Distinct mechanisms control genome recognition by p53 at its target genes linked to different cell f

The tumor suppressor p53 integrates stress response pathways by selectively engaging one of several potential transcriptomes, thereby triggering cell fate decisions (e.g., cell cycle arrest, apoptosis). Foundational to this process is the binding of tetrameric p53 to 20-bp response elements (REs) in the genome (RRRCWWGYYYN0-13RRRCWWGYYY). In general, REs at cell cycle arrest targets (e.g. p21) are of higher affinity than those at apoptosis targets (e.g., BAX). However, the RE sequence code underlying selectivity remains undeciphered. Here, we identify molecular mechanisms mediating p53 binding to high- and low-affinity REs by showing that key determinants of the code are embedded in the DNA shape. We further demonstrate that differences in minor/major groove widths, encoded by G/C or A/T bp content at positions 3, 8, 13, and 18 in the RE, determine distinct p53 DNA-binding modes by inducing different Arg248 and Lys120 conformations and interactions. The predictive capacity of this code was confirmed in vivo using genome editing at the BAX RE to interconvert the DNA-binding modes, transcription pattern, and cell fate outcome.

In response to different forms of stress, the tumor suppressor p53 selectively induces one of a variety of distinct transcriptional programs, each linked to a unique biological outcome (e.g., cell cycle arrest, apoptosis, DNA repair, ferroptosis)1,2. The central event in this process is the binding of tetrameric p53 to regulatory DNA motifs in the genome (i.e., response elements) of the consensus sequence RRRCWWGYYYN0–13RRRCWWGYYY (R = A or G, W = A or T, Y = C or T, N = any base)3,4. Focusing on this initial step, the prevailing model holds that p53 exhibits dose-dependent target gene activation, with REs linked to cell cycle arrest targets exhibiting higher affinity and REs linked to pro-apoptotic targets exhibiting lower affinity5,6,7. This has provided a conceptually satisfying model in which cells with modest DNA damage induce low levels of p53 that are capable of binding only at the high-affinity REs, thus providing the cell with the opportunity to repair its genome prior to advancing through the cell cycle. However, if levels of DNA damage exceed the cell’s repair capacity, sufficiently high quantities of p53 are induced and even the low-affinity REs linked to pro-apoptotic target genes are occupied and activated.

It has been previously proposed that the number of mismatched bps within a given RE determines its binding affinity8,9. However, REs with equal numbers of mismatches still show different affinities for p53, suggesting that mechanisms beyond the degree of a mismatch from the consensus are responsible. This is of physiological relevance because only 10% of 2183 in vivo verified p53 REs match the consensus sequence across all 20 bps10. p53 REs exhibit the highest sequence variability at the triplets RRR/YYY, which flank the core (CWWG) sequence11, and p53 tetramers interact with REs at those bases via Arg248 from the DNA minor groove and Lys120 from the DNA major groove12. Of note, different crystal structures and molecular dynamics simulations of p53 bound to DNA reveal that the Arg248 and especially Lys120 side chains are highly dynamic and can adopt multiple conformations when in complex with DNA12,13,14,15. This set of observations triggered our hypothesis that the interplay of Lys120 and Arg248 residues with variable sequences of REs might be a potential source of p53 selectivity.

The role of Lys120 in regulating p53 target gene selectivity has been previously explored. Acetylation of this residue positively correlates with the increase in the expression of pro-apoptotic genes16,17, while a conservative, cancer-derived mutation of Lys120 to Arg (K120R) selectively impairs the induction of apoptosis and binding of p53 to pro-apoptotic REs without impacting induction of cell cycle arrest18. However, the molecular mechanism responsible for p53 dose-dependent and Lys120-dependent target gene selectivity has not been identified. In our study, we describe a mechanism that explains the differential binding of p53 to two major sets of genomic targets—high-affinity and low-affinity REs, linked to cell cycle arrest and pro-apoptotic genes, respectively. Specifically, we show that RE minor groove width, which is encoded by G/C or A/T bp content at positions 3, 8, 13, and 18 in the RE and sensed by Arg248, determines two distinct p53 DNA-binding modes. Our findings advance a p53 RE code that mechanistically links RE sequence, DNA shape, DNA-binding mode, and biological output.

