Nanoparticle-mediated rhodopsin cDNA but not intron-containing DNA delivery causes transgene silencing in a rhodopsin knockout model

Abstract

Previously, we compared the efficacy of nanoparticle (NP)-mediated intron-containing rhodopsin (sgRho) vs. intronless cDNA in ameliorating retinal disease phenotypes in a rhodopsin knockout (RKO) mouse model of retinitis pigmentosa. We showed that NP-mediated sgRho delivery achieved long-term expression and phenotypic improvement in RKO mice, but not NP housing cDNA. However, the protein level of the NP-sgRho construct was only 5–10% of wild-type at 8 mo postinjection. To have a better understanding of the reduced levels of long-term expression of the vectors, in the present study, we evaluated the epigenetic changes of subretinal delivering NP-cDNA vs. NP-sgRho in the RKO mouse eyes. Following the administration, DNA methylation and histone status of specific regions (bacteria plasmid backbone, promoter, rhodopsin gene, and scaffold/matrix attachment region) of the vectors were evaluated at various time points. We documented that epigenetic transgene silencing occurred in vector-mediated gene transfer, which were caused by the plasmid backbone and the cDNA of the transgene, but not the intron-containing transgene. No toxicity or inflammation was found in the treated eyes. Our results suggest that cDNA of the rhodopsin transgene and bacteria backbone interfered with the host defense mechanism of DNA methylation-mediated transgene silencing through heterochromatin-associated modifications.—Zheng, M., Mitra, R. N., Filonov, N. A., Han, Z. Nanoparticle-mediated rhodopsin cDNA but not intron-containing DNA delivery causes transgene silencing in a rhodopsin knockout model.

Current strategies use cDNA for gene targeting, but genomic DNA locus (physical nature), which contains all endogenous regulatory sequences, may better preserve the stability of the message and normal gene regulation (1–4). This approach may be a more effective method to elicit therapeutic rescue after gene supplementation. The use of genomic DNA can dramatically alter gene expression by providing physiologic levels of expression (3–5). Epigenetics, such as DNA methylation and heterochromatinization, are identified as some of the obstacles for the development of next-generation gene targeting. DNA methylation and histone molecules can regulate gene expression by turning genes on and off in cells in a heritable fashion without changing the DNA sequence. Transgene silencing brought on by the host genome defense system can typically cause partial or permanent transgene silencing. As a result, epigenetic therapy is emerging as a new therapeutic target to combat gene silencing to achieve activation of beneficial genes. However, silence can be golden. DNA methylation that causes genes to be turned on or off is a routine and essential biological process in normal genome regulation and development in cells. Therefore, the exogenous genetic code of the delivered gene must be recognized by the host system; if not, it will be treated as an invader, undoubtedly, and the consequences might be unwanted epigenetic reprogramming.

The use of cDNA transgene and plasmid backbone often causes transgene silencing as overexpression or underexpression is known to affect protein function (6–10). In contrast, large genomic DNA transgenes have shown position-independent, copy number-dependent expression (6, 11–13). Precise control of rhodopsin expression is critical in maintaining viable photoreceptors. In our previous studies (14), we have delivered nanoparticle (NP) carrying rhodopsin gene constructs in a rhodopsin knockout (RKO) mouse model (15). Improved long-term expression is observed by using an intron-containing short-form genomic DNA construct (named NP-sgRho) in comparison with an intronless cDNA construct (named NP-cRho) employing the natural mouse opsin promoter (MOP). Protection from retinal degeneration and rescue of outer segment formation is observed in RKO mice treated with the NP-sgRho construct but not NP-housing cDNA construct up to 8 mo postinjection (PI). Although NP-sgRho mediates improved retinal function, levels of expressed protein is only 5–10% of wild-type (WT) at 8 mo PI, an amount that could not fully rescue the phenotype. Following this initial proof-of-principle study, in the current study, we have compared the epigenetic modifications of the same constructs in the RKO model through NP delivery. We have shown that plasmid bacteria backbone and the cDNA of the transgene but not the intron-containing short-form genomic DNA was associated with heterochromatin-associated specific DNA methylation. We documented that delivery of sgRho using these NPs reduced transgene silencing with decreased levels of DNA methylation.

MATERIALS AND METHODS

Animals

All experiments and animal maintenance were approved by the University of North Carolina at Chapel Hill Animal Care and Use Committee and adhered to the Association for Research in Vision and Opthamology Statement for the Use of Animals in Ophthalmic and Vision Research. All mice (RKO and WT) used in this study were in the C57BL/6 background. RKO mice were kindly provided by Janis Lem (Tufts New England Medical Center, Boston, MA, USA). Animals were maintained in the breeding colony under cyclic light (30 lux, 12 h light-dark) conditions. Subretinal injections were performed as previously described (14).

