p53 isoforms differentially impact on the POLι dependent DNA damage tolerance pathway

The recently discovered p53-dependent DNA damage tolerance (DDT) pathway relies on its biochemical activities in DNA-binding, oligomerization, as well as complex formation with the translesion synthesis (TLS) polymerase iota (POLι). These p53-POLι complexes slow down nascent DNA synthesis for safe, homology-directed bypass of DNA replication barriers. In this study, we demonstrate that the alternative p53-isoforms p53β, p53γ, Δ40p53α, Δ133p53α, and Δ160p53α differentially affect this p53-POLι-dependent DDT pathway originally described for canonical p53α. We show that the C-terminal isoforms p53β and p53γ, comprising a truncated oligomerization domain (OD), bind PCNA. Conversely, N-terminally truncated isoforms have a reduced capacity to engage in this interaction. Regardless of the specific loss of biochemical activities required for this DDT pathway, all alternative isoforms were impaired in promoting POLι recruitment to PCNA in the chromatin and in decelerating DNA replication under conditions of enforced replication stress after Mitomycin C (MMC) treatment. Consistent with this, all alternative p53-isoforms no longer stimulated recombination, i.e., bypass of endogenous replication barriers. Different from the other isoforms, Δ133p53α and Δ160p53α caused a severe DNA replication problem, namely fork stalling even in untreated cells. Co-expression of each alternative p53-isoform together with p53α exacerbated the DDT pathway defects, unveiling impaired POLι recruitment and replication deceleration already under unperturbed conditions. Such an inhibitory effect on p53α was particularly pronounced in cells co-expressing Δ133p53α or Δ160p53α. Notably, this effect became evident after the expression of the isoforms in tumor cells, as well as after the knockdown of endogenous isoforms in human hematopoietic stem and progenitor cells. In summary, mimicking the situation found to be associated with many cancer types and stem cells, i.e., co-expression of alternative p53-isoforms with p53α, carved out interference with p53α functions in the p53-POLι-dependent DDT pathway.


INTRODUCTION
Already in the nineties, shortly after p53 was recognized to be a tumor suppressor [1,2], it became clear that it is a key player in the protection of the genome integrity [3]. Simultaneously, with the discovery of p53′s canonical functions in transcriptionally activating cell cycle-regulatory and pro-apoptotic genes through sequence-specific DNA-binding [4][5][6][7], non-canonical functions of p53 in DNA repair, recombination, and replication, not requiring its transcriptional activity, emerged [8][9][10][11][12][13][14]. After more than 25 years of research, it came as a surprise when twelve TP53-isoforms were identified, which are generated by the use of different promoters, alternative splicing, and the internal ribosome entry site [15,16]. TP53 gene products other than the canonical p53α lack N-and C-terminal domains of the human protein with welldefined biochemical functions like the first transcriptional transactivation domain (TAD1, amino acids [aa] 1 to 39 in p53α) in Δ40p53α (Fig. 1A). These alternative p53-isoforms were then found to be differentially expressed in normal and cancer tissues, revealing pro-survival features of N-terminally truncated Δ133p53 and Δ160p53 in a p53α-dependent and -independent manner [15]. Therefore, co-expression of alternative p53-isoforms provides an additional mechanism to modulate p53α´s tumor suppressor functions beyond TP53 gene mutations mostly affecting the DNAbinding domain (DBD) [17].
While the canonical activities of p53-isoforms have been investigated quite intensively [18], non-canonical functions in stabilizing the genome have remained largely obscure. Recently, we discovered that p53α, via interactions with proliferating cell nuclear antigen (PCNA) and the translesion synthesis (TLS) polymerase iota (POLι), regulates a homology-directed DDTpathway [19,20]. This p53-POLι-dependent DDT-pathway serves to bypass barriers during DNA replication and may confer a prosurvival effect to stem cells from which, however, cancer stem cells also benefit [19][20][21][22]. In this study, we analyzed the role of the p53-isoforms p53β, p53γ, Δ40p53α, Δ133p53α, and Δ160p53α in this p53-POLι-dependent DDT through the analyses of replicationassociated recombination, DNA replication dynamics, and key protein interactions. We provide evidence that these alternative p53-isoforms differentially lost biochemical functions in the homology-directed resolution of replication barriers. When coexpressed with p53α, thereby mimicking their status in cancer tissues, alternative p53-isoforms block the p53-POLι-dependent DDT-pathway which unlocks faster and likely more mutagenic bypass mechanisms.
