Gene activation by a CRISPR-assisted trans enhancer

The deactivated CRISPR/Cas9 (dCas9) is now the most widely used gene activator. However, current dCas9-based gene activators are still limited by their unsatisfactory activity. In this study, we developed a new strategy, the CRISPR-assisted trans enhancer, for activating gene expression at high efficiency by combining dCas9-VP64/sgRNA with the widely used strong CMV enhancer. In this strategy, CMV enhancer DNA was recruited to target genes in trans by two systems: dCas9-VP64/csgRNA-sCMV and dCas9-VP64-GAL4/sgRNA-UAS-CMV. The former recruited trans enhancer by annealing between two short complementary oligonucleotides at the ends of the sgRNA and trans enhancer. The latter recruited trans enhancer by binding between GAL4 fused to dCas9 and UAS sequence of trans enhancer. The trans enhancer activated gene transcription as the natural looped cis enhancer. The trans enhancer could activate both exogenous reporter genes and variant endogenous genes in various cells, with much higher activation efficiency than that of current dCas9 activators.

Although variant dCas9-based activators have been developed (Chen and Qi, 2017), the current dCas9-based transcriptional activators are relatively inefficient in endogenous gene activation and cell reprogramming (Gao et al., 2014). By a systematic comparison of relative potency and effectiveness across various cell types and species (human, mouse, and fly) (Chavez et al., 2016), it was found that the majority of second-generation activators had higher activity than that of dCas9-VP64, with the three most potent activators being VPR, SAM, and Suntag. The three activators were consistently better than VP64 across a range of target genes and cellular environments. Moreover, the three activators showed a similar level of activity, and fusing their elements did not yield more potent activators (Chavez et al., 2016). Novel, more potent dCas9-based activators might be built by creating other architectures.
Almost three decades ago, the human cytomegalovirus (CMV) enhancer/promoter (referred to as CMV enhancer hereafter) was found. It is a natural mammalian promoter with high transcriptional activity (Boshart et al., 1985). Later studies showed the CMV enhancer to be a strong enhancer in various mammalian cells (Boshart et al., 1985;Foecking and Hofstetter, 1986;Ho et al., 2015;Kim et al., 1990). This enhancer has been widely used to drive ectopic expression of various genes in a wide range of mammalian cells, and to drive ectopic expression of exogenous genes in broad tissues in transgenic animals (Furth et al., 1991;Schmidt et al., 1990), protein production by gene engineering, and gene therapy. We have recently improved the transcriptional activity of the CMV enhancer by changing the natural NF-kB binding sites into artificially selected NF-kB binding sequences with high binding affinity . Therefore, we conceived that a unique architecture may be constructed to improve dCas9-based activators using the CMV enhancer.
In this study, mimicking the natural enhancer activating gene expression by a loop structure (Carter et al., 2002;Tolhuis et al., 2002), we developed a new dCas9-based activator by combining dCas9/sgRNA with CMV enhancer. The 3 0 end of sgRNA was redesigned to contain a short capture sequence complementary to a stick-end of a double-stranded CMV enhancer. The CMV enhancer was anchored to the promoter region of a target gene by dCas9/sgRNA. The dCas9/sgRNArecruited CMV enhancer thus functioned like a natural looped cis enhancer in a trans form. We found that the new activator could efficiently activate exogenous and endogenous genes in various cells. More importantly, the CMV enhancer could be also recruited to a target gene in trans using another system consisting of dCas9-VP64-GAL4/sgRNA and UAS-CMV.