An RE code distinguishes p53 effector pathways

To gain insight into the role of Lys120 integrity in the recognition of specific RE sequences, we performed a chromatin-immunoprecipitation experiment and analyzed genome-wide occupancy of conditionally-expressed wild-type (WT) and K120R forms of p53 in the isogenic human H1299 lung cancer cell line that is null for endogenous p53 (Fig. 1a). In our ChIP-seq data we identified 1358 unique genomic-binding sites for WT p53 that were associated with previously identified, 20-bp REs (Fig. 1b)10. Of these 1358 REs, only 622 (45.8%) retained p53 binding when Lys120 was converted to arginine (termed “K120R-bound”). The remaining 736 (54.2%) sites lost p53 binding upon conversion of Lys120 to arginine (termed “K120R-unbound”) and exhibited generally lower WT p53 occupancy (Supplementary Fig. 1a, b). In agreement with previous biological observations18, the vast majority of REs at genes linked to the positive regulation of apoptosis were K120R-unbound in our studies. In contrast, REs at genes crucial for activation of processes such as cell cycle arrest or DNA repair were still bound by the K120R p53 variant (Fig. 1c and Supplementary Fig. 1c).

Further analysis of the nucleotide sequence of REs present within our ChIP-seq peaks revealed that the generic number of mismatches from the consensus RE did not directly correlate with in vivo occupancy by WT p53 or the dependence on Lys120 integrity (Supplementary Fig. 2). We therefore postulated that sequence identity at specific positions within the RE contributes to these two, presumably linked characteristics. To interrogate this hypothesis, consensus sequences were generated for the K120R-bound and K120R-unbound REs in our datasets. This analysis demonstrated that the major difference in the sequence of these two groups of REs resides at positions adjacent to the core (CWWG) sequence (Fig. 1d). More specifically, it demonstrated a preponderance of A/T bps at positions 3, 8, 13, and 18 in the K120R-bound group and G/C bps in the K120R-unbound group of REs. Additional analysis of the nucleotide content specifically at those positions confirmed this observation by showing that ~80% of the sites enriched with G/C bps were deficient in K120R-p53 binding (Fig. 1e).

As the role of Lys120 integrity in distinguishing between REs is linked to distinct cell fates18, we assessed whether the G/C vs. A/T bp content at positions 3, 8, 13, and 18 could similarly distinguish targets linked to distinct cell fates. We analyzed REs from our ChIP-seq data, which were linked to the genes previously identified as high-confidence p53 targets through the comprehensive meta-analysis of 319 previously published gene studies19. Remarkably, analysis of the REs connected to 16 pro-apoptotic targets revealed enrichment of G/C bps at these positions, similar to the sequence developed in Fig. 1d for the entire K120R-unbound group of REs (Fig. 2a,   b, bottom). An analysis of REs of sixteen, high-confidence, pro-survival targets revealed a consensus sequence with predominantly A/T bps at these positions, similar to the sequence developed in Fig. 1d for all K120R-bound REs (Fig. 2a, b, top). Additionally, REs linked to pro-apoptotic targets averaged more mismatches at the core (CWWG) sequence, relative to pro-survival REs. However, further testing showed that these mismatches only marginally decrease the p53 occupancy in vivo (Supplementary Fig. 3a), and do not regulate K120-dependency (Supplementary Fig. 3b). Interestingly, the G/C vs. A/T bp content at positions directly adjacent to the CWWG core still distinguished K120R-bound and K120R-unbound REs, independent of the status of the CWWG sequence itself (Supplementary Fig. 3c). Collectively, these findings suggest that differential recognition by p53 of distinct classes of REs is encoded, at least in part, by the A/T vs. G/C bp content at positions 3, 8, 13, and 18 within the primary DNA sequence.

p53 uses two distinct RE-binding modes in vitro

To quantify p53 binding to K120R-bound and K120R-unbound REs in vitro, we measured the dissociation constants (Kd) of natural and artificial REs, employing fluorescence polarization at 275 mM NaCl, which would best reflect in vivo conditions and which rendered non-specific DNA binding undetectable (Table 1 and Supplementary Fig. 4)14,20,21. For these experiments, we used a thermostable recombinant p53 protein lacking the unstructured N-terminal transactivation and proline-rich regions. This ΔN-p53 construct (residues 94-393) has previously been demonstrated to retain the same DNA-binding specificity as a full-length p53 protein14,22. Under the salt conditions described above, WT p53 exhibited high affinity for K120R-bound REs (23–97 nM), but one order of magnitude lower affinity for K120R-unbound REs (254–783 nM). This supports previous reports of higher affinity for cell cycle arrest REs such as p21 RE and lower affinity for pro-apoptotic REs such as BAX RE6,21. Conversion of Lys120 to Arg had relatively little impact on binding affinity for K120R-bound REs (43–80 nM), while significantly reducing binding affinity for K120R-unbound REs (800 nM to non-detectable). Together, these in vitro data recapitulate the K120R binding selectivity observed in vivo by ChIP-seq (Fig. 1b), and elsewhere18.

