Machine Learning Discoveries of PLK4-X Synergy in ETC-1922159 Treated Colorectal Cancer Cells

Machine learning discoveries of PLK4-X synergy in ETC-1922159 treated colorectal cancer cells shriprakash sinha Independent Researcher; Orcid ID : orcid.org/0000-0001-7027-5788 104-Madhurisha Heights Phase 1, Risali, Bhilai-490006, India Abstract Polo like kinase 4 (PLK4) is a serine/threonine-protein kinase that localizes to centrioles and regulates centriole duplication during the cell cycle. Overexpression of PLK4 causes centrosome amplification, and its knockdown leads to loss of centrosomes. In colorectal cancer (CRC) cells treated with ETC-1922159, PLK4 was found to be down regulated along with other genes. A recently developed search engine ranked combinations of PLK4-X (X, a particular gene/protein) at 2nd order level after drug administration. Some of these combinations have been tested in wet lab, however many have been pointed out by the search engine that are yet to be explored/tested. These rankings reveal which PLK4-X combinations might be working synergistically in CRC. In this research work, I cover combinations of PLK4 with possible members of aurora kinase (AURK), centrosomal protein (CEP), ubiquitin specific peptidase (USP), E2F transcription factor (E2F), epithelial cell transforming 2 (ECT2), STIL centriolar assembly protein (STIL), cell division cycle (CDC), cell division cycle associated (CDCA), cyclin dependent kinase (CDK), forkhead box (FOX), kinesin family member (KIF) and small nucleolar RNA host gene (SNHG) family. Keywords: PLK4, Porcupine inhibitor ETC-1922159, Sensitivity analysis, Machine learning, Colorectal cancer. 1. Introduction 1.1. PLK4 PLK4 as a key regulator of centriole duplication (Habedanck et al. [1]). Centriole duplication is tightly controlled to prevent cells from developing multipolar spindles (which promotes chromosomal instability). To limit centriole duplication, PLK4 levels ✩ML dicoveries of PLK4-X synergy in ETC-1922159 treated CRC cells Email address: [email protected] (shriprakash sinha) 1Aspects of unpublished work were presented in a poster session at the first Wnt Gordon Research Conference, from 6-11 August 2017, held in Stowe, VT 05672, USA. Preprint submitted to Preprint November 21, 2025 are controlled through trans-autophosphorylation. Slevin et al. [2] studied its structure and find that it possesses a central region called the cryptic polo box, which reveals two homodimerized polo boxes, PB1-PB2. Thus, with its C-terminal polo box (PB3), PLK4 has a triple polo box architecture that promotes trans-autophosphorylation, limiting centriole duplication to once per cell cycle and facilitates oligomerization and targeting. High frequency of centrosome amplification has been reported in more aggressive tumors. Godinho et al. [3] show that centrosome amplification induced by PLK4 resulted in the formation of invasive protrusions that invade the surrounding matrix. Chan [4] provide a review of mechanisms of centrosome amplification in various cancers and also provide references that implicate the role of PLK4 in centrosome amplification. More recently, Liao et al. [5] show that overexpression of PLK4 promotes tumor progression and induces EMT by regulating the WNT/β-catenin pathway in colorectal cancer. In colorectal cancer (CRC) cells treated with ETC-1922159, PLK4 was found to be down regulated along with other genes. Some combinations of PLK4 have been confirmed in wet lab, however, many of the combinations have not been explored/tested or are known. To reveal these combinations, I use a modification of a recently published machine learning based search engine, details of which are given in the next section. 