p53-isoforms differentially affect DNA replication dynamics after mock-and MMC-treatment Activation of homology-directed DDT by p53α has been linked to a replication slow-down detectable by DNA fiber spreading assay [19,20]. Accordingly, we investigated the effect of the p53isoforms on DNA replication speed in K562 cells (Fig. 2). After sequential incubation with 5-Chloro-2-deoxyuridine (CldU) and 5iodo-2-deoxyuridine (IdU) (experimental overview and representative fibers in left panels of Fig. 2), we monitored a similar shortening of the DNA track lengths after expression of p53α or p53β compared to ctrl after mock-treatment ( Fig. 2A). Expression of p53γ generated intermediate track lengths. After treatment with the replication stress-inducing agent MMC only p53α, but not p53β or p53γ, decreased track lengths (Fig. 2B). To investigate, if deceleration of DNA replication was associated with stalling, we Fig. 1 Analysis of the p53 isoforms p53α, p53β, p53γ, Δ40p53α, Δ133p53α, and Δ160p53α in replication-associated recombination. K562 (HR-EGFP/3′-EGFP) cells were transfected with 10 µg expression plasmid for p53α and corresponding amounts for p53β, p53γ, Δ40p53α, Δ133p53α, Δ160p53α or empty vector (EV) in controls (ctrl) as indicated in the graphs. Seventy-two hours after transfection FACS analysis (left panels of B, C) and protein harvesting (right panels of B, C) were performed. Recombination (rec.) fold changes were analyzed flow cytometrically by quantification of EGFP-positive cells among living cells. Mean values from p53α-expressing cells were set to 1 (absolute mean frequencies: 4*10 −5 ). Data were collected from ≥18 individual measurements each. For graphic presentation, calculation of SEM and statistically significant differences via Kruskal-Wallis test followed by two-tailed Mann−Whitney U test GraphPadPrism8.4 software was used. # indicates a statistically significant difference between ctrl and the respective p53-isoform data. ****(# # # #) P < 0.0001. Quantification of protein levels was carried out using ImageLab software, normalized to β-actin, and indicated above the representative immunoblot. A Schematic overview of different domains of p53 isoforms. p53α (highlighted in black) consists of transactivation domain I (TAD I),  transactivation domain II (TAD II), proline-rich domain (PRD), DNA-binding domain, hinge domain (HD), oligomerization domain (OD), C-terminal domain (CTD) while parts of the other p53-isoforms are replaced by other sequences or missing. Note that the color code in the scheme was used in the subsequent graphs. B Role of p53β and p53γ in replication-associated recombination. Left panel shows Recombination (Rec.) fold changes. Right panel shows representative Western Blot analysis of the indicated p53 isoforms. β-Actin served as a loading control. C Role of p53α, Δ40p53α, Δ133p53α, and Δ160p53α in replication-associated recombination. Left panel shows Rec. fold changes. Right panel shows representative Western Blot analysis of the indicated p53 isoforms. β-Actin served as loading control. measured the ratio of two IdU track lengths emanating from the same CldU track (fork asymmetry [FA]; graphic presentations shown on top of right panels in Fig. 2) [19,27]. While MMCtreatment increased FA, no further changes were seen after p53α, p53β, or p53γ expression ( Fig. 2A, B).
Next, we performed DNA fiber spreading assays after the expression of p53α-or ΔN-isoforms. Similar as with the C-terminal isoforms, we observed differences between mock-and MMC-treatments: after mock-treatment expression of p53α and all ΔNisoforms decelerated DNA replication (Fig. 2C). Interestingly, FA analysis revealed increased values in Δ133p53αand Δ160p53αexpressing cells compared to all other samples. After MMCtreatment only p53α shortened DNA track lengths, while FA values did not differ between the samples (Fig. 2D). Hence, whereas track shortening in Δ133p53α and Δ160p53α-expressing cells was associated with stalling after mock-treatment, shortening in p53α or Δ40p53α-expressing cells did not, compatible with a role in continuous deceleration of nascent DNA synthesis. As such a role of p53α was demonstrated to depend on POLι [19], we examined POLι expression levels in K562 cells after mock-and MMC-treatment. However, no significant differences were observed ( Supplementary Fig. S1B, C).