Principle of gene activation by a CRISPR-assisted trans enhancer
The principle of activating gene expression by a CRISPR-assisted trans enhancer is schematically illustrated in Figure 1a. A capture sgRNA (csgRNA) was produced by adding a capture sequence to the 3 0 end of a normal sgRNA sequence. A linear stick-end CMV (sCMV) enhancer was produced by adding a 3 0 end single-strand overhang. The overhang can anneal with the csgRNA capture sequence. When dCas9 protein was guided to the promoter of the target gene by csgRNA, sCMV could be recruited by csgRNA. The recruited sCMV may activate the transcription of the target gene like a natural looped cis enhancer. Because the dCas9/csgRNA-anchored sCMV functions as a transcription factors in trans, we named it a trans enhancer to distinguish it from the natural cis enhancer.

Effect of capture sequence on the function of sgRNA
To determine whether the capture sequence affects the function of sgRNA, we prepared a normal sgRNA and three csgRNAs targeting the same site of the HNF4a promoter. The three csgRNAs had different capture sequences. We used these sgRNAs to associate with the Cas9 endonuclease to cut a 732 bp HNF4a promoter fragment. The results indicated that the target DNA could be digested by all sgRNAs (Figure 1b), indicating that the capture sequence did not affect the sgRNA function.

Activation of exogenous reporter gene by trans enhancer
To determine whether the CRISPR-assisted trans enhancer activates gene expression, we constructed a reporter construct of HNF4a promoter (pEZX-HP-ZsGreen). 293 T cells were then transfected with various vectors (Figure 2a, Figure 2-figure supplement 1). The transfection indicated that ZsGreen expression could be successfully activated by dCas9/csgRNA2-sCMV but not activated by dCas9/csgRNA2-blunt CMV (bCMV). Although the dCas9/csgRNA2-sCMV showed a similar activation level to Cas9-VP64/sgRNA, it was far inferior to cis CMV enhancer. To improve the A capture sequence is added to the 3 0 end of sgRNA, which is used to capture a trans CMV enhancer with a single-stranded overhang that can anneal with the capture sequence of sgRNA. The captured trans CMV enhancer may function like the natural looped cis enhancer to activate transcription of the gene of interest, including exogenous and endogenous genes. (b) In vitro target DNA cutting by the Cas9-csgRNA complex. DNA fragments (732 bp) amplified from the HNF4a promoter region were, respectively, cut by the Cas9/ csgRNA and Cas9/sgRNA complexes. csgRNA1, csgRNA2 and csgRNA3 had the same target sequence but different capture sequences. DOI: https://doi.org/10.7554/eLife.45973.002   performance of trans CMV, we tried transfecting 293 T cells with dCas9-VP64/csgRNA2-sCMV. The results indicated that ZsGreen expression was highly activated by the transfection. In contrast, the dCas9-VP64/csgRNA2-bCMV showed a similar activation level to dCas9-VP64/sgRNA. These data revealed that the trans CMV not only truly functioned in trans via dCas9/csgRNA, but also synergistically interacted with dCas9-fused VP64. Subsequent transfections indicated that ZsGreen expression could also be highly activated by combination of dCas9-VP64, sCMV and other two csgRNAs, csgRNA1 and csgRNA3.