1.2. Combinatorial search problem and a possible solution In a recently published work Sinha [6], a frame work of a search engine was developed which can rank combinations of factors (genes/proteins) in a signaling pathway. Readers are requested to go through the adaptation of the above mentioned work for gaining deeper insight into the working of the pipeline and its use of published data set generated after administration of ETC-1922159, Sinha [7]. The work uses SVM package by Joachims [8] in https://www.cs.cornell.edu/people/tj/svm_light/svm_ rank.html. I use the adaptation to rank 2nd order gene combinations. 2. Results & Discussion 2.1. PLK4 related synergies 2.1.1. PLK4 - AURK Luo et al. [9] observe that CEP192, PLK4 and AURKB/C associate with the WNT-PCP protein DVL2 and DVL2 recruits CEP192/PLK4/AURKB module to promote protrusive activity and cell motility by mediating a kinase-dependent switch of DAAM1 for DAAM2. Thus there is synergy between PLK4 and AURK family. In colorectal cancer cells treated with ETC-1922159, AURK family members and PLK4, were found to be down regulated and their regulation was recorded independently. I was able to rank 2nd order combination of these AURK members along with PLK4. Table 1 shows rankings of these combinations. Followed by this is the unexplored combinatorial hypotheses in table 2 generated from analysis of the ranks in table 1. The table 1 shows rankings of AURK members w.r.t PLK4. AURKB - PLK4 shows low ranking of 1339 (laplace), 1268 (linear) and 586 (rbf). These rankings point to the 2 synergy existing between the two components, which have been down regulated after the drug treatment. Further, AURKA showed high ranking with PLK4, thus indicating that they might not be working synergistically with PLK4, before the drug treatment. RANKING AURK FAMILY VS PLK4 RANKING OF AURK FAMILY W.R.TPLK4 laplace linear rbf AURKB - PLK4 1339 1268 586 AURKA - PLK4 2247 617 1930 Table 1: 2nd order interaction ranking between PLK4 VS AURK members. One can also interpret the results of the table 1 graphically, with the following influences - •AURK members w.r.t PLK4 with PLK4 −>AURK-B. UNEXPLORED COMBINATORIAL HYPOTHESES AURK members w.r.t PLK4 AURK-B PLK4 Table 2: 2nd order combinatorial hypotheses between PLK4 and AURK members. 2.1.2. PLK4 - CEP From the previous section, Luo et al. [9] also show that upon ACM (active-conditionedmedium) treatment, CEP192 isoform-1 specifically recruits active PLK4, while other CEP192 isoforms can recruit active AURKB. Further, Sullenberger et al. [10] dissected centrosomal localization of PLK4 in G1 and S phase and find that CEP57, CEP63, CEP44, and CEO192 localize in ninefold symmetry. During centriole maturation, they propose that CEP152 (a receptor of PLK4) molecular arrangement creates flexibility for PLK4 and procentriole placement during centriole initiation. Thus there is synergy between PLK4 and CEP family. In colorectal cancer cells treated with ETC-1922159, CEP family members and PLK4, were found to be down regulated and their regulation was recorded independently. I was able to rank 2nd order combination of these CEP members along with PLK4. Table 3 shows rankings of these combinations. Followed by this is the unexplored combinatorial hypotheses in table 4 generated from analysis of the ranks in table 3. The table 3 shows rankings of CEP members w.r.t PLK4. CEP70 - PLK4 shows low ranking 3 of 42 (laplace) and 660 (rbf). CEP152 - PLK4 shows low ranking of 337 (laplace) and 1017 (rbf). CEP76 - PLK4 shows low ranking of 395 (laplace) and 272 (rbf). CEP68 - PLK4 shows low ranking of 458 (laplace), 1537 (linear) and 251 (rbf). CEP55 - PLK4 shows low ranking of 357 (laplace), 474 (linear) and 676 (rbf). CEP57 - PLK4 shows low ranking of 945 (laplace) and 