To examine nascent DNA synthesis as a function of the p53isoforms in another p53-negative cell type, we engaged Saos-2 ( Supplementary Fig. S2). Here, p21 and MDM2 were detectable in p53α-expressing cells only ( Supplementary Fig. S2A). Results from DNA fiber spreading assays were consistent with the results in K562 except that in mock-treated cells p53γ was not only intermediate, but fully functional in track shortening (Supplementary Fig. S2B) and Δ40p53 only partially functional ( Supplementary  Fig. S2D).
Since the p53 targets p21 and MDM2 were reported to play roles in the regulation of replication [28][29][30], we expressed exogenous p21 and MDM2 in cells with representative C-terminal and ΔN-isoforms. DNA fiber spreading analyses revealed that these proteins did not reconstitute the replication track shortening phenotype of p53α in p53γ, Δ133p53α or Δ160p53αexpressing cells treated with MMC ( Supplementary Fig. S3). From this, it is unlikely that changes in p21 and MDM2 expression explain dysfunction of p53-isoforms in the p53-POLι DDTpathway.
Taken together, C-terminal and ΔN-isoforms of p53 show at least partial function after mock-but full loss-of-function after MMC-treatment in track shortening. Therefore, failure to decelerate replication under conditions of enforced replication stress mirrors their defect in recombination stimulation (Fig. 1B, C) indicating loss of key biochemical activities in the homologydirected DDT-pathway.
C-terminal and ΔN-isoforms no longer promote ternary p53-POLι-PCNA complex formation in MMC-treated cells Our previous work demonstrated that POLι forms a complex with p53 and PCNA and that this p53-POLι idling complex is required to slow down replication [19,20]. Therefore, we investigated the subnuclear distribution of POLι and PCNA as well as their association after the expression of different p53-isoforms in K562 (Fig. 3). In mock-treated cells, we observed increased accumulation of POLι-foci in all p53-expressing cells except for Δ160p53α and of PCNA-foci except for p53α and Δ40p53α (Fig.  3A, B, D, E). Moreover, we observed a 2-to 4-fold increase of colocalized POLι-PCNA-foci in all p53-expressing cells (Fig. 3C, F). However, after MMC-treatment, increased numbers of POLι-as well as co-localized POLι-PCNA-foci seen after p53α expression was lost with all other p53-isoforms ( Fig. 3A, C, D, F). This defect was similarly seen regarding PCNA-foci in cells expressing C-terminal p53-isoforms but not ΔN-isoforms (Fig. 3B, E). Specific POLι detection was verified by POLι-silencing ( Supplementary Fig.  S4). Moreover, we analyzed the percentages of MMC-treated K562 cells in different cell cycle phases and in apoptosis as indicated by a sub-G1 content ( Supplementary Fig. S5). This analysis did not reveal significant differences in cells expressing different p53isoforms, though a trend of enhanced apoptosis was noticed in p53α-expressing cells.
Next, we investigated physical interactions between p53, POLι, and PCNA in co-immunoprecipitation studies after the preparation of crosslinked chromatin. After pull-down of POLι, we detected less than half of the input levels of p53β and p53γ in the coprecipitate as compared with p53α ( Fig. 4A). In PCNA coprecipitates p53β and p53γ levels were not reduced. After pulldown of p53α without crosslink, we also observed an interaction with PCNA, as reported before [19]. However, PCNA band intensities were reduced by 50−90% relative to input in coprecipitates of the N-terminally truncated p53-isoforms (Fig. 4B). Previously, we showed that p53α binds POLι via its N-terminus [20]. Consistently, p53α but not Δ40p53α co-immunoprecipitated POLι ( Supplementary Fig. S6). Altogether, all alternative p53isoforms are defective in supporting the formation of POLι-PCNAcomplexes after MMC-treatment in situ. In co-precipitation experiments, POLι-p53-complex formation is defective in cells expressing any alternative isoform and PCNA−p53 interactions are compromised in cells expressing ΔN-isoforms of p53.

