various combinations that might be of interest in the working mechanism of the pathway. For example, in a range of fourth order combinations among the various factors of the Wnt pathway, it would be easy to assess the influence of the destruction complex formed by APC, AXIN, CSKI and GSK3 interaction. But the effect of these combinations vary over time as measurements of fold changes and deviations in fold changes vary. So it is imperative to know how an interaction or a combination of the involved factors behave in time and Sinha [2] develops a procedure to track the behaviour by exploiting the influences of these involved factors. 2.2. A possible solution In this work, after estimating the individual effects of factors for a higher order combination, the individual indices are considered as discriminative features. A combination, 2 then, is a feature set in higher order (≥2 ,i.e multivariate). With an excessively large number of factors involved in the pathway, it is difficult to search for important combinations in a wide search space over different orders. Exploiting the analogy with the issues of prioritizing webpages using ranking algorithms, for a particular order, a full set of combinations of interactions can then be prioritized based on these features using a powerful ranking algorithm via support vectors Joachims [3]. Recording the changing rankings of the combinations over time reveals how higher order interactions behave within the pathway and when an intervention might be necessary to influence the interaction within the pathway. 2.3. frequently rearranged in advanced T-cell lymphomas FRAT regulator of WNT signaling pathway 1 (FRAT1) Embi et al. [4] purified a glycogen synthase kinase (GSK) from rabbit skeletal muscle by precipitation with ammonium sulphate, chromatography on DEAE-cellulose and chromatography on hydroxy-apatite. This enzyme was highly specific for glycogen synthase and was separated from virtually all phosvitin kinase and casein kinase activity by thechromatography on DEAE-cellulose. Further,the activity of this enzyme was unaffected by cyclic AMP, cyclic GMP, calcium ions, calcium ions plus calmodulin, and the specific protein inhibitor of cyclic-AMP-dependent protein kinase. They observed that the properties of the enzyme demonstrated that it was distinct from both cyclic-AMP-dependent protein kinase and phosphorylase kinase, the two well characterized glycogen synthasekinases in skeletal muscle. This enzyme was therefore termed GSK3. The phosphorylation of glycogen synthase by GSK3 reached a pleateau near 1.5 molecule phosphate incorporated per subunit under optimal conditions. Finally, glycogen synthase was phosphorylated by cyclic-AMP-dependent protein kinase, phosphor-ylase kinase and GSK3, using conditions where the phosphorylation by anyone protein kinase reached a plateau near one molecule of phosphate incorporated per subunit. These results implied that each protein kinase preferentially phosphorylated a different site(s) on glycogen synthase, which was confirmed by the aminoacid sequence analysis later on. Acceleration of lymphomagenesis in oncogene-bearing transgenic mice by slowtransforming retroviruses has proven a valuable tool in identifying cooperating oncogenes. Jonkers et al. [5] modified this protocol to search for genes that could collaborate with the transgene in later stages of tumor development. Propagation of tumors induced by Moloney murine leukemia virus (M-MuLV) in Eµ-PIM1 or H2-K-MYC transgenic mice by transplantation to syngeneic hosts permitted proviral tagging of ’progression’ genes. They identified a novel gene, designated Frat1 (in this article mFRAT1), upon molecular cloning of common proviral insertion sites that were detected preferentially in transplanted tumors. They further cloned and sequenced both the mouse Frat1 (mFRAT1) gene and its human counterpart (hFRAT1), the encoded protiens of which were highly homologous. Upon infection with a mFRAT1-IRESlacZ retrovirus, they observed that tumor cell lines with high expression of MYC and PIM1 acquired an additional selective advantage in vivo, which underscored the role of mFRAT1 in tumor progression, and the ability of mFRAT1 to collaborate with PIM1 3 and MYC in lymphomagenesis. Dorsal accumulation of β-catenin in early Xenopus embryos is required for body axis formation. Evidence has indicated that β-catenin is dorsally stabilized by the localized inhibition of the kinase X-GSK3, utilizing a novel WNT ligand-independent mechanism. Using a two-hybrid screen, Yost et al. [6] identified GBP, a maternal X-GSK3-binding protein that is homologous to a T cell protooncogene in three wellconserved domains. They observed that GBP inhibits in vivo phosphorylation by XGSK3, and ectopic GBP expression induced an axis by stabilizing β-catenin within Xenopus embryos. Using the similarity of GBP to T cell proto-oncogene FRAT1 by Jonkers et al. [5] and a human EST, they demonstrate that GBP contains a highly conserved domain required for GSK3 binding and inhibition. They observed that both the mouse and human proteins contained three regions that are well-conserved within GBP. A comparison of GBP and Frat1/FRAT1 showed that domain I was 63% identical, II was 70% identical, and III was 83% identical. Their studies with GBP provide a basis for understanding the proliferation of lymphoma cells, that overexpress Frat1. Because Frat1 contains the conserved domain III, it is likely to bind and inhibit GSK3, and they predicted that it stimulated T cell lymphoma proliferation by decreasing GSK3 function. In support of this, Beals et al. [7] have shown that GSK3 acts as a negative regulator of NF-ATc (nuclear factor activated in T cells). GSK3 activity can also be inhibited in response to Wnt growth factor signaling. Cytoplasmic β-catenin is central to the transmission of Wnt signals to the nucleus; in the absence of Wnt signaling, β-catenin is phosphorylated by GSK3, which targets β-catenin for degradation via ubiquitin-dependent proteolysis. In response to Wnt signals, however, GSK3 activity toward β-catenin is inhibited, resulting in the stabilization and accumulation of β-catenin and the activation of TCF/LEF-1 transcription factors. Further, genetic studies have shown that DVL, a cytoplasmic protein downstream of Wnt receptors, is essential for transmitting WNT signals to GSK3. AXIN interacts with DVL and, in addition, binds both GSK3 and β-catenin, thus facilitating GSK3-mediated phosphorylation of β-catenin. Consequently, dissociation of the GSK3/AXIN/β-catenin complex prevents GSK3-mediated phosphorylation of βcatenin, resulting in its stabilization. Jonkers et al. [5] have a proposed mechanism for WNT-induced dissociation of the GSK3/AXIN/β-catenin complex which involves the mammalian protein FRAT1. To understand the mechanism by which DVL acts through GSK to regulate LEF-1, Li et al. [8] investigated the roles of AXIN and FRAT1 in WNT-mediated activation of LEF-1 in mammalian cells. They found that DVL interacted with AXIN and with FRAT1, both of which interacted with GSK. They also found that DVL, AXIN and GSK can form a ternary complex bridged by AXIN, and that FRAT1 could be recruited into this complex probably by DVL. Their observation that the DVL-binding domain of either