|Home | About | Journals | Submit | Contact Us | Français|
At the present time, no efficient in vivo method for gene transfer to skin stem cells exists. In this study, we hypothesized that early in gestation, specific epidermal stem cell populations may be accessible for gene transfer. To test this hypothesis, we injected lentiviral vectors encoding the green fluorescence protein marker gene driven by either the cytomegalovirus promoter or the keratin 5 (K5) promoter into the murine amniotic space at early developmental stages between embryonic days 8 and 12. This resulted in sustained green fluorescent protein (GFP) expression in both basal epidermal stem cells and bulge cells in the hair follicles of the skin. Transduction of stem cell populations was dependent on the developmental stage, and confirmed by the prolonged duration of GFP expression in all skin elements into adulthood. In addition, transduced stem cell populations responded to regenerative signals after wounding and actively participated in wound healing. Finally, we quantified the fraction of epidermal stem cells transduced, and the distribution of transduction related to the promoters utilized, confirming improved efficiency with the K5 promoter. This simple approach has possible biological applications in our study of gene functions in skin, and perhaps future clinical applications for treatment of skin based disorders.
Molecular genetics has identified many of the genes responsible for intractable skin diseases.1,2 Although the skin might appear to be an ideal target tissue for gene correction, the superficial keratinized epithelium acts as a formidable barrier, limiting access to epidermal stem and progenitor cells.3 Strategies employed include ex vivo and in vivo approaches.3,4 The ex vivo approach requires harvesting skin cells from the patient, correcting the gene defect, expanding the gene-corrected cells, and then transplanting them back onto the patient. The first clinical success of gene therapy for an inherited skin disorder in adult patients was recently reported with epidermolysis bullosa (EB) using the ex vivo approach.5 While there are a number of drawbacks to the ex vivo approach including high cost, and the need for surgery and complex wound care,6,7 ex vivo gene transfer currently represents the only successful approach to skin gene therapy and will undoubtedly play a role in future treatment of EB and other skin disorders. However, this approach may not be efficient enough for treatment of severe phenotype EB patients that present in the neonatal period with diffuse disease. In these patients, an in vivo approach that achieves gene transfer to the entire skin stem cell/progenitor compartment and provides permanent correction of the gene defect would seem ideal.8,9 Thus far, efforts using non-viral and viral vector based strategies for in vivo gene transfer have failed to achieve widespread, or durable gene expression primarily due to inefficient targeting of stem cells.4,10
Effective gene therapy for skin disorders will require gene transfer to skin stem cells due to the rapid cellular turnover inherent to the epidermis. Efficient methods of accomplishing transgene integration in vivo are currently limited to viral vector technology. The most effective and broadly studied integrating vectors that efficiently target both dividing cells and quiescent stem cell populations are lentiviral vectors.11,12 Lentiviral vectors have been specifically shown to transduce both proliferating and quiescent keratinocytes in vivo13 and in vitro.14
Potential advantages of in utero gene therapy include the small size of the fetus, high frequency and accessibility of stem cell populations, and fetal immunologic immaturity that may facilitate tolerance for transgene encoded novel proteins. Skin-specific developmental advantages for gene transfer include the simple, single layer organization of early gestational skin, as well as its exposure to the amniotic fluid throughout development. However, previous efforts at intraamniotic gene transfer (IAGT) at relatively late gestational time points have resulted in very inefficient gene transfer to skin15–18 unless potentially toxic mechanical methods to overcome the barrier of the stratified epithelium were utilized.19 Consideration of the above, led to our hypothesis that a very early gestational “window of opportunity” exists during which specific skin stem cell/progenitor populations are accessible for IAGT. We tested our hypothesis by injecting lentiviral vectors encoding the enhanced green fluorescent protein (EGFP) marker gene into the amniotic fluid of post coital days 8 (E8) to E12 mice. Our results confirm that the precursors that give rise to epidermal and hair follicle stem cell populations are accessible and can be efficiently transduced with lentiviral vectors at early developmental time points.