673 (rbf). CEP250 - PLK4 shows low ranking of 1057 (laplace), 1134 (linear) and 642 (rbf). CEP192 - PLK4 shows low ranking of 1109 (laplace), 1146 (linear) and 742 (rbf). CEP44 - PLK4 shows low ranking of 1127 (laplace) and 1003 (rbf). These rankings point to the synergy existing between the two components, which have been down regulated after the drug treatment. Of importance is the fact that the engine pointed out correct/appropriate results for CEP192, CEP152, CEP57 and CEP44 compbination with PLK4, as cited in literature above. Further, CEP164P1, CEP128, CDP41 and CEP78 showed high ranking with PLK4, thus indicating that they might not be working synergistically with PLK4, before the drug treatment. RANKING CEP FAMILY VS PLK4 RANKING OF CEP FAMILY W.R.TPLK4 laplace linear rbf laplace linear rbf CEP70 - PLK4 42 2078 660 CEP152 - PLK4 337 2095 1017 CEP76 - PLK4 395 2444 272 CEP68 - PLK4 458 1537 251 CEP55 - PLK4 357 474 676 CEP57 - PLK4 945 2286 673 CEP250 - PLK4 1057 1134 642 CEP192 - PLK4 1109 1146 742 CEP44 - PLK4 1127 1633 1003 CEP164P1 - PLK4 1951 2319 993 CEP128 - PLK4 2208 130 2502 CEP41 - PLK4 2525 552 2283 CEP78 - PLK4 2594 84 2477 Table 3: 2nd order interaction ranking between PLK4 VS CEP members. One can also interpret the results of the table 3 graphically, with the following influences - •CEP members w.r.t PLK4 with PLK4 −>CEP-70/152/76/68/55/57/250/192/44. UNEXPLORED COMBINATORIAL HYPOTHESES CEP members w.r.t PLK4 CEP-70/152/76/68/55/57/250/192/44 PLK4 Table 4: 2nd order combinatorial hypotheses between PLK4 and CEP members. 2.1.3. PLK4 - USP The role of centrosomal protein in centrosome replication is established and their overexpression can lead to centrosome duplicate abnormality, which is closely associated with tumorigenesis and development. Zhang et al. [11] have demonstrated that CEP120 promotes centrosome amplification and gastric cancer (GC) progression by 4 USP54-mediated deubiquitination of PLK4. In colorectal cancer cells treated with ETC-1922159, USP family members and PLK4, were found to be down regulated and their regulation was recorded independently. I was able to rank 2nd order combination of these USP members along with PLK4. Table 5 shows rankings of these combinations. Followed by this is the unexplored combinatorial hypotheses in table 6 generated from analysis of the ranks in table 5. The table 5 shows rankings of USP members w.r.t PLK4. USP39 - PLK4 shows low ranking of 14 (laplace) and 881 (rbf). USP10 - PLK4 shows low ranking of 325 (laplace) and 931 (rbf). USP1 - PLK4 shows low ranking of 592 (laplace), 1415 (linear) and 938 (rbf). USP28 - PLK4 shows low ranking of 645 (linear) and 1451 (rbf). These rankings point to the synergy existing between the two components, which have been down regulated after the drug treatment. Further, USP36 abnd USP13 showed high ranking with PLK4, thus indicating that they might not be working synergistically with PLK4, before the drug treatment. RANKING USP FAMILY VS PLK4 RANKING OF USP FAMILY W.R.TPLK4 laplace linear rbf USP39 - PLK4 14 2042 881 USP10 - PLK4 325 2285 931 USP1 - PLK4 592 1415 938 USP28 - PLK4 1725 645 1451 USP36 - PLK4 1813 2478 1202 USP13 - PLK4 2191 148 2630 Table 5: 2nd order interaction ranking between PLK4 VS USP members. One can also interpret the results of the table 5 graphically, with the following influences - •USP members w.r.t PLK4 with PLK4 −>USP-39/10/1/28. 