FRAT1 or AXIN was able to inhibit WNT1-induced LEF-1 activation suggested that the interactions between DVL and AXIN and between DVL and FRAT may be important for WNT signaling pathway. Furthermore, WNT1 appeared to promote the disintegration of the FRAT1-DVL-GSK-AXIN complex, resulting in the dissociation of GSK from AXIN. Thus, formation of the quaternary complex may be an important step in WNT signaling, by which DVL recruits FRAT1, leading to FRAT1-mediated dissociation of GSK from AXIN. 4 Now, AXIN is a negative regulator of embryonic axis formation in vertebrates, which acts through a WNT signal transduction pathway involving the serine/threonine kinase GSK3 and β-catenin. It has been shown to have distinct binding sites for GSK3 and β-catenin and to promote the phosphorylation of β-catenin and its consequent degradation thus inhibiting signaling through β-catenin. Hsu et al. [9] report the results of a yeast two-hybrid screen for proteins that interact with the C-terminal third of AXIN, a region in which no binding sites for other proteins have previously been identified. They found that AXIN can bind to the catalytic subunit of the serine/threonine protein phosphatase 2A (PP2A) through a domain between amino acids 632 and 836. Their results suggested that PP2A might interact with the AXIN-APC-GSK3-β-catenin complex, where it could modulate the effect of GSK3 on β-catenin or other proteins in the complex. Willert et al. [10] show that AXIN is dephosphorylated in response to WNT signaling and dephosphorylated AXIN binds β-catenin less efficiently than the phosphorylated form, thus disengaging β-catenin from the degradation machinery. The AXINdependent phosphorylation of β-catenin catalysed by GSK3, is inhibited during embryogenesis. This protects β-catenin against ubiquitin-dependent proteolysis, leading to its accumulation in the nucleus, where it controls the expression of genes important for development. Thomas et al. [11] demonstrated that FRATtide (a peptide corresponding to residues 188226 of FRAT1) bound to GSK3 and prevented GSK3 from interacting with AXIN. They also observed that FRATtide blocked the GSK3-catalysed phosphorylation of AXIN and β-catenin, thus pointing to a potential mechanism by which GBP could trigger axis formation. In contrast, they also found that FRATtide did not suppress GSK3 activity towards other substrates, such as glycogen synthase and eIF2B (whose phosphorylation is independent of AXIN but dependent on a ’priming’ phosphorylation). This might explain how the essential cellular functions of GSK3 can continue, despite the suppression of β-catenin phosphorylation. Hedgepeth et al. [12] defined a 25-amino-acid sequence in AXIN that is comprised of GSK3βinteraction domain (GID). In contrast to full-length AXIN, which has been shown to antagonize WNT signaling, they found that the GID inhibited GSK3βin vivo and activated WNT signaling. They proposed that the AXIN complex might directly regulate GSK3βenzymatic activity in vivo. Although it is known that GID and FRATtide peptides have a similar functionality, the GSK3 binding domains in FRAT and AXIN do not appear to have homologous amino acid sequences. Bax et al. [13] described the crystal structures of isolated tyrosine 216-phosphorylated GSK3 and of a complex of GSK3 with FRATtide and a sulfate ion. Structures of uncomplexed Tyr216 phosphorylated GSK3 and of its complex with a peptide and a sulfate ion both showed the activation loop adopting a conformation similar to that in the phosphorylated and active forms of the related kinases CDK2 and ERK2. The sulfate ion, adjacent to Val214 on the activation