Original Article

Sequence-Modified Antibiotic Resistance GenesProvide Sustained Plasmid-Mediated TransgeneExpression in MammalsJiamiao Lu,1,2 Feijie Zhang,1,2 Andrew Z. Fire,2,3 and Mark A. Kay1,2

1Department of Pediatrics, Stanford University School of Medicine, Palo Alto, CA 94305, USA; 2Department of Genetics, Stanford University School of Medicine, Palo Alto,

CA 94305, USA; 3Department of Pathology, Stanford University School of Medicine, Palo Alto, CA 94305, USA

Conventional plasmid vectors are incapable of achieving sus-tained levels of transgene expression in vivo even in quiescentmammalian tissues because the transgene expression cassetteis silenced. Transcriptional silencing results from the presenceof the bacterial plasmid backbone or virtually any DNAsequence of >1 kb in length placed outside of the expressioncassette. Here, we show that transcriptional silencing can besubstantially forestalled by increasing the An/Tn sequencecomposition in the plasmid bacterial backbone. Increasingnumbers of An/Tn sequences increased sustained transcriptionof both backbone sequences and adjacent expression cassettes.In order to recapitulate these expression profiles in compactand portable plasmid DNA backbones, we engineered thestandard kanamycin or ampicillin antibiotic resistance genes,optimizing the number of An/Tn sequence without alteringthe encoded amino acids. The resulting vector backbones yieldsustained transgene expression from mouse liver, providinggeneric DNA vectors capable of sustained transgene expressionwithout additional genes or mammalian regulatory elements.

Received 30 August 2016; accepted 2 March 2017;http://dx.doi.org/10.1016/j.ymthe.2017.03.003.

Correspondence: Mark A. Kay, Departments of Pediatrics and Genetics, StanfordUniversity School of Medicine, Palo Alto, CA 94305, USA.E-mail: markay@stanford.edu

INTRODUCTIONGeneric plasmid DNA vectors containing a classical promoter, codingsequence, and polyadenylation signal (polyA) are often used for ex-pressing therapeutic, reporter, or proteins of interest in amyriad of cellsand tissues. Even in quiescent tissues in which plasmid DNA levelsremain static, transgene expression is transcriptionally extinguishedover a period of several weeks.We have previously referred and definedthis as transcription silencing.1 Having explored the mechanism oftranscriptional silencing in themouse liver following hydrodynamic in-jection, we have found that CpG content and DNA methylation havelittle or no influence in transcription silencing.2 Instead, the length ofthe DNA outside of the eukaryotic expression cassette (between the50 end of the promoter and the 30 end of the polyA site) was determinedto be a critical factor that dictates transcriptional silencing in this sys-tem.1,3 Silencing begins to be observedwith�1 kb ormore of backboneDNA, even if the bacterial plasmid DNA sequences are replaced withrandom DNA sequences.1 These observations led to the productionof alternative plasmid vectors commonly referred to as minicircleDNAs (MCs)4–9 (devoid of a bacterial backbone) or mini-intronicplasmid DNAs (MIPs),3 in which the bacterial components required

Molecular Therapy Vol. 25 No 5 May

for plasmid propagation in bacteria are engineered into themammalianexpression cassette. However, each of these products requires addi-tional steps and/or specific reagents for their production.

Previous work has shown nucleosomal patterns to be a key element inthe establishment and maintenance of silenced chromatin.10 A set ofAn/Tn sequence motifs is known to substantially modify nucleosomalpositioning and mobility.11 The An/Tn tracts are thought to decreasethe structural stability to the basic 147-bp nucleosome core wrappedaround a histone octamer, which allows for greater accessibilityof RNA polymerase II (RNAPII) complexes.11–17 Such sequences,sometimes referred to as nucleosome exclusion but perhaps moreappropriately as nucleosome relaxation or loosing, remain a mysteryin terms of their mode of action.

Certain sequences with a high content of An/Tn segments have beenshown to counteract silencing in an invertebrate system,18 althoughwe note that the particular sequences used in the nematode systeminvolved shorter An/Tn segments and a phased occurrence ratherthan the longer An/Tn runs described as exclusionary in vertebrateand yeast systems.

As part of our efforts to further explore plasmid-mediated transgenesilencing,we asked if the addition of shortAn/Tn tractswould influenceplasmid-mediated transgene expression. Ultimately, we used the resultsof these experiments to design sequence-modified bacterial antibioticresistance genes to abrogate transcriptional silencing, providing pro-longed transgene expression from canonical plasmid vectors in vivo.

RESULTSIncreasing An/Tn Sequence Composition in Plasmid Bacterial

Backbone Abrogated Plasmid-Mediated Transcriptional

Silencing In Vivo

In order to establish how the sequences outside of the canonicaleukaryotic expression cassette affect plasmid-mediated transgene

2017 ª 2017 The American Society of Gene and Cell Therapy. 1187

Molecular Therapy

expression, we replaced the bacterial plasmid backbone with a 2.2 kbrandom DNA (RD) fragment using minicircle DNA technology (Fig-ure 1). To study how the An/Tn motifs affected transgene expression,we cloned a stretch of 20 “T” nucleotides into every 60 bp of the RDsequence (RD-An/Tn) (Figure S1) and then compared these vectorsfor transgene expression in a mouse liver.

Plasmids consisting of the human alpha 1-antitrypsin (hAAT) cDNAdriven by the Rous sarcoma virus (RSV) promoter containing the RDor RD-An/Tn fragment cloned in either direction were delivered intothe livers of 6- to 8-week-old C57BL/6J mice by hydrodynamic trans-fection (Figures 1A and 1B). A conventional plasmid vector (pRHB)and minicircle vector (MC.RHB) were used as the silencing and non-silencing controls, respectively. Transgene expression monitored byserial serum hAATmeasurements is shown in Figure 1B.We followedtransgene expression levels over time. Generally, by 3 to 5 weeks afterinfusion, the steady-state expression level is observed. To ensuresteady-state conditions, we used the 2-month time point (day 63 orlater if the results for day 63 were not available) for statistical compar-isons (Figure 1B). The RD backbone containing and canonical plas-mids were quickly silenced, whereas the RD-An/Tn backbone con-taining and canonical minicircles resulted in sustained expressionlevels. We found that the effect on transgene expression was indepen-dent of the orientation of the backbone DNA (Figures 1A and 1B) andthat even when two copies of the RD-An/Tn segment (4.4 kb) wereused to replace the plasmid backbone, the transgene expression levelswere maintained (Figures 1C and 1D). To establish these results werenot dependent on the expression cassette, the RD-An/Tn backbonewas also tested in a human Factor IX (hFIX) reporter system drivenby the EF1-alpha promoter. Consistent with the previous results,the hFIX expression was sustained at high levels in minicircles con-taining the RD-An/Tn backbone (Figures 1E and 1F).

