Use of MYB as a new synthetic activator to enhance transgene expression within repressed Polycomb chromatin

Background Epigenetic silencing of transgenes through chromatin packaging has been a persistent issue for the development of transgenic mammalian cell lines. Endogenous mechanisms are known to induce a closed chromatin state around foreign DNA before and after it has been integrated into a host cell’s genome. Scientists are interested in reversing this silencing, but a lack of a priori knowledge of the chromatin features at transgenes hinders the rational design and application of effective strategies for transcriptional activation. Results Here, we systematically tested activation-associated DNA elements and proteins in transfected plasmid DNA and at epigenetically-silenced chromosomal transgenes. We demonstrated that placing DNA elements that are targeted by MYB (c-myb) and p65 upstream of a minimal promoter enhance expression from transfected plasmid DNA. To regulate the expression of chromosomally-integrated transgenes, we used proteins fused to the Gal4 DNA binding domain or dCas9/sgRNA. Three activation-associated peptides, p65, VP64, and MYB, sustained reactivation of transgene expression over 15 cell divisions in an immortalized human cell line (HEK293). Activity of the MYB fusion was inhibited by celastrol, a drug that blocks interactions between MYB and the p300/CBP histone acetyltransferase complex. Single-site targeting via dCas9-MYB was sufficient to activate transgenes within ectopic Polycomb heterochromatin and at a different site that had undergone position effect silencing. Conclusion Here we demonstrate the utility and flexibility of cis-regulatory elements and fusion proteins derived from natural gene regulation systems to enhance expression from epigenetically silenced transgenes. DNA motifs for p65 and MYB can be added to the transgene itself, or the activating proteins can be targeted to transgenes without enhancers to stimulate gene activation. This work has implications for determining the most appropriate strategy to enhance gene expression specifically in Polycomb-repressed chromatin.


BACKGROUND
The advancement of cell engineering requires robust and reliable control of endogenous and synthetic genetic material within living cells. A lack of tools for enhancing the expression of transgenes in mammalian cells currently limits effective gene regulation across contexts. The rapid formation of heterochromatin around transgenic material in mammalian cells limits our ability to express foreign DNA for the production of therapeutic proteins and the development of engineered mammalian systems for biosensing and computing [1,2]. Integrated transgenes are often silenced by the same mechanisms that serve as a cellular defense against viral insertion into the genome [3][4][5]. Nucleation of heterochromatin around transgenic material can be initiated and sustained by both promoter methylation [1,5] and various histone modifications [2,4]. For example, MyD88 pathway-mediated silencing of transgenes leads to an accumulation of repressive H3K9me on newly bound histones [2,6]. Silencing of transgenes may also be Polycomb-mediated, where Polycomb repressive complexes deposit H3K27me3 on histones to establish a silenced state [7][8][9]. The diversity and persistence of transgene silencing has led to the development of tools for mammalian cell engineering specifically aimed at combating heterochromatin.
Recruiting activators to a specific locus in order to reverse epigenetic silencing can be achieved either by including an activation-associated cis-regulatory DNA sequence within the construct itself, or through the targeting of engineered fusion proteins to the silenced transgene.
Both natural and synthetic cis-regulatory motifs that recruit activators have been used [10-13] to help increase transgene expression as an alternative to viral promoters that are prone to methylation and silencing [1]. Previous screens by ourselves and other groups [11,14,15] have identified mammalian activation-associated cis-regulatory elements that recruit endogenous factors to increase the expression of epigenetically silenced transgenes, including motifs for nuclear factor Y, CTCF, and elongation factor alpha (EF1-) [12,13]. The underlying regulatory mechanisms are not entirely understood, since in this case efficient screening for functional sequences has been prioritized over dissecting the mechanism of individual elements.
Likewise, p65-based systems are very effective at restoring both endogenous [16,21] and transgenic [22] gene expression, but have an undetermined effect on chromatin structure and accessibility.
Significant progress towards transgene reactivation has been made so far, but several important gaps remain. First, several natural mechanisms of activation are still under-investigated by biological engineers. For instance, chromatin remodelers shift, remove, or exchange nucleosomes [23], and pioneer factors increase DNA accessibility in closed chromatin by displacing linker histones [23][24][25]. Second, the parameters for stable transgene activation are not yet fully defined. So far, at least two studies have demonstrated prolonged enhancement of transgenes (10 to 25 days) via targeted fusion proteins alone [26] or in combination with flanking anti-repressor DNA elements [27]. Neither study evaluated the chromatin features at the target genes prior to their reactivation, therefore the context in which expression enhancement occurred is uncertain. Finally, the performance of targeted activators can be context-dependent. Catalytic domains used for site-specific chromatin remodeling [27][28][29], may be inhibited by pre-existing chromatin features that vary across loci. For example, Cano-Rodriguez et al. constructed a targeted histone methyltransferase fusion and found that the endogenous chromatin microenvironment, including DNA methylation and H3K79me, impacted the ability of their fusion to deposit H3K4me and induce activation [30]. Similarly inconsistent performance has been shown for other fusions that generate H3K79me and H3K9me [31,32]. Systematic studies at loci with welldefined chromatin compositions are needed to fully understand mechanisms of chromatin state switching.
Here, we expand previous work where we had identified cis-regulatory sequences that enhanced expression from plasmid-borne transgenes [12]. To regulate expression of chromosomally-inserted transgenes, we compare targeted proteins that represent diverse activities: transcriptional activation through cofactor recruitment, direct histone modification, and nucleosome repositioning and displacement. We focus on reversal of silencing within Polycomb heterochromatin, which is known to accumulate at transgenes that are integrated into chromosomes [7][8][9] and is widely distributed across hundreds or thousands of endogenous mammalian genes that play critical roles in normal development and disease [9, 33,34]. We report that recruitment of p65 and MYB-associated components via a cis-regulatory element or fusion proteins enhances expression from epigenetically silenced transgenes. MYB-mediated activation within Polycomb heterochromatin relies on interactions with p300 and CBP. Our results have implications for determining the most appropriate strategy to enhance gene expression, specifically within Polycomb-repressed chromatin.

