Paradoxical gene regulation explained by competition for genomic sites

Paradoxical gene regulation explained by competition for genomic sites | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Paradoxical gene regulation explained by competition for genomic sites Dhruv D. Jatkar , View ORCID Profile Krishna Manoj Aravind , View ORCID Profile Eduardo D. Sontag , View ORCID Profile Domitilla Del Vecchio doi: https://doi.org/10.1101/2025.11.27.691022 Dhruv D. Jatkar 1 Northeastern University , Boston, MA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Krishna Manoj Aravind 2 Massachusetts Institute of Technology , Cambridge, MA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Krishna Manoj Aravind Eduardo D. Sontag 1 Northeastern University , Boston, MA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Eduardo D. Sontag For correspondence: e.sontag{at}northeastern.edu ddv{at}mit.edu Domitilla Del Vecchio 2 Massachusetts Institute of Technology , Cambridge, MA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Domitilla Del Vecchio For correspondence: e.sontag{at}northeastern.edu ddv{at}mit.edu Abstract Full Text Info/History Metrics Preview PDF ABSTRACT Understanding how opposing regulatory factors shape gene expression is essential for interpreting complex biological systems. A motivating observation, drawn from cancer epigenetics, is that removing an activating factor can sometimes lead to higher, not lower, expression of a gene that is also subject to repression. This counterintuitive behavior suggests that competition between activators and repressors for limited genomic binding sites may produce unexpected transcriptional outcomes. Prior theoretical work proposed this mechanism, but it has been difficult to test directly in natural systems, where layers of chromatin regulation obscure causal relationships. This paper introduces a fully synthetic, tunable genetic platform in a prokaryotic model system that isolates this competition mechanism in a clean and interpretable setting. The engineered construct contains a target gene with binding sites for both an activator and a repressor, together with a separate decoy region that carries overlapping binding sites for the same regulators. Activator and repressor functions are implemented using CRISPRa and CRISPRi, which permit independent control of regulator expression levels and binding affinities. Using this minimal system, the paper shows that increasing activator expression can reduce expression of the target gene when both regulators are present, consistent with the prediction that additional activator molecules displace the repressor from decoy sites and allow it to more effectively repress the target. By demonstrating how competition alone can invert expected regulatory responses, this synthetic framework provides a validated model for understanding similar paradoxical behaviors in natural regulatory networks and establishes a foundation for future studies in more complex mammalian contexts. Introduction Metastasis is the primary cause of cancer mortality, and its onset is strongly influenced by the epithelial mesenchymal transition (EMT), which equips carcinoma cells with migratory, invasive, and drug-resistant traits. Transcription factors such as ZEB1 function both as key effectors of EMT and as sensitive readouts reflecting how upstream signaling and regulatory circuits reshape cellular state. Although EMT is often framed primarily in terms of transcription factors, it is increasingly evident that chromatin-modifying complexes also play essential roles. PRC1 and PRC2 deposit the repressive histone mark H3K27me3 and help maintain epithelial identity by silencing genes such as ZEB1, ZEB2 , and TWIST1 . In contrast, Trithorax and COMPASS complexes, including KMT2D/MLL4, deposit H3K4me1 and activate lineage-specific enhancers. Genome-wide CRISPR screens show that both PRC2 and KMT2D act as “epithelial guardians,” and that removing either complex can unlock EMT subcircuits and produce unexpected changes in the expression of EMT transcription factors. Notably, perturbations sometimes lead to higher, not lower, levels of ZEB1, ZEB2, TWIST1 , and other EMT-related genes when activators are removed 1 , 2 . These observations raise a fundamental question about how competition for a limited pool of genomic binding sites can invert expected regulatory outcomes in complex mammalian regulation. Here, we investigate how such paradoxical responses can arise from competitive binding between an activator and a repressor. Prior theoretical work suggests that counterintuitive outcomes occur when regulators share a restricted set of binding sites and when decoy sites