Plasmids and NPs

Endotoxin-free plasmids (16) were amplified in Escherichia coli DH10B cells (Life Technologies, Grand Island, NY, USA). DNA NPs were compacted at the principal investigator’s lab oratory (Carolina Institute for NanoMedicine) as previously described (16–19).

Quantification of vector genome after subretinal injections

To determine the levels of vector genome in retinas of the NP-cRho- vs. NP-sgRho-treated mice, quantitative PCR for kanamycin-resistance gene region was performed in triplicate on isolated retina genomic DNA at 1 and 8 mo PI. To quantify the vector genome number in the DNA samples, a standard curve of serial 10-fold dilutions of the plasmid pEPI-MOP-sgRho (10661 bp) 0.1–100 pg were prepared (Fig. 1A, B). Age matched saline or uninjected RKO animals were used as negative control. At least 3 retinas from each group were analyzed. Each 25 µl reaction contained 2 µl (∼1/50 of total amount in each retina) genomic DNA. All reactions were performed in an ABI StepOne Plus7500 thermocycler (Life Technologies). The amplified products were run on a 2% agarose gel after the amplification. β-actin was used as a reference control (Fig. 1C).

Figure 1.

Persistence of vector genome following subretinal injection. A) A standard curve was generated using known DNA concentrations as described in the Materials and Methods. Quantitative PCR was performed using primers for kanamycin resistance gene in the vector at 1 (PI-1m) and 8 mo PI (PI-8m). B) The quantity of DNA was calculated as copy numbers. Data are expressed as mean ± sd (n = 4). Significance was set at the 0.05 level. No statistically significant differences were observed. C) Quantitative PCR products were loaded in 2% agarose gel. β-actin was used a reference gene control.

Southern blot analysis

Total retina DNA was extracted at 1 mo PI as previously described (16). Total (20 µg) retina DNA (∼1/10 of total amount) was digested with EcoRI restriction enzyme, separated on 0.8 agarose gels, and blotted onto nylon membranes (BrighStar-Plus, Ambion, Austin, TX, USA). An 1182-bp rhodopsin cDNA fragment was labeled with (α[³²P])dCTP and used as a probe. The random primer DNA labeling kit (catalog no. 18187-013) was purchased from Invitrogen (Carlsbad, CA, USA). The blots were prehybridized in NorthernMax prehybridization/hybridization buffer for 4 h and hybridized overnight at 45°C with the [³²P]-labeled probe for rhodopsin accordingly to the manufacturer’s guidelines. After hybridization, membranes were washed in 0.2 × saline sodium citrate buffer, 0.1% sodium dodecyl sulfate (SDS) buffer at 45°C and exposed to Kodak (Rochester, NY, USA) X-Omat film for 6–72 h at −70°C.

Methylation-specific PCR and bisulfite DNA sequencing

The methylation status of the bacteria, MOP promoter, rhodopsin gene, and scaffold/matrix attachment region (S/MAR) regions were analyzed by methylation-specific PCR (MSP). Six hundred nanograms of genomic DNA from the NP-cRho- and NP-sgRho-treated retinas were treated with sodium bisulfite using an Epitect Bisulfite kit (Qiagen, Valencia, CA, USA) according to manufacturer's instruction. The resulting PCR products were separated on 2% agarose gels and visualized by ethidium bromide staining. For bisulfite sequencing, bisulfite treatment of DNAs (2 µg of genomic NP-cRho- or NP-sgRho-treated retinas) was performed according to the manufacturer’s protocol. The products were amplified with a set of primers corresponding to the methylated or unmethylated reaction (Table 1). For PCR, 1 µl of bisulfite-treated DNA was used for PCR using bisulfite primers. The products were purified by agarose gel purification, cloned into a pCR2.1-TOPO vector (Life Technologies) and then were sent for DNA sequencing at University of North Carolina at Chapel Hill Genome Analysis Facility.

TABLE 1.