Plasmid Vectors Carrying An/Tn-Sequence-Modified Antibiotic

Resistance Genes Provide Sustained Plasmid-Mediated

Transgene Expression In Vivo

We hypothesized that perhaps optimizing the An/Tn content of ca-nonical plasmids might result in vectors providing the same benefitof sustained transgene expression. Therefore, we engineered two anti-biotic resistance genes commonly used in plasmid propagation vec-tors for maximal An/Tn content. The following rules were followedwhile modifying the sequences: (1) select the codons that incorporatemore “An/Tn” but maintain the same amino acid sequence; (2) createas many “An/Tn” tracts as possible; (3) avoid RNAPII pause orcleavage/polyA-addition sequences, such as “AATAAA,” “TTTATT,”“AATAA,” or “TTATT”; and (4) avoid the use of rare codons. About30% of the kanamycin resistance gene bases were modified (J-Kan).After modification, the GC content dropped from 60.6% in wildtype to 39.0% in J-Kan (Figure S2A). The codon adaptation index(CAI), a measure of codon usage bias, remained the same (Fig-ure S2B). Similarly, for the ampicillin resistance gene, the GC contentdropped from 49.8% in wild type to 34.8% in J-Amp, with the CAIremaining nearly constant (0.67 from 0.63) after modification (Fig-ures S3A and S3B).

1188 Molecular Therapy Vol. 25 No 5 May 2017

The plasmid vector with J-Kan sequences was compared in parallelwith the plasmid vector containing the wild-type kanamycin resis-tance gene (Kan) sequences in animals for their abilities to sustaintransgene expression. The incorporation of J-Kan sequences into theplasmid backbone resulted in enhanced duration of hAAT transgeneexpression in vivo to a level similar to that expressed from the corre-sponding minicircle vector and about six times higher than that ex-pressed from a plasmid with wild-type Kan (Figures 2A and 2B). Like-wise, a plasmid (pRHB.pUC.J-Amp.J-Kan) with a modified “An/Tn”rich ampicillin resistance gene (J-Amp) was also able to maintaintransgene expression to the same level as the minicircle vector andenhanced transgene expression to about ten times higher comparedto the corresponding plasmid containing the wild-type ampicillinresistance gene pRHB.pUC.Amp.J-Kan (Figures 2C and 2D).

The modified antibiotic resistance genes did not alter the ability of thecorresponding plasmids to be propagated in standard bacterial cul-ture (Figure 2E). To make the modified plasmid suited for routinecloning of any expression cassette, we constructed universal cloningvectors (p.pUC.J-Kan and p.pUC.J-Kan.J-Amp) containing the mul-tiple cloning sites (MCSs) (Figure 2F).

Abrogation of Transcriptional Silencing by An/Tn Sequences Is

Associated with Enhanced Plasmid Backbone Transcriptional

Activity

The fact that nucleosome relaxation sequences allow for persistentexpression led us to hypothesize that maintenance of transcriptionmight require RNAPII activity to be maintained on the DNA plasmidtemplates. Recent studies have provided evidence of widespreaddivergent transcription at protein-encoding gene promoters.19 Ithas been proposed that transcription factors must first nucleate asense-oriented preinitiation complex at the transcription start site(TSS) but can initiate both sense and antisense-oriented transcrip-tion.19 The RNAPII pauses at the AAUAAA hexamer in the polyAsignal.20,21 RNAPII does not fall off the DNA template immediatelyupon reaching the hexamer. In many situations, RNAPII moleculespresent well over 1 kb downstream of the hexamer and continue toexhibit delayed progress down the template in vivo.20 One possiblesource for a sustainable transcriptional state on the plasmid DNAwould be a condition in which transcription would continue pastthe canonical expression cassette into the backbone, with thecontinuing transcription serving to keep the plasmid in an open(and accessible) chromatin configuration. To explore the distributionof plasmid-derived RNA transcripts within the plasmid sequence, weused reverse transcription and real-time PCR (RT-qPCR) to comparethe entire transcription activities of silencing and non-silencingplasmid vectors.

In Figure 3A, separate primer pairs were designed to detect tran-scripts from the expression cassette and backbone regions of thepRHB.pUC.Kan and pRHB.pUC.J-Kan vectors. These two vectorswere infused into the livers of 6- to 8-week-old C57BL/6J, and totalliver RNA samples were extracted on day 1 and day 21 after infusion(Figure 3B). As shown in Figure 3C, and consistent with protein

Figure 1. Effects of RD-An/Tn Sequence in Maintaining

Transgene Expression In Vivo

(A) Schematic of the hAAT DNA plasmid constructs. The RD-

An/Tn segment was inserted as the backbone in either

forward (MC.RHB.2.2kb F-RD-An/Tn) or backward orientation

(MC.RHB.2.2kb B-RD-An/Tn). (B) Serum hAAT levels measured

by ELISA at various time points after equimolar infusion of the DNA

vectors into the mouse liver shown in (A) (n = 5/group). Error bars

represent the SD in all figures. Serum hAAT levels on day 84

post infusion were used to calculate p values. The p value of each

group was calculated by comparing with the non-silencing control

group (MC.RHB). If p < 0.005, the tested group is considered as

silenced. Otherwise the tested group is considered as non-

silenced. For the pRHB group, p = 0.002. For the MC.RHB.2.2kb

F-RD-An/Tn group, p = 0.06. For the MC.RHB.2.2kb B-RD-An/Tn

group, p = 0.2. For the MC.RHB.2.2kb RD group, p = 0.0008. (C)

Schematic of vectors used for the experiments shown in (D). Two

copies of 2.2-kb RD-An/Tn or RD sequences were placed after

the bpA sequence as the backbone. (D) Serum hAAT levels after

equimolar transfection of vectors presented in (C). Serum hAAT

levels on day 63 post infusion were used to calculate p values. The

MC.RHB group was used as the non-silencing control. For the

pRHB group, p = 0.002. For MC.RHB.4.4kb RD-An/Tn, p = 0.3.

For MC.RHB.4.4kb RD, p = 0.002. (E) Schematic representation

of the DNA constructs expressing hFIX. 2.2-kb RD-An/Tn or

2.2-kb RD sequence was used as backbone in these constructs.