redistribute factors across the genome 2 . Classical transcriptional biophysics predicts that decoy sites can sequester regulators 3 , reshape input and output curves 4 , and even allow increased activator levels to lower the expression of a target gene. Directly testing these principles in an endogenous setting is difficult because many potential binding sites and feedback loops obscure causal mechanisms 5 . To address this challenge, we constructed a minimal synthetic genetic circuit in bacterial cells as a proof-of-concept system. The circuit uses CRISPRa and CRISPRi as synthetic activator and repressor molecules and enables independent control of binding affinities and competitive interactions. We developed a mechanistic reaction network model with activator and repressor molecules that regulate a target gene while also binding to decoy sites that represent other genomic locations. The model, based on our previous work 6 , predicts that, under specific constraints involving exclusive binding and decoy-site affinity, increasing the concentration of an activator can repress its target. We then build an experimental system satisfying these conditions using a CRISPR system in which overlapping binding sequences at decoy sites enforce competitive binding. By tuning the activator-target affinity through engineered mismatches in the binding sequence, we achieved the parameter regime that produces the paradoxical effect. The experiments validate the theoretical predictions and establish a minimal framework for understanding how simple molecular competition mechanisms could underlie the unexpected regulatory outcomes observed in EMT control and cancer progression. Results To probe the counterintuitive possibility that higher concentrations of activator A can suppress its own target, we formulated a reaction-network model incorporating competition for limiting resources and binding sites. The model includes activator ( A ) and repressor ( R ) molecules that transcriptionally regulate production of a protein encoded by a target gene. Sequestration of these regulators by the rest of the genome is represented through the introduction of decoy binding sites. Within this framework, we outline a possible mechanism for paradoxical repression by an activator ( Figure 1a ). Download figure Open in new tab Figure 1. Competition and retroactivity lead to repression by an activator. (a) Schematic diagram showing the proposed mechanism behind the paradoxical effect, where increasing activator leads to target repression. (b) Numerical simulations showing the paradoxical behavior where the target expression levels decrease as the activator is increased. (c) Schematic diagram after removing the overlapping binding sites at the decoy, (d) Numerical simulations showing the removal of the paradoxical behavior where the target expression levels increases as the activator is increased. In the low-activator regime, most accessible sites for R are occupied, so R remains bound to both the target and the decoys, leading to low target expression. As the concentration of A increases, activator molecules increasingly bind to overlapping sites on the decoys. When A binds to a decoy site previously occupied by R , it displaces the repressor, releasing free R into the system. The liberated R can then bind to the target gene. Thus, at high activator levels, activators are sequestered by decoy sites, whereas the repressors displaced from the decoys bind the target and drive unintended repression. Consequently, the model’s input–output curve exhibits a paradoxical decrease in target expression as the activator concentration rises ( Figure 1b ). However, the paradoxical effect is not observed in the absence of competitive binding at the decoy sites ( Figure 1c ). We tested this by designing orthogonal binding sites for activators and repressors on the decoys. In this configuration, increasing the level of A can no longer displace R from the decoy sites. As a result, the input–output curve shows the expected monotonic activation of target gene expression ( Figure 1d ). Thus, competitive binding of A and R to overlapping sites on the decoys is a key requirement that enables activator-dependent displacement of the repressor and gives rise to the paradoxical effect. Next, we carry out model-guided experiments in bacterial cells to test the theoretical predictions outlined in Figure 1 . The synthetic construct is implemented as shown in Figure 2a . The genetic circuit diagram encodes the following components. Download figure Open in new tab Figure 2. CRISPRa and CRISPRi allow combinatorial regulation of gene targets via competitive binding. (a) Genetic construct