Primer sequences

Name	Sense primer, 5′ to 3′	Antisense primer, 5′ to 3′	Product size (bp)
PCR primers
IL-2	CCTGAGCAGGATGGAGAATTACA	TCCAGAACATGCCGCAGAG	141
IL-6	CAGAATTGCCATCGTACAACTCTTTTCTCA	AAGTGCATCATCGTTGTTCATACA	141
TNF-α	CATCTTCTCAAAATTCGAGTGACAA	TGGGAGTAGACAAGGTACAACCC	175
INF-γ	TCAAGTGGCATAGATGTGGAAGAA	TGGCTCTGCAGGATTTTCATG	92
Kanamycin	TGCTCCTGCCGAGAAAGTAT	GCTCTTCGTCCAGATCATCC	164
MOP	TCCTGGGAAGAGATGGGATA	GGTGGAGGCCCTTAGGTAAA	195
Rhodopsin	GAGGGCCCCAATTTTTATGT	TGAGGAAGTTGATGGGGAAG	160
S/MAR	CTTCCTCCCCAAACCCTTAC	AATGGCTTTGCCGTGTTAAT	297
β-actin	TGTTACCAACTGGGACGACA	CTTTTCACGGTTGGCCTTAG	130
MSP primers
Kan-M	GTTTAAGGCGAGTATGTTCGAC	AATCGATAAATCCAAAAAAACGAC	107
Kan-U	AGGTTTAAGGTGAGTATGTTTGATG	CAATCAATAAATCCAAAAAAACAAC	110
MOP-M	TTTTTGTAAGTTAATTAGGTTTCGG	TACTTAACGAACTCAACCACTAACG	176
MOP-U	TTTTTGTAAGTTAATTAGGTTTTGG	CTTAACAAACTCAACCACTAACAAC	174
Rho-M	GAGTAGTCGTAGTATTATTTGGCGG	ATTACAACCTATAAACCCAAAAACG	261
Rho-U	GAGTAGTTGTAGTATTATTTGGTGG	ACAACCTATAAACCCAAAAACAAA	258
S/MAR-M	TTTTAATTTGTTGATGGGATAAATTAC	AAACCAACATTACCTTAATACCGAA	237
S/MAR-U	TTTTTAATTTGTTGATGGGATAAATTAT	AAACCAACATTACCTTAATACCAAA	238

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) was performed using the EZ ChIP Chromatin Immunoprecipitation Kit (Millipore, Billerica, MA) according to the manufacturer's protocol. Samples from 6 pooled NP-cRho- or -sgRho-injected RKO retinas were prepared using standard ChIP protocols to crosslink DNA and protein. Briefly, formaldehyde was added to the cells to a final concentration of 1% to crosslink protein to DNA and incubated at 37°C for 15 min. The cells were washed in cold PBS with proteinase inhibitors and suspended in SDS lysis buffer. Lysates were sonicated for 10 s 9 times on ice and centrifuged at 15,000 rpm for 10 min at 4°C. The sonicated samples were precleaned with salmon sperm DNA/protein A agarose beads (Upstate Biotechnology, Lake Placid, NY, USA). Sonicated chromatin was immunoprecipitated with H3K4m2, H3K9m1, and H3K27m1 and incubated overnight on a rotator at 4°C. Normal mouse IgG in the immunoprecipitation reactions was used as the negative control and input (without immunoprecipitation reactions) as the positive control in later PCR reactions. After rotation, chromatin-antibody (Ab) complexes were collected using salmon sperm DNA/protein A agarose beads and washed according to the manufacturer's protocol. The immune complexes were eluted with 1% SDS and 0.1 M NaHCO₃; samples were treated with proteinase K for 1 h, and DNA was purified by phenol:chloroform:isoamyl alcohol (25:24:1, Life Technologies) extraction and ethanol precipitation and resuspended in 30 µl of H₂O. Quantitative PCR validation of Ab enrichment was performed. Primers (Table 1) designed to separately amplify 4 regions (Kan, promoter, rhodopsin, and S/MAR) in the constructs. Mouse alpha-albumin (AFM) (Simple ChIP Mouse AFM Intron 2 Primers, Cell Signaling, Inc., Danvers, MA, USA) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; SimpleChIP Mouse GAPDH Intron 2 Primers; Cell Signaling, Inc.) were used as heterochromatin and euchromatin controls, respectively. The results of 3 biological replicates are shown as the absolute enrichment compared with the input.

Immunohistochemistry

Tissue fixation and sectioning were performed as described previously (19). Briefly, eyes from mice at 1 mo PI were enucleated and fixed with PBS containing 4% paraformaldehyde at 4°C, dissected to remove the cornea and lens, then sequentially immersed in 10, 20, and 30% (w/v) sucrose. Eyecups were embedded in M1 embedding medium (Thermo Electron Corp., PA, USA) and frozen on dry ice; the cryostat sections of the eyes were immunostained with Abs for macrophage (rabbit polyclonal to F4/80, 1:1000; Abcam, Cambridge, MA, USA) and mouse monoclonal rhodopsin (1D4, 1:1000; Santa Cruz Biotechnology, Dallas, TX, USA). Sections were stained with Alexa Fluor 488 goat anti-mouse (for 1D4) and donkey-anti-rabbit Alexa Fluor 555 (for F4/80) as secondary Ab (1:1,000; Life Technologies). Sections were imaged using a Zeiss CLSM 710 Spectral Confocal Laser Scanning Microscope (Carl Zeiss GmbH, Jena, Germany).