To gain further insight into the nature of K120R-bound and K120R-unbound-binding modes, we determined the Kd values at a lower salt concentration (200 mM NaCl). Under this condition, binding affinity for K120R-unbound REs was increased, when compared to the higher salt concentration, for both WT p53 (8–45 nM) and the K120R mutant (23–280 nM), suggesting an electrostatically-driven interaction. By contrast, recognition of K120R-bound REs increased only slightly at lower ionic strength for both WT (6–26 nM) and K120R (33–37 nM) proteins, consistent with predominantly non-polar interactions. We also analyzed an ideal, G/C-rich, presumably K120R-unbound RE termed “Pro-apoptotic RE” because the natural REs that we used in our binding assay (BAX RE and NOXA RE) contained mismatches with respect to the p53 consensus sequence, including the core (CWWG) sequence (indicated by lower-case letters in Table 1). The ideal RE confirmed the observed characteristics of the natural K120R-unbound REs, indicating that the mismatches had no effect on the binding mode or K120R sensitivity. Altogether, these in vitro binding data suggest that p53 employs a high-affinity, non-electrostatic effective binding mode for K120R-bound REs, while utilizing a lower affinity, electrostatic effective binding mode for K120R-unbound REs.

p53 uses structurally distinct DNA-binding modes for different classes of REs

To investigate whether the two distinct p53-binding modes observed in vitro have correlates on the atomic level, we analyzed p53-DNA crystal structures available in the Protein Data Bank (PDB) and classified them into two groups based on Lys120 being ordered or disordered, as judged by an omit electron density map (Supplementary Table 1). In the first group, represented by a p53 structure complexed with a BAX-like RE (PDB: 6FJ523, Fig. 3a and Supplementary Fig. 5), Lys120 protrudes into the major groove in an extended conformation and forms direct hydrogen bonds with the purine bases R12 and R13. On the opposite side of the double helix (minor groove), Arg248 adopts a bent conformation in which its guanidinium group engages in a direct or water-mediated electrostatic interaction with the phosphate(s) of Y8 and/or Y9, which base-pair with R12 and R13 (Fig. 3a and Supplementary Fig. 5). In the second group represented by a p53 structure complexed with a p21-like RE (PDB: 3Q0514, Fig. 3b and Supplementary Fig. 6), the Lys120 side chain is disordered, which is a result of its Cα atom within the p53 L1 loop being farther away from the DNA (outside of the major groove), precluding the Lys120 Nζ atom from hydrogen bonding with the purine bases. Arg248 is extended and penetrates into the minor groove undergoing primarily hydrophobic interactions14,22. We confirmed that the DNA sequences of the two representative PDB structures, which only differ at 4 bps, recapitulated the characteristics of the two distinct binding modes observed in vitro (Table 1). Thus, p53 indeed adopts structurally distinct binding modes at high-affinity, K120R-bound and low-affinity, K120R-unbound REs.

Although distinct Lys120 and Arg248 conformations have previously been reported in complex with different REs12,15, the current findings provide a mechanistic understanding of the roles played by these two key DNA contact residues in providing specificity to genome occupancy by p53. For example, the conversion of Lys120 to arginine exclusively affects the binding affinity to the low-affinity, K120R-unbound REs, since the arginine side chain causes a steric clash with these REs (Fig. 3c). Our analysis also provides a structural basis for the different salt sensitivities observed in vitro (Table 1), as the salt-dependent interaction at low-affinity, K120R-unbound REs can be ascribed to the electrostatic interactions of the four Arg248 residues in tetrameric p53 with the DNA backbone phosphates, while these arginine residues convert to predominantly non-polar, hydrophobic interactions in the salt-insensitive binding mode at high-affinity, K120R-bound REs.