of the target gene (GFP) activated through CRISPRa and repressed through CRISPRi. The sRNA for CRISPRa is expressed from two sites and recruits constitutively expressed dCas9 and RBP-AD to activate GFP. The gRNA for CRISPRi recruits dCas9 to repress GFP. To decrease the affinity of the activator 3 mismatches (3 MM) were introduced in the target-activator binding site. More details about the effect of mismatches is provided in Figure 4 . (b) (Top) Schematic diagram of the CRISPRi complex consisting of dCas9 and gRNA binding near the promoter of the target physically hindering transcription. (Bottom) Schematic diagram of the CRISPRa complex consisting of dCas9, RBP-AD, and sRNA binding upstream of the transcription start site (TSS) of the target allowing activation. RBP-AD recruits RNAP activating the expression of the target gene. (c) Bar chart representation of the experimental data showing GFP levels for varying combinations of the sRNA and gRNA. The combinations are {sRNA,gRNA}: {+, − }, { −, − } and { − , +}. Here + sRNA refers to 100 nM aTc induction and the presence of the sRNA casettes with J119 constitutive promoter, whereas + gRNA corresponds to 10 uM IPTG induction. − sRNA correspond to the complete absence of both sRNA cassettes, while − gRNA correspond to the complete removal of the gRNA cassette. (d) Genetic construct with RFP placed downstream of the decoy sites activated by sRNA and repressed by gRNA. (e) Schematic diagram showing the implementation of overlapping binding sites at the decoy to obtain competitive binding. (f) Bar chart representation of the experimental data showing RFP levels for varying combinations of the sRNA and gRNA. The combinations are {sRNA,gRNA}: {+, − }, { −, − } and { − , +}. Here + sRNA refers to 100 nM aTc induction, + gRNA corresponds to 10 uM IPTG induction and − sRNA or gRNA correspond to the complete absence of sRNA or gRNA cassettes, respectively. (g) Input output response of RFP with varying sRNA (through aTc induction) in the presence of constant repressor (through 10 uM IPTG induction). The dashed lines is the extension of the respective values from (f). In (a) and (d), all plasmids are introduced into the Marionette strain of E. coli 7 , where the corresponding regulators are endogenously expressed. We use CRISPR-based gene regulation, CRISPR interference (CRISPRi) 8 and CRISPR activation (CRISPRa) 9 , to function as the repressor and activator, respectively ( Figure 2b ). During CRISPRi, the catalytically inactive Cas9 (dCas9) protein binds to a guide RNA (gRNA) forming the dCas9/gRNA complex. This complex then binds to the target DNA sequence, physically obstructing the binding of RNA Polymerase (RNAP), thereby repressing target expression ( Figure 2b - top). In CRISPRa systems, an RNA-binding protein fused to an activation domain (RBP-AD) and dCas9 are both recruited to the DNA by a scaffold RNA (sRNA). Once bound, the RBP-AD interacts with the RNAP, enhancing the rate of transcriptional initiation ( Figure 2b - bottom). By employing these two systems simultaneously, the circuit allows for the independent control of activating and repressing signals. To test the efficiency of CRISPRi and CRISPRa on target expression, we implement the genetic circuit shown in Figure 2a . Here, the target gene encodes green fluorescent protein (GFP), and the gRNA is produced at a constant level. The sRNA is produced from two different cassettes to obtain stronger activation. We measure the activated (+ sRNA and - gRNA), basal (- sRNA and - gRNA) and repressed (- sRNA and + gRNA) expression of GFP. The GFP levels are higher than basal levels in the presence of the sRNA, whereas it is lower than basal in the presence of gRNA ( Figure 2c ). From Figure 1 , we identified competitive binding as a key requirement for achieving the paradoxical effect. To verify the competitive binding at the decoy sites, we employed a genetic circuit with red fluorescent protein (RFP) downstream of the decoy sites ( Figure 2d ). Mutually exclusive binding of gRNA and sRNA at the decoy sites is implemented by engineering their individual binding sequences to overlap ( Figure 2e ). To elaborate, the gRNA (guiding the repressor) and the sRNA (recruiting the activator) recognize specific 20 base pair (bp) sequences. To enforce competition, these binding sites are positioned such that they overlap by 10 bp. This significant overlap ensures that the binding of the CRISPRi complex physically precludes the binding of the CRISPRa complex, and vice-versa, thereby allowing the mutual exclusion required for the displacement mechanism. To verify the effectiveness of this methodology in ensuring competitive