Quantification of proinflammatory cytokines (quantitative real-time PCR)

The levels of IL-2, IL-6, TNF-α, and INF-γ obtained from mice injected with NP-cRho vs. NP-sgRho at 1 mo PI were assessed by RT-PCR (16, 20, 21). RNA was extracted from retinas following the TRIzol protocol (Life Technologies) according to the manufacturer’s instructions. Samples were subjected to RT-PCR and normalized to β-actin to verify the efficiency of RNA extraction and for the presence of procytokines. Primers (Table 1) targeting the proinflammatory cytokines (IL-6, IL-2, TNF-α, and IFN-γ) were used. Bacillus cereus-injected eyes were used as positive controls. Saline and uninjected samples were used as negative controls as previously described (19). All experiments were performed in triplicate. At least 4 eyes from each group were analyzed. Relative expression = 2(ΔΔC_t), where ΔC_t = gene C_t − β-actin C_t. All reactions were performed in an ABI StepOne Plus7500 thermocycler (Life Technologies). The parameters for the amplification of proinflammatory cytokines were as follows: 95°C for 2 min, 94°C for 45 s, 56°C for 30 s, 72°C for 30 s for 39 cycles, and 72°C for 5 min.

Statistical analysis

Data are expressed as the mean ± sd of at least 3 independent experiments. The results were analyzed by 2-way analysis of variance followed by Bonferroni's post hoc test for multiple comparisons, and values of P < 0.05 were considered significant. All tests were carried out using GraphPad prism (La Jolla, CA, USA).

RESULTS

Detection and quantification of DNA levels after gene delivery

We have previously shown that similar levels of mRNA were observed in eyes treated with NP-cRho or -sgRho at 1 mo PI by Northern blot, indicating that the primary transcript was correctly initiated, and the mRNA was correctly translated to yield a properly expressed rhodopsin protein at that time (14). To find out whether the gradually declining gene expression was due to transgene silencing or vector degradation, in this study, total retina genomic DNA were isolated and real-time quantitative PCR for vector sequences at 1 and 8 mo PI was performed. The analysis produced a linear standard curve allowing the quantification of vector copies in the unknown DNA samples derived from retinas of the treated animals (Fig. 1A, B). The presence of the PCR products following quantitative PCR was visualized in 2% agarose gel with ethidium bromide staining (Fig. 1C). The quantitative PCR analysis showed a persistence of vector genome in both the NP-cRho- and NP-sgRho-injected eyes (Fig. 1B). No genome vector information was detected in saline or uninjected animals. Consistence with our previous finding by our group, although we found that the majority of the DNA, placed in the subretinal space, might have been lost or degraded at very early time (22), no significant reduction in DNA copy numbers was observed from 1 to 8 mo PI as revealed by this study (Fig. 1B, C) as well as a previous report (17). At 8 mo PI, the total genome amount treated with NP-cRho was still comparable to that of its -sgRho DNA counterpart, although the transgene expression is largely undetectable at this point. The loss of transgene expression in eyes treated with NP-cRho at late points strongly indicates that the transgene was suffering an impaired transcriptional shut down.

Episomal state persistence of DNA

To test whether the transgene persisted in the nucleus of the eye in an episomal state, at 1 mo PI, retinas were collected and DNA was isolated for Southern blotting. The point 1 mo PI was chosen because of results from previous studies that showed gene expression arrived at a relatively stable peak (17). This time also keeps away from the surgical stress triggered by subretinal injection, which could cause up-regulation of inflammatory cytokines or macrophage infiltrate (19). As a result, we chose 1 mo PI for most of the studies. DNA was digested with EcoRI (which has a double cut site to remove the rhodopsin gene from the vector, Fig. 2A). The resultant blots were probed with rhodopsin cDNA (Fig. 2B) or the S/MAR DNA sequence (Fig. 2C). A single band corresponding to the vector size was detected (Fig. 2B, C), suggesting that both vectors remain episomal. Hybridization to undigested high molecular weight DNA was observed with all the injected samples (Fig. 2B, arrows). However, the bands are distinct and the sizes are exactly double or triple size of the plasmids, suggesting that these are concatemers or trimeric concatemers of the released plasmids. We did not observe multiple or different size hybridization bands, indicating that there were no unexpected integrations. Quantification of band intensities reveals that transgene copy numbers were within a similar range in both vectors with the same amount of DNA (4.3 µg/retina) injected (Fig. 2D). Because there is only one gene copy in the WT genome, no band was detected in WT digested control. No hybridization band was observed to the uninjected RKO mice. The total copy number per retina [(1 × 10⁶ ∼ 1.2 × 10⁶) × 10 = (1 × 10⁷ ∼ 1.2 × 10⁷)] obtained from Southern blot (Fig. 2C) was consistent with that of the quantitative PCR results [(2.3 × 10⁵ ∼ 2.5 × 10⁵) × 50 = (1.15 × 10⁷ ∼ 1.25 × 10⁷)] (Fig. 1B) at 1 mo PI.