We performed IAGT on a total of 470 BALB/c fetuses (61 pregnant mice) from E8 to E17 (Table 1). The overall survival rate was 40.4% (190/470). The mice reported in this study injected with the lentiviral-CMV vector overlap with those reported in a previous study that examined ocular stem cell transduction.20 The mice injected with the lentiviral-K5 vector were generated specifically for this study. Survival rates increased with gestational age. Surviving mice appeared healthy. The high mortality at early gestational times appeared to be due to rupture of the amnion caused by the relatively high volume of injection (350 nl) rather than any toxicity of the vector.
A total of 134/190 live born mice that underwent intraamniotic injection of lentiviral vectors were analyzed by stereoscopic fluorescence microscopy with the results summarized in Figure 1a. Mice injected after E12 showed no GFP expression and are not described further. In mice injected at E8 and E9 we found widespread skin GFP expression, including the digits, with the greatest intensity of fluorescence in the scalp region. The skin of the body and trunk demonstrated a striped distribution of fluorescence consistent with transduction of early somatic epidermal progenitors. Skin appendages were also GFP positive including the hair and nails. In the IAGT mice injected at E10, GFP-positive cells were still clearly visible at the scalp and body region. However the pattern of expression was punctate rather than striped with minimal expression in the digits or accessory structures. In mice injected at E11 and E12, GFP-positive cells were present in a cross shaped distribution only on the scalp (Figure 1b–f). We investigated the duration of exogenous gene expression in the skin by analyzing the mice serially by stereoscopic fluorescence microscopy from postnatal day 1 (P1) to 6 months of age. IAGT at E8 resulted in expression beyond 6 months (Figure 1g and h). Figure 1g and i demonstrate the similar gross patterns of transduction using the cytomegalovirus (CMV) or keratin 5 (K5) promoter, respectively. In addition to skin transduction we observed transduction of multiple other fetal tissues and the amniotic membranes with both the CMV and K5 promoters after early gestational IAGT. A complete description of the age dependent distribution of expression after this mode of administration is beyond the scope of this study and will be described in a separate manuscript.
To confirm the transduction of stem cell populations, we assessed the participation of transduced cells in adult wound healing. If stem cells were transduced, GFP-positive cells should proliferate and migrate toward the wound during wound repair. We created 4 mm diameter full-thickness wounds on the backs of 8-week-old IAGT mice injected with lentiviral CMV vector from E8 to E11 when the hair follicles were in the telogen or resting phase of the hair cycle. One week after wounding, GFP-positive cells could clearly be seen streaming from the margin of the wound from individual hair follicles consistent with transduction of bulge stem cells (Figure 2). In addition, sheets of cells could be seen migrating inward from outside of the wound margin, a pattern consistent with basal stem cells. Participation of transduced cells in the wound healing process is also dependent on gestational age and is consistent with the requirement for stem cell transduction.
While the CMV promoter is active in most tissues, it has been demonstrated to specify expression primarily in the suprabasalar layers of the skin.14 In contrast, the K5 promoter is strongly active in the basal layer of the skin.21,22 To compare the relative efficacy of the two promoter constructs for different skin strata, we performed a histologic analysis of the skin following IAGT at E8 with each vector. Both promoters resulted in GFP expression in the epidermis and appendages (Figure 3a–j). However, the pattern of expression differed for the CMV and K5 promoters, resulting in expression predominantly in the differentiated layers of the dermis or the basal layer, respectively (Figure 3a and b). To quantify the relative efficiency of the two vectors for transgene expression in basal or bulge keratinocytes, we isolated each population from the whole skin of IAGT mice during the telogen stage of the hair cycle (8 weeks of age) and used flow cytometry with appropriate lineage markers to assess the frequency of GFP-positive cells in the gated populations. The frequency of GFP-positive cells among total skin cells ranged from 13.8 to 38.6% (n = 3) using the K5 promoter and from 1.06 to 1.76% using the CMV promoter (n = 4). Among α6-integrin positive cells, which comprise the basal layer keratinocytes, an average of 48.2% of the cells were GFP positive using the K5 promoter, compared to only 1.23% using the CMV promoter. In the CD34+/α6-integrin+ subset of cells, which defines the bulge stem cell population, an average of 13.4% versus only 0.41% of cells were GFP positive using the K5 and CMV promoters, respectively (Figure 3k).
