Inactivation of RB1 , CDKN2A , and TP53 have distinct effects on genomic stability at side‐by‐side comparison in karyotypically normal cells

Abstract Chromosomal instability is a common feature in malignant tumors. Previous studies have indicated that inactivation of the classical tumor suppressor genes RB1, CDKN2A, and TP53 may contribute to chromosomal aberrations in cancer by disrupting different aspects of the cell cycle and DNA damage checkpoint machinery. We performed a side‐by‐side comparison of how inactivation of each of these genes affected chromosomal stability in vitro. Using CRISPR‐Cas9 technology, RB1, CDKN2A, and TP53 were independently knocked out in karyotypically normal immortalized cells, after which these cells were followed over time. Bulk RNA sequencing revealed a distinct phenotype with upregulation of pathways related to cell cycle control and proliferation in all three knockouts. Surprisingly, the RB1 and CDKN2A knocked out cell lines did not harbor more copy number aberrations than wild‐type cells, despite culturing for months. The TP53‐knocked out cells, in contrast, showed a massive amount of copy number alterations and saltatory evolution through whole genome duplication. This side‐by‐side comparison indicated that the effects on chromosomal stability from inactivation of RB1 and CDKN2A are negligible compared to inactivation of TP53, under the same conditions in a nonstressful environment, even though partly overlapping regulatory pathways are affected. Our data suggest that loss of RB1 and CDKN2A alone is not enough to trigger surviving detectable aneuploid clones while inactivation of TP53 on its own caused massive CIN leading to saltatory clonal evolution in vitro and clonal selection.

Mechanisms causing CIN include defects in chromosome cohesion, the spindle assembly checkpoint (SAC), kinetochore-microtubule attachment, cell-cycle regulation, and an increased number of centrosomes (inducing merotely). 6 TP53, essential for cell cycle control in eukaryotic cells, are commonly found across several types of neoplasia. 2,[9][10][11] RB1 is a tumor suppressor whose homozygous inactivation catalyzes development of the rare tumor retinoblastoma. 12,13 Its protein product pRb is a constituent of the G1/S cell cycle checkpoint, hindering progression to S-phase in presence of faulty double strand break repair caused by defective canonical nonhomologous end joining (cNHEJ). 14 Activated pRb physically binds to the E2F-DP heterodimer protein and remodulates chromatin, resulting in an inhibition of E2F-DP activity. Non pRb bound E2F-DP activates cyclins, cyclin dependent kinases and PCNA, aiding the transition from G1 to S-phase. The protein pRb also inhibits the production of cyclin E as well as MAD2.
Conversely, loss of pRb results in an overexpression of MAD2, which has been shown to induce CIN. 15,16 Loss of pRb has previously been shown to increase chromosomal instability and cause aneuploidy. [16][17][18][19] CDKN2A encodes two different proteins: p14/ARF and p16/ INK4a. The protein p14/ARF can downregulate E2F-dependent transcription, causing G1/S arrest. It also inhibits MDM2, which controls the activity and stability of p53. Loss of p14/ARF hence has a similar effect as loss of p53, that is, abrogation of cell cycle arrest in G2, leading to apoptosis. The protein p16/INK4a binds to CDK4/6 and inhibits its ability to phosphorylate pRb, keeping pRb bound to E2F1 and causing G1/S arrest. CDKN2A dysregulation has been shown to cause aneuploidy and CIN. 20 Both loss of p14/ARF and p16/INK4a may generate an increased incidence of aneuploidy and supernumerary centrosomes through centriole pair splitting, which in turn drive aneuploidy through unequal segregation of the genomic material during mitosis. 21,22 However, the evidence so far for CDKN2A and RB1, respectively causing aneuploidy and CIN, is modest and merely performed with older cytogenetic techniques such as metaphase spreads. In addition, many of the cell lines used are known to be genetically unstable in themselves. 23,24 TP53 is a tumor suppressor encoding the protein p53 involved in pathways encompassing hundreds of genes, acting as a response to a variety of stress signals, inducing apoptosis, cellular senescence, or cell cycle arrest. If the stress is removed, p53 causes an upregulation of MDM2 and thereby induces its own degradation, resulting in a half-life between 5 and 20 min. p53 loss of function may facilitate aneuploidy and enable cells to survive otherwise lethal chromosomal imbalances. 25,26 There is, however, some evidence that loss of p53 by itself may not be a primary cause of aneuploidy, but may synergize with other alterations to promote aneuploidy and facilitate chromosomal imbalances through indirect mechanisms. 27 TP53-alterations are often accompanied by other genetic alterations and seen late in the evolution of a tumor, in which case aneuploidy is already present, 5 but this may vary across cancers.
In this study, we sought to disentangle the effect the three classical tumor suppressor genes RB1, CDKN2A and TP53 has on chromosomal instability. Karyotypically normal hTERT immortalized human fibroblasts were subjected to CRISPR-Cas9 mediated knock-out of RB1, CDKN2A, and TP53, respectively, and the resulting clones were cultured and analyzed under close to identical conditions. In addition, a dataset of prolonged passaging of the wild type cell line was analyzed for comparison to its intrinsic rate of CIN. High-resolution copy number profiling and bulk RNA sequencing was performed at multiple passaging times for each cell line. Our data suggest that loss of RB1 and CDKN2A alone is not enough to trigger surviving detectable aneuploid clones while inactivation of TP53 on its own caused massive CIN leading to saltatory clonal evolution in vitro and clonal selection.