Activation of endogenous genes by trans enhancer
To further evaluate the activity of CRISPR-assisted trans enhancer, we activated endogenous genes with trans sCMV. csgRNAs targeting ten different genes was designed and their linear expression vectors were produced. Seven different cell lines were transfected with dCas9-VP64/csgRNA2-sCMV, dCas9-VP64/sgRNA and dCas9/csgRNA2-sCMV (Figure 3-figure supplement 1). The quantitative PCR (qPCR) detection of gene expression revealed that almost all genes were most significantly activated by dCas9-VP64/csgRNA2-sCMV in all cells. Moreover, most genes were more significantly activated by dCas9/csgRNA-sCMV than dCas9-VP64/sgRNA in all cells. These results suggest that the CRISPR-assisted trans enhancer could be used to activate variant endogenous genes in various cells. In addition, by activating the HNF4a gene in 293 T cells, we found that dCas9-VP64/csgRNA-sCMV had better activity than dCas9-VPR/csgRNA-sCMV in activating endogenous genes (Figure 3-figure supplement 2).
It has been reported that the cancer cells HepG2 and PANC1 can be differentiated into normal liver-and pancreas-like cells by exogenously expressing transcription factor HNF4a and E47. In the above assays, we found that the endogenous HNF4a and E47 genes were highly activated by the CRISPR-assisted trans enhancer in HepG2 and PANC1 cells (Figure 3). To further confirm the cellular effects of HNF4a and E47 activation, we detected expressions of other genes related to the differentiation of the two cancer cells (Figure 4). The results indicated that the genes related to stemness (CD133 and CD90) and pluripotency (Oct3/4, Sox2, Nanog, c-Myc, LIN28, and Klf4) were down-regulated, but those related to normal liver (GS, BR, ALDOB, CYP1a2, PEPCK, APOCIII, G-6-P, and HPD) and pancreas (MIST1, PRSS2, CELA3A, and CPA2) functions were highly up-regulated in HepG2 and PANC1 cells. Additionally, the cell cycle arrest-related gene p21 (HepG2 and PANC1) and TP53INP1 (PANC1) were highly up-regulated.

Activation of genes by other trans enhancers
To explore whether other enhancers could be also used as trans enhancers, we fabricated the bluntand stick-end versions of two other widely used promoters EF1a and PGK (bEF1a, bPGK, sEF1a, and sPGK). 293 T cells were co-transfected with these trans enhancers and dCas9-VP64/csgRNA and reporter construct. The results indicated that the ZsGreen expression was also activated by the two trans enhancers; however, the activation levels were lower than that of sCMV (Figure 5a, Figure 5figure supplement 1). All other transfections as controls did not activate ZsGreen expression (Figure 5a, Figure 5-figure supplement 1). The qPCR detection of HNF4a expression in the same transfected 293 T cells revealed that the endogenous HNF4a gene expression was also significantly activated by three stick-end trans enhancers, but not activated by all blunt-ended trans enhancers (Figure 5b). Subsequent HepG2 cell transfections with the same trans enhancers and dCas9-VP64/ csgRNA indicated that the endogenous HNF4a gene expression could also be significantly activated by all stick-ended trans enhancers, but not activated by all blunt-ended trans enhancers (Figure 5b). These results indicate that the variant enhancers could be used as the CRISPR-assisted trans enhancer.

Activation of genes by the GAL4/UAS-based trans enhancer
To further improve in vivo application of the CRISPR-assisted trans enhancer, we tried realizing the trans enhancer with the GAL4-UAS system. A dCas9-VP64-GAL4 expression vector and a UAS-CMV trans enhancer was constructed. Two forms of trans UAS-CMV enhancers, linear UAS-CMV (LUAS-CMV) and circular UAS-CMV (CUAS-CMV), were expected to be recruited to the target gene by the dCas9-VP64-fused GAL4 (Figure 6a). By transfecting 293 T cells with dCas9-VP64-GAL4/sgRNA-LUAS-CMV/CUAS-CMV and reporter construct, the ZsGreen expression of the exogenous reporter gene was significantly activated by both LUAS-CMV and CUAS-CMV, but not activated by all transfections as controls (Figure 6b, Figure 6-figure supplement 1). By transfecting 293T and HepG2 cells with dCas9-VP64-GAL4/sgRNA-LUAS-CMV/CUAS-CMV, the expression of endogenous HNF4a gene was highly activated in the two cells (Figure 6c). More importantly, both trans LUAS-CMV and CUAS-CMV enhancers showed significantly higher activity than the trans sCMV ( Figure 6c). In contrast, all transfections as controls did not activate the expression of endogenous HNF4a gene in the two cells (Figure 6c). These results reveal that the CRISPR-assisted trans enhancer could be better realized with the GAL4-UAS system.