Minor groove width is sensed by Arg248 and determines the p53 RE-binding mode

Comparison of the DNA sequences in the structurally distinct p53-DNA complexes (Supplementary Table 1) revealed a preponderance of A/T bps in the flanking RRR/YYY regions in the group featuring a disordered Lys120 side chain. Since A/T bps can induce a narrow minor groove and a wider major groove in B-form DNA24,25, we examined the shape of the DNA in the representative p53-DNA structures. We found that the width of the major and minor grooves in the regions adjacent to the CWWG core motif—the sites where Arg248 interacts with DNA— are indeed altered in these two structures (Fig. 3d). In A/T-rich, p21-like, K120R-bound RE, the minor groove is narrow and recognized by the extended Arg248 side chain in the primarily hydrophobic mode (Fig. 3b and Supplementary Fig. 6). In G/C-rich, BAX-like, K120R-unbound RE, the minor groove is wider and sensed by Arg248 in a bent conformation, interacting electrostatically with DNA (Fig. 3a and Supplementary Fig. 5). Thus, only two bps adjacent to the CWWG motifs (positions 8 and 18, Fig. 3d) suffice to switch the p53-binding mode in these examples, regardless of the base composition of the CWWG core motif (Supplementary Fig. 7)11,26. Our hypothesis that the Arg248 side chain conformation is influenced by minor groove width is confirmed by a double mutation in p53 (R273H/T284R) that widens the minor groove width, resulting in a change of the Arg248 side chain conformation (Supplementary Fig. 8)27,28. These findings support the conclusion that minor groove width at bps 3, 8, 13, and 18 in p53 REs is critical in determining different DNA-binding modes because it induces distinct Arg248 conformations and interactions.

Rewriting the RE code reprograms p53 function

Our model predicts that single or dual bp conversions at positions 3, 8, 13, and/or 18 will impact the minor groove width of an RE, alter interactions of Arg248 and Lys120, and ultimately dictate the binding mode utilized by p53. To rigorously test this model, targeted G/C to A/T, along with A/T to G/C bp conversions were introduced in specific REs in vitro and in vivo. Using the in vitro DNA-binding assay, we measured Kd values of the modified BAX_8/13 RE that carried G/C to A/T conversions at positions 8 and 13 in a natural BAX RE (Fig. 4a). Consistent with the model, these conversions increased the affinity of WT p53 for this RE (from 460 nM to 40 nM) and restored K120R-p53 binding. An increase in WT and K120R-p53 binding was similarly observed for the NOXA_3/13 RE, which contained G/C to A/T conversions at positions 3 and 13 (Supplementary Fig. 9a). We also measured Kd values of the p21_8/13 RE, which carried reciprocal A/T to G/C conversions, at positions 8 and 13. As expected, these conversions decreased the affinity of the WT p53 (from 23 nM to 99 nM) and more markedly K120R-p53 (from 43 to 479 nM) for this RE (Supplementary Fig. 9b). Next, we experimentally manipulated the minor groove width of the RE in vitro by introducing G/C to inosine-cytosine (I/C) conversions, which narrows the minor groove without impacting major groove width29. These changes are predicted to impact the interaction of Arg248 with DNA by increasing the non-polar nature of the binding while having no impact on K120R sensitivity. Supporting this scenario, WT p53 exhibited an increased affinity for the BAX_8/13 I/C RE (from 460 nM to 83 nM), but K120R-p53 binding remained undetectable (Fig. 4b). Increased WT affinity and retention of K120R binding was also observed after introducing G/C to I/C substitutions at the analogous positions in the artificial 6FJ5 RE (Supplementary Fig. 9c).

To assess whether the RE code could be predictably manipulated in vivo, we examined whether the in vitro binding data correlated with p53-binding behavior in human cells. Initially, this strategy relied on a panel of luciferase reporters driven by specific variants of the p53 RE. Natural and artificial high-affinity and A/T-rich REs (p21 and 3Q05) exhibited full activation by K120R-p53, whereas low-affinity and G/C-rich REs (BAX and 6FJ5) were significantly defective for activation by K120R-p53 (Fig. 4c), in agreement with in vitro results (Table 1). Notably, converting the BAX RE positions 8 and 13 from G/C to A/T or fully converting positions 8 through 13 to resemble the p21 RE (from CCCGGG to TCCCAA, “BAX_4bp”), fully restored K120R-p53 responsiveness. In contrast, A/T to G/C conversions in the p21_8/13 RE decreased activation by K120R-p53, again consistent with the dual-binding mode model proposed here (Fig. 4d).