Figure 2.

Southern blot showing episomal status of vectors. RKO mice were subretinally injected at postnatal day 3 (P3) with NP-cRho or NP-sgRho and retinas were collected at 1 mo PI. Total genomic DNA (15 µg) was digested with EcoRI, which has 2 restriction sties. A) Schematic of pEPI-MOP-cRho and pEPE-MOP-sgRho vectors with EcoRI restriction sites. Rho-ex1: rhodopsin exon1. B) The results showed distinct signals of either a 1.4 kb (cRho) or 5 kb (sgRho) rhodopsin fragment, which corresponds to the expected single band sizes, respectively. C) Membrane was striped and incubated with an S/MAR-specific probe. The results show a single distinct band of 5 kb of vector backbone, which matches with the size of the vector backbone. D, digested; U, undigested. D) Densitometric analysis of the Southern blot was used to assess mean copy number with S/MAR probe (standard curve was generated using the calculated copy numbers listed on the right).

Comparison of DNA methylation profiles

To determine whether transgene DNA methylation may be the cause of gene silencing, we examined the methylation changes of the constructs in the regions of bacterial backbone (kanamycin-resistant gene), MOP promoter, rhodopsin gene, and S/MAR by MSP. After treatment with sodium bisulfite, cytosine cytosine but not 5-methylcytosine can be selectively deaminated (C to U). After the treatment, the methylated and unmethylated cytosines can be characterized by DNA sequencing. We found that the NP-cRho vector, including the regions of bacteria backbone, MOP, cDNA, and S/MAR, was associated with DNA methylation. Whereas, in NP-sgRho, we found that only the regions of plasmid backbone, MOP promoter, and S/MAR region were associated with DNA methylation, but less than that of the NP-cRho vector as revealed by band intensities (Fig. 3A). Interestingly, we did not find DNA methylation within the intron-containing rhodopsin gene (Fig. 3A). These DNA methylation patterns were further confirmed by DNA bisulfite sequencing (Fig. 3B and Table 2). These results correlated with observed expression levels in both Western blot and PCR assays as shown previously (14), further suggesting that the NP-cRho construct was subjected to global DNA methylation.

Figure 3.

DNA methylation analysis. RKO animals were injected at postnatal day 3 with NP-cRho and NP-sgRho, and retinas were collected at 1 mo PI for MSP followed by DNA sequencing. Regions of Kan, MOP, Rho, and S/MAR of the vectors were amplified, cloned, and sequenced. A) Methylation-specific PCR was performed on bisulfite-treated DNA prepared from NP-cRho and -sgRho or saline-treated retinal extracts. M indicates primers specific for the methylated DNA form, and U indicates primers for the unmethylated form. B) Schematic DNA bisulfite sequencing showed that the tested regions of MOP, rhodopsin, S/MAR, and Kan were methylated in NP-cRho-treated eyes, but not in the same regions of NP-sgRho-treated eyes.

TABLE 2.

The number of sequences with DNA methylation change

Name	NP-cRho	NP-sgRho
Kan	5/6	4/5
MOP	4/6	3/6
Rho	5/6	0/6
S/MAR	4/6	3/6

Number of methylated (or partially methylated) sequences was determined for at least 6 clones. Values are expressed as number of methylations/number of sequences.