Regulation of PCNA ubiquitination by p53-isoforms
Deceleration of nascent DNA synthesis in the p53-POLι DDTpathway buys time for the cell to polyubiquitinate PCNA and recruitment of helicase-like transcription factor (HLTF) and Zinc Finger Ran-Binding Domain-Containing Protein 3 (ZRANB3) to bypass the replication barrier [19,20]. Western Blot analysis revealed a statistically significant 1.5-to 1.8-fold increase (P ≤ 0.0156) of PCNA-monoubiquitination in mock-treated cells after expression of any of the p53-isoforms ( Fig. 5A-C). PCNApolyubiquitination showed a similar pattern ( Fig. 5A-C), but the statistical significance of the increase in polyubiquitination was reached only with p53α-expressing cells (P = 0.0156). After MMCtreatment we detected reduced levels of mono-and polyubiquitinated PCNA in cells expressing C-terminal or ΔN-isoforms as compared with p53α ( Fig. 5A, B, D), which reached statistical significance (P = 0.0078) with p53β and p53γ samples (Fig. 5D). We conclude that alternative p53-isoforms are compromised in supporting PCNA-ubiquitination after MMC-treatment.
Subsequently, we silenced HLTF and ZRANB3 to investigate a potential impact on PCNA-ubiquitination ( Supplementary Fig. S7) as both proteins were found to exert critical roles in the POLι DDTpathway, most likely mediating PCNA-polyubiquitination and fork reversal, respectively [19]. In MMC-treated cells expressing p53α, we found that knockdown of HLTF reduced PCNA-poly-but not -monoubiquitination ( Supplementary Fig. S7A, C, E), while knockdown of ZRANB3 had no effect (Supplementary. Fig. S7B, D, F). In MMC-treated cells, neither HLTF nor ZRANB3 affected PCNApolyubiquitination after p53γ or Δ133p53α expression strengthening the notion of their reduced capacity to stimulate PCNApolyubiquitination. Fig. 2 p53α and its isoforms modulate nascent DNA synthesis. K562 cells were transfected with expression plasmids for p53α, alternative p53 isoforms [A, B p53β, p53γ or C, D Δ40p53α, Δ133p53α, Δ160p53α] or EV in controls (ctrl). Forty-eight hours after transfection DNA fiber spreading assay was performed. Graphic overviews in the left panel show the experimental outline and representative fibers. Cells were subsequently incubated with 5-chloro-2′-deoxyuridine (CldU, 20 µM) and 5-iodo-2′-deoxyuridine (IdU, 200 µM) for 20 min. During IdUincorporation cells were either mock-treated (A, C) or treated with 3 µM MMC (B, D). Both, CldU-and IdU-tracks were measured but for clarity graphic presentations in the middle panel focus on IdU-tracks in ongoing forks (≥361 to ≥423 fibers in two independent biological experiments). Right panels show the graphic presentation of FA with the respective schematic overview on top. DNA fibers were reanalyzed comparing track lengths of IdU incorporation (red) originating from the same CldU track (green). Graphs represent ≥66 to ≥121 fibers from two independent biological experiments. For graphic presentation, calculation of SEM, and statistically significant differences GraphPadPrism8.4 software was used. Statistically significant differences among groups were calculated by Dunn's multiple comparisons test. # indicates a statistically significant difference between ctrl and the respective p53 isoform. (#)P < 0.05, ***P < 0.001, ****(# # # #), P < 0.0001 (scale bar: 5 µm).
Strikingly, we obtained almost the same results in Saos-2-cells: Co-expression of each alternative p53-isoform did not decelerate nascent DNA synthesis as seen in p53α + p53α samples versus ctrl independently of treatment, supporting the loss-of-function concept ( Supplementary Fig. S8A, B). Moreover, compared with p53α + EV longer tracks were measured in p53α + Δ133p53α mock-treated cells and p53α + p53β, p53α + Δ133p53α, and p53α + Δ160p53α MMC-treated cells. These Saos-2 data strengthened the concept of an inhibitory effect of the alternative p53isoform Δ133p53α in particular.
Next, we also investigated the subnuclear distribution of POLι and PCNA, as well as their co-localization after co-expression of p53α and alternative isoforms ( Fig. 6C−E). Both, after mock-and MMC-treatment, co-expression of p53α together with any alternative p53-isoform reduced accumulation of POLι and POLι-PCNA colocalizing foci observed after expression of p53α alone (Fig. 6C, E). It also abrogated p53α-mediated PCNA-foci accumulation after MMC-treatment in p53α + p53γ, p53α + Δ40p53α, and p53α + Δ133p53α samples (Fig. 6D). Notably, p53α + p53β co-expression causing intermediate DNA track lengths (Fig. 6B) also showed intermediate enhancement of POLι-PCNA-co-localization after MMC-treatment (Fig. 6E). Altogether, co-expression of alternative isoforms blocked p53α-mediated recruitment of POLι into PCNAcomplexes and consequently slow-down of DNA replication, which was still observed in mock-treated cells expressing individual isoforms. In conclusion, co-expression of p53-isoforms compromises the p53-POLι DDT not only via loss-of-function but also via inhibitory effects.