2.1.4. PLK4 - E2F In breast cancer cells, Lee et al. [12] observed that overexpression of E2F1, E2F2, or E2F3 increased centrosome amplification and revealed that E2Fs affect the expression of PLK4. They also show that PLK4 is strongly correlated with E2F factors and is a direct transcriptional traget of E2F activators. In colorectal cancer cells treated with ETC-1922159, E2F family members and PLK4, were found to be down regulated and 5 UNEXPLORED COMBINATORIAL HYPOTHESES USP members w.r.t PLK4 USP-39/10/1/28 PLK4 Table 6: 2nd order combinatorial hypotheses between PLK4 and USP members. their regulation was recorded independently. I was able to rank 2nd order combination of these E2F members along with PLK4. Table 7 shows rankings of these combinations. Followed by this is the unexplored combinatorial hypotheses in table 8 generated from analysis of the ranks in table 7. The table 7 shows rankings of E2F members w.r.t PLK4. E2F5 - PLK4 shows low ranking of 84 (laplace) and 898 (rbf). E2F7 - PLK4 shows low ranking of 269 (laplace) and 629 (linear). E2F8 - PLK4 shows low ranking of 973 (laplace), 606 (linear) and 747 (rbf). E2F1 - PLK4 shows low ranking of 1193 (laplace) and 304 (linear). These rankings point to the synergy existing between the two components, which have been down regulated after the drug treatment. Further, E2F2 showed high ranking with PLK4, thus indicating that they might not be working synergistically with PLK4, before the drug treatment. RANKING E2F FAMILY VS PLK4 RANKING OF E2F FAMILY W.R.TPLK4 laplace linear rbf laplace linear rbf E2F5 - PLK4 84 2574 898 E2F7 - PLK4 269 629 1720 E2F8 - PLK4 973 606 747 E2F1 - PLK4 1193 304 1765 E2F2 - PLK4 2152 499 2295 Table 7: 2nd order interaction ranking between PLK4 VS E2F members. One can also interpret the results of the table 7 graphically, with the following influences - •E2F members w.r.t PLK4 with PLK4 −>E2F-5/7/8/1. UNEXPLORED COMBINATORIAL HYPOTHESES E2F members w.r.t PLK4 E2F-5/7/8/1 PLK4 Table 8: 2nd order combinatorial hypotheses between PLK4 and E2F members. 6 2.1.5. PLK4 - ECT2 / STIL Rosario et al. [13] show that PLK4 normally localizes to the midbody and binds to and phosphorylates ECT2. Further, they show that loss of heterozygosity (LOH) occurs at the Plk4 locus in 50% of human hepatocellular carcinomas (HCC) and is associated with reduced PLK4 expression in HCC tumors, which causes failure to localize the ECT2 to the spindle midbody, as one of the phenotypes. Moyer et al. [14] have discovered a molecular basis for the timing of PLK4 activation through accumulation of STIL and show that direct binding of STIL activates PLK4 by promoting selfphosphorylation of the activation loop of the kinase. In colorectal cancer cells treated with ETC-1922159, these individual members and PLK4, were found to be down regulated and their regulation was recorded independently. I was able to rank 2nd order combination of these individual members along with PLK4. Table 9 shows rankings of these combinations. Followed by this is the unexplored combinatorial hypotheses in table 10 generated from analysis of the ranks in table 9. The table 9 shows rankings of INDIVIDUAL members w.r.t PLK4. ECT2 - PLK4 shows low ranking of 382 (laplace), 1493 (linear) and 582 (rbf). STIL - PLK4 shows low ranking of 132 (laplace) and 171 (rbf). These rankings point to the synergy existing between the two components, which have been down regulated after the drug treatment. It is important to note that in the light of the observations in the above literature about the estabilished connection of a member (here ECT2/STIL), we find that the combination of the same with PLK4 gets appropriate ranking in colorectal cancer cells also, thus confirming the possible existence of synergy. RANKING INDIVIDUAL FAMILY VS PLK4 RANKING OF INDIVIDUAL FAMILY W.R.TPLK4 laplace linear rbf ECT2 - PLK4 382 1493 582 STIL - PLK4 132 2010 171 Table 9: 2nd order interaction ranking between PLK4 VS individual members. One can also interpret the results of the table 9 graphically, with the following influences - •individual members w.r.t PLK4 with PLK4 −>ECT2/STIL. UNEXPLORED COMBINATORIAL HYPOTHESES Individual members w.r.t PLK4 ECT2/STIL PLK4 Table 10: 2nd order combinatorial hypotheses between PLK4 and individual members. 