binding, we compare the repressed, activated, and basal states of RFP, both in isolation ( Figure 2f ) and when both gRNA and sRNA are present ( Figure 2g ). Similar to the target, we observe that the RFP levels are higher in the presence of sRNA alone (+ sRNA and - gRNA) and lower in the presence of gRNA alone (- sRNA and + gRNA) when compared to basal expression (- sRNA and - gRNA). Next, we implement concurrent sRNA and gRNA with 10 mM induction of IPTG while varying the levels of aTc ( Figure 2d ). The constant amount of gRNA brings RFP down towards the repressed level at low aTc (indicated by the lower dashed line in Figure 2g ). Then, as the amount of sRNA is progressively increased (titrated via the inducer aTc), RFP level increases crossing the basal expression level (indicated by the middle dashed line in Figure 2g ), ultimately reaching the maximally activated level (indicated by the upper dashed line in Figure 2g ). This recovery demonstrates that the sRNA is capable of out competing and displacing the gRNA from the designed decoy sites. Now that we have the individual model-driven requirements implemented and verified experimentally, we move on to exploring the paradoxical behavior of the system. Towards this, we design a construct that allows constant expression of the repressor, and overlapping binding sites at the decoy by combining the plasmids in Figure 2 (a) and (d) , and vary the activator levels ( Figure 3a ). We observe that as the activator is increased, GFP levels decrease showing an unintended target repression ( Figure 3b ). For low scRA the GFP values are slightly below 1 due to the presence of the repressor. As we increase sRNA, the GFP levels drop to approximately 0.6 (40% drop), exhibiting a paradoxical behavior. Hence, through model-driven experiments we verify our theoretical hypothesis for the emergence of the paradoxical effect. Download figure Open in new tab Figure 3. Paradoxical repression of the target by an activator due to competitive binding at decoy sites. (a) Genetic circuit implemented to observe the paradoxical behavior. The target and decoy binding sites are the same as in Figure 2 . Competitive binding is established at the decoy sites by overlapping binding sites as shown in Figure 2(e) . The amount of gRNA is kept constant with 10 uM IPTG induction. The sRNA is produced from two distinct casettes, one with an aTc inducible promoter and the other with a constitutive J119 promoter. Both plasmids are co-transformed into the Marionette strain of E. coli 7 , where the corresponding regulators are endogenously expressed. (b) Input output response of normalized GFP with increasing sRNA. The GFP value is normalized using the basal expression levels obtained in Figure 2(c) . The sRNA levels in increasing order correspond to varying aTc induction in the absence of the secondary constitutive sRNA cassette, followed by full aTc induction with the secondary constitutive sRNA cassette. Next, we explore the effect of binding affinity of the activator to the target, a model identified key parameter, on the paradoxical behavior of the system. The model predicts that the paradoxical effect is observed only when the dissociation constant between the CRISPRa complex and the target binding site (denoted as K ta ) is sufficiently large, i.e., when the activator–target affinity is relatively weak ( Figure 4a ). For low values of K ta , we observe activation of the target gene ( Figure 4a-1,2 ), whereas we observe the paradoxical repression by the activator ( Figure 4a-3,4 ) for high values of K ta , i.e. low binding affinity between the target and the activator. Download figure Open in new tab Figure 4. Paradoxical repression with increasing activator is observed for low target-activator affinity. (a) Schematic diagram and numerical simulations showing the target expression as the amount of activator is increased for different values of K ta . We observe activation for low K ta values and paradoxical repression for high K ta values. The genetic diagram shows the behavior for high activator levels. (b-Left) Genetic circuit diagram showing the implementation of varying K ta through the addition of mismatches at the activator binding site on the target. 