loop, represented the binding site for the phosphoserine residue on ’primed’ substrates. The peptide FRATtide formed a helixturn-helix motif in binding to the C-terminal lobe of the kinase domain, at which FRAT and AXIN compete for binding to GSK3. FRATtide (and FRAT1) does not inhibit the activity of GSK3 toward glycogen synthase (GS). GSK3 sequentially phosphorylates four serine residues on GS, in the sequence SxxxSxxxSxxxSxxxS(p), by recognizing and phosphorylating the first serine in the sequence motif SxxxS(P) (where S(p) rep5 resents a phosphoserine). Protein inhibitors such as FRAT, which appear to regulate kinase activity by modulating protein interactions within signaling complexes, might well inhibit activity toward a selected subset of substrates and might be the target for structure-based ligand design. I present 3rd order combinations of FRAT1 with other genes, that the machine learning based search engine points to, as possible synergistic combinations that might be working in time. 3. Methods Please refer to sections of Sinha [2] for methods, design of study and analysis of data for 2nd order combinations. The same method and design of study is used to generate results for 3rd order combinations presented in this study. 4. Time series data Gujral and MacBeath [1] present a set of 71 WNT-related gene expression values for 6 different times points over a range of 24-hour period using qPCR. The changes represent the fold-change in the expression levels of genes in 200 ng/mL WNT3A-stimulated HEK 293 cells in time relative to their levels in unstimulated, serum-starved cells at 0hour. Gujral and MacBeath [1] state that qPCR data are the means of three biological replicates. Only genes whose mean transcript levels changed by more than two-fold at one or more time points during the 24-hour time course were considered significant. Positive (negative) numbers represent up (down) -regulation. We have already covered the issues related to these data sets in detail in Sinha [14]. Readers are requested to go through them in the pointed reference. The tools of study which are used here have been published in another foundational work in Sinha [14]. 5. Design of experiment 5.1. Pipeline for time series data For the case of time series data, interactions among the contributing factors are studied by comparing triplets of fold-changes at single time points. The prodecure begins with the generation of distribution around measurements at single time points with added noise is done to estimate the indices. A distribution is generated for the fold changes at single time points. Then for every gene, there is a vector of values representing fold changes as well as deviations in fold changes for different time points and durations between time points, respectively. Next a listing of all Cn kcombinations for knumber of genes from a total of ngenes is generated. kis ≥2 and ≤(n−1). Each of the combination of order krepresents a unique set of interaction between the involved genetic factors. After this, the datasets are combined in a specifed format which go as input as per the requirement of a particular sensitivity analysis method. Thus for each pth combination in Cn kcombinations, the dataset is prepared in the required format from 6 the distributions for two separate cases which have been discussed above. (See .R code in mainScript-1-1.R). After the data has been transformed, vectorized programming is employed for density based sensitivity analysis and looping is employed for variance based sensitivity analysis to compute the required sensitivity indices for each of the pcombinations. This procedure is