Discussion
In this study, we developed a new dCas9-based gene activation strategy, the CRISPR-assisted trans enhancer, in which a trans enhancer could be recruited to target promoters by dCas9-VP64/csgRNA or dCas9-VP64-GAL4/sgRNA. The results revealed that expression of variant exogenous and endogenous genes could be highly activated by CRISPR-assisted trans enhancers in various mammalian cells, more efficiently than with current widely used dCas9-VP64 and dCas9-VPR. This strategy has unique advantages over the current dCas9-based gene activation systems.
The capture sequences of csgRNA can be easily designed. We originally designed three different capture sequences. All functioned in the CRISPR-assisted trans enhancer; however, csgRNA2 showed the best performance. The capture sequences were artificially designed short sequences, they have no complementary sequences in human cells, which is important for their specific annealing with sCMV at high efficiency. This study demonstrated that sCMV could be efficiently captured by csgRNA in the nucleus of human cells. To our knowledge, this is the first report of a gene being activated by an artificial DNA in trans.
In this study, we realized the CRISPR-assisted trans enhancer with two forms: csgRNA-sCMV and GAL4-UAS. Two forms can be easily used to activate genes in in vitro cultivated cells. As to the in vivo applications, the csgRNA-sCMV-based trans enhancer can be used via nanoparticle gene carriers, while the GAL4-UAS-based trans enhancer can be easily transferred by virus vectors such as AAV, with AAV already being approved for use as a gene vector in clinics. We found that the GAL4-UAS-based trans enhancer had better performance than the csgRNA-sCMV-based trans enhancer, especially the linear UAS-CMV. In in vivo applications, the linear UAS-CMV can be easily transferred by AAV vector.
As a typical application, dCas9-based transcriptional activators are used to reprogram cells in vitro and in vivo for biomedical applications by activating endogenous genes. For example, fibroblasts were reprogramed into induced pluripotent stem (iPS) cells by endogenous activation of the Oct4 and Sox2 genes with dCas9-SunTag-VP64 . Mouse embryonic fibroblasts were converted into neuronal cells by endogenous activation of the Brn2, Ascl1, and Myt1l genes with VP64 dCas9 VP64 (Black et al., 2016). In vivo target genes were activated by MPH to ameliorate disease phenotypes in mouse models of type I diabetes, acute kidney injury, and muscular dystrophy (Liao et al., 2017). Brain astrocytes were converted into functional neurons in vivo by activating the Ascl1, Neurog2 and Neurod1 genes with SPH (Zhou et al., 2018). These studies make CRISPR therapies the grade not the cut (Burgess, 2018).
In this study, we selected 10 endogenous genes to be activated by the CRISPR-assisted trans enhancer. Most of these genes code transcription factors, including HNF4a, E47, Ascl1, Ngn2, Sox2, Oct4, and Nanog. Ascl1, Ngn2, and Sox2 are used to directly reprogram fibroblasts into nerve cells (Zhao et al., 2015). Oct4, Sox2, and Nanog are widely used to reprogram fibroblasts into iPS cells (Takahashi et al., 2007;Takahashi and Yamanaka, 2016;Yu et al., 2007). HNF4a and E47 are used to differentiate liver and panaceas cancer cells into normal cells (Kim et al., 2015;Yin et al., 2008). TNFAIP3 is a well-known natural NF-kB inhibitor (Cooper et al., 1996), having the potential to treat NF-kB-overactivated diseases such as inflammation and cancers. Caspase9 is a key gene making cell apoptosis (Li et al., 2017). CSF3 codes granulocyte-colony stimulating factor (G-CSF), a glycoprotein that stimulates the bone marrow to produce granulocytes and stem cells and release them into bloodstream (Cetean et al., 2015), and is widely used in chemotherapy to enhance the immunity of cancer patients. We selected these genes for exploring the future in vitro and in vivo applications of the CRISPR-assisted trans enhancer.
The CMV enhancer fragment was amplified from pEGFP-N1 using primers CMV-F and CMV-1-R/ CMV-2-R/CMV-3-R. The PCR products were purified with PCR clean kit and used as linear blunt-end CMV (bCMV). To prepare stick-end CMV (sCMV), the PCR products were firstly digested with Nt. BbvCI and then added with complementary oligonucleotide CS-1-R/CS-2-R/CS-3-R. The PCR products were denatured at 85˚C for 10 min and then naturally cooled to room temperature. The PCR products was purified with PCR clean kit and used as linear sCMV. The blunt-end EF1a/PGK promoters were all amplified from pEF1a-FB-dCas9-puro (Addgene) using primers EF1-a-F/R and PGK-F/R. The stick-end EF1a/PGK promoters were similarly prepared by treating blunt-end EF1a/PGK promoters.
The sequences of all PCR primers used in the above vector construction are shown in the Supplementary file 1- DNA cutting with Cas9-csgRNA A sgRNA targeting the HNF4a promoter sequence was selected. The sgRNAs were prepared by in vitro transcription using T7 RNA polymerase (NEB). The sgRNA transcription template was amplified from pEASY-csgRNA using primers HNF4a-T7-F and U6-R/U6-1-R/U6-2-R/U6-3-R. A normal sgRNA (HNF4a-sgRNA) and three csgRNAs (HNF4a-csgRNAs) were prepared. A 732 bp HNF4a promoter fragment was amplified from pEZX-HP-ZsGreen using primers HNF4a-sP-F and HNF4a-sP-R. The sequences of PCR primers are shown in the Supplementary file 1- Table 1. The Cas9 digestion reaction (30 mL) consisted of 1 Â Cas9 Nuclease Reaction Buffer, 1 mM Cas9 Nuclease (NEB), and 300 nM HNF4a-sgRNA or HNF4a-csgRNA. The reaction was incubated at 25˚C for 10 min. Then 400 ng of purified 732 bp HNF4a promoter fragment was added to the reaction and incubated at 37˚C for 15 min. Finally, the Cas9 nuclease was inactivated at 65˚C for 10 min. The reaction was run with 1.5% agarose gel electrophoresis.