0 MM correspond to no mismatches, such that sequence at the target is exactly complementary to the target binding sequence on the sRNA. The red nucleotides show the introduced mismatches in the case of 1 MM, 3 MM and 20 MM (full mismatch). Note the gRNA binding sequence at the target is unaffected by these changes. The sRNA is produced from two distinct casettes, one with an aTc inducible promoter and the other with a constitutive J119 promoter. (b-Right) Bar chart showing fold activation in GFP for different amounts of mismatches in the target-activator binding site. The fold activation is the ratio of GFP in the presence of the sRNA cassette with respect to the GFP in the absense of the sRNA cassette. (c) Genetic circuit implemented to investigate the effect of target-activator binding affinity on the paradoxical repression. The binding affinity is varied by introducing mismatches as shown in (b). (d) Input output response of normalized GFP normalized with increasing sRNA, for different number of mismatches. The GFP value is normalized using the basal expression levels obtained in Figure 2(c) . The amount of gRNA is kept constant with 10 uM IPTG induction and the sRNA levels are varied through aTc induction. In (b) and (c), all plasmids are introduced into the Marionette strain of E. coli 7 , where the corresponding regulators are endogenously expressed. To vary the dissociation constant of the CRISPRa complex to the target, we mutated the DNA binding site sequence at the target, by systematically introducing mismatches between the sRNA and the binding site ( Figure 4b ). A perfect match (0 mismatches denoted as 0 MM) results in high affinity (low K ta ), while introducing mismatches (e.g., 3 mismatches) weakens the interaction, increasing K ta . The experimental results validate this tuning strategy. GFP is measured at a saturating concentration of the sRNA for different number of mismatches (MM) and normalized with the GFP levels in the absence of the sRNA to obtain the fold activation values. With zero mismatches (0 MM), activation is strong (greater than 5 fold). As the number of mismatches increases (1 MM, 3 MM), the level of activation progressively decreases (approximately 4 fold and 1.3 fold, respectively), such that with 20 mismatches (20 MM), the activator no longer binds to the target (approximately 1 fold activation, i.e. no change in GFP). This demonstrates that the activator–target affinity, and thus K ta , can be modulated, enabling experimental exploration of the parameter regime required for paradoxical behavior. Now, we design a construct that allows constant expression of the repressor, and overlapping binding sites at the decoy, while varying the activator levels for different target-activator binding affinities ( Figure 4c ). We observe that for high binding affinities (0 MM and 1 MM), increasing the activator does not show paradoxical effect as predicted by the model ( Figure 4d ). As the sRNA (varied by the amount of aTc) is increased GFP expression increases. On the other hand for low binding affinities (3 MM and 20 MM), as the activator is increased, GFP levels decrease showing the paradoxical effect ( Figure 4d ). Next, we investigate other key conditions under which the paradoxical repression effect is observed by systematically validating the critical requirements predicted by the theoretical model. By utilizing the genetic construct established in Figure 4(c) , the following experimental validation involve modifications of the construct to no longer meet individual requirements, followed by measurement of the input-output response curves (GFP as a function of sRNA, induced by aTc). First, we explore the requirement of off-target sequestration to observe the paradoxical effect by removing the decoy cassettes from the genetic circuits ( Figure 5a ). The configuration retains the repressor and the low activator-target affinity (high K ta ). The input-output curve of GFP in the absence of decoy shows that the paradoxical effect is removed. As aTc increases, GFP increases monotonically ( Figure 5b ), confirming the theoretical prediction that without the decoy sites to serve as a reservoir, the dynamics are dictated by direct competition between the activator and repressor at the target. Download figure Open in new tab Figure 5. The presence of decoy sites with overlapping binding and repressors are critical to observe the paradoxical behavior. (a) Genetic construct implemented to observe the paradoxical effect in the absence of decoy sites. (b) Input output response of GFP with increasing aTc for the construct in (a) for 10 uM IPTG induction. (c) Genetic construct implemented to observe the paradoxical effect in the absence of overlapping binding sites at the decoy. (d) Input output response of GFP with increasing aTc for the construct in (c) for 10 uM IPTG induction. (e) Genetic construct implemented to observe the paradoxical effect in the absence of the repressor. (f) Input output response of GFP with increasing aTc for the construct in (e). In (a), (c) and (e), all plasmids are introduced into the Marionette strain of E. coli 7 , where the corresponding regulators are endogenously expressed. To verify the necessity of the displacement mechanism ensuring competitive binding at the decoy, we modified the binding sites to be non-overlapping. Here, the decoy sites and repressor are present, and K ta is high, but the binding sites on the decoys are redesigned to be far apart from each other ( Figure 5c ). The input-output curve of GFP without overlapping decoy sites again shows no paradoxical effect ( Figure 5d ). GFP increases as aTc is increased, aligning with the theoretical prediction ( Figure 1b ). Without overlapping binding sites, the activator cannot displace the repressor from the decoys. The increased activator concentration therefore leads to standard activation, albeit lower due to sequestration of activators by the decoys. Finally, we investigate the requirement for the presence of repressor by removing the gRNA cassette from the genetic circuit ( Figure 5e ). The configuration includes the overlapping binding at the decoy sites and high K ta . The input-output curve shows very low levels of GFP that remain relatively constant at a moderately activated state as the aTc is increased ( Figure 5f ). Due to low target-activator affinity compared to affinity between decoys and activators, the majority of the activator molecules are sequestered by the decoys, resulting in a flattened GFP response. This experiment highlights that the paradoxical repression requires the presence of the repressor molecule that is being displaced. Discussion Metastasis remains the leading cause of cancer mortality, and its initiation depends critically on the ability of carcinoma cells to adopt more migratory, invasive, and drug-resistant phenotypes. A central regulator of this transition is the epithelial mesenchymal transition (EMT), a process controlled by intertwined networks of transcription factors and chromatin-modifying complexes. Among EMT-inducing transcription factors, ZEB1 plays a particularly prominent role: it regulates cell-cell adhesion, drives motility, and confers broad resistance to therapy. Because ZEB1 expression responds sensitively to alterations in upstream regulatory circuits, it provides a powerful readout of how molecular perturbations shift EMT state. Although EMT is often framed in terms of transcription factors alone, it is now clear that chromatin regulators strongly shape EMT stability and plasticity. Two major chromatin-modifying systems act at EMT-relevant loci. PRC1/PRC2 complexes deposit the repressive histone mark H3K27me3 and maintain epithelial identity by silencing mesenchymal genes, including ZEB1 and ZEB2 . In contrast, Trithorax/COMPASS complexes such as KMT2D/MLL4 deposit the enhancer-associated mark H3K4me1 and activate lineage-specific regulatory programs that promote or stabilize mesenchymal traits. Because these complexes influence enhancer logic, promoter accessibility, and transcription factor binding, they serve as central determinants of cell identity during tumor progression. Recent genome-wide CRISPR screens underscore this duality by identifying both PRC2 and KMT2D as “epithelial guardians.” In epithelial cancer cells, loss of PRC2 weakens repression of EMT transcription factors, while loss of KMT2D rewires enhancer landscapes that normally constrain EMT-promoting signals. Perturbations to either system do not simply shift expression uniformly upward or downward; instead, they activate distinct EMT subcircuits and can even invert expected regulatory outcomes. For example, PRC2 occupancy at the ZEB1 and ZEB2 promoters changes dynamically under perturbations, and in some contexts, knocking out an activating factor such as KMT2D paradoxically increases ZEB1 expression despite intact repressive machinery. These findings highlight a deeper challenge: understanding why modifying an activator or repressor can produce expression changes that defy naive intuition. Chromatin regulators share genomic territories, compete for access to enhancers and promoters, and influence one another’s binding patterns through steric and kinetic effects. Such interactions make it difficult to isolate causal principles in native chromatin environments. This motivates the synthetic approach used in the present study. By reconstructing a minimal activator and repressor competition module in a prokaryotic system, this work provides a clean test of a mechanistic hypothesis that may help explain paradoxical ZEB1 responses observed in cancer epigenetics. The synthetic platform does not replicate chromatin biology, but it clarifies a fundamental competitive interaction that could operate upstream of those chromatin-level phenomena, thereby