Editing the RE code interconverts p53-driven cell fates

As the p53 RE consensus sequence is extremely degenerate, there are few examples in the human genome of active REs, which differ only by a single nucleotide at positions 3, 8, 13, and/or 18. To test our model in the context of native chromatin, this limited number of examples were examined for p53 occupancy. In one of these examples, relying on a comparison of the validated but uncharacterized p53 target genes APBB2 and SSUH2, a dual nucleotide conversion at positions 8 and 13 from A/T bps, in the RE linked to APBB2 gene, to G/C bps, in the RE linked to SSUH2 gene, strongly decreased WT p53 binding and completely diminished binding of K120R-p53 protein, again supporting the model (Supplementary Fig. 10).

To formally test the proposed p53 RE code in the context of native chromatin, we utilized CRISPR/Cas9 to edit the sequence of the endogenous BAX RE in human cells. We edited the wild-type BAX RE to increase its A/T content, by converting positions 8, 11, 12, and 13, from CCCGGG to TCCCAA (“BAX_4bp”, identical to the sequence used in the Fig. 4d). Our model predicts that this edited BAX_4bp RE should be recognized by WT p53 through a non-polar, K120R-bound binding mode. Remarkably, when compared to a G/C-rich WT BAX RE, WT p53 exhibited higher occupancy at the edited BAX_4bp RE locus and K120R-p53 binding was completely restored (Fig. 5a and Supplementary Fig. 11a). Furthermore, the restoration of K120R-p53 binding observed by ChIP analysis was of functional relevance as it resulted in rescued BAX mRNA transcription and protein expression, in both WT and K120R-p53-expressing cells (Fig. 5b, c and Supplementary Fig. 11b). Finally, targeted conversions of the BAX RE within native chromatin resulted in a restoration of the pro-apoptotic functions of the cancer-derived K120R-p53 mutant, under conditions where unedited cells undergo cell cycle arrest (Fig. 5d, e)18. This data strongly supports the model in vivo and highlights DNA-binding as a crucial step in p53’s ability to selectively activate distinct effector functions. In summary, by introducing this targeted 4-bp change in a BAX RE and converting this binding site from lower affinity, K120R-unbound to higher affinity, K120R-bound, p53-dependent cell fate outcomes were transformed.

Cell fate decisions are frequently regulated by the presence or absence of key transcription factors that control gene expression programs ultimately responsible for the cell phenotype. In some cases, a critical transcription factor can “decide” between multiple cell fate options, depending on the cellular context. For example, in response to different developmental cues, individual members of the SOX and WNT/β-catenin families of transcription factors can drive cell differentiation along several distinct lineage pathways30,31. Similarly, the oncogenic transcription factor MYC can trigger either cell cycle progression or cell death, also in a context-dependent manner32. Like MYC, the tumor suppressor p53 can trigger multiple cellular outcomes, often depending on the form of stress that initially drove p53 activation4. The ability of a single, sequence-specific transcription factor to “choose” to activate only one of several of its gene expression programs raises a number of biochemical questions. Central among these is how recognition of genomic targets linked to different effector functions can be discriminated by a conserved DNA-binding domain. Here, we report a new mechanism of DNA-binding specificity of p53 protein, by showing that a nucleotide code at positions 3, 8, 13, and 18 in the p53 RE creates a DNA shape recognition difference between distinct categories of genomic targets, with these differences being sensed by two key DNA contact residues—Lys120 and Arg248. The code our findings establish is predictive and, when manipulated in vivo, can be utilized to interconvert p53 target gene transcription patterns and cell fate.

As previously mentioned, p53 RE is unusually long and tolerates a substantially higher level of degeneracy than what is typically observed for a sequence-specific DNA-binding protein. Out of twenty nucleotides, G/C bps at positions 4, 7, 14, and 17 (RRRCWWGYYYRRRCWWGYYY) are the only invariant positions in the p53 RE. Guanines at those positions are directly recognized by four Arg280 residues, and this base-specific readout is a well-defined mechanism by which tetrameric p53 recognizes specific genomic targets33,34,35. The number of mismatched bps within a RE, in particular within a core (CWWG) sequence, generally lowers the affinity of individual REs11. However, identical numbers of mismatches to the consensus sequence, including mismatches in the crucial G/C bps at positions 4, 7, 14, and 17, are present in both high- and low-affinity-binding sites. This finding suggests that the degree of a mismatch from the consensus RE fails to fully explain the differential binding of p53 to high- and low-affinity REs and that a mechanism beyond base-specific recognition must account for the exquisite ability of p53 to distinguish between different targets.