(F) The same molar amounts of DNA constructs pictured in (E)

were infused into mice, and the plasma hFIX levels were

measured by ELISA at various time points. Serum hFIX levels on

day 63 post infusion were used to calculate p values. The

MC.EF1alpha.hFIX.hGpA group was used as non-silencing

control. For the pEF1alpha.hFIX.hGHpA group, p = 0.0001. For

MC.EF1alpha.hFIX.hGHpA.2.2kb RD, p = 0.0001. For the

MC.EF1alpha.hFIX.hGHpA.2.2kb RD-An/Tn group, p = 0.007

(hAAT and hFIX have been defined in the text; hGHpA, human

growth hormone polyA; bpA, bovine growth hormone polyA).

www.moleculartherapy.org

Molecular Therapy Vol. 25 No 5 May 2017 1189

Figure 2. Effect of Sequence-Modified Antibiotic Resistance Genes on

Transgene Expression

(A) Schematic of plasmid vectors used for infusion pictured in (B). (B) Serum hAAT

levels at various time points after equimolar infusion into the mouse liver. Serum

hAAT levels on day 63 post infusion were used to calculate p values. The MC.RHB

group was used as the non-silencing control. For the pRHB.pUC.J-Kan group,

p = 0.06. For the pRHB.pUC.Kan group, p = 0.0001. (C) Schematic of plasmid

vectors with either wild-type Ampicillin resistance gene (Amp) or modified Amplicillin

(J-Amp) as part of the plasmid backbone sequence. (D) Serum levels of transgene

hAAT at various time points after the infusion of plasmids pictured in (C). Serum

hAAT levels on day 77 post infusion were used to calculate p values. The MC.RHB

group was used as the non-silencing control. For the pRHB.pUC.Amp.J-Kan

group, p = 0.0049. For pRHB.pUC.J-Amp.J-Kan, p = 0.3. (E) The yield of plasmid

vectors with modified antibiotic resistance genes as part of the plasmid backbone

Molecular Therapy

1190 Molecular Therapy Vol. 25 No 5 May 2017

expression (Figure 3B), substantially higher hAATmRNA concentra-tions from pRHB.pUC.J-Kan (non-silenced) versus pRHB.pUC.Kan(silenced) infused animals was observed on day 21 after infusion.Interestingly, both sense and antisense transcripts originating fromthe plasmid backbone regions were present. The non-silencing vectorbackbone (pRHB.pUC.J-Kan) generated significantly higher tran-scription activities at both sense and antisense orientations whencompared with transcription from the silencing vector backbone(pRHB.pUC.Kan) on day 21 (Figures 3D and 3E).

The silencing MC.RHB.2.2kb RD and non-silencing MC.RHB.2.2kbRD-An/Tn vectors gave the same pattern of sense and antisense tran-scripts from the expression cassette and plasmid backbone (Fig-ure S4), as did the pRHB.pUC.Kan and pRHB.pUC.J-Kan vectors(Figure 3), respectively. This further supports the idea that the substi-tution of J-Kan for Kan overcomes silencing by the same mechanismas including nucleosome-relaxation sequences in the plasmid DNAbackbone.

Incorporation of RNAPII Arrest Sites in Plasmid Backbone

Silenced the Transgene Expression

To further provide support for the transcriptional dependence of theplasmid backbone sequences on persistence of transgene expression,we asked whether transcriptional pausing could alter expression pro-files from the non-silenced plasmids. Transcription elongation byRNAPII does not proceed uniformly, and polymerase pausing isthought to be a mechanism to regulate transcription. RNAPII arrestsites identified by studies performed in cell-free systems are primarilylocalized across the coding region. Pause sites block a fraction oftranscribing polymerases, making them unable to continue RNA syn-thesis.22 In addition, RNAPII arrest sites function in an orientation-dependent manner.23,24 Because pause sites are frequently locatedin the coding sequence, it has been difficult to manipulate and studythem in living cells. In this system, we have the ability to place thesesequences into plasmid backbones for in vivo study. We hypothesizedthat if the plasmid backbone transcription activities influence theexpression of transgene, the incorporation of an RNAPII arrest sitein the plasmid backbone of non-silencing vectors should block back-bone transcription, causing silencing of the transgene expressioncassette. To test this hypothesis, we incorporated the mammalian his-tone H3.3 RNAPII arrest site “TTTTTTTCCCTTTTTT”25 into thenon-silencing backbone sequence RD-An/Tn. This RNAPII arrestsite was chosen because it contains a Tn sequence that is consistentwith the RD-An/Tn design used in our experiments described above.

Because RNAPII arrest sites function in an orientation-dependentmanner, one copy of the H3.3 RNAPII arrest site was first

sequence and conventional plasmid vectors (n = 4/vector). The yield was derived

from quadruplicate 100-mL overnight cultures. The following formula was used

to convert the yield of mg/L to mol/L: mol/L = [yield (mg/L)� 10E�3 g/L]/[size (kb)�1,000 � 330 � 2 g/mol], where 330 is the average molecular weight of dNTP. (F)

Structures of the ready-to-use cloning vector with sequence-modified antibiotic

resistance genes.

Figure 3. Detection of Plasmid Backbone Transcription fromBoth

Sense and Antisense Orientation

(A) Schematic of primer designs used for RT and qPCR experiments to

detect transcripts from the expression cassette and backbone regions.

The actin transcript was used as a standardization control. (B) Serum

levels of the hAAT product on days 1, 3, and 21 after infusion of

pRHB.pUC.Kan and pRHB.pUC.J-Kan plasmids (n = 6/group). (C) The

hAAT transcripts detected from RNA samples extracted from animals

shown in (B) on day 1 and day 21 after infusion. Primers used for the RT

and qPCR experiment were shown in (A). (D) Short and long backbone

transcripts in the sense orientation detected by using primers listed in (A)

from RNA samples extracted on day 1 and day 21 after infusion (SS,

short sense; LS, long sense). (E) Short and long backbone transcripts in

the antisense orientation are detected by using primers shown in (A) from

RNA samples extracted on day 1 and day 21 after infusion (SA, short

antisense; LA, long antisense). Three liver RNA samples from each group

were extracted on day 1 and day 21. Each sample was tested with qPCR

in duplicate. Error bars represent standard deviation of six qPCR test

tubes of each sample.

www.moleculartherapy.org

Molecular Therapy Vol. 25 No 5 May 2017 1191

Figure 4. Influence of Incorporating the H3.3 RNAPII Arrest Site in the Backbone on Transgene Expression

(A) Schematic of plasmid vectors used for the infusion pictured in (B). The arrow indicated the orientation of transcription that the RNAPII arrest site was designed to block

(antisense). The short line indicated the location of the H3.3 RNAPII arrest site. The number indicated the distance of the RNAPII arrest site from the RSV promoter. (B) Serum

hAAT levels at various time points after equimolar transfection into the mouse liver (n = 5/group). Serum hAAT levels on day 63 post infusion were used to calculate p values.