Cell lines
All cells including 293T, HepG2, PANC1, A549, HeLa, SKOV3, and HT29 were obtained from the Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. The identity was authenticated by STR profiling. Mycoplasma contamination testing was performed and no mycoplasma contamination was ensured.

Flow cytometry
The fluorescence intensity of cells was quantified with flow cytometry (Calibur). Ten-thousand cells were measured for each transfection. Flow cytometry data analysis and figure preparation were done using BD software.

Quantitative PCR
The total RNA was extracted from cells using TRIzol (Invitrogen). The complementary DNA (cDNA) was synthesized with 3 mg of total RNA using the Hifair III SuperMix (Yeasen). Gene transcription was detected with quantitative PCR (qPCR) using the Hieff qPCR SYBR Green Master Mix (Yeasen) according to the manufacturer's instructions. GAPDH was used as an internal reference to analyze the relative mRNA expression of different genes. The sequences of PCR primers are shown in the Supplementary file 1- Table 3 and 4. The qPCR programs were run on StepOne Plus (Applied Biosystems). Each qPCR detection was performed in at least three technical replicates. Melting curve analysis was performed. Data analysis was performed using the Applied Biosystems StepOne software v2.3, and C t values were normalized with that of GAPDH. The relative expression level of target mRNAs was calculated as relative quantity (RQ) according to the equation: RQ = 2 -DDCt .

Statistical analyses
Each cell transfection for detecting gene expression activation by trans enhancer was performed in three biological replicates. In each biological replicate, at least three technical replicates (three replicate wells) were performed. In qPCR detection of gene expression, the mean RQ value of technical replicates was used as the RQ value of one biological replicate. The mean RQ value of three biological replicates was used to calculate the final mean and standard deviation (SD). Data were analyzed by Student's t test when comparing two groups. Data are shown as mean ± SD, and differences were considered significant at p<0.05.