informing future efforts to interpret EMT regulation in mammalian cells. The results presented in this work provide a possible mechanistic understanding of the paradoxical regulatory behaviors observed in cancer epigenetics. The experiments show that, under mild constraints that are common in natural gene regulatory systems, transcription factors can shape gene expression in unexpected ways. A key scenario arises when two effectors share multiple binding sites in an exclusive manner. Under such conditions the system exhibits retroactivity 3 , 4 , 10 , and the activity of one regulator can strongly influence the genomic availability of the other. A simple mechanistic model we developed predicted that, when exclusive binding and specific decoy-site affinities are present, a system containing one activator and one repressor can produce target gene repression in response to induction of the activator ( Figure 1 ). This prediction closely parallels the counterintuitive responses seen at EMT transcription factor loci such as ZEB1 when chromatin regulators are perturbed in cancer cells. We tested the model’s prediction in a minimal bacterial circuit using CRISPRa as the activator and CRISPRi as the repressor. Competitive binding at decoy sites was implemented by directly overlapping the corresponding binding sequences ( Figure 2c ). To tune the affinity between the activator and the target, which the model identified as a critical parameter, we introduced defined mismatches in the activator’s DNA binding site ( Figure 4b ). The experiments confirmed that at low target-activator affinity, induction of the activator consistently represses the target gene. Finally, systematic perturbations of individual components demonstrated that the conditions predicted by the model are indeed essential for producing the paradoxical response ( Figure 5 ). These findings illustrate how molecular competition, when implemented in a controllable synthetic system, can reproduce regulatory behaviors that resemble the unexpected expression patterns seen in EMT control and cancer progression. Methods Bacterial strain construction and manipulation Plasmids used in this study were cloned using standard molecular biology protocols. The bacterial strain E. coli NEB5 α (NEB, C2987H) grown in LB Broth Lennox (240230) with appropriate antibiotics were used for genetic circuit construction. The sequence-verified plasmids are co-transformed or retransformed to E. coli Marionette strain 7 for all presented results. The experiment’s growth medium is an M9 minimal medium composed of M9 salts (1X), 0.4% glucose (SIGMA-ALDRICH, 49159), 0.2% casamino acid (VWR, TS61204-5000), and 1mM thiamine hydrochloride (SIGMA-ALDRICH, T4625-25G) supplemented by appropriate antibiotics. The final concentrations of antibiotics, ampicillin (SIGMA-ALDRICH, A0166-5G) and kanamycin (SIGMA-ALDRICH, K1377-5G) are 100, and 50 ug mL −1 , respectively. Isopropyl β -D-1-thiogalactopyranoside (IPTG, SIGMA-ALDRICH I6758-5G) and anhydrotetracycline (aTc, SIGMA-ALDRICH 37919-100MG-R) are used as the inducers. Genetic circuit construction The genetic circuit construction was based on Gibson assembly methods using plasmid 1 and plasmid 2a from Aravind et al. 9 as the backbone. DNA fragments to be assembled were amplified by PCR using Phusion High-Fidelity PCR Master Mix with GC Buffer (NEB, M0532S). The PCR mix underwent gel electrophoresis for purification. Zymoclean Gel DNA Recovery Kit (Zymo Research, D4002) was used for gel extraction. Later, the plasmid is assembled with Gibson assembly protocol using NEBuilder HiFi DNA Assembly Master Mix (NEB, E2621S). Assembled DNA was transformed into competent cells. Overnight cultures of singular colonies are grown in LB at 37°C. Plasmid DNA is then extracted from the overnight cultured using the plasmid miniprep-classic kit (Zymo Research, D4015). DNA sequencing is outsourced and is performed by Primordium Labs. Plate reader experiments The glycerol stocks (stored at −80°C) are used to prepare overnight cultures. These cultures are grown in M9 media supplemented with appropriate antibiotics at 37 °C, shaking at 250 rpm in a horizontal orbiting shaker for 13-16 hours in 15 mL culture tubes (VWR, 60818-667). These overnight cultures are diluted to an optical density at 600nm (OD600) measurement of 0.01 in 200 uL growth medium per well in a transparent 96-well plate with a flat bottom (FalconR 96-Well Cell Culture Plates, Corning, 15705-066, VWR) with appropriate antibiotics and inducers. The 96-well plate was incubated at 30°C in a Tecan Infinite 200 PRO microplate reader in static condition and was shaken at a fast speed for 5 seconds right before