The MC.RHB.2.2kb RD-An/Tn group was used as the non-silencing control. For the pRHB.pUC.Kan group, p = 0.0002. For the MC.RHB.2.2kb RD-An/Tn Antisense 0.5kb

group, p = 0.66. For theMC.RHB.2.2kb RD-An/Tn Antisense 1.1kb group, p = 0.004. For theMC.RHB.2.2kb RD-An/Tn Antisense 1.5kb group, p = 0.0003. (C) Schematic of

plasmid vectors used for injection pictured in (D). Each copy of the H3.3 RNAPII arrest site was incorporated into two locations on the backbone (0.7 kb and 1.7 kb away from

bpA) at the sense orientation. (D) Serum hAAT levels at various time points after equimolar transfection into the mouse liver (n = 5/group). Serum hAAT levels on day 91 post

infusion were used to calculate p values. The MC.RHB.2.2kb RD-An/Tn group was used as the non-silencing control. For the pRHB.pUC.Kan group, p = 0.0002. For the

MC.RHB.2.2kb RD-An/Tn Sense 0.7kb+1.7kb group, p = 0.0002.

Molecular Therapy

incorporated at different locations in the backbone at the antisenseorientation (Figure 4A) to block backbone transcription in theantisense orientation and tested the expression profiles of thesevectors in mice. We constructed vectors with this site at various dis-

1192 Molecular Therapy Vol. 25 No 5 May 2017

tances from the RSV promoter. “MC.RHB.2.2kb RD-An/Tn Anti-sense 0.5kb,” “MC.RHB.2.2kb RD-An/Tn Antisense 1.1kb,” and“MC.RHB.2.2kb RD-An/Tn Antisense 1.5kb” had one copy of theH3.3 RNAPII arrest site incorporated into the backbone at the

Figure 5. Effect of Additional bpA Signal Inserted into

the Plasmid Backbone on Transgene Expression

(A) Schematic of plasmid constructs infused into animals for

experiment pictured in (B). The arrow indicated the tran-

scription orientation that the additional bpA signal was de-

signed to block. The short line indicated the location of the

additional bpA signal. The number indicated the distance of

additional bpA signal from the expression cassette end at

the same orientation. (B) Serum hAAT levels at various time

points after equimolar transfection into the mouse liver

(n = 5/group). Serum hAAT levels on day 63 post infusion

were used to calculate p values. The MC.RHB.2.2kb RD-

An/Tn group was used as the non-silencing control. For the

pRHB.pUC.Kan group, p = 0.0001. For the MC.RHB.2.2kb

RD-An/Tn Sense bpA 0.5kb group, p = 0.0001. For

the MC.RHB.2.2kb RD-An/Tn Antisense bpA 0.5kb group,

p = 0.0001.

www.moleculartherapy.org

antisense orientation at 0.5 kb, 1.1 kb, and 1.5 kb away from the RSVpromoter, respectively. As shown by protein expression in Figure 4B,transgene silencing was observed once the arrest site was placedfurther away from the promoter (1.1 kb and 1.5 kb) in the antisenseorientation. Thus, the H3.3 RNAPII arrest site at certain positions inthe backbone is capable of blocking backbone transcription andsilencing transgene expression.

When one copy of the same RNAPII arrest site was incorporated intovarious locations of the backbone at the sense orientation, the trans-gene expression profile was not affected (Figures S5A and S5B). How-ever, when two copies of the H3.3 RNAPII arrest site were incorpo-rated into the MC.RHB.2.2kb RD-An/Tn backbone in the senseorientation (0.7 kb and 1.7 kb away from polyA) (Figure 4C), trans-gene expression was silenced (Figure 4D). Nevertheless, pausing ofantisense backbone transcription had a more substantial effect ontransgene silencing, indicating weaker backbone transcription atthe antisense orientation than at the sense orientation. Establishingthis result was not just due to the specific sequences; a second An/Tn-rich c-Myc RNAPII arrest site, “TTTTAATTTATTTTTTTAT,”was also tested and found to give similar results (Figures S5Cand S5D). These results support the idea that transcription inboth directions may be important for long-term maintenance ofexpression.26–28

Finally, because polyA signals are efficient in inducing an RNAPIIpause, one copy of 230-bp-long bpA was inserted into the

RD-An/Tn backbone at each orientation(Figure 5A). As expected, that additionalcopy of bpA signal at either orientation inthe RD-An/Tn backbone was sufficient tosilence transgene expression (Figure 5B). Takentogether, our data are consistent with theconcept that maintaining non-canonical tran-scription from regions outside of the transgene

expression cassette may indeed be required to maintain persistentexpression.

DISCUSSIONThe persistence of episomal vector-mediated transgene expression isdictated by a number of parameters. In the case of plasmid-mediatedgene transfer, transgene expression is dependent on the length of theplasmid backbone makeup and, in some cases, the CpG content. Ca-nonical plasmids have a high CpG content, which can trigger innateimmune responses that promote the loss of the vector DNA.2,29–32

However, the CpG-mediated loss of transgene expression is onlyobserved when lipid-based vehicles are used for DNA delivery.2,33

The hydrodynamic transfection without lipids used in this workavoids the innate response.2,33 Consistent with the expectation of aCpG-independent foreign DNA response and of identifying a meanstoward its mitigation, in the present study, we compared plasmidswith or without AT-rich RNAPII arrest sites in their backboneregions (Figures 4, S5C, and S5D). These plasmids have identicalCpG content. However, the plasmids with RNAPII arrest sites in theirbackbone silenced transgene, whereas the plasmids without RNAPIIarrest sites were able to sustain transgene expression. These resultsare consistent with our previous findings that CpG content has littleor no influence in transcription silencing.2

From our previous study, we demonstrated that the length andnot the sequence of the intervening DNA contained outside of theexpression cassette was responsible for plasmid-mediated transgene

Molecular Therapy Vol. 25 No 5 May 2017 1193

Figure 6. Proposed Model That the Ability to Maintain RNAPII on the DNA Sequences Outside of the Expression Cassette Is the Primary Factor That

Determines Transgene Persistence

(A) Schematic of transgene silencing model for the silenced plasmids. When the plasmid backbone is long enough to sufficiently recruit enough numbers of nucleosomes (left

panel) or contains RNAPII pause signals (middle panel), the backbone transcription is blocked. This consequently prevents an efficient reinitiation of transcription in the

plasmid expression cassette region and causes plasmid silencing (right panel). (B) Schematic of transgene persisting model for the non-silenced plasmids.When the plasmid

backbone is short enough to avoid recruiting a sufficient amount of nucleosomes (left panel) or contains An/Tn rich sequences (middle panel), the backbone transcription is

sustained. Therefore, RNAPII is able to be reinitiated efficiently in the plasmid expression cassette region to maintain persistent transcription (right panel).