done for different kinds of sensitivity analysis methods. After the above sensitivity indices have been stored for each of the pth combination, the next step in the design of experiment is conducted. Since there is only one recording of sensitivity index per combination, each combination forms a training example which is alloted a training index and the sensitivity indices of the individual genetic factors form the training example. Thus there are Cn ktraining examples for kth order interaction. Using this training set SVMRank learn Joachims [3] is used to generate a model on default value Cvalue of 20. In the current experiment on toy model Cvalue has not been tunned. The training set helps in the generation of the model as the different gene combinations are numbered in order which are used as rank indices. The model is then used to generate score on the observations in the testing set using the SV MRank classi f y Joachims [3]. Note that due to availability of only one example per combination, after the model has been built, the same training data is used as test data to generates the scores. This procedure is executed for each and every sensitivity analysis method. This is followed by sorting of these scores along with the rank indices (i.e the training indices) already assigned to the gene combinations. The end result is a sorted order of the gene combinations based on the ranking score learned by the SV MRank algorithm. Finally, this entire procedure is computed for sensitivity indices generated for each and every fold change at time point and deviations in fold change at different durations. Observing the changing rank of a particular combination at different times and different time periods will reveal how a combination is behaving. Note that the following is the order in which the files should be executed in R, in order, for obtaining the desired results (Note that the code will not be explained here) - • use source(”mainScript-1-1.R”) with arguments for Dynamic data •source(”SVMRankResults-D.R”), to rank the interactions (again this needs to be done separately for different kinds of SA methods), •use source(”Combine-Time-files.R”), if computing indices separately via previous file, •source(”Sort-n-Plot-D.R”) to sort the interactions. Note that the sorting is chages the interaction ranking in time. Thus •use source(”Interaction-Priority-Intime.R”) to find the prioritized ranking of each and every interaction over the different time points and finally •use source(”Print-RankingAND-Interaction-Rank.R”) to print individual ranking of the required input factor with other interaction factors. 6. Results & Discussion 6.1. Time series data by Gujral and MacBeath [1] NOTE - Ranking was assigned on scores that were sorted in DECREASING values. So, 1 was assigned to highest score and vice versa. Results for the 3rd order interactions are presented here. The results first discuss 7 the behaviour of interactions across the snapshots of time using the computed sensitivities on fold change measurements per time snapshot. The analysis was done using 4 different sensitivity indices. Out of the 71C3combinations, I consider/present only those combinations that show a ranking within first 10,000 out of 57,155. This choice is liberal and biologists/oncologists can have a more stricter choice as per need. Two observations are made, •the ranking of a particular combination is conserved (i.e within the 10,000 range) in a particular time point or in the early phase or late phase of WNT3A stimulation, across the majority of the four sensitivity methods, which is a strict criteria of assessment or •the ranking