Introns affect chromatin modification and gene expression

DNA and its binding protein (histones) are always working together to regulate gene expression. Bacterial DNA backbone elements have previously been shown to bind heterochromatin elements and thus contribute to gene silencing (9, 23). To evaluate why NP-sgRho mediates partial phenotype rescue but NP-cRho was associated with transgene shutdown at 8 mo PI in RKO mice, we conducted an extensive series of ChIP experiments using Abs for histone proteins associated with active regions (euchromatin and H3K4me2 Ab) or inactive regions (heterochromatin, H3K9m1 Ab, and H3K27m1 Ab). These studies distinguish whether the vector DNA is associated with heterochromatin or euchromatin structures. Samples from pooled NP-cRho- or -sgRho -injected RKO retinas were prepared using standard ChIP protocols. PCR was performed using primers specific for the MOP promoter, rhodopsin coding region, bacterial backbone (kanamycin-resistant gene, Kan), and S/MAR region. AFM, known to be transcribed only in liver, and GAPDH, known to be actively transcribed in all cell types, were used as heterochromatin (negative) and euchromatin (positive) controls, respectively (Fig. 4). Virtually all GAPDH DNA was associated with the euchromatin marker, and the AFM was detected in the heterochromatic (H3K9m1 and H3K27m1) pool. No amplification was detected in any sample without Ab. For each primer set, the quantity of DNA in the 3 pull-downs (H3K4m2, H3K9m1, and H3K27m1) was assessed relative to the input. In the case of rhodopsin, the MOP promoter, the bacterial backbone, and the S/MAR, the DNA was barely distributed in the euchromatin pools in NP-cRho-treated eyes. Our results were consistent with predicted interactions with the rhodopsin region, which was preferentially found in the euchromatin (H3K4m1) fraction in NP-sgRho-treated eyes except the bacterial backbone (Kan), the MOP promoter, and the S/MAR regions, which were mainly associated with heterochromatin or evenly distributed between the euchromatin and heterochromatin pools. Our findings suggest that noncoding elements, such as introns, are associated with transcriptional regulation via histone modifications. In addition, although H3K9 and H3K27 are both markers for heterochromatin, we observed different band intensities for the same region pooled by these Abs (Fig. 4, H3K9 and H3K27 gels), suggesting that they interacted with different heterochromatin domains.

Figure 4.

Epigenetic regulation of rhodopsin. A) ChIP assay extracted from 6 NP-treated eyes/group at postnatal day 3 were pooled at 1 mo PI and underwent ChIP for H3K4m2, H3K9m1, and H3K27m1 or no Ab followed by quantitative PCR with primers from different regions of the vector. GAPDH intron 2 was used as euchromatin (H3K4m2) control. Mouse AFM intron 2 was used as heterochromatin (H3K9m1 or H3K27m1) control. B) Samples were normalized to the input from the same primer set.

No production of proinflammatory cytokines and macrophages

To examine whether transcription silencing was activated by inflammatory cell (neutrophils, eosinophils, mononuclear phagocytes, and macrophages) activation caused by the vectors, injected retinas and their frozen sections were tested by RT-PCR for proinflammatory cytokines and by immunohistochemistry (IHC) for macrophages at 1 mo PI (Fig. 5). Consistent with previous studies (19), our data showed that no specific toxicity was observed in the experimental groups as detected by both proinflammatory cytokines via RT-PCR (Fig. 5A) and macrophages by IHC (Fig. 5B) using a macrophage marker (F4/80 Ab) in the treated eyes.

Figure 5.

Inflammatory response to the injection. A) Cytokine mRNA quantification by real-time PCR. Retinas were collected and assayed to quantify cytokine gene expression (IL-2, IL-6, TNF-α, and IFN-γ) by RT-PCR at 1 mo PI. Levels of cytokine expression are presented as means ± sd of different samples in triplicate (n = 4). Retinas from untreated and saline-injected mice were used as negative controls. B. cereus-injected eyes were chosen as the positive control. No elevated cytokines were found after the injections in the eyes. B) Immunohistochemistry assessment of infiltrating macrophages after subretinal injection. Immunolabeling with Abs for the macrophage (F4/80, red) and rhodopsin (1D4, green) was performed at 30 d PI in the RKO eyes injected with NP-cRho or -sgRho at P3. Age-matched B. cereus-injected WT eyes were used as positive control. Saline-injected (inj.) and uninjected (uninj.) RKO or WT eyes were used as negative controls. n = 3. ONL, outer nuclear layer; INL, inner nuclear layer. Scale bar, 20 µm.

DISCUSSION

Gene therapy has great potential. However, in the past, we have seen more theoretical promises than practical realities. Chemical modification of delivered genes, particularly DNA methylation, and DNA binding protein, particularly histones, are the major factors in controlling epigenetic transgene regulation and expression. DNA methylation is a normal process that maintains higher organism development. Aberrant DNA methylation can cause transcription silencing. Histone modification maintains DNA accessibility for controlling gene activity. In most cases, DNA hypermethylation is mainly associated with gene repression, whereas DNA hypomethylation, particular at the promoter region, allows for gene expression. DNA methylation followed by chromatin modification plays a causative role in controlling gene transcription (24, 25).