Molecular Therapy

silencing.1 Our experiments here show that including a high propor-tion of An/Tn in the plasmid bacterial backbone allows for the main-tenance of transgene transcription, and this is associated withenhanced backbone transcripts. Because A-T base pairs have fewerhydrogen bonds than G-C base pairs, the An/Tn rich region requiresfewer forces for unzipping the double-stranded DNA.34 Such featurescan enhance the AT-rich region in bacterial replication origin to formreplication complexes.35 In terms of plasmid backbone transcription,a possible source of the observations would also come from an abilityof An/Tn sequences to make the nucleosome unwinding easier forRNAPII through the backbone, permitting the RNAPII complex toremain on plasmid and continue to re-initiate transcription (Figure 6).In the absence of sufficient An/Tn sequences, RNAPII could stochas-tically stall or pause, eventually falling off the plasmid DNA andleading to an unprotected DNA that could be silenced. We proposea model whereby the ability to maintain RNAPII on the DNA se-quences outside of the expression cassette is the primary factor that

1194 Molecular Therapy Vol. 25 No 5 May 2017

determines transgene persistence (Figure 6). If the DNA is unfavor-able and RNAPII engagement is lost, then the plasmid vectors takeon a “heterochromatic” chromatin state.36 Expression of basally tran-scribed plasmid vectors would then persist under conditions in whicheuchromatic chromatin is the default state for transcribed DNA.However, further studies are required to establish the stepwisemechanism and whether the degree of RNAPII occupancy based onthe backbone An/Tn sequence composition is the direct cause ofplasmid-mediated transcriptional silencing.

The pEPI-1 vector that contains a chromosomal scaffold/matrix-attached region (S/MAR) has been shown to have a major influenceon transcriptional rate and might influence plasmid DNA replicationwhen tightly associated with the nuclear matrix.37–39 The matrix-associated regions (MARs) are defined as AT-rich DNA sequencesthat are preferentially retained by the nuclear matrix.40,41 This raisesa question of whether the An/Tn-rich backbone sequences would

www.moleculartherapy.org

serve as MAR elements. Arguing against this as the primarymechanism are observations in which the insertion of an RNAPIIarrest site resulted in plasmids that are identical, except for theinsertion of this site (RD-An/Tn backbone with and without theRNAPII arrest site; Figures 4, S5C, and S5D). The plasmid with anRNAPII arrest site has an even higher content of An/Tn se-quences, and both of these backbones should have the potentialto serve as MARs. However, our results demonstrate that theRD-An/Tn backbone with RNAPII arrest site silences transgeneexpression (Figures 4, S5C, and S5D). These data point away froma single protective element (such as a MAR) in sustaining plasmidfunction in this system, leading us to the working model of contin-uous baseline transcription as a major determinant for sustainedactivity.

To be suitable in molecular engineering applications, plasmid vectorswould need to be stable for amplification in bacterial hosts. Althoughcertain unusual sequences can affect growth or integrity in E. coli(such as GT-rich “recombinant islands” that locally increase theRecA-mediated DNA recombination42), we found no evidence thatthe modified antibiotic resistance genes (J-Kan and J-Amp) describedin this study had any such effects. Indeed, experiences in construction,growth, yield, and homogeneity of the expression clones based on theJ-Kan and J-Amp vectors were comparable to our extensive experi-ences with the parental pUC vectors.

In our study, all vectors were infused into C57BL/6J mice through hy-drodynamic tail vein injection. The data collected through our exper-iments thus only represent plasmid expression levels from the mouselivers. Although additional testing in other tissues will be required forevaluation of suitability for diverse tissue types, there is a precedentfor generalization of anti-silencing effects beyond liver and hydrody-namic injection.43–45 Importantly, we applied our mechanistic find-ings to make a canonical plasmid that is resilient to transcriptionalsilencing. Our previous approaches to enhance transgene expressionused minicircle4 or MIP3 plasmid vector technology. Although eachof these have their advantages, the fact that special DNAs, bacterialcell lines, and/or purification schemes were required limited theiruse by some laboratories as well as limited their production inlarge-scale manufacturing protocols. The reagents described in thisstudy will be beneficial tomany in order to optimize transgene expres-sion studies.

MATERIALS AND METHODSMaterials

The RD-An/Tn and modified J-Kan and J-Amp sequences were syn-thesized by Thermo Fisher Scientific. A SacI site and an XbaI site wereincorporated to the 50 and 30 sequences of J-Kan, respectively. AnXbaI site was incorporated to both the 50 and 30 sequences of J-Amp.

Vector Construction

pRHB andMC.RHBwere previously described.3,4 To generate vectorsusing synthesized RD-An/Tn sequence and Random DNA sequenceas backbone, minicircle vectors MC.RHB.2.2kb RD-An/Tn and

MC.RHB.2.2kb RD were created. Because neither RD-An/Tn norRD is able to support DNA replication and selection in bacterialhost strains, conventional plasmid vectors are not suitable for ourpurpose. The RD-An/Tn sequence was inserted into the SpeI site ofminicircle MC.RHB, producing parental plasmid on both orienta-tions. The resulting minicircle parental plasmids were further pre-pared into minicircle vectors MC.RHB.2.2kb F-RD-An/Tn (forward)and MC.RHB.2.2kb B-RD-An/Tn (backward). The RD sequence wasalso inserted into the SpeI site of minicircle MC.RHB, producingparental plasmid to produce MC.RHB.2.2kb RD. In order to buildthe minicircle vector carrying 4.4 kb of RD-An/Tn as the backbone,an additional copy of 2.2 kb of RD-An/Tn sequence was insertedinto the XbaI site of the MC.RHB.2.2kb RD-An/Tn parental plasmid.The same restriction enzymewas used to insert a second copy of 2.2 kbof RD to generate the MC.RHB.4.4kb RD vector. To generate thepRHB.pUC.J-Kan vector, the pRHB was digested by SacI and XbaIto remove the wild-type kanamycin resistance gene sequence, followedby ligating to the SacI-XbaI double-digested synthesized J-Kanfragment. To build the pRHB.pUC.Amp.J-Kan vector, the XbaI sitecontaining forward primer 50 ATATATTCTAGAATGAGTATTCAACATTTCCGTG 30 and the XbaI site containing backward primer50 ATATATTCTAGATTACCAATGCTTAATCAGTGAG 30 wereused to amplify the wild-type ampicillin resistance gene sequencethrough PCR. This PCR product was further treated with XbaI, andthen ligated with the XbaI-digested pRHB.pUC.J-Kan vector. ThepRHB.pUC.J-Amp.J-Kan vector was constructed through ligatingtheXbaI-digested pRHB.pUC.J-Kan vector andXbaI-digested synthe-sized J-Amp fragment. The orientation of insertion was confirmedthrough DNA sequencing. The RNAPII arrest site was inserted intoRD-An/Tn backbone through site-directed mutagenesis by usingQuikChange II XL Site-Directed Mutagenesis Kit (Agilent Tech-nologies). The following primers were used. The forward primer50 GATTCCACCGGGCGCAGCAAAAAAGGGAAAAAAAGTTCCTCTCGAGCGGCGG 30 and the backward primer 50 CCGCCGCTCGAGAGGAACTTTTTTTCCCTTTTTTGCTGCGCCCGGTGGAATC 30 were used to generate MC.RHB.2.2kb RD-An/Tn Anti-sense 0.5kb. The forward primer 50 GCCAGGCAGGAGTCTCAAAAAAAAGGGAAAAAAAGTTCTCGCGAGTATCTGC 30 and thebackward primer 50 GCAGATACTCGCGAGAACTTTTTTTCCCTTTTTTTTGAGACTCCTGCCTGGC 30 were used to constructMC.RHB.2.2kb RD-An/Tn Antisense 1.1kb. The forward primer 50