Identification of Activation Associated Peptides
We surveyed public data to identify epigenetic enzymes and other proteins that are associated with transcriptional activation, and therefore might effectively disrupt repressive Polycomb chromatin.
The AAPs fall into six general categories. The transcriptional activation group, (NFkB)-p65 and the MYB (c-myb) transcriptional activation domain (TAD), includes proteins that recruit RNA Polymerase II (PolII) and p300/CBP, respectively. These AAPs have no known intrinsic gene regulation activity, and therefore rely upon the recruitment of other proteins to stimulate transcription [37][38][39]. We also included the recombinant TAD VP64 (four tandem copies of VP16), a popular component for synthetic activators. Histone modifications deposited by the co-activators that are recruited by these three domains are primarily associated with activation.
The histone acetylation (HAT) group includes ATF2, P300, and KAT2B. These peptides acetylate H3K27. In particular, p300 is associated with the recruitment of CBP and other coactivators that generate the activation associated mark H3K4me [40]. The histone H3 methyltransferase (H3 MT) group and the H4 methyltransferase (H4 MT) group include proteins that are either Mixed-Lineage Leukemia (MLL) complex components or SET proteins. SETD7 deposits the activation associated modification H3K4me, but its regulatory impact may vary based on local DNA methylation, which can enhance or impede co-recruitment of repressive cofactors.
The histone H4 methyltransferase PRMT5 induces histone acetylation that is associated with DNA methylation in some contexts [41]. Still, PRMT5 primarily acts as an activator.
The final two groups, chromatin remodelers (CR) and pioneer factors (PF) represent activities that are relatively underexplored in the design of fusion-protein regulators. SMARCA4 is a chromatin remodeler that relies on an ATP-dependent reaction to shift the position of nucleosomes at a target site [42]. It does not mediate the deposition of histone modifications, but is associated with CBP recruitment that evicts Polycomb-associated histone modifications [43]. PFs are represented in our library by FOXA1, a winged-helix protein that displaces linker histones from DNA to facilitate a transition to open chromatin [44]. In general, PFs bind to DNA within heterochromatin and do not catalyze histone post-translational modifications [25].
Several of the AAPs in our panel are associated with the eviction of Polycomb repressive complexes (PRCs) from endogenous genes. Accumulation of the chromatin remodeling protein SMARCA4 (BRG1) leads to the loss of PRCs at Pou5f1 in mouse cells [45] and at INK4b-ARF-INK4a in human malignant rhabdoid tumor cells [46]. In the latter case, KMT2A (MLL1) also participates in PRC depletion. ATF2 interacts with a kinase that generates H3S28p, which antagonizes PRC binding [47][48][49]. Acetylation and methylation at H3K27 are mutually exclusive [50,51], therefore the AAPs associated with H3K27ac (p65, MYB, ATF2, P300, KAT2B) might contribute to PRC eviction (Fig. 1). None of the AAPs in our panel are associated with enzymatic erasure of H3K27me3.  Table S1.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Cis-regulatory elements recognized by transcriptional activators p65 and MYB enhance expression from an extra-chromosomal transgene
First, we used enhancer DNA elements to regulate expression from transiently-transfected plasmid DNA. Work from our group [52] and others [53,54] has shown that plasmid DNA becomes occupied by histones, which may contribute to transgene silencing in human cells. In a previous study, we used DNA sequences that were known targets of endogenous activation-associated proteins to reduce silencing of a luciferase reporter gene [12]. Here, we tested additional motifs ( Fig. 2a) that are recognized by AAPs from the transcriptional activator group in our panel: MYB and p65 (Fig. 1).
One of three MYB enhancer variants or the p65 enhancer was placed in either a forward or reverse orientation upstream of an EF1a promoter and a luciferase reporter (Fig. 2b). PC-3 (human prostate cancer) cells were transfected with each plasmid as described previously [12]. The highest levels of enhanced expression were observed for MYB variant A (4.5-fold, p = 0.03) or p65 (5-fold, p = 0.08) placed in the reverse orientation (Fig. 2c). Interestingly, switching the orientation of these motifs eliminated the enhancement effects. Nonetheless, these results suggest that cisregulatory elements from the p65 and MYB systems can be used to attract endogenous transcriptional activators to a synthetic promoter to drive transgene expression. was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/487736 doi: bioRxiv preprint first posted online Dec. 5, 2018; represent the average of three transfections. Asterisks (*) = p < 0.05 for the experimental average, relative to the Control average.