To test the influence of endogenously expressed p53-isoforms, we performed silencing experiments engaging siRNA targeting either all p53 isoforms or siRNA targeting Δ133p53α (and Δ160p53α), i.e., the isoform most robustly affecting DNA replication speed in the presence of p53α (Fig. 6A, B and Supplementary  Fig. S8A, B). We chose U2OS cells, as well as primary hematopoietic stem and progenitor cells (HSPCs) derived from human cord blood, as both carry wild-type TP53 and express p53α and Δ133p53α (Fig. 7A, B). When performing DNA fiber spreading assays, we found that DNA replication was accelerated both in mock-and MMC-treated U2OS cells and HSPCs after silencing all p53-isoforms (Fig. 7A, B). Selective knockdown of Δ133p53/ Δ160p53 (siΔ133p53) caused replication slowing in MMC-treated HSPCs (Fig. 7B), matching lack of replication deceleration after ectopic expression of Δ133p53α or Δ160p53α as compared to p53α in MMC-treated K562 cells (Fig. 2D). For comparison, replication track lengths were unaffected by siΔ133p53 in U2OS cells, most likely due to a lower Δ133p53α/p53α ratio as compared to HSPCs as deduced from the exposure times necessary to detect p53α and Δ133p53α in immunoblots (Fig. 7A, B). Altogether, the data in HSPCs confirmed an antagonistic role of endogenous Δ133p53 towards the DNA replication decelerating role of p53α.

DISCUSSION
In this study, we provide evidence that both C-terminal and ΔNisoforms show loss of the p53α-specific functions in the p53-POLι-dependent DDT-pathway [19,20]. Underscoring the biological relevance, we find interference of alternative p53isoforms, particularly of Δ133p53α and Δ160p53α, with p53α′s functions in DNA replication both in human tumor and stem cells. ) and then processed for immunoblotting using ubiquityl PCNA (Lys164, D5C7P, Cell Signaling, Massachusetts, USA) antibody, recognizing PCNA protein only when ubiquitinated at Lys164, as well as antibodies against PCNA and p53. β-Actin was used as loading control. "ub" indicates ubiquitination. Quantification of respective protein expression level was carried out using ImageLab software. Levels of PCNA mono-/polyubiquitination were corrected for PCNA and normalized to ctrl (C, D) which was set to 1 on each blot. Statistically significant differences among groups were calculated by Friedman test followed by Wilcoxon-signed ranks test in case of statistical significance. Representative Western Blots from cells expressing ctrl, p53α, and C-terminal isoforms (A) or cells expressing ctrl, p53α, and ΔN-isoforms (B). Heatmap of mean values from ≥ 5 independent experiments (C) or ≥ 4 independent experiments (D).
Alternative p53-isoforms show loss-of-function in the p53-POLι-dependent DDT-pathway mediating homology-directed bypass of replication barriers Treatment with MMC causes DNA intra-and interstrand-crosslinks, which represent the ultimate roadblock for nascent DNA synthesis [34,35]. The p53-POLι-dependent DDT-pathway can overcome these obstacles [20]. Hallmarks of this pathway, namely (i) deceleration of DNA replication, (ii) increased POLι-foci assembly and PCNA-POLι-co-localization, as well as (iii) PCNA-ubiquitination were absent in cells expressing alternative p53-isoforms after Y. Guo et al.
MMC-treatment. Yet, under unperturbed growth conditions, these key functions were retained at least partially. Another hallmark, namely replication-associated recombination was undetectable even in untreated cells expressing alternative p53-isoforms. Such sensitive detection of loss-of-function can be explained because this approach selectively identifies homology-directed bypass at hard-to-replicate sites including extended secondary structures or DNA-crosslinks stemming from naturally occurring aldehydes [21,36]. For comparison, DNA Fiber Spreading Assays reflect nascent DNA synthesis at each progressing replication fork all over the genome irrespective of a barrier, why enforced replication stress is required to uncover defects.