OD and fluorescence measurements. The sampling interval was 5 min. Excitation and emission wavelengths to monitor RFP fluorescence were 584 and 619 nm, and for GFP fluorescence were 485 and 530 nm, respectively. To ensure continued exponential growth, cell cultures were diluted with fresh growth medium (with antibiotics and inducers) to OD600 of 0.01 when OD600 approaches 0.08 at the end of one batch. Multiple batches were conducted until gene expression reached a steady state. The reported steady-state fluorescent values are normalized with the OD600 and are reported when each culture is at an OD600 of 0.06 in the final batch. Acknowledgements This work was partially supported by grants AFOSR FA9550-22-1-0316 and NSF/DMS-2052455 Funder Information Declared NSF CCF FET , Award 2007674 United States Air Force Office of Scientific Research, https://ror.org/011e9bt93 , AFOSR FA9550-22-1-0316 References 1. ↵ Zhang , Y. , Donaher , J. L. , Das , S. , Li , X. , Reinhardt , F. , Krall , J. A. , Lambert , A. W. , Thiru , P. , Keys , H. R. , Khan , M. , Hofree , M. , Wilson , M. M. , Yedier-Bayram , O. , Lack , N. A. , Onder , T. T. , Bagci-Onder , T. , Tyler , M. , Tirosh , I. , Regev , A. , Lees , J. A. & Weinberg , R. A. Genome-wide CRISPR screen identifies PRC2 and KMT2D-COMPASS as regulators of distinct EMT trajectories that contribute differentially to metastasis . Nat. Cell Biol . 24 , 554 – 564 , DOI: 10.1038/s41556-022-00877-0 ( 2022 ). OpenUrl CrossRef PubMed 2. ↵ Al-Radhawi , M. , Tripathi , S. , Zhang , Y. , Sontag , E. & Levine , H. Epigenetic factor competition reshapes the EMT landscape . Proc Natl Acad Sci USA 119 , e2210844119 ( 2022 ). OpenUrl CrossRef PubMed 3. ↵ Gyorgy , A. & Del Vecchio , D. Modular composition of gene transcription networks . PLoS computational biology 10 , e1003486 ( 2014 ). OpenUrl 4. ↵ Jayanthi , S. , Nilgiriwala , K. S. & Del Vecchio , D. Retroactivity controls the temporal dynamics of gene transcription . ACS synthetic biology 2 , 431 – 441 ( 2013 ). OpenUrl PubMed 5. ↵ Friedlander , T. , Prizak , R. , Guet , C. C. , Barton , N. H. & Tkačik , G. Intrinsic limits to gene regulation by global crosstalk . Nat. communications 7 , 12307 ( 2016 ). OpenUrl 6. ↵ Al-Radhawi , M. A. , Manoj , K. , Jatkar , D. , Duvall , A. , Del Vecchio , D. & Sontag , E. Competition for binding targets results in paradoxical effects for simultaneous activator and repressor action . In Proc. 63rd IEEE Conference on Decision and Control (CDC) , 5579 – 5585 ( 2024 ). 7. ↵ Meyer , A. J. , Segall-Shapiro , T. H. , Glassey , E. , Zhang , J. & Voigt , C. A. Escherichia coli “marionette” strains with 12 highly optimized small-molecule sensors . Nat. chemical biology 15 , 196 – 204 ( 2019 ). OpenUrl PubMed 8. ↵ Huang , H.-H. , Bellato , M. , Qian , Y. , Cárdenas , P. , Pasotti , L. , Magni , P. & Del Vecchio , D. dCas9 regulator to neutralize competition in CRISPRi circuits . Nat. communications 12 , 1692 ( 2021 ). OpenUrl 9. ↵ Manoj , K. & Del Vecchio , D. Resource competition in CRISPRa genetic circuits . bioRxiv 2024 – 07 ( 2024 ). 10. ↵ Del Vecchio , D. , Ninfa , A. & Sontag , E. Modular cell biology: Retroactivity and insulation . Mol. Syst. Biol . 4 , 161 ( 2008 ). OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted November 27, 2025. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Paradoxical gene regulation explained by competition for genomic sites Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Paradoxical gene regulation explained by competition for genomic sites Dhruv D. Jatkar , Krishna Manoj Aravind , Eduardo D. Sontag , Domitilla Del Vecchio bioRxiv 2025.11.27.691022; doi: https://doi.org/10.1101/2025.11.27.691022 Share This Article: Copy Citation Tools Paradoxical gene regulation explained by competition for genomic sites Dhruv D. Jatkar , Krishna Manoj Aravind , Eduardo D. Sontag , Domitilla Del Vecchio bioRxiv 2025.11.27.691022; doi: https://doi.org/10.1101/2025.11.27.691022 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Synthetic Biology Subject Areas All Articles Animal Behavior and Cognition (7635) Biochemistry (17697) Bioengineering (13894) Bioinformatics (41951) Biophysics (21455) Cancer Biology (18592) Cell Biology (25507) Clinical Trials (138) Developmental Biology (13380) Ecology (19903) Epidemiology (2067) Evolutionary Biology (24321) Genetics (15610) Genomics (22509) Immunology (17737) Microbiology (40398) Molecular Biology (17182) Neuroscience (88618) Paleontology (667) Pathology (2833) Pharmacology and Toxicology (4825) Physiology (7641) Plant Biology (15158) Scientific Communication and Education (2046) Synthetic Biology (4296) Systems Biology (9825) Zoology (2271)