7 2.1.6. PLK4 - CDC Bonni et al. [15] show that CDC25C is a substrate of PLK4 as both wild type and kinase active forms of the latter were able to phosphorylate CDC25C. In colorectal cancer cells treated with ETC-1922159, CDC family members and PLK4, were found to be down regulated and their regulation was recorded independently. Additionally, i report the combinations of cell division cycle associated (CDCA) with PLK4 also. I was able to rank 2nd order combination of these CDC/CDCA members along with PLK4. Table 11 shows rankings of these combinations. Followed by this is the unexplored combinatorial hypotheses in table 12 generated from analysis of the ranks in table 11. The table 11 shows rankings of CDC members w.r.t PLK4. CDC123 - PLK4 shows low ranking of 16 (laplace) and 735 (rbf). CDC25C - PLK4 shows low ranking of 321 (laplace), 424 (linear) and 648 (rbf). CDC23 - PLK4 shows low ranking of 322 (laplace), 1389 (linear) and 485 (rbf). CDC25A - PLK4 shows low ranking of 622 (laplace), 171 (linear) and 730 (rbf). CDC45 - PLK4 shows low ranking of 990 (laplace), 1469 (linear) and 1512 (rbf). CDC6 - PLK4 shows low ranking of 1548 (laplace) and 517 (linear) CDC20 - PLK4 shows low ranking of 701 (linear) and 1515 (rbf). These rankings point to the synergy existing between the two components, which have been down regulated after the drug treatment. Additionally, in line with the above literature reference, we find appropriate ranking of CDC25C with PLK4 in colorectal cancer case treated with ETC-1922159, thus pointing towards possible synergy in CRC. Further, CDC7 showed high ranking with PLK4, thus indicating that they might not be working synergistically with PLK4, before the drug treatment. The table 11 also shows rankings of CDCA members w.r.t PLK4. CDCA3 - PLK4 shows low ranking of 208 (laplace) and 149 (rbf) CDCA7 - PLK4 shows low ranking of 585 (laplace) and 186 (rbf) CDCA7L - PLK4 shows low ranking of 1165 (laplace), 626 (linear) and 1243 (rbf) CDCA5 - PLK4 shows low ranking of 1206 (laplace) and 615 (rbf). These rankings point to the synergy existing between the two components, which have been down regulated after the drug treatment. Further, CDCA8, CDCA2 and CDCA4 showed high ranking with PLK4, thus indicating that they might not be working synergistically with PLK4, before the drug treatment. One can also interpret the results of the table 11 graphically, with the following influences - •CDC members w.r.t PLK4 with PLK4 −>CDC-123/25C/23/25A/45/6/20 and •CDCA members w.r.t PLK4 with PLK4 −>CDCA-3/7/7L/5. 2.1.7. PLK4 - CDK Franck et al. [16] found that PLK4 and CEP192, showed reduced levels at centrosomes of mitotic CDK11-depleted cells. CDK11p58, which accumulates only in the vicinity of mitotic centrosomes, directly interacts with the centriole-associated PLK4. In colorectal cancer cells treated with ETC-1922159, CDK family members and PLK4, were found to be down regulated and their regulation was recorded independently. I was able to rank 2nd order combination of these CDK members along with PLK4. Table 13 shows rankings of these combinations. Followed by this is the unexplored 8 RANKING CDC FAMILY VS PLK4 RANKING OF CDC FAMILY W.R.TPLK4 laplace linear rbf CDC123 - PLK4 16 2506 735 CDC25C - PLK4 321 424 648 CDC23 - PLK4 322 1389 485 CDC25A - PLK4 622 171 730 CDC45 - PLK4 990 1469 1512 CDC6 - PLK4 1548 517 2224 CDC20 - PLK4 1677 701 1515 CDC7 - PLK4 1834 2039 1913 RANKING CDCA FAMILY VS PLK4 RANKING OF CDCA FAMILY W.R.TPLK4 laplace linear rbf CDCA3 - PLK4 208 2363 149 