of a particular combination is conserved across time points/phase (i.e they are within the 10,000 range) and the majority of the four sensitivity methods, which is relaxed criteria of assessment. Applying this filter helps reveal important combinations of interest that might be working synergistically at a higher order level in the cell. Regarding technical points of implementation, the rankings were generated without scaling/normalizing the time series data provided by Gujral and MacBeath [1]. For estimating the sensitivity indices, a small gaussian distribution using the function rnorm that generates a vector of normally distributed random variables given a vector length n (here 9, the 10th one is the mean/recorded gene regulation itself), a population mean µand population standard deviation σ. The syntax for using rnorm is as follows: rnorm(n, mean, sd). Further, I use the jitter funtion to add a little bit of noise to the data. This helps to see if the generated rankings are robust or not. 6.2. Enumeration and ranking of 2415 FRAT1-X-X combinations from Gujral and MacBeath [1] In the supplementary section, I present four files, each containing the rankings of 3rd order combinations, that wary in time (shown for 5 time points). Each file represents the rankings computed using a particular sensitivity method. The changing rankings in time for a particular combination represents the importance of contribution/role that combination plays in the cell stimulated with WNT3A. The sensitivity methods used are Hilbert Schmidt Independence Criterion indices (HSIC) indices (with rbf and linear kernel in Da Veiga [15]) and Sobol indicies (with 2002 implementation in Saltelli [16] and martinez implementation in Martinez [17] and Baudin et al. [18]). 6.3. Conserved machine learning rankings for tested FRAT1-X-X combinations A total of 2415, 3rd order combinations involving FRAT1 were obtained from a full set of 71C3= 57155 combinations. Further, from this selected set, using the above criteria for conserved rankings, I report/tabulate the meaningful combinations that might be working synergistically. Tables 2, 3 and 4 show the rankings for the same combinations as in table 1, but using rbf kernel for HSIC, 2002 implementation for SOBOL and martinez implementation for SOBOL, respectively. As one tallies the rankings of across these tables for a particular combination, one finds that the role of the combination of interest is conserved. This conservation points to the existence of the biological 8 RANKING @tiUSING HSIC - LINEAR 3rd order comb. t1t3t6t12 t24 3rd order comb. t1t3t6t12 t24 FOSL1-FRAT1-SENP2 24 25444 38488 23401 24549 CXXC4-FRAT1-FRZB 64 19925 35266 282 20625 AES-AXIN1-FRAT1 153 39308 39740 48864 34865 CSNK1D-FGF4-FRAT1 154 4058 50213 49903 6296 CXXC4-FRAT1-TLE2 212 18126 46011 57 25507 CXXC4-FRAT1-FZD1 296 10717 21805 2575 15715 CXXC4-FRAT1-FBXW4 304 26037 23203 450 10918 FRAT1-NLK-SENP2 308 14795 5137 12639 16597 FOSL1-FRAT1-WNT4 372 34237 49971 11736 48629 FRAT1-FZD1-SFRP4 507 39272 1172 47699 40072 CXXC4-FRAT1-TCF7L1 519 13366 30568 2083 18784 FRAT1-NLK-WNT4 534 1131 4271 13789 39755 APC-BCL9-FRAT1 599 22924 30930 45217 55567 DVL1-FOXN1-FRAT1 613 42124 2382 9631 35079 DKK1-FRAT1-SENP2 669 29229 40235 2545 51616 FOSL1-FRAT1-TLE2 675 26389 36474 6816 48298 FRAT1-JUN-WNT2 676 35197 23478 4515 23224 CXXC4-FOSL1-FRAT1 731 20058 29538 14504 27891 CTNNBIP1-FRAT1-WNT5A 734 14176 29898 41568 9680 FBXW2-FGF4-FRAT1 741 34437 49723 51069 4028 FRAT1-JUN-PITX2 748 50033 38113 50412 44660 CTNNBIP1-FRAT1-KREMEN1 758 28255 42529 10768 13750 FRAT1-PORCN-SFRP4 816 23261 18955 23358 33807 CXXC4-FRAT1-FZD8 