With base-specific recognition ruled out as the major determinant of RE recognition by p53, other biophysical mechanisms must be considered. Relevant to the findings reported here, minor groove shape recognition is a commonly used mechanism in protein-DNA interactions, utilized by a variety of DNA-binding proteins24,36. In particular, recognition of a narrow minor groove by arginine residues has a thermodynamically beneficial impact on protein-DNA interactions, by increasing binding affinity24,37. However, differential arginine-minor groove interactions that modulate DNA binding without base-specific recognition, and a DNA-binding protein that uses two binding modes through distinct minor groove interactions, have not been previously described to our knowledge. Our findings identify two distinct molecular mechanisms used by p53 for differential recognition of its genomic targets as a direct consequence of that phenomenon. We find that minor groove width, encoded by the presence of G/C or A/T bps at positions 3, 8, 13, and 18 in the RE, determines the two distinct p53 DNA-binding modes, which are in turn linked to different cell fates.

The RE code proposed here explains initial DNA recognition by p53. However, other factors, such as chromatin structure or central insertions in the RE, may impact or even override this code. For example, discontinuous REs with inserts between two half-sites generally have a lower affinity, independent of their DNA sequence38. Furthermore, some genomic sites containing a consensus RE sequence, do not exhibit p53 binding in vivo, perhaps due to a suppressive effect of the aforementioned chromatin landscape9. These exceptions provide an important caveat to our model. However, for the vast majority of established p53 targets10,19, RE recognition is regulated primarily by DNA sequence, and for these genomic targets, selective DNA-binding, as the first, fundamental step in the activation of downstream p53 targets, can be largely explained via the use of the two distinct DNA-binding modes described in our model.

These two binding modes constitute an additional layer of gene regulation by the p53 protein to create, via Arg248, qualitatively (predominantly hydrophobic vs. predominantly electrostatic) and quantitatively (high-affinity vs. low-affinity) different interactions with distinct classes of REs. This unique role of Arg248 in target gene selectivity builds on its already established and crucial role in p53 biology, most evident from Arg248 being the most frequently mutated residue in human cancer39. The observation reported here regarding the ability of Lys120 to sense differences in the RE shape relates to two previous findings of the role this residue plays in DNA recognition and cell fate. First, the Halazonetis group has reported structural findings that the L1 loop, which contains Lys120, can adopt two different conformations when complexed with DNA14,22. These conformations are referred to as extended and recessed, reflecting the position of the L1 loop relative to the DNA major groove. Similarly, Ethayathulla et al.40 have reported that Lys138, which is the Lys120 analog in the L1 loop of the p53 family member p73, displays different conformations at different REs. The combination of biophysical, genomic, and biological evidence reported here has established a DNA code and p53 recognition mechanisms that unifies these observations. Second, we and others have shown that the Lys120 residue is post-translationally modified by acetylation, whose levels correlate with apoptosis but not cell cycle arrest16,17. Ultimately, the acetylation of Lys120 may be a mechanism by which DNA recognition by the p53 is stabilized in one mode or the other and future studies should elucidate its potential role.

In summary, our discovery of two distinct binding modes as a means of conferring p53 RE selectivity through differential recognition of minor groove shape offers an explanation for a major unresolved mystery in p53 biology—the mechanism by which p53 achieves dose-dependent RE recognition and differential target gene activation. Since the existence of four p53 monomers in a complex with four RE quarter-sites allows for a wide range of p53-RE interactions (from strongly polar to strongly non-polar), the two biochemically distinct binding mechanisms identified here allow p53 to impose a subtle “fine-tuning” of its DNA recognition. Additionally, our model describes DNA recognition by p53 via two distinct binding modes as initial and foundational step in p53-dependent activation of downstream targets. Other factors, such as the specific stress signaling pathway responsible for p53 activation, post-translational modification of residues beyond Lys120 in the p53 DNA binding domain or elsewhere in the protein, differential co-factor binding, or pulsatile expression dynamics, are likely to play a role in modulating selective target gene activation by p53 and enacting an ultimate cell fate outcome4. Having established a paradigm for a single transcription factor using different DNA-binding modes determined by different arginine-minor groove interactions to selectively effect distinct downstream pathways, future studies may identify other DNA-binding proteins involved in cell fate decisions, that use similar mechanisms to gain target gene selectivity.