CTTGAATTCTTCTATTCCAAAAAAGGGAAAAAAACGCCGCGGGATGAGGCCG 30 and the backward primer 50 CGGCCTCATCCCGCGGCGTTTTTTTCCCTTTTTTGGAATAGAAGAATTCAAG 30 were used to build MC.RHB.2.2kb RD-An/Tn Antisense1.5kb.

Animal Studies

6- to 8-week-old female C57BL/6J mice purchased from Jackson Lab-oratory were used for DNA or rAAV vector infusion. To ensure thesame molar amount of DNA was injected into each animal,3.63 mg/kb DNA was used for various sized constructs. Each DNAconstruct was diluted into 1.8 mL of 0.9% NaCl for each animal,and was delivered through hydrodynamic tail vein injection. After

Molecular Therapy Vol. 25 No 5 May 2017 1195

Molecular Therapy

infusion, blood samples were collected periodically by a retro-orbitaltechnique. The serum hAAT and the plasma hFIX levels were quan-tified by ELISA measurements. All animal studies performed withinthis paper were oversighted by the Stanford review board.

ELISA

ELISA for hAAT and hFIX was performed as previously described46

with the following antibodies: goat anti-human alpha-1 antitrypsinprimary antibody at 1:1,000 (Strategic Biosolutions S0311G000-S4),goat pAB to alpha-1 antitrypsin (HRP) (Abcam ab7635), mouseanti-human F9 IgG primary antibody at 1:1,000 (Sigma F2645),and polyclonal goat anti-human F9 peroxidase-conjugated IgG sec-ondary antibody at 1:4,200 (Enzyme Research GAFIX-APHRP).The detection limit of ELISA for hAAT and hFIX under our experi-mental condition is 1 ng/mL, and the background for non-infusednegative control mouse serum is zero.

RT-qPCR

Liver samples of animals tested in Figure 3B were used for RT-qPCRexperiments in Figures 3C–3E. Liver total RNA samples were ex-tracted by using the mirVana miRNA Isolation Kit (Ambion). 5 mgof total RNA of each sample was used to generate cDNA throughthe SuperScriptIII First-Strand Synthesis System for RT-PCR (Invi-trogen). The RT1 primer 50 TTACCAGTGGCTGCTGCCA 30 wasused to amplify short antisense backbone transcripts. The RT2 primer50 CAGGCATGCTGGGGATGCGGT 30 was used to amplify longantisense backbone transcripts. The RT3 primer 50 CGCCGCATTGCATCAGCCAT 30 was used to amplify short sense backbone tran-scripts. The RT4 primer 50 ATACCTGTCCGCCTTTCTCC 30 wasused to amplify long sense backbone transcripts. The standard OdTwas used to amplify hAAT and actin cDNA. 5 mL of each cDNA prod-uct was further used for each 20 mL qPCR reaction. qPCR was per-formed by using the QuantiFast SYBR Green PCR Kit (QIAGEN)with the Corbett Research RG6000 PCR instrument (CorbettResearch). Forward primer (qPCR2) 50 AAGGCAAATGGGAGAGACCT 30 and reverse primer (qPCR1) 50 TACCCAGCTGGACAGCTTCT 30 oligos were used to amplify about a 150 bp fragmentfrom the hAAT cDNA region. Forward primer 50 TTGCTGACAGGATGCAGAAG 30 and reverse primer 50 TGATCCAC ATCTGCTGGAAG 30 oligos were used to amplify a 150 bp fragmentfrom b-actin as the loading control. Forward primer (qPCR3) 50

AGTCGTGTCTTACCGGGTTG 30 and reverse primer (qPCR4) 50

GGCGCTTTCTCATAGCTCAC 30 were used to amplify about150 bp of short antisense and long sense backbone transcription sig-nals. Forward primer (qPCR5) 50 ATGGACAGCAAGCGAACCG 30

and the reverse primer (qPCR6) 50 GCGAAACGATCCTCATCCTG30 were used to amplify about 150 bp of short sense and long antisensebackbone transcription signals. The tested transgene signal was thennormalized to the b-actin signal. Liver samples of animals testedin Figure S4A were used for RT-qPCR experiments in Figures S4B–S4F. The hAAT and actin signals were amplified as described inFigure 3. The RT primers 50 CCCTCTCCTATTCTGGCTC30,50TGACTCCCGAAGGCCGGT30, and 50CGACTATATTATCAGCTTACGA30 were used to amplify short, medium, and long sense tran-

1196 Molecular Therapy Vol. 25 No 5 May 2017

scripts from the RD-An/Tn backbone, respectively. The RT primers50ACGAACTCCGACGTCTGTAT30, 50AAAGCCGCATGGAAAGTTGAG30, and 50TGTTTAATGGCTGGAGTACAAA30 were usedto amplify short, medium, and long antisense transcripts from theRD-An/Tn backbone, respectively. For the RD backbone, RT primers50CCAGTCAATTATGAGTTACTCC30 and 50TGCGCGTACACCACCAGC30 were used to amplify short and long sense transcripts,whereas RT primers 50TGGAGCCGCGCCCGGC30 and 50TTAGCCGCGAGGGAGATACA30 were used to amplify short and long anti-sense transcripts from the RD backbone region. Forward primer50TTGGTGGCTCCCTAAAGTTG30 and reverse primer 50TCACGCGTGGAAAACATAAA30 were used to amplify about 150 bp ofshort sense and long antisense transcription signals from the RD-An/Tn backbone. Forward primer 50GCGAGCGAATATGCTTTGTT30 and reverse primer 50GTATAGGGATGGTAGGCGCA30

were used to amplify about 150 bp of short antisense and long sensetranscription signals from the RD-An/Tn backbone. Forward primer50GGAGCCAGAATAGGAGAGGG30 and reverse primer 50TCCGGAATCCGTAGTACGTC30 were used to amplify about 150 bp ofmedium sense and medium antisense transcription signals from theRD-An/Tn backbone. Forward primer 50TAAAGACTGCGCCGGTAACT30 and reverse primer 50GCGTGGTGAATATCTCGGTT30

were used to amplify about 150 bp of short sense transcription signalsfrom the RD backbone. Forward primer 50GAAGCCGGAAGAACTGTCTG30 and reverse primer 50GCGCCTACCATCCCTATACA30 were used to amplify about 150 bp of long sense and shortantisense transcription signals from the RD backbone. Forwardprimer 50AACCGAGATATTCACCACGC30 and reverse primer50TCCCTCGTAGCTCTCCTCAA30 were used to amplify about150 bp of long antisense transcription signals from the RD backbone.