Since alternative p53-isoforms still promoted DNA replication slow-down and PCNA-POLι-co-localization under unperturbed conditions, we conclude that alternative p53-isoforms can build p53-PCNA-POLι-complexes at replication sites at least conditionally. Interestingly, these intermediate DDT-pathway defects originate from different combinations of deficiencies and proficiencies of key biochemical activities linked with specific protein domains (Fig. 8). In case of C-terminal p53-isoforms, both p53β and p53γ possess an intact N-terminus and consequently still interacted with PCNA and partially with POLι in immunoprecipitation experiments [20]. p53β and p53γ also possess an intact DBD [23] and retain the interaction site with topoisomerase-I (aa 302-321 [37]), another critical interaction partner in the pathway [20,25]. However, they lost half of the OD, which is required to enhance binding to three-way DNA junctions in the tetramer conformation [38] and accordingly to trigger p53-POLι-dependent DDT [20]. p53β was reported to still bind DNA, however, in a dimeric conformation [39]. We propose that C-terminal p53isoforms are transiently recruited to replication sites via physical interactions with DDT factors. This may suffice to promote PCNAubiquitination (a pre-requisite for DDT [40]) and deceleration of nascent DNA synthesis under unperturbed conditions, but no longer when MMC-roadblocks require tight association to DNA.
The ΔN-isoforms possess an intact OD and more than two-thirds of the DBD. They are completely devoid of the PCNA and POLι interaction sites within the N-terminal 40 aa of p53α [20,41]. Consistently, we observed that Δ40p53α was defective in coimmunoprecipitating these two proteins. ΔN-isoforms can no longer contact the basal transcription machinery through the TAD I [42]. As a transcriptional activator, p53 prominently upregulates p21 and MDM2 [43][44][45], i.e., two proteins, which exhibit p53-independent functions in DNA replication [29,30,46]. Here, we confirmed reduced p21 and MDM2 levels in cells expressing alternative p53isoforms, particularly in p53γ, Δ133p53α, and Δ160p53α-expressing cells. However, re-expression of p21 and MDM2 did not rescue deceleration of nascent DNA synthesis, excluding that the p53 target gene products are the missing components in cells expressing alternative p53-isoforms. Fig. 6 Co-expression of p53 isoforms affects the POLι-dependent DDT pathway. K562 cells were transfected with a total amount of 10 μg plasmid DNA, containing either EV only in controls (ctrl) or expression plasmid for p53α plus EV (p53α + EV) or for p53α plus p53α (p53α + p53α) or for p53α plus one of the alternative isoforms (p53α + p53β, p53α + p53γ, p53α + Δ40p53α, p53α + Δ133p53α, and p53α + Δ160p53α). For graphic presentation, calculation of SEM and statistically significant differences via Dunn's multiple comparison test GraphPadPrism8.4 software was used. # indicates a statistically significant difference between ctrl and the respective p53-isoform values. *(#)P < 0.05, **(# #)P < 0.01, ***(# # #)P < 0.001, ****P(# # # #) < 0.0001. A, B: DNA fiber-spreading assays were performed 48 h after transfection. The experimental design was as described in the legend to Fig. 2. Both mock-treated CldU-and IdU-tracks were measured but for clarity graphic presentations focus on IdU-tracks in ongoing forks (≥421 fibers (A) and ≥320 fibers (B) in two independent biological experiments). POLι  Both CldU-and IdU-tracks were measured while only IdU-tracks in ongoing forks (≥257 fibers from two independent biological experiments) are graphically presented for clarity. Knockdown of p53-isoforms was verified by immunoblot staining using anti-p53 (DO-11, MCA1704, Biorad) shown in the right panel of (A, B) each. Quantification of Δ133p53α relative to α-Tubulin levels was achieved by Imagelab and is indicated above the blots. The dashed frame in (A) marks an unspecific band stained by DO-11 in U2OS cells that had to be cut out, followed by antibody reincubation and a long exposure for 300 s to permit immunodetection of the Δ133p53α-band. Statistically significant differences between groups were calculated by Dunn's multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