827 10881 39520 3401 24688 FOSL1-FRAT1-FBXW4 832 28714 21515 30364 38569 FRAT1-JUN-SENP2 876 27146 34459 14780 13983 FRAT1-FZD1-FZD7 888 26968 7671 38732 41048 FRAT1-PORCN-SENP2 915 19198 9986 14436 18176 AXIN1-FOXN1-FRAT1 990 1738 4384 15268 22220 FRAT1-NLK-WNT3A 1005 19771 7884 9479 10249 FRAT1-FZD1-LRP5 1051 34518 43038 31699 24508 FRAT1-PORCN-RHOU 1077 31453 10861 21530 17749 CCND1-FGF4-FRAT1 1105 52382 24371 40896 23694 FRAT1-JUN-SFRP4 1177 38200 26202 16198 34929 FOSL1-FRAT1-TCF7L1 1246 18255 19691 45038 20105 DKK1-FRAT1-FRZB 1283 35064 37595 2698 45083 APC-FOXN1-FRAT1 1418 596 2065 13524 33264 CXXC4-FRAT1-WNT3A 1421 24998 29856 17406 1342 DKK1-FGF4-FRAT1 1462 21926 23959 17544 31567 DIXDC1-FGF4-FRAT1 1533 9050 41138 51810 43824 FBXW11-FOXN1-FRAT1 1538 7360 6561 15328 24410 FRAT1-NLK-FBXW4 1570 6400 25113 13146 24103 EP300-FGF4-FRAT1 1588 28152 39282 54116 41791 FOSL1-FRAT1-LRP5 1606 10313 45532 19498 50080 FRAT1-JUN-TCF7L1 1626 39276 40450 30425 12391 FRAT1-PORCN-WNT2B 1796 26246 45830 36444 11708 CXXC4-FRAT1-LRP5 1810 3252 48463 383 12590 EP300-FOXN1-FRAT1 1814 1637 5045 2015 815 FGF4-FOSL1-FRAT1 1842 39537 26432 55090 52175 BCL9-FGF4-FRAT1 1875 5789 48461 56251 35603 FRAT1-JUN-KREMEN1 1937 45988 42307 30915 9629 DKK1-FRAT1-FZD1 1941 7211 29023 2680 47286 DAAM1-FOXN1-FRAT1 1981 44098 29909 11384 47997 FRAT1-PORCN-WNT4 1985 2835 18017 16053 23730 FZD5-FGF4-FRAT1 1988 17464 37325 36960 11187 APC-BTRC-FRAT1 1993 4234 3760 19668 20988 CTNNB1-FRAT1-TLE1 2116 31975 20159 13284 55089 FRAT1-GSK3A-RHOU 2147 48585 3413 51070 8121 DAAM1-FGF4-FRAT1 2191 52861 50322 53931 30508 FRAT1-WNT1-WNT2B 2208 3244 1753 8833 56838 FRAT1-GSK3A-SFRP4 2238 47737 14694 38145 13213 DKK1-FOSL1-FRAT1 2248 20750 40597 26220 53540 FRAT1-FZD7-SFRP4 2265 6984 42886 29035 50691 CSNK2A1-CTNNB1-FRAT1 2318 44804 22708 39543 10094 FRAT1-JUN-PPP2R1A 2465 21726 41887 23096 41465 CSNK1G1-CXXC4-FRAT1 2496 24384 49297 7946 49881 FRAT1-NLK-TLE2 2530 6281 12625 8358 38149 FRAT1-JUN-PPP2CA 2547 32152 28002 1650 1276 CTNNB1-FRAT1-WIF1 2625 18691 37238 24416 4230 FRAT1-GSK3A-PPP2CA 2638 38412 1286 35232 13767 DVL1-FGF4-FRAT1 2664 45629 31255 34356 19416 FRAT1-JUN-SLC9A3R1 2692 5241 18995 27536 35730 CSNK1D-CXXC4-FRAT1 2744 18327 34701 8594 2661 CTBP1-FRAT1-WNT4 2821 37523 56525 1417 51685 CXXC4-FGF4-FRAT1 2842 9594 30978 23790 51147 FRAT1-JUN-LRP5 2861 38548 27221 17003 13998 FRAT1-JUN-WNT3A 2877 41865 29807 34170 11250 CXXC4-FRAT1-WNT2 2894 23698 18152 32 32646 CTNNBIP1-FRAT1-RHOU 2910 38029 54971 19935 51723 CTNNBIP1-FRAT1-FZD1 2935 24095 30151 22445 24470 FRAT1-FBXW4-WNT3A 2950 1626 150 55124 17849 FRAT1-PORCN-TCF7 3001 18766 42465 28851 20276 FRAT1-JUN-TCF7 3032 30172 40041 39760 42770 CTNNB1-FRAT1-WNT4 3049 31688 29436 13852 32929 CSNK1G1-FGF4-FRAT1 3087 3833 51164 49600 29524 FRAT1-JUN-PYGO1 3094 24225 44587 43882 21321 CCND1-FRAT1-FZD8 3095 44826 16342 57124 29952 DAAM1-FRAT1-FZD1 3098 48916 26712 40838 53076 FOSL1-FRAT1-FZD7 3142 34552 21454 20799 55831 FRAT1-WIF1-WNT4 3157 3802 25184 32202 1523 FZD5-CTNNBIP1-FRAT1 3164 40626 31787 13769 20789 FRAT1-GSK3A-KREMEN1 3217 49686 3676 49905 14302 FZD5-FOXN1-FRAT1 3239 1614 3233 20068 5777 DIXDC1-DVL2-FRAT1 3310 15749 16154 12064 34806 FRAT1-WIF1-WNT3A 3328 15377 26382 55074 7682 FGF4-FRAT1-LEF1 3378 28956 25427 26696 11917 FRAT1-JUN-TLE2 3416 30940 881 14082 28776 FGF4-FRAT1-T 3423 44186 35087 36000 23427 CXXC4-FRAT1-KREMEN1 3542 41044 29584 5685 51990 CSNK1D-FRAT1-FZD1 3569 4219 52992 30046 47083 FRAT1-PORCN-PPP2R1A 3641 17091 37502 41949 11356 BTRC-FOXN1-FRAT1 3675 26618 6170 28752 42168 FRAT1-FZD1-TCF7 3848 22218 19562 49291 40637 CSNK1G1-CTNNBIP1-FRAT1 3861 2169 25070 15462 45122 