Identification of fusions with robust activity within Polycomb heterochromatin
Next, we asked whether the individual peptides MYB and p65, as well as other AAPs could enhance transgene expression in the absence of a specific enhancer sequence. To determine AAP activity within silenced chromatin, we targeted AAP fusion proteins (Fig. 1) to a chromosomal luciferase reporter that had been previously targeted by Polycomb repressive complexes (PRCs).
The AAP open reading frames (ORFs) encode catalytic subunits or full length proteins (Fig. 3) that have been shown to support an epigenetically active state in various prior studies [37,38,42,44,[56][57][58][59][60][61][62]. All of these ORFs exclude DNA binding and histone binding domains, except for the ORF encoding FOXA1 which has a catalytic domain that requires histone interactions. We cloned each ORF into mammalian vector 14 (MV14) (Fig. 3) to express a Gal4-mCherry-AAP fusion. The Gal4 DNA binding domain serves as a module to target AAPs to UAS sequences in the transgene, while the mCherry tag allows for protein visualization and quantification of the activator fusion.  Table S2).
ORFs were cloned into MV14 to express a Gal4-AAP fusion protein from a cytomegalovirus (CMV) promoter.
Gal4-AAPs are expressed with a C-terminal nuclear localization signal (NLS) and a 6X histidine tag. MV14 expresses puromycin resistance to enable selection of Gal4-AAP positive cells.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
Expression of luciferase is switched from active to silenced through accumulation of polycomb chromatin features, which have been detected by chromatin immunoprecipitation (ChIP) experiments: EZH2, Suz12, CBX8, depletion of H3K4me [63], and gain of H3K27me3 [22,63]. This system allows us to test the activity of Gal4-AAPs with a priori knowledge of the chromatin environment at the target gene.
Gal4-EED/luc cells were treated with dox for 48 hours to induce heterochromatin at the luciferase transgene. Afterwards, dox was removed and cells were grown for four days without dox to allow for Gal4-EED depletion. Cells were then transfected with individual Gal4-AAP plasmids.
Luciferase expression was measured 72 hours post transfection.
Three of the sixteen Gal4-AAP-expressing samples showed increased luciferase levels compared to a mock-transfected control (Lipofectamine reagent only) (p < 0.05) (Fig. 4a). Lack of enhanced luciferase expression for the other fusions could have been due to strong inhibition by PRC complexes or failure of the AAPs to function as Gal4 fusions at the UAS site. Therefore, we also tested the activities of the fusion proteins within open chromatin. We used a parental HEK293 cell line, Luc14, that carries the firefly luciferase construct (Gal4UAS-Tk-luciferase) but lacks the TetO-CMV-Gal4EED repression cassette (Fig. 4b) [63]. Luciferase is constitutively expressed at high levels in Luc14.
We found a similar trend of expression enhancement at open chromatin in Gal4-AAPexpressing cells (Fig. 4b), where only three Gal4-AAP fusions were able to stimulate expression when positioned at the promoter-proximal UAS (Fig. 4b). In both chromatin states, AAPs from the transcriptional activation group (Fig. 1, Fig. 3) significantly increased expression compared to a mock transfection control (p < 0.05) by up to five fold. Our results are consistent with previous studies where p65, VP64, or MYB stimulated gene expression from a promoter-proximal site [37][38][39]. Here, we have demonstrated activities of these proteins within highly PRC-enriched chromatin.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which .