The p53-isoforms Δ133p53α and Δ160p53α were special in that they showed signs of a replication defect already under unperturbed conditions. Thus, we noticed increased replication fork stalling (RFS) in the presence of these p53-isoforms to the extent elicited by MMC-treatment. Shortening of DNA tracks can be explained by exonucleolytic degradation, continuous slowdown of DNA synthesis or by RFS [19]. We, therefore, propose that replication deceleration observed in cells expressing Δ133p53α and Δ160p53α is due to increased RFS and not due to p53α-POLιdependent DDT [19]. As this pathway is inactive in cells expressing exonuclease-deficient p53(H115N) [47], it was proposed that p53-POLι complexes slow down DNA synthesis via iterative idling that gives time for HLTF-mediated PCNA-polyubiquitination and ZRANB3-mediated fork reversal, i.e., homology-directed bypass of the barrier [19]. Compatible with loss of idling, Δ133p53α and Δ160p53α not only lack the N-terminal PCNA-and POLιinteraction site, but additionally and distinct from Δ40p53α, the binding site (aa 38-58) for human single-stranded DNA-binding protein RPA [9,48] and parts of the DBD enabling specific recognition and exonucleolytic attack of three-way junctions [11,38]. Given the propensity of Δ133p53α and Δ160p53α to form aggregates [39], it is conceivable that these p53-isoforms nonproductively occupy DNA replication sites through the residual DBD and OD supported by the C-terminus recognizing DNA lesions [10,49] and by interacting with topoisomerase I [37,50] and PARylated PARP1 [51]. In support of the view that Δ133p53α and Δ160p53α interact with replication forks, we noticed an increase of PCNA foci without a concomitant rise in POLι-PCNA Fig. 8 Model for the impact of different p53 isoforms on POLι-dependent DDT. A Hetero-oligomerization or aggregation of p53α with alternative p53 isoforms, particularly with p53α + Δ133p53α or p53α + Δ160p53α, can antagonize p53α functions in the POLι-dependent DDT. B C-terminal p53 isoforms still interact with PCNA [20,41] but are compromised in binding to three-stranded DNA junctions like replication forks due to shortened OD [38]. Intact DBD, OD, and RPA-interaction sites [8], but compromised interactions with POLι and PCNA due to lack of the N-terminal end in Δ40p53α permit limited replication fork recognition only. Δ133p53α or Δ160p53α are compromised in DNA-binding due to the N-terminally truncated DBD and in RPA-, POLι and PCNA-binding due to the missing N-terminus. However, all N-terminal p53 isoforms retain the OD, enabling hetero-oligomerization with p53α and therefore dominant-negative interference with p53α functions in performing idling events in complex with POLι and PCNA, giving time for PCNA ubiquitination and FR events by HLTF and ZRANB3 [19]. Black arrows, DNA-binding; blue arrows, PCNA-binding; stippled arrow, compromised interaction. colocalizing foci. Additionally, Δ133p53α and Δ160p53α can no longer compete for RPA70 subunit-binding with other potentially harmful proteins such as the nuclease MRE11, known to act on stalled and deprotected forks [19,[52][53][54][55].
Altogether, p53β, p53γ and Δ40p53α display residual performance in the p53-POLι-dependent DDT-pathway, as at least one of the underlying biochemical activities are retained each (Fig. 8). However, Δ133p53α and Δ160p53α, with full proficiency to form tetramers, but devoid of regions providing specificity for this DDTpathway cause non-specific deceleration of DNA replication by RFS.
Alternative p53-isoforms exert an inhibitory effect in the p53α-POLι-dependent DDT-pathway, which impacts on tumor suppressor functions Mutant p53 proteins were described to experience loss-offunction, acquire gain-of-function and interfere dominantnegatively with canonical and non-canonical wild-type p53 functions through hetero-oligomerization and aggregation [56][57][58][59][60]. Squelching out of factors required for wild-type p53 function represents even one further dominant-negative mechanism. In this work we found three pieces of evidence indicating inhibitory effects of alternative p53-isoforms. First, all alternative isoforms abrogated DNA replication slow-down by p53α under unperturbed conditions. Second, Δ133p53α and Δ160p53α elongated DNA tracks after ectopic co-expression with p53α in tumor cells.
A dominant-negative effect of ΔN-isoforms with intact OD could be explained by the formation of mixed hetero-oligomers with less specific DNA-binding properties [32], as the N-terminal TAD of p53 is known to directly interact with the DBD promoting specific DNA interactions [61]. Having shown that Δ133p53α and Δ160p53α cause RFS already on their own, we propose that these ΔN-isoforms additionally squelch out C-terminal p53α-interaction partners.