To assess the possibility of gene silencing with each promoter we performed quantitative polymerase chain reaction (quantitative PCR) for the transgene in fluorescence-activated cell sorting (FACS) purified GFP+ and GFP− basal layer cells and bulge stem cells after IAGT. Figure 3l shows the normalized data derived from genomic DNA extracted from these four cell populations. Although the EGFP transgene was quantitatively higher with both promoters in the GFP-positive populations relative to the GFP-negative populations, EGFP transgene was detectable in all four cell populations, suggesting that a significant amount of in vivo gene silencing occurs with either promoter. This data supports an even higher efficiency of gene transfer than represented by the quantitative FACS data and suggests that further optimization of the vector systems employed might improve results.
In this murine study, we demonstrate that IAGT using lentiviral vectors can result in efficient, diffuse and sustained gene transfer to cutaneous epithelial lineages. With appropriate timing of vector administration, gene transfer to stem cell populations that generate and maintain the skin and skin appendages can be achieved. To our knowledge, this is the first report that demonstrates widespread transduction of skin stem cell/progenitor populations by any in vivo gene transfer strategy.
Our results are best interpreted in the context of normal skin development.23 During early murine development until E9, the skin is comprised of only a single cell layer. Focal thickening of the epidermis next occurs, creating the epidermal placodes.24 The epidermal placodes then signal the mesenchyme to aggregate beneath the placodes.25 Once aggregated, mesenchymal–epidermal interaction induces the development of the skin appendages.24 In our study, we achieved gene transfer to all of the skin appendages by early IAGT. This finding suggests that early IAGT results in gene transfer to epidermal stem cells prior to the differentiation of the epidermal placodes. A second event in fetal skin development is formation of the periderm, an epithelial layer covering the emerging epidermis in early embryogenesis, around E9.5.23,26 The periderm is a transient layer of cells, interfacing between the developing epidermis and amniotic fluid. The periderm differentiates itself in tandem with epidermal development, completes its formation by E12, and is sloughed into amniotic fluid by E17 as the skin becomes fully keratinized.27 The epidermis starts to stratify at the end of E9 and is fully differentiated by E18.5. Thus, our inability to transduce skin stem cells after E9, and inability to transduce any skin epidermal cell population after E12, corresponds with the formation of the periderm and stratification of the epidermis that presumably acts as a physical barrier. Similarly, the pattern of loss of transduction with gestational age corresponds with the pattern of periderm formation. The periderm completes its formation on the limb buds first, followed by the dorsal and ventral trunk regions and scalp.28,29 These observations support our hypothesis of a unique developmental “window of opportunity” during which skin associated stem cells are accessible for gene transfer.
The transduction of stem cells is of obvious importance with respect to the experimental and therapeutic potential of this approach. There have been two types of stem cells/progenitors identified in the skin; the epidermal basal cell and the bulge stem cell.30–34 These populations play important, but distinct roles in skin maintenance, regeneration, and repair.30,31 Recently, using transgenic mice with an inducible reporter gene under control of the K15 promoter, Ito et al. demonstrated that bulge stem cells contribute to early stages of wound healing but not to homeostasis of the epidermis, and that stem cells in the basal layers contribute to maintain the homeostasis of the epidermis.30 Several observations in our study indicate that we transduced both populations of stem cells. First, the longevity of epidermal GFP expression argues strongly for stem cell transduction. The maintenance of epidermal GFP expression beyond 6 months is clearly derived from transduction of nascent epidermal basal stem/progenitor cells. Second, after making a full-thickness wound in the skin, GFP-positive cells migrated from single hair follicles at the wound edge towards the center of the wound. This pattern is identical to that described by Ito et al. in observing the bulge cell response to a similar wound. Finally, we found significant levels of gene transduction in sorted cell populations highly enriched for each stem/progenitor cell type.