Fusion-induced activation is sustained after loss of the Gal4-AAP transactivator
The results so far were obtained at a single time point after Gal4-AAP expression. We were interested in determining whether transgene activation within polycomb chromatin is stable or is transient and susceptible to eventual re-silencing [64]. To investigate this question, we performed time-course experiments to measure expression from re-activated luciferase over time. We was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

MYB-mediated activation within closed chromatin requires interactions with a histone acetyltransferase
Next, we used specific chemical inhibitors to probe the mechanism of MYB-driven enhancement.
The TAD core acidic domain of human MYB (D286-L309) included in our Gal4-MYB fusion construct is known to interact with a protein heterodimer of p300 and CBP (Supplemental Fig. S2).
A single base pair mutation within the MYB TAD domain (M303V) disrupts p300 recruitment and subsequent activation by MYB indicating that this recruitment is crucial to activation by MYB [65,66]. The p300/CBP histone acetylation complex deposits H3K27ac in opposition to H3K27me3 induced by PRC2 [67,68]. Therefore, induced activation within Polycomb heterochromatin may be driven by histone acetylation.
To test this idea, we treated cells with two compounds that are known to disrupt the activity of the MYB/p300/CBP complex. Celastrol is a minimally toxic pentacyclic triterpenoid that directly inhibits the MYB/p300 interaction, by binding to the KIX-domain of CBP which serves as a docking site for the formation of the MYB/p300/CBP complex [69][70][71][72] (Fig. 6a). C646, a pyrazolonecontaining small molecule, binds the p300 catalytic domain and thus directly and selectively inhibits p300 HAT activity regardless of its association with MYB (Fig. 6a) [73][74][75]. These compounds allow us to resolve the roles of complex assembly and p300-mediated histone acetylation during Gal4-MYB-mediated activation.
Gal4-EED/luc cells were treated with dox to induce polycomb chromatin and transfected with Gal4-MYB as described for previous experiments. We treated these cells with 5 μ M celastrol or 5 μ M C646 for six hours. MTT assays indicated no toxicity to HEK293 cells at this concentration (Supplemental Fig. S3). We expected luciferase assays to show a decrease in Gal4-MYB-induced expression in drug-treated cells compared to an untreated control. We observed a significant (p < 0.05) decrease in luciferase expression in celastrol-treated cells, but not in C646-treated cells (Fig.   6b). This result suggests that Gal4-MYB activity requires MYB TAD and p300/CBP assembly, while p300 HAT activity is dispensable. The other two strong activators, Gal4-VP64 and -p65, were insensitive to celastrol and C646 (Fig. 6b), indicating a p300/CBP-independent mechanism for these two fusions.
In a time-course experiment using celastrol, we observed that Gal4-MYB-mediated activation is reversible. Eighteen hours after removal of celastrol from Gal4-MYB-treated cells, luciferase expression levels increased (p < 0.05 compared to repression at t = 6), nearly restoring expression to original levels (t = 0) (Fig. 6c). Re-addition of celastrol led to a loss of Gal4-MYB induced expression (Fig. 6c).
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