SUPPLEMENTAL INFORMATIONSupplemental Information includes five figures and can be found withthis article online at http://dx.doi.org/10.1016/j.ymthe.2017.03.003.

AUTHOR CONTRIBUTIONSJ.L. came up with the initial idea, designed, performed and analyzedexperiments. F.Z. performed experiments. A.Z.F. designed experi-ments and contributed to manuscript preparation in conjunctionwith J.L. and M.A.K. M.A.K. designed experiments, supervised thework and contributed to manuscript preparation in conjunctionwith J.L. and A.Z.F.

ACKNOWLEDGMENTSWe wish to express our sincere thanks to Professor Roger Kornberg,Dr. Christian Frøkjaer Jensen, Dr. Lia Gracey Maniar, and Dr. LorenHansen for their suggestions.

REFERENCES

1. Lu, J., Zhang, F., Xu, S., Fire, A.Z., and Kay, M.A. (2012). The extragenic spacer lengthbetween the 50 and 30 ends of the transgene expression cassette affects transgenesilencing from plasmid-based vectors. Mol. Ther. 20, 2111–2119.

www.moleculartherapy.org

2. Chen, Z.Y., Riu, E., He, C.Y., Xu, H., and Kay, M.A. (2008). Silencing of episomaltransgene expression in liver by plasmid bacterial backbone DNA is independentof CpG methylation. Mol. Ther. 16, 548–556.

3. Lu, J., Zhang, F., and Kay, M.A. (2013). A mini-intronic plasmid (MIP): a novelrobust transgene expression vector in vivo and in vitro. Mol. Ther. 21, 954–963.

4. Chen, Z.Y., He, C.Y., Ehrhardt, A., and Kay, M.A. (2003). Minicircle DNA vectorsdevoid of bacterial DNA result in persistent and high-level transgene expressionin vivo. Mol. Ther. 8, 495–500.

5. Chen, Z.Y., He, C.Y., and Kay, M.A. (2005). Improved production and purification ofminicircle DNA vector free of plasmid bacterial sequences and capable of persistenttransgene expression in vivo. Hum. Gene Ther. 16, 126–131.

6. Bigger, B.W., Tolmachov, O., Collombet, J.M., Fragkos, M., Palaszewski, I., andCoutelle, C. (2001). An araC-controlled bacterial cre expression system to produceDNA minicircle vectors for nuclear and mitochondrial gene therapy. J. Biol. Chem.276, 23018–23027.

7. Darquet, A.M., Cameron, B., Wils, P., Scherman, D., and Crouzet, J. (1997). A newDNA vehicle for nonviral gene delivery: supercoiled minicircle. Gene Ther. 4,1341–1349.

8. Mayrhofer, P., Blaesen, M., Schleef, M., and Jechlinger, W. (2008). Minicircle-DNAproduction by site specific recombination and protein-DNA interaction chromatog-raphy. J. Gene Med. 10, 1253–1269.

9. Schakowski, F., Gorschlüter, M., Buttgereit, P., Märten, A., Lilienfeld-Toal, M.V.,Junghans, C., Schroff, M., König-Merediz, S.A., Ziske, C., Strehl, J., et al. (2007).Minimal size MIDGE vectors improve transgene expression in vivo. In Vivo 21,17–23.

10. Beisel, C., and Paro, R. (2011). Silencing chromatin: comparing modes and mecha-nisms. Nat. Rev. Genet. 12, 123–135.

11. Iyer, V., and Struhl, K. (1995). Poly(dA:dT), a ubiquitous promoter element thatstimulates transcription via its intrinsic DNA structure. EMBO J. 14, 2570–2579.

12. Segal, E., and Widom, J. (2009). Poly(dA:dT) tracts: major determinants of nucleo-some organization. Curr. Opin. Struct. Biol. 19, 65–71.

13. Sekinger, E.A., Moqtaderi, Z., and Struhl, K. (2005). Intrinsic histone-DNA interac-tions and low nucleosome density are important for preferential accessibility of pro-moter regions in yeast. Mol. Cell 18, 735–748.

14. Field, Y., Kaplan, N., Fondufe-Mittendorf, Y., Moore, I.K., Sharon, E., Lubling, Y.,Widom, J., and Segal, E. (2008). Distinct modes of regulation by chromatin encodedthrough nucleosome positioning signals. PLoS Comput Biol 4, e1000216.

15. Yuan, G.C., Liu, Y.J., Dion, M.F., Slack, M.D., Wu, L.F., Altschuler, S.J., and Rando,O.J. (2005). Genome-scale identification of nucleosome positions in S. cerevisiae.Science 309, 626–630.

16. Mavrich, T.N., Jiang, C., Ioshikhes, I.P., Li, X., Venters, B.J., Zanton, S.J., Tomsho,L.P., Qi, J., Glaser, R.L., Schuster, S.C., et al. (2008). Nucleosome organization inthe Drosophila genome. Nature 453, 358–362.

17. Schones, D.E., Cui, K., Cuddapah, S., Roh, T.Y., Barski, A., Wang, Z., Wei, G., andZhao, K. (2008). Dynamic regulation of nucleosome positioning in the humangenome. Cell 132, 887–898.

18. Frøkjær-Jensen, C., Jain, N., Hansen, L., Davis, M.W., Li, Y., Zhao, D., Rebora, K.,Millet, J.R., Liu, X., Kim, S.K., et al. (2016). An abundant class of non-codingDNA can prevent stochastic gene silencing in the C. elegans germline. Cell 166,343–357.

19. Seila, A.C., Calabrese, J.M., Levine, S.S., Yeo, G.W., Rahl, P.B., Flynn, R.A., Young,R.A., and Sharp, P.A. (2008). Divergent transcription from active promoters.Science 322, 1849–1851.

20. Orozco, I.J., Kim, S.J., and Martinson, H.G. (2002). The poly(A) signal, withoutthe assistance of any downstream element, directs RNA polymerase II to pausein vivo and then to release stochastically from the template. J. Biol. Chem. 277,42899–42911.

21. Park, N.J., Tsao, D.C., and Martinson, H.G. (2004). The two steps of poly(A)-depen-dent termination, pausing and release, can be uncoupled by truncation of the RNApolymerase II carboxyl-terminal repeat domain. Mol. Cell. Biol. 24, 4092–4103.