Our initial observations using a lentiviral vector driven by the CMV promoter demonstrated diffuse expression in the suprabasal layers of the epithelium, and a robust contribution of transduced stem cells to wound healing. This was inconsistent with our quantitative finding of only 0.41 and 1.23% of isolated bulge and basal stem cells respectively, expressing GFP. A possible explanation for this discrepancy could be that the stem cell populations were efficiently targeted but that promoter specificity limited transgene expression to the superficial epithelial strata. This is consistent with previous data that examined transduction of keratinocytes in regenerated human skin with a lentiviral vector driven by the CMV promoter; here, GFP expression was seen only in the suprabasal layers, which arose from transduced basal stem cells.14 In the hopes of achieving greater basal layer gene expression, we compared vectors driven by the CMV and the K5 promoters. In contrast to CMV, the K5 promoter is expressed only in ectodermal tissues with high expression in the basal layer of the epidermis.21 In this study, there were distinct differences in the distribution of epidermal expression between the vectors. Expression with the K5 promoter driven vector was very high in the basal epithelial cells of the epidermis, whereas expression driven by the CMV promoter was primarily in the superficial epidermal layer. Our FACS data quantitatively confirms this difference, as most of the collected cells are α6-integrin positive basal layer cells (63–71% of total cells).
Another factor that may result in underestimation of the true transduction efficiency is in vivo silencing. Mechanisms of silencing for retroviral vectors are thought to be both genetic and epigenetic. Epigenetic mechanisms may silence gene expression after expression is established by de novo DNA methylation within the promoter region or other mechanisms resulting in a repressive chromatin configuration.35–37 In contrast to epigenetic silencing, genetic mechanisms result in primary failure of expression and relate to the integration site but not to epigenetic changes of the integrated element.38,39 It has been recently documented in a study designed to ensure only single round infection kinetics and the expression of only integrated transgene, that only a minority of human immunodeficiency virus-1-based vector integrations express transgene40 but that once transgene expression is established it is likely to remain stable favoring a predominantly genetic mechanism of silencing, at least in this defined system. This would support the concept that the likelihood of gene expression as well as the strength of expression would depend upon the number of copies integrated. Consistent with these concepts we found the EGFP transgene in all of the cell populations to be independent of GFP expression, suggesting better transduction efficiency than represented by the histology and FACS analysis, and significant levels of in vivo gene silencing with both promoter constructs. Higher levels of expression in stem cell populations using the K5 promoter reflect the relative specificity of the K5 promoter for expression in basal layer cells. Therefore the choice of promoter for therapeutic application to congenital skin diseases would depend upon which level of the epidermis one wished to target. For EB, that affects the basal layer, the K5 promoter would appear to be a better choice.41,42
Our demonstration of high efficiency transduction of skin stem cell compartments using the IAGT approach addresses the need for an in vivo methodology for manipulating skin gene expression. From an experimental and biological perspective, combination of this approach with technologies such as RNAi, Cre/loxP, and drug inducible On/Off systems will provide versatile systems for elucidating mechanisms of skin differentiation, regeneration, wound healing, and scarring. From a therapeutic perspective, this approach may allow prenatal treatment of genetic skin disorders such as EB in which defective protein synthesis occurs at the dermal, epidermal interface. However, there are a number of hurdles that would need to be overcome prior to any clinical application of this approach. First, although we have chosen to focus on skin gene transfer in this study, IAGT is not specific for skin. While some selectivity occurs due to the limited interface between amniotic fluid and other fetal tissues, we have documented multiple other epidermal tissues that are transduced by IAGT.20 This raises obvious concerns about the potential for insertional mutagenesis, developmental effects, and the potential for germ line alteration that exists for lentiviral vector based approaches. These dangers are only heightened by early gestational transduction.43 While greater tissue specificity and safety can probably be accomplished by the use of tissue specific promoters, or regulated transgene expression, safer gene transfer techniques will need to be developed to alleviate these concerns. The second major impediment is that stage for stage, the timing of our early gestational injections between E8 and E11 in the mouse, corresponds to the 21st to 55th day of gestation in human fetal development,23 a time in pregnancy that precedes current capabilities for prenatal diagnosis. Nevertheless, in the foreseeable future, prenatal diagnosis may allow for diagnosis of genetic disorders during this period of gestation. Finally, because of the early and relatively high efficiency of gene transfer to stem cell populations using this approach, it may offer a relatively rigorous assay for safety studies on vector based gene transfer.