MYB-mediated activation in Polycomb heterochromatin relies upon proximity to the transcriptional start site
Next we asked whether MYB-mediated activation at transgenes is context dependent. We leveraged the flexible dCas9/sgRNA system to target the MYB TAD to several sites along the luciferase transgene (Fig. 7a). To do so, we targeted sites at different positions within the Tkluciferase gene. We also tested the MYB TAD at a different transgene, CMV-GFP in HEK293, that had become silenced after several passages (C. Liu, unpublished).
We induced Polycomb heterochromatin in HEK 293 Gal4EED/luc cells with dox, followed by washout of dox to allow Gal4-EED depletion as described above. We transfected the cells with one of four dCas9-MYB constructs, each carrying a different sgRNA targeted at the luciferase . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/487736 doi: bioRxiv preprint first posted online Dec. 5, 2018; transgene. After 72 hours, we tested luciferase expression and found that dCas9-MYB targeted nearest the transcription start site (+9) was able to restore levels of expression similarly to Gal4-MYB (Fig. 7b). In induced Polycomb heterochromatin we observed clear position effects, as the downstream target sites show levels of activation significantly lower than Gal4-MYB (p < 0.05).
After determining the viability of dCas9-MYB to act as an activator for silenced transgenes in a defined chromatin environment, we wanted to test this domain against endogenous heterochromatin at the CMV-GFP transgene. The construct, GFP under the control of a CMV promoter, was inserted via Cas9-mediated HDR into a non-protein-coding region of the HEK293 genome (HEK293 site 3 [76]). We transiently transfected the cells with dCas9-MYB constructs, each carrying one of four different sgRNAs targeted upstream, within the promoter, or in the coding region of the transgene. Seventy-two hours post transfection, we used flow cytometry to measure GFP fluorescence compared to a mock-transfected control (Lipofectamine reagent only). We found that GFP fluorescence was significantly higher (p < 0.05) in all dCas9-MYB-expressing cells regardless of gRNA position (Fig. 7c), indicating that MYB-mediated activation does not require proximity to the TSS in all contexts.  Table S3) were cloned into the BbsI sites and expressed from a hU6 promoter on the same vector. B) We targeted dCas9-MYB to four locations (g46, g32, g31, g25) across the Tk-luciferase transgene in silenced Gal4-EED/luc cells. Mean luciferase signal per cell is presented as described for was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/487736 doi: bioRxiv preprint first posted online Dec. 5, 2018; Circle = median GFP fluorescence value from one transfection, 10,000 cells; bars = means of three transfections. In B and C, asterisks (*) = p < 0.05 for experimental mean compared to the NA control mean.