Most recently, we have compared the expression levels of NP-mediated rhodopsin cDNA (NP-cRho) with a short form of genomic DNA transgene expression (NP-sgRho) in a RKO mouse model. The latter showed consistent transgene expression and improved structural and functional rescue in RKO mice up to 8 mo postinjection, whereas the former showed a marked decrease in transgene expression with no structural and/or functional rescue. As both vectors have the same pEPI plasmid backbones, core MOP promoter, and short 3′ flanking sequence, the only difference is that the -sgRho vector contains all 4 full-length introns found in the genomic sequence (Fig. 2A), our finding suggests that noncoding elements, such as introns, may be critical for meaningful expression of the therapeutic transgene. Although we have demonstrated phenotypic improvement in RKO mice, these vectors maximally produced rhodopsin protein in photoreceptors at levels ∼10% of WT, an amount insufficient to mediate complete rescue. To determine whether transgene methylation may be the cause of gene silencing, in the current study, we have evaluated DNA methylation and histone status of the transgenes on the regions of bacteria plasmid backbone, MOP promoter, rhodopsin gene, and S/MAR of the vectors by MSP and ChIP. By combining the bisulfite genomic sequencing to the analysis of 4 regions of the delivered vectors, we found dramatic differences in methylated CpG dinucleotides between NP-cRho and NP-sgRho. Our studies indicate that transgenic silencing in NP-cRho correlates with a high degree of methylation at global levels but NP-sgRho only correlates with a relatively high degree of methylation with its plasmid backbone, promoter, and S/MAR regions but not the sgRho of the transgene. Our findings have interestingly shown that the same region of the promoter or gene functions differently in gene expression. This phenomenon indicates that noncoding DNA, such as introns, is critically correlated to gene expression through methylation specific histone modification.

Methylation differences caused by intron-containing and intronless constructs indicate that noncoding portions of the gene are directly or indirectly coordinated with gene regulation. Studies have shown that genes that contain introns are more efficiently transcribed, and mRNA generated by splicing is more efficiently released for nuclear export (1–4, 26–28). Examples of introns commonly used are the rabbit β-globin intron, which has important effects on mRNA stabilization and leads to significantly greater transgene expression (29). Some noncoding elements are highly conserved. Mutations in these elements can cause aberrant expression of linked genes, causing diseases such as cancer and genetic disorders (30, 31). Studies have shown that increasing numbers of genetic disorders are due to intron or intron–exon junction mutations (32–37). Indeed, gene expression is controlled by multiple regulatory elements, which are located both near and far from the gene. Studies have shown that the noncoding portion of the genome, such as cis-acting elements, are censoriously involved in histone modifications and further effect regulation of gene expression (38–40). For example, studies have shown that sequences that are far from genes (up to 2 kb away region) were correlated with DNA methylation and gene expression. Similarly, enhancer elements are able to activate genes in cis at a considerable distance (41) (42), suggesting that potential regulatory regions are, not only directly connected to the marked promoter or genes, but linked with other genomic elements. Exons are enriched in nucleosome content relative to introns and have a distinctively higher level of methylation than introns (43, 44). Studies have shown that DNA methylation interconnects with the splicing machinery during the exon selection process, indicating a potential role of DNA methylation in the regulation of alternative splicing (45, 46). This correlates with the level of the histone modifications. Increasing evidence has shown that exon–intron boundary elements in the genomic sequence have played an important role in maintaining an active chromatin conformation in controlling gene expression (43, 47). In our study, we assume that the histone mark of the nucleosome might act to guide DNA methyltransferases resulting in higher methylation levels on the “naked” exons of transgene (cDNA vector), whereas inclusion of noncoding introns shows a potential ability to protect transcripts from silencing (48, 49). This phenomenon might explain why intron-containing rhodopsin shows reduced methylation and a suppression of repressive histone modifications when associated with the genomic rhodopsin as opposed to the intronless rhodopsin cDNA construct.

The use of genomic DNA instead of cDNA vectors in gene therapy has made a way for vital accurate gene expression, as these vectors contain all endogenous control elements in terms of achieving a close to physiologic expression level. The majority of studies have relied on transgenes using cDNA, which actually does not faithfully mimic physiologic gene expression, because it does not contain most of the endogenous cis-acting regulatory sequences (50). To date, little effort has gone into detailed analyses of the genomic environments of genetic element-containing transgenes. A better understanding of the processes of epigenetic reprogramming is extremely needed. One practical reason is that the vector selected has to be large enough to carry the gene. In our current study, we take advantage of our NP, which carries large DNA for delivery. Nonviral products are known to avoid unwanted immunogenicity and insertional mutagenesis, which are concerns for viral-based gene transfer. However, one of the major hurdles in nonviral gene delivering is transcription silencing of the DNA vector. In this study, we have used nonviral NP as a vehicle for gene delivery. Our results showed that both vector presented in the nucleus of the photoreceptor cells in an episomal state and the vector genome remained persistent in the retinas of the treated animals with no inflammatory responses. However, Western blot, IHC, and RT-PCR analysis have shown a transgene shutdown in NP-cRho-injected animals and a declined transgene expression in NP-sgRho-injected animal eyes at 8 mo PI (16). We further demonstrated that the transgene silencing was due to the cDNA of the transgene and the plasmid bacteria backbone. These observations suggest that these genes were transcribed but were gradually partly or totally inactivated. Although we do not rule out that DNA methylation may be connected with retinal degeneration disease itself to some extent (51, 52), our results strongly suggest that endogenous elements, such as introns, play an important role in regulating DNA methylation patterns, which in turn affect gene expression through histone modifications.