Balb/c mice were mated in our breeding colony (breeding stock purchased from Jackson Laboratories, Bar Harbor, ME) to achieve accurate time-dated pregnancies. Animals were inspected daily and the day of appearance of the vaginal plug was taken as E0. Pregnant mice from E8 to E18 were utilized for IAGT. Animals were housed in the Laboratory Animal Facility of the Abramson Pediatric Research Center at The Children’s Hospital of Philadelphia and were maintained in sterilized plastic micro isolator cages and given sterilized standard laboratory chow and tap water ad libitum. All experimental protocols were reviewed and approved by The Institutional Animal Care and Use Committee at The Children’s Hospital of Philadelphia, and followed guidelines set forth in the “Guide for Care and Use of Laboratory Animals” by the National Institutes of Health.
We used an ultrasound guided injection system (Vevo 660; VisualSonics, Toronto, Canada) for intraamniotic vector injection of E8–E12 pregnant mice. The anesthetic and surgical methods are as we have previously described.20,44
We used stereoscopic fluorescence microscopy (MZ16FA; Leica, Heerburg, Switzerland) for analyzing the GFP expression sites in the vector-injected mice. For this study the skins were first examined in situ and then the hair over the whole skin was removed using a chemical hair remover (Nair Church & Dwight Co., Princeton, NJ). Skins were analyzed at serial time points after birth; the newborn period (P1–P7), neonatal period (P8–P30) and adult period (P30~). To allow comparison of the fluorescence intensity, we maintained the exposure time at 1 second for all photographs.
Tissue specimens collected for histology and immunohistochemistry were fixed in 10% buffered formalin solution and embedded in paraffin. To evaluate and localize GFP protein in the harvested tissues, 4 mm sections were obtained using a paraffin microtome (Leica RM2035; Leica Instrument, Germany). Paraffin sections were incubated overnight at 55 °C and then deparaffinized in serial xylene washes, followed by rehydration through a graded alcohol series to deionized water. To quench autofluorescence caused by free aldehydes, slides were placed in 1% phosphate-buffered saline (PBS) for 10–20 minutes. After rinsing in PBS, slides were blocked for specific protein with goat serum (1:10 dilution) for 30 minutes at room temperature followed by 30 minutes incubation with monoclonal rabbit anti-GFP immunoglobulin G fraction (1:200 dilution; Molecular Probes, Eugene, OR) at 4 °C. The slides were then washed with PBS followed by a peroxidase blocking step with Dakocytomation (S-2001; Dako, Carpenteria, CA) for 30 minutes at room temperature. Slides were rinsed with deionized water, then PBS, followed by incubation with biotinylated goat antirabbit immunoglobulin G (1:200 dilution, Vector Lab PK-4001; Vector Lab, Burlingame, CA) for 30 minutes at room temperature. The slides were washed with PBS, and avidin–biotin complex (1:200 dilution; Vector Lab, Burlingame, CA) was added for 30 minutes at room temperature. The slides were rinsed well in PBS, developed with peroxidase substrate kit (SK-4100; Vector Lab, Burlingame, CA) and lightly stained with Harris hematoxylin, dehydrated in alcohol, cleared in xylene and mounted using Acrymount (Statlab Medical Products, Lewisville, TX).