DISCUSSION
We have demonstrated that DNA enhancer elements and fusion proteins derived from endogenous mammalian systems can be used to support strong expression from transgenes. Furthermore, we have successfully demonstrated long-term reactivation of a transgene that had been previously Our results also suggest that the mechanism of artificial transgene reactivation within Polycomb heterochromatin requires assembly of transcription initiation complexes. From STRING analysis, we found no clear pattern of histone modifications to distinguish the inactive Gal4-AAPs from activators that were able to enhance expression in Polycomb heterochromatin (Fig. 1). We observed that several of the fusion proteins did not restore expression from the Polycombrepressed luciferase transgene in HEK293 cells (Fig. 4a). Thus, Gal4-tethered proteins might be functional but not sufficient, are non-functional (sterically hindered), or require positioning within non-coding DNA such as enhancer elements. For instance, a p300 fusion has shown strong activation of MyoD and Oct4 when positioned 5-20 kb upstream at an enhancer [29]. Future work could be done to systematically test the AAPs at endogenous enhancers.
Upon further investigation we determined that assembly of the MYB TAD with P300/CBP is critical for Gal4-MYB-mediated activation within Polycomb chromatin. Inhibition of p300 HAT activity via C646 did not disrupt Gal4-MYB function (Fig. 6b). Furthermore, the Gal4 fusion that included only the p300 HAT domain failed to activate Polycomb-repressed Tk-luciferase (Fig. 3,   Fig. 4). Therefore the p300 catalytic domain alone is neither necessary nor sufficient to reverse epigenetic silencing under the conditions tested here. CBP, which is also a histone acetyltransferase, might compensate for p300 in C646-treated cells [68]. Celastrol inhibits the interaction of p300/CBP with MYB by binding to the CBP KIX domain [69][70][71][72], and completely reduces Gal4-MYB activity (Fig. 6b). In contrast to C646, celastrol may disrupt the recruitment of both HAT enzymes, p300 and CBP.
In contrast to Gal4-MYB, the Gal4-p65 and -VP64 fusions showed robust activation of PRCsilenced luciferase in the presence of both inhibitors (Fig. 6b). Although VP64 (VP16) and p65 are known to interact with p300/CBP, they also interact with the large multi-subunit Mediator complex . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/487736 doi: bioRxiv preprint first posted online Dec. 5, 2018; to initiate transcription [77][78][79]. Multiple interactions of Gal4-p65 and -VP64 with Mediator may allow these proteins to function independently of p300/CBP [80]. However in the case of Gal4-MYB, cooperative interactions between p300/CBP and Mediator [81,82] may be necessary for gene activation. In our study, Mediator complex recruitment arises as a particularly potent mechanism of transgene reactivation in Polycomb heterochromatin [82]. Mediator is known to cooperatively regulate PRC2 repression [83] and certain Mediator subunits are directly involved in the removal of PRC2 from endogenous promoters [84]. Similarly, Mediator has an antagonistic relationship with the PRC1 repression complex [85].
The inhibitor experiments also suggest a novel technique for chemically-inducible gene regulation in mammalian cells. The ability to quickly toggle between enhanced and repressed states is a cornerstone technology for the control of engineered transgenic systems [26, 86,87].
Current methods for toggling gene expression in mammalian cells employ drug-mediated transactivator localization, such as allosteric modulation of DNA-binding protein domains [26, 86,88], blue light-responsive CRY proteins [89], and chemically induced dimerization (CID) systems [90][91][92], or RNA interference to deplete the regulator [87]. To our knowledge, no systems currently exist where the transactivation module's activity (i.e., MYB-CBP binding) is modulated by a small molecule drug. Celastrol has a low toxicity and is in fact being explored as a therapeutic due to its positive effects on the immune system [93][94][95]. The concentration of celastrol that is sufficient to toggle Gal4-MYB activity in polycomb chromatin is well below any reported LD50 values for celastrol [96][97][98][99][100].

CONCLUSION
In conclusion, we have determined a predominant role for p300/CBP-recruiting transcriptional activators in the reversal of Polycomb-mediated expression in the context of synthetic transgene . CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/487736 doi: bioRxiv preprint first posted online Dec. 5, 2018; regulation. In particular, we have expanded the characterization of the transcriptional activator protein MYB and its associated enhancer DNA sequence for applications in artificial gene regulation in mammalian cells.

Construction and Testing of Plasmids containing MYB-and p65 Motifs
Plasmid construction, transfection of PC-3 cells, and luciferase assays were carried out as described previously [12]. Briefly, cloning of double-stranded oligos was used to insert motifs 222 bp upstream of the transcription start site of an EF1a promoter at XbaI/SphI. Plasmids were then transfected into PC-3 cells (ATCC, CRL-1435) using Lipofectamine LTX™ following the manufacturer's recommended protocols. Luciferase expression was measured 48 hours after transfection using a luciferase assay kit (Promega, Madison, WI). All luciferase values were normalized relative to the native plasmid control, which contained an unaltered EF1a promoter.

Construction of MV14 and Gal4-AAP Plasmids
We constructed mammalian expression vector 14 (MV14) for the overexpression of Gal4-mCherry-AAP fusion proteins in-frame with a nuclear localization sequence and 6X-histidine tag. First,  Table S4)