FRAT1-FZD1-WNT4 3864 17916 29682 40787 38438 FRAT1-GSK3A-WNT2B 3884 47252 4540 55696 35586 CCND1-FRAT1-NLK 3929 35214 17635 44956 38660 CCND3-FGF4-FRAT1 3979 45710 31230 52987 44235 FOSL1-FRAT1-WNT5A 3989 11766 16292 43205 16137 CXXC4-FRAT1-WNT2B 4026 15233 43434 56478 23693 DKK1-FOXN1-FRAT1 4029 12980 50095 12148 39677 FOSL1-FRAT1-WNT3A 4035 2654 13405 47524 18665 FBXW2-FOXN1-FRAT1 4037 3650 18095 22100 31935 APC-CCND1-FRAT1 4041 16166 3819 56694 45787 FRAT1-NLK-RHOU 4074 18571 5441 22722 18400 FRAT1-NLK-PPP2R1A 4096 7150 14882 4380 42041 FRAT1-JUN-SFRP1 4122 39180 11075 7077 30063 FRAT1-PYGO1-WNT3A 4132 14097 12408 18181 22475 FRAT1-NKD1-WNT2B 4145 29853 29809 31718 30950 CSNK1G1-DVL2-FRAT1 4160 46593 24765 28311 27782 FRAT1-PORCN-PPP2CA 4204 33092 39327 31761 14446 CCND3-FOXN1-FRAT1 4217 35968 4531 12204 31006 CTNNBIP1-FRAT1-PPP2R1A 4272 14548 33170 16600 48598 FRAT1-PORCN-FBXW4 4346 15930 19660 14012 11065 FRAT1-GSK3A-PITX2 4363 56251 2247 18588 44926 FRAT1-GSK3A-SLC9A3R1 4367 13258 9450 48551 44914 Table 1: Rankings of RHOU-X-X. A list of approximately first 125 combinations with rankings below 10,000 out of 57,155. SA - HSIC; Kernel - linear synergy, whether the combination has been tested or unexplored/untested. 9 [9] W. Hsu, L. Zeng, F. Costantini, Identification of a domain of axin that binds to the serine/threonine protein phosphatase 2a and a self-binding domain, Journal of Biological Chemistry 274 (1999) 3439–3445. [10] K. Willert, S. Shibamoto, R. Nusse, Wnt-induced dephosphorylation of axin releases β-catenin from the axin complex, Genes & development 13 (1999) 1768–1773. [11] G. M. Thomas, S. Frame, M. Goedert, I. Nathke, P. Polakis, P. Cohen, A gsk3-binding peptide from frat1 selectively inhibits the gsk3-catalysed phosphorylation of axin and β-catenin, FEBS letters 458 (1999) 247–251. [12] C. M. Hedgepeth, M. A. Deardorff, K. Rankin, P. S. Klein, Regulation of glycogen synthase kinase 3βand downstream wnt signaling by axin, Molecular and cellular biology 19 (1999) 7147–7157. [13] B. Bax, P. S. Carter, C. Lewis, A. R. Guy, A. Bridges, R. Tanner, G. Pettman, C. Mannix, A. A. Culbert, M. J. Brown, et al., The structure of phosphorylated gsk-3βcomplexed with a peptide, frattide, that inhibits β-catenin phosphorylation, Structure 9 (2001) 1143–1152. [14] S. Sinha, Hilbert-schmidt and sobol sensitivity indices for static and time series wnt signaling measurements in colorectal cancer-part a, BMC systems biology 11 (2017) 120. [15] S. Da Veiga, Global sensitivity analysis with dependence measures, Journal of Statistical Computation and Simulation 85 (2015) 1283–1305. [16] A. Saltelli, Making best use of model evaluations to compute sensitivity indices, Computer physics communications 145 (2002) 280–297. [17] J. Martinez, Analyse de sensibilite globale par decomposition de la variance, Presentation in Journ´ ee des GdR Ondes & Mascot 13 (2011) 207. [18] M. Baudin, K. Boumhaout, T. Delage, B. Iooss, J.-M. Martinez, Numerical stability of sobol’indices estimation formula, in: Proceedings of the 8th International Conference on Sensitivity Analysis of Model Output (SAMO 2016), volume 30, 2016, pp. 50–51. [19] S.-i. Hino, T. Michiue, M. Asashima, A. Kikuchi, Casein kinase iεenhances the binding of dvl-1 to frat-1 and is essential for wnt-3a-induced accumulation of β-catenin, Journal of Biological Chemistry 278 (2003) 14066–14073. [20] E. Hay, C. Faucheu, I. Suc-Royer, R. Touitou, V. Stiot, B. Vayssi` ere, R. Baron, S. Roman-Roman, G. Rawadi, Interaction between lrp5 and frat1 mediates the activation of the wnt canonical pathway, Journal of Biological Chemistry 280 (2005) 13616–13623. 16