The basic starting materials11,45 for generating a self-inactivating human immunodeficiency virus-1-based vector were kindly provided by Inder Verma (Salk Institute, La Jolla, CA). Modifications of the CS-CG vector included deletion of the remaining right U3 region except for 23 nucleotides (nt) downstream of the 3′ppt, deletion of the residual envelope and ancillary gag/pol sequences, insertion of the central DNA FLAP,46 insertion of the Rev response element, and insertion of the Woodchuck hepatitis virus posttranscriptional regulatory element,47 which is modified eliminating the initiation codon for the Woodchuck hepatitis X protein. The EGFP (Clontech Laboratories, Palo Alto, CA) was located downstream of the human CMV immediate early promoter and modified so that all stop codons from the transcription start site to the translation initiation site were removed. Viral vectors pseudotyped with the vesicular stomatitis virus envelope were generated as previously reported.48
A human K5 promoter (−812 to +94) was amplified from genomic DNA with primers 5′-ACGCGTGATCCCCGGGTTTCCTAAACC-3′ and 5′-GCGGCCGCGGCTTGTTCCTGGTGGAGCAAGAGAAC-3′ in which an MluI site was introduced into the 5′- end and an NotI site was introduced into the 3′-end to simplify insertion in the expression vector.48
We used 8-week-old mice after IAGT with injection of the lentiviral CMV vector at E8–E11 for this experiment. Under general isoforane anesthesia, mice were clipped of hair and a 4 mm diameter punch biopsy was performed on the skin of their back. One week later, under general anesthesia, the healing wound was excised and after removing the subcutaneous tissue,30 both surfaces of the wound were analyzed under fluorescence stereomicroscopy for the presence of GFP-positive cells.
Keratinocytes were harvested from the dorsal skin of 50 to 60-day-old E8 IAGT mice following previously described protocols.49 We used anti-human CD49f-PE (α6-integrin), biotinylated anti-CD34 (BD Pharmingen, San Diego, CA) and streptavidin-perCP5.5 to select basal keratinocytes or hair follicle bulge cells. Antibodies and cells were incubated at 4 °C for 25 minutes. Cell sorting was carried out on a FACS Vantage cell sorter (Becton Dickinson, Mt. View, CA). Sorted populations were gated on the bulge stem cell population (CD49f+/CD34+) with or without GFP and the basal layer population (CD49f+/CD34−) with or without GFP. The purity of sorted populations routinely exceeded 99%. Data were analyzed using Cell Quest software (Becton Dickinson, Mt. View, CA).
For relative quantification of EGFP gene in skin stem cells, primer-probe sets and TaqMan MGB probes using the dyes 6-carboxyfluorescein (excitation, 494 nm), both of the target gene (EGFP) and the internal control gene (GAPDH) were made using the Applied Biosystems design service (Foster City, CA). The EGFP sequences are as follows: carboxyfluorescein probe, 5′-ACA GCC ACA ACG TCT-3′; forward primer, 5′-GGG CAC AAG CTG GAG TAC AAC T-3′; reverse primer, 5′-TCT GCT TGT CGG CCA TGA-3′. Real-time PCR was performed using 12.5 µl of TaqMan 2x universal master mix (Applied Biosystems, CA), 1 µl of primer-probe (22.5 and 6.25 µmol/l, respectively), 4.5 µl of DNase-free water (Sigma), and 5 µl of sample genomic DNA, in a total volume of 25 µl per single tube reaction using Applied Biosystems 7000 Real-Time PCR system (Applied Biosystems, Foster City, CA). Three wells of a 96-well plate (Applied Biosystems, Foster City, CA) were used for each sample. DNase-free water was used as a non-template negative control and genomic DNA of skin of GFP transgenic mice was used as a positive control and was included in each assay run. Assay conditions were 2 minutes at 50 °C, 10 minutes at 95 °C, and 40 cycles of 95 °C for 15 seconds and 60 °C for 1 minute. The relative quantification assay was set up using SDS, version 2.1 software (Applied Biosystems, Foster City, CA) and post assay analysis was also performed using the SDS software as described previously.50 We normalized the EGFP amount in each sample relative to 2−ΔΔCT of skin of EGFP transgenic mice. The error bars represent relative quantification maximum.
In the relative quantification PCR assay, three independent samples were performed for each of the four cell populations. Data for the relative quantification of each population were normalized by 2−ΔΔCT of skin of EGFP transgenic mice. The error bars represent relative quantification maximum.
This study was supported in part by the Ruth and Tristram C. Colket Jr. Chair in Pediatric Surgery (A.W.F.). We thank Keith Alcorn (Children’s Hospital of Philadelphia) for animal support, Eric Rappaport (Children’s Hospital of Philadelphia) for quantitative polymerase chain reaction and Tiago Henriques-Coello (Children’s Hospital of Philadelphia) for ultrasound technical support.