Cell Culturing and Transfections
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which . http://dx.doi.org/10.1101/487736 doi: bioRxiv preprint first posted online Dec. 5, 2018; Luc14 and Gal4-EED/luc HEK293 cells were grown in Gibco DMEM high glucose 1× (Life Technologies) with 10% Tet-free Fetal Bovine Serum (FBS) (Omega Scientific), 1% penicillin streptomycin (ATCC) at 37 °C in a humidified 5% CO 2 incubator. Gal4-EED/luc cells were treated with 1 µg/mL doxycycline (Santa Cruz Biotechnology) for 2 days to induce stable polycomb repression. Dox was removed and cells were cultured for another four days before being seeded in 12-well plates. Luc14 cells and dox-induced Gal4-EED/luc cells were seeded in 12-well plates such that cells reached 90% confluency for lipid-mediated transfection. Transfections were performed with 1 µg plasmid per well, 3 µL Lipofectamine LTX, and 1 µL Plus Reagent (Life Technologies) per the manufacturer's protocol. Seventy-two hours post transfection, cells were either collected for analysis or passaged further.
Puromycin selection was carried out on Gal4-AAP-expressing cells for the experiments represented in Figure 5 and Supplemental Figure S1. Dox-treated Gal4-EED/luc cells were transfected in 12-well plates and then grown for 24 hours before the addition of 10 µg/mL puromycin (Santa Cruz Biotechnology) to Gibco DMEM high glucose 1× (Life Technologies) with 10% Tet-free Fetal Bovine Serum (FBS) (Omega Scientific), 1% penicillin streptomycin (ATCC).
Cells were grown in puromycin containing media for two days before wash out.

Luciferase Assays
Luciferase assays were performed as previously described in Tekel et al. [101]. In brief, a single well of cells from a 12 well tissue culture plate was collected per independent transfection in 1.5mL 1X PBS. Cells were loaded into 9 wells of a Black Costar Clear Bottom 96 Well Plates (Corning #3631). Three wells of cells were used to detect mCherry in order to quantify Gal4-AAP proteins. A 2X Hoechst 33342 stain (Invitrogen #H3570) was loaded into three more wells to stain nuclear DNA in order to quantify cell density. The final three wells were prepared with Luciferase Assay Buffer (Biotium #30085). Plates were scanned in a microplate reader (Biotek Synergy H1) to detect RFP (580 nm -610 nm), Hoechst 33342 fluorescence (360 nm -460 nm) and chemiluminescence from the same sample in parallel.

RT-qPCR
We prepared total RNA from ~1.0 x 10 6 cells (Qiagen RNeasy Mini kit 74104) and generated cDNA from 2 µg of total RNA and the SuperScript III First Strand Synthesis system (Invitrogen was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Construction of dCas9-MYB and Design of sgRNAs
We modified the vector pX330A_dCas9-1 × 4 (a gift from Takashi Yamamoto, Addgene plasmid #63598) by inserting a gBlock Gene Fragment (Integrated DNA Technologies) encoding the MYB TAD followed by a p2A signal [102] and mCherry after the dCas9 ORF. The resulting vector expresses a dCas9-MYB fusion and mCherry as separate peptides from a single mRNA transcript.
The vector and gBlock were digested with FseI (New England BioLabs) and FastDigest EcoRI (ThermoFisher Scientific) and ligated using T4 DNA Ligase (New England BioLabs). We named this new vector pX330g_dCas9-MYB. SgRNAs used in the study (Supplemental Table S3) were designed using the CRISPR design tool at crispr.mit.edu. DNA oligos were synthesized with BbsI overhangs for cloning into pX330g_dCas9-MYB (Integrative DNA Technology). Drop-in of sgRNAs followed the cloning protocol described in Cong et al. [103].

Celastrol and C646 Treatments
Gal4-EED/luc cells were induced with dox and transfected as described above. Three days post transfection, cells were treated with either C646 (Selleck Chemicals) or Celastrol (Selleck Chemicals) diluted to a concentration of 5μM in Gibco DMEM high glucose 1× (Life Technologies) with 10% Tet-free Fetal Bovine Serum (FBS) (Omega Scientific). Cells were incubated with the drug for six hours before being washed and either harvested for a luciferase assay or grown further in drug-free media.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Statistical Analyses
The differences of means were calculated using the two sample, one-tailed Student's t test. For p < 0.05, confidence was 95% for 2 degrees of freedom and a test statistic of t (0.05,2) = 2.920. To evaluate significance of Gal4-MYB induced activation after the removal of celastrol and its subsequent re-addition, a nest one-way ANOVA was used with 95% confidence and two degrees of freedom. Vargas for early efforts on this work.

LIST OF ABBREVIATIONS
Availability of data and material: Sequences of plasmids used in this study are publically available as listed in Supplemental Table S4 as well as physically available in the DNASU plasmid repository.
Competing interests: The authors declare no competing interest.
Funding: This project was supported by NSF CBET grant 1403214. KAH was supported by National Institutes of Health NCI grant K01 CA188164.
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.