Accumulating evidence has shown that an increased ratio of ΔN-isoforms/p53α is associated with poorer overall survival or cancer aggressiveness, particularly for Δ133p53 in esophageal squamous cell carcinoma [62], Cholangiocarcinoma [63], and prostate cancer [64]. Here, we showed, that alternative p53isoforms, Δ133p53α and Δ160p53α, in particular, interfere with coexpressed p53α in DDT. Since homology-directed DDT-pathways, involving HLTF-and ZRANB3-mediated fork reversal, protect the genome from unrestrained and mutagenic bypass mechanisms like TLS [65,66], we anticipate that these isoforms increase genomic instability. Besides, the same ΔN-isoforms stimulate proliferation and inhibit apoptosis [15,23], all-in-all creating a dangerous combination of pro-mutagenic and pro-survival features, known to drive carcinogenesis. Intriguingly, p53α evolved from N-terminally truncated Δ40p53α [15]. Hence, p53 gained the N-terminus which is critical for its canonical transcriptional activities but also for its non-canonical function in regulating replication dynamics by interacting with PCNA and POLι [20]. Therefore, our observations may serve to better understand the role of p53-isoforms in cancer and stem cells regarding the maintenance of human genome stability. washed and re-incubated with fresh media for additional 3 h. Cell cultures were free from mycoplasma contamination as verified by PCR. HSPCs were isolated from cord blood samples and cultivated as described in [22,67].

DNA fiber spreading assay
To measure DNA replication track lengths for assessment of replication speed the DNA fiber spreading technique was performed as detailed in [20]. CldU (5-Chloro-2-deoxyuridine, Sigma-Aldrich by Merck, Darmstadt, Germany) was incorporated for 20 min. After washing, IdU (5-Iodo-2deoxyuridine, Sigma-Aldrich by Merck, Darmstadt, Germany) was incorporated in ten times higher concentration for another 20 min. During the IdU pulse, cells were either treated with MMC (3 µM, Sigma-Aldrich by Merck, Darmstadt, Germany) or its solvent H 2 O. Then, cells were washed, harvested, and resuspended in PBS. 2500 cells were put on a slide, lysed with 6 μl of 0.5% SDS, 200 mM Tris-HCl, pH7.4, 50 mM EDTA, and incubated at room temperature for 6 min. Afterwards, they were tilted about 20°to allow DNA-spreading via gravity. Subsequently, slides were covered with aluminum foil, air-dried for 6 min, fixed for 5 min with newly prepared 3:1 methanol:acetic acid, air-dried for 7 min, and stored in 70% ethanol at 4°C. Twenty-four hours later, the slides were processed for denaturation/deproteination in 2.5 N HCl for 1 h, followed by immunofluorescence staining and microscopy. To determine the speed of ongoing replication, the track lengths of bi-colored forks (CldU: green and IdU: red) were measured in the microscopic images using ImageJ. These track lengths in μm obtained during a 20 min pulse each reflect DNA replication fork speed, which can be calculated with the formula: DNA replication fork speed (kb/min) = 2.59 (kb/μm) × track length (μm) ÷ pulse time (min) [20]. For clarity lengths of the red IdU tracks are shown only. For fork asymmetry analysis, track lengths of IdU-incorporation (red) departing from the same origin (CldU, green) were measured and the respective fork asymmetries show the ratio of longer tracks versus shorter track lengths.
Western blotting for expression analysis without immunoprecipitation followed the protocols described in [19]. Conditions of protein expression were based on preceding titration experiments to ensure comparable p53 levels.

Cell cycle and apoptosis analysis
Cell cycle distribution and apoptosis in K562 cells were analyzed as described in [19].

Statistics
Graphic presentation of data, statistical analysis, calculation of mean values, standard deviation, and standard error of the mean were performed with GraphPadPrism8.4 software (San Diego, CA). For calculation of statistically significant differences in recombination measurements, the Kruskal−Wallis test followed by Mann−Whitney two-tailed test was used. For calculation of statistically significant differences in DNA fiber spreading analysis and immunofluorescence experiments Dunns-multiple comparison test was used. For calculation of statistically significant differences in Western Blot analysis, ANOVA followed by paired t-test or Friedmann test followed by Wilcoxon-matched pair signed-rank test was used.

DATA AVAILABILITY
The datasets generated during and/or analyzed during the current study are available from the corresponding authors upon request.