CDCA7 - PLK4 585 2152 186 CDCA7L - PLK4 1165 626 1243 CDCA5 - PLK4 1206 1941 615 CDCA8 - PLK4 1557 2092 1041 CDCA2 - PLK4 2103 274 2233 CDCA4 - PLK4 2278 1298 1966 Table 11: 2nd order interaction ranking between PLK4 VS CDC members. combinatorial hypotheses in table 14 generated from analysis of the ranks in table 13. The table 13 shows rankings of CDK members w.r.t PLK4. CDK5RAP2 - PLK4 shows low ranking of 156 (laplace) and 73 (rbf). CDK6 - PLK4 shows low ranking of 552 (laplace), 1329 (linear) and 322 (rbf). CDK20 - PLK4 shows low ranking of 797 9 [6] S. Sinha, Machine learning ranking of plausible (un) explored synergistic gene combinations using sensitivity indices of time series measurements of wnt signaling pathway, Integrative Biology 16 (2024) zyae020. [7] S. Sinha, Sensitivity analysis based ranking reveals unknown biological hypotheses for down regulated genes in time buffer during administration of porcn-wnt inhibitor etc-1922159 in crc, bioRxiv (2017) 180927. [8] T. Joachims, Training linear svms in linear time, in: Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining, ACM, 2006, pp. 217–226. [9] Y. Luo, M. Barrios-Rodiles, G. D. Gupta, Y. Y. Zhang, A. A. Ogunjimi, M. Bashkurov, J. M. Tkach, A. Q. Underhill, L. Zhang, M. Bourmoum, et al., Atypical function of a centrosomal module in wnt signalling drives contextual cancer cell motility, Nature communications 10 (2019) 2356. [10] C. Sullenberger, D. Kong, P. Avazpour, D. Luvsanjav, J. Loncarek, Centrosomal organization of cep152 provides flexibility in plk4 and procentriole positioning, Journal of Cell Biology 222 (2023). [11] C. Zhang, X. Ma, G. Wei, X. Zhu, P. Hu, X. Chen, D. Wang, Y. Li, T. Ruan, W. Zhang, et al., Centrosomal protein 120 promotes centrosome amplification and gastric cancer progression via usp54-mediated deubiquitination of plk4, IScience 26 (2023). [12] M.-Y. Lee, C. S. Moreno, H. I. Saavedra, E2f activators signal and maintain centrosome amplification in breast cancer cells, Molecular and Cellular Biology (2014). [13] C. O. Rosario, M. A. Ko, Y. Z. Haffani, R. A. Gladdy, J. Paderova, A. Pollett, J. A. Squire, J. W. Dennis, C. J. Swallow, Plk4 is required for cytokinesis and maintenance of chromosomal stability, Proceedings of the National Academy of Sciences 107 (2010) 6888–6893. [14] T. C. Moyer, K. M. Clutario, B. G. Lambrus, V. Daggubati, A. J. Holland, Binding of stil to plk4 activates kinase activity to promote centriole assembly, Journal of Cell Biology 209 (2015) 863–878. [15] S. Bonni, M. L. Ganuelas, S. Petrinac, J. W. Hudson, Human plk4 phosphorylates cdc25c, Cell cycle 7 (2008) 545–547. [16] N. Franck, E. Montembault, P. Rome, A. Pascal, J.-Y. Cremet, R. Giet, Cdk11p58 is required for centriole duplication and plk4 recruitment to mitotic centrosomes, PloS one 6 (2011) e14600. [17] R. Penailillo, V. Velasquez, S. Acuna-Gallardo, F. Garcia, M. Sanchez, G. Nardocci, S. E. Illanes, L. J. Monteiro, Foxm1 participates in trophoblast migration and early trophoblast invasion: Potential role in blastocyst implantation, International Journal of Molecular Sciences 25 (2024) 1678. [18] K. Zhou, Y. He, X. Lin, H. Zhou, X. Xu, J. Xu, Kifc1 depends on trim37-mediated ubiquitination of plk4 to promote centrosome amplification in endometrial cancer, Cell Death Discovery 10 (2024) 419. [19] Z. Xu, Y. Sun, D. Wang, H. Sun, X. Liu, Snhg16 promotes tumorigenesis and cisplatin resistance by regulating mir-338-3p/plk4 pathway in neuroblastoma cells, Cancer Cell International 20 (2020) 1–13. [20] B. Madan, Z. Ke, N. Harmston, S. Y. Ho, A. Frois, J. Alam, D. A. Jeyaraj, V. Pendharkar, K. Ghosh, I. H. Virshup, et al., Wnt addiction of genetically defined cancers reversed by porcn inhibition, Oncogene 35 (2016) 2197. 16