| Clonal evolution under prolonged passaging
The cell line Bj-5ta consists of fibroblasts with a normal karyotype and is known to lack tumorigenic characteristics. 28 The cells have been transfected with an hTERT-expressing plasmid resulting in a constantly active telomerase, continuously sustaining its telomeres, thus allowing Bj-5ta cells to proliferate for a prolonged time compared to normal cells. Few studies have analyzed the genetic profile of this cell line after passaging beyond the Hayflick limit, restraining cells with regular cellular senescence. In a previous study, Bj-5ta cells were subcultured for a total of 45 passages at two different laboratories ( Figure 1A,B). 29 That study illuminated the retained evolutionary capacity of Bj-5ta, as the initial clone was replaced with a new, genetically distinct, subclone after approximately 20 passages. 29 This type of inherent clonal replacement in Bj-5ta was confirmed in our study through phylogenetic analysis of the same original dataset (Dataset S1, Figure 1C,D, and S1). Prolonged culturing by passaging the cells more than 20 times resulted in a significant bottleneck. This baseline F I G U R E 1 Experimental setup and copy number analysis (A) The experimental setup. Wild type Bj-5ta cells were cultured for 45 passages and samples for SNP-array were extracted at 7 time points. The experiment was repeated at two different laboratories, one at Lund University (LU) and one at Ben Gurion University (BGU). 29 CRISPR-technique was used to knock out RB1, CDKN2A, and TP53 in three different cell populations. Samples for SNP-array and RNA-seq were extracted at three different passages. One wild type sample and one TP53 knocked-out cell line from 2016 from which one single sample was available were included. (B) The number of genetic alterations (Events) as a function of passages for the cells cultured at LU and BGU. The middle graph illustrates the total number of genetic alterations found in each sample and the rightmost graph the number of events larger than 1 Mbp across the samples. Significant differences are annotated (Dataset S1). (C) Phylogenetic subclone tree based on the SNP-array data for all samples (Dataset S1). The legend indicates the included samples and their corresponding color code. Darker colors imply a higher passaging number. Pie charts illustrate the proportion of cells in a particular sample that has a specific set of genetic alterations. The chromosomal alterations encompassed by a specific branch are shown above the branches. (D) Fishplots for the TP53, RB1, and CDKN2A-knocked-out cell lines as well as the WT cells cultured at either LU or BGU (Dataset S1)     CDKN2A knocked-out cell lines, that showed no capability to generate new surviving detectable clones over time ( Figure S4).

| RNA sequencing reveals distinct phenotypical representations
To verify the phenotype of the respective knockouts, RNA sequencing was performed followed by differential expression analysis (Dataset S2, Figure S5a). The expressions levels of RB1 and TP53 were low in the corresponding knocked out cell lines, compared to the empty vector samples  It has previously been shown that defective TP53 affects the function of homologous recombination repair and nonhomologous end joining. Hence the mending of double strand breaks will be deficit, resulting in an increased incidence of structural chromosome changes. 38 Neither RB1 nor CDKN2A seem to have a strong scientifically proven connection to defects in these reparation mechanisms.
Notably, the G1/S and G2/M checkpoints do not react to copy number imbalances per se, but merely DNA-damage such as single strand breaks, double strand breaks, oxidations, alkylations, deaminations and mismatches. Many copy number aberrations, such as intrachromosomal aberrations or copy number alterations affecting entire chromosomes or arms, will not be halted at these checkpoints. Hence, they are allowed to continue to the spindle assembly checkpoint (SAC). Also, the SAC has been shown to be affected by loss of TP53, 39 but the effect may vary between cell types. 26 Consequently, aneuploid p53 defective cells are allowed to continue the cell cycle, which increases the risk of missegregation and further aneuploidy. 39  A possible risk with our experimental setup is that protein products are still prevailing in the cell due to slow degradation or that the gene is not adequately knocked out. When performing CRISPR-Cas9 mediated knockout, thorough assessment of the actual knock-out or depletion of the gene product should always be performed. 40 We have here shown through copy number analysis and deep targeted sequencing of cDNA that the RNA-products are nonfunctional. Cell culture was also performed for multiple months, making prevailing protein products extremely unlikely, especially for p53 which has a half-life of merely 20 min.
It is also possible that prolonged culture could result in a form of clonal adaptation where the stable optimum for some cell populations is a diploid cell state. If this is true, the cells should be aneuploid in the initial stages, but eventually converge toward their diploid ancestor. In this study we did not see any aneuploid subclones for neither the RB1 The present study also stresses the impact of clonal evolution in in vitro settings. Even the ancestral Bj-5ta cell line showed an extensive subclonal evolution with an entirely new clone taking over the sample after prolonged passaging. This was replicated in two different laboratories with parallel cultures of the same cell line. 29 Interestingly the cells that had been knocked out for RB1 or CDKN2A exhibited less genomic alterations than the wild type cells cultured at one of the sites. Hence, the manipulation of the cells by knocking out RB1, CDKN2A, and TP53 managed to affect the inherent evolutionary trajectory of the cell line seen when culturing it without manipulation.
This also stresses the need to consider the clonal evolution of the cell line itself when using them for research, to not impede the inferred results of the experiments.
In conclusion, using a very pure model system, the present study questions the long-held notion that RB1 and CDKN2A depleted cells exhibit CIN and puts them in stark contrast to TP53-depleted cells.