22. Hawryluk, P.J., Ujvári, A., and Luse, D.S. (2004). Characterization of a novel RNApolymerase II arrest site which lacks a weak 30 RNA-DNA hybrid. Nucleic AcidsRes. 32, 1904–1916.

23. Wiest, D.K., and Hawley, D.K. (1990). In vitro analysis of a transcription terminationsite for RNA polymerase II. Mol. Cell. Biol. 10, 5782–5795.

24. Kash, S.F., Innis, J.W., Jackson, A.U., and Kellems, R.E. (1993). Functional analysis ofa stable transcription arrest site in the first intron of the murine adenosine deaminasegene. Mol. Cell. Biol. 13, 2718–2729.

25. Reines, D., Wells, D., Chamberlin, M.J., and Kane, C.M. (1987). Identification ofintrinsic termination sites in vitro for RNA polymerase II within eukaryotic genesequences. J. Mol. Biol. 196, 299–312.

26. Xu, Z., Wei, W., Gagneur, J., Perocchi, F., Clauder-Münster, S., Camblong, J.,Guffanti, E., Stutz, F., Huber, W., and Steinmetz, L.M. (2009). Bidirectional pro-moters generate pervasive transcription in yeast. Nature 457, 1033–1037.

27. Kowalczyk, M.S., Higgs, D.R., and Gingeras, T.R. (2012). Molecular biology: RNAdiscrimination. Nature 482, 310–311.

28. Wang, G.Z., Lercher, M.J., and Hurst, L.D. (2011). Transcriptional coupling of neigh-boring genes and gene expression noise: evidence that gene orientation and noncod-ing transcripts are modulators of noise. Genome Biol. Evol. 3, 320–331.

29. Tan, Y., Li, S., Pitt, B.R., and Huang, L. (1999). The inhibitory role of CpG immunos-timulatory motifs in cationic lipid vector-mediated transgene expression in vivo.Hum. Gene Ther. 10, 2153–2161.

30. Yew, N.S., Wang, K.X., Przybylska, M., Bagley, R.G., Stedman, M., Marshall, J.,Scheule, R.K., and Cheng, S.H. (1999). Contribution of plasmid DNA to inflamma-tion in the lung after administration of cationic lipid:pDNA complexes. Hum.Gene Ther. 10, 223–234.

31. Yew, N.S., and Cheng, S.H. (2004). Reducing the immunostimulatory activity ofCpG-containing plasmid DNA vectors for non-viral gene therapy. Expert Opin.Drug Deliv. 1, 115–125.

32. Chen, Z.Y., He, C.Y., Meuse, L., and Kay, M.A. (2004). Silencing of episomal trans-gene expression by plasmid bacterial DNA elements in vivo. Gene Ther. 11, 856–864.

33. Yew, N.S., Zhao, H., Przybylska, M., Wu, I.H., Tousignant, J.D., Scheule, R.K., andCheng, S.H. (2002). CpG-depleted plasmid DNA vectors with enhanced safety andlong-term gene expression in vivo. Mol. Ther. 5, 731–738.

34. Essevaz-Roulet, B., Bockelmann, U., and Heslot, F. (1997). Mechanical separation ofthe complementary strands of DNA. Proc. Natl. Acad. Sci. USA 94, 11935–11940.

35. Rajewska, M., Wegrzyn, K., and Konieczny, I. (2012). AT-rich region and repeatedsequences - the essential elements of replication origins of bacterial replicons.FEMS Microbiol. Rev. 36, 408–434.

36. GraceyManiar, L.E., Maniar, J.M., Chen, Z.Y., Lu, J., Fire, A.Z., and Kay, M.A. (2013).Minicircle DNA vectors achieve sustained expression reflected by active chromatinand transcriptional level. Mol. Ther. 21, 131–138.

37. Piechaczek, C., Fetzer, C., Baiker, A., Bode, J., and Lipps, H.J. (1999). A vector basedon the SV40 origin of replication and chromosomal S/MARs replicates episomally inCHO cells. Nucleic Acids Res. 27, 426–428.

38. Cook, G.A., Wilkinson, D.A., Crossno, J.T., Jr., Raghow, R., and Jennings, L.K. (1999).The tetraspanin CD9 influences the adhesion, spreading, and pericellular fibronectinmatrix assembly of Chinese hamster ovary cells on human plasma fibronectin. Exp.Cell Res. 251, 356–371.

39. Baiker, A., Maercker, C., Piechaczek, C., Schmidt, S.B., Bode, J., Benham, C., andLipps, H.J. (2000). Mitotic stability of an episomal vector containing a humanscaffold/matrix-attached region is provided by association with nuclear matrix.Nat. Cell Biol. 2, 182–184.

40. Cockerill, P.N., and Garrard, W.T. (1986). Chromosomal loop anchorage of thekappa immunoglobulin gene occurs next to the enhancer in a region containingtopoisomerase II sites. Cell 44, 273–282.

41. Gasser, S.M., and Laemmli, U.K. (1986). Cohabitation of scaffold binding regionswith upstream/enhancer elements of three developmentally regulated genes of D.melanogaster. Cell 46, 521–530.

42. Tracy, R.B., Chédin, F., and Kowalczykowski, S.C. (1997). The recombination hotspot chi is embedded within islands of preferred DNA pairing sequences in theE. coli genome. Cell 90, 205–206.

Molecular Therapy Vol. 25 No 5 May 2017 1197

Molecular Therapy

43. Dietz, W.M., Skinner, N.E., Hamilton, S.E., Jund, M.D., Heitfeld, S.M., Litterman,A.J., Hwu, P., Chen, Z.Y., Salazar, A.M., and Ohlfest, J.R. (2013). Minicircle DNAis superior to plasmid DNA in eliciting antigen-specific CD8+ T-cell responses.Mol. Ther. 21, 1526–1535.

44. Chang, T.Y., Chung, C.Y., Chuang, W.M., Li, L.Y., Jeng, L.B., and Ma, W.L. (2014).Durable expression of minicircle DNA-liposome-delivered androgen receptorcDNA in mice with hepatocellular carcinoma. BioMed Res. Int. 2014, 156356.

1198 Molecular Therapy Vol. 25 No 5 May 2017

45. Huang, M., Chen, Z., Hu, S., Jia, F., Li, Z., Hoyt, G., Robbins, R.C., Kay, M.A., andWu,J.C. (2009). Novel minicircle vector for gene therapy in murine myocardial infarction.Circulation 120, S230–S237.

46. Kay, M.A., Graham, F., Leland, F., and Woo, S.L. (1995). Therapeutic serum concen-trations of human alpha-1-antitrypsin after adenoviral-mediated gene transfer intomouse hepatocytes. Hepatology 21, 815–819.