Transcription initiation requires promoter and upstream regulatory regions (the cis-acting elements). The large capacity for foreign DNA, allowing transfer of large regulatory elements and even a proportion of genes in their natural genomic context, is a significant advantage over the other current gene therapy options. Our current results indicate significant advantages of regulated gene expression using intron-containing DNA for gene transfer.

Currently, much attention is focused on antiepigenetic therapy that influences histone modifications. Our studies have shown that adding cis-acting noncoding genomic introns can optimize (shape) rhodopsin expression and reduce transgene silencing. sgRho delivery offers several additional key advantages over the current prevailing cDNA transgene expression methodologies. In contrast to the cDNA construct, intron-containing DNA constructs show relative genome stability, as assayed by MSP and ChIP analysis. sgRho contains the actual genomic locus coding for the protein to be expressed. This genetic information is very important for the synthesis of mRNAs and generation of proteins. Herein, we demonstrate for the first time that reduced levels of long-term transgene expression was associated with heterochromatin-associated methylation by the plasmid bacteria backbone and the cDNA of the transgene. Although this study focuses on the rhodopsin gene delivery in the RKO mouse, the logical principles are likely the same for other strains.

It should be noted that adeno-associated virus carrying rhodopsin cDNA has successfully been used in retinitis pigmentosa mouse models (53, 54). Although the studies did not directly focus on the biological consequences of DNA methylation, long-term effects of in vivo gene therapy have been demonstrated. One such possibility is that adeno-associated virus may take advantage of viral machinery, in part, via the effects of their inverted terminal repeats, which may have the ability to protect the therapeutic DNA from unwanted degradation in the host cell (55). In higher eukaryotes, gene expression is regulated autonomously through its entire system. For example, by methylating a unique identifying sequences, such as the promoter region, instead of changing the DNA sequence, the internal and external environment of the chromosome can be altered (so-called histone modifications), and thus a specific gene function can be controlled. Similarly, an enhancer element, which is interrupted by unrelated independently expressed genes, can activate a gene in cis at a relative distance to control transcription and other chromosomal processes. More convincing evidence now indicates that communication between the genes and the regulatory elements are far more complex than we previously thought. Although using cDNA as a transgene for gene targeting has been proven to be useful, it is a way without considering the physical nature of DNA. In any nonphysical manner, DNA methylation/histone modification is supposed to develop unanticipated toxicity and transgene silencing might be one of the results. Although this perception is changing, more important questions still remain. Indeed, artificial chromosomal vectors, such as yeast artificial chromosomal vector, bacterial artificial chromosomal vector, and human artificial chromosomal vector, have been developed over the past decades to carry the genomic locus of a gene and have shown very accurate long-term transgene expression in vitro and in vivo. Yet, a number of challenges are still associated with the use of these artificial chromosomal vector, including insertional mutagenesis, immune response, complex structure, and the safety and efficiency of delivery procedures, which have proven to be very difficult in clinical practice (39–42). Interestingly, we have observed that the delivery of an intron-containing construct could reduce transcriptional silencing caused by DNA methylation/histone modification.

In summary, our study provides new insights regarding transgene expression. We have shown that the delivery of an intron-containing construct could reduce transcriptional silencing caused by DNA methylation/histone modification. The different fates of NP-cRho and NP-sgRho may thus be caused by a consequence of altered chromatin structure. The NP-cRho transgene was placed in a heterochromatic environment, whereas the NP-sgRho transgene was placed in a euchromatic environment that scattered the regulatory DNA sequence into neighboring euchromatic regions that transcription factors initiated to activate the transgene. These findings support the possibility of using the intact genomic locus of a transgene as a means of gene targeting. Future experiments will be aimed at testing our hypothesis that the development of the genomic DNA transgene that lacks bacteria or viral coding sequences might be able to provide efficient delivery and long-term regulated transgene expression.

Glossary

Ab

antibody

AFM

alpha-albumin

ChIP

chromatin immunoprecipitation

cRho

intronless rhodopsin complementary DNA

IHC

immunohistochemistry

MOP

mouse opsin promoter

MSP

methylation-specific PCR

NP

nanoparticle

PI

postinjection

RKO

rhodopsin knockout

SDS

sodium dodecyl sulfate

sgRho

intron-containing short-form rhodopsin genomic DNA

WT

wild-type