¹Department of Microbiology and Immunology, University of Texas Medical Branch, Galveston, TX, USA

²Department of Biological Science, University of Delaware, Newark, DE, USA

³Gene-Cell Inc., Houston, TX, USA

Correspondence to: BR Davis, Gene-Cell Inc, 1010 Hercules Avenue, Houston, TX 77058, USA

Chimeric oligonucleotides have been used successfully to correct point and frameshift mutations in several cell types, as well as in animal and plant models. However, their application to primitive human blood cells has been limited. In this investigation, chimeric oligonucleotides designed to direct a site-specific nucleotide exchange in the human -globin gene (an A to T substitution within codon 6) were introduced into normal human CD34⁺ and Lin^-CD38^- cells via microinjection. This A to T nucleotide exchange introduces the single site mutation responsible for sickle cell anemia. In 23% of experimental samples, gene conversion was detected in the progeny of microinjected CD34⁺ and Lin^-CD38^- cells that were cultured for at least 4 weeks. In addition, gene conversion was detected in the erythroid progeny of Lin^-CD38^- cells at the mRNA level. Conversion rates as high as 10-15% in 11% (five of 44) of experimental samples were confirmed by allele-specific PCR and sequence analysis of genomic DNA from the progeny of microinjected Lin^-CD38^- cells. Given that as few as 10% normal hematopoietic cells are sufficient to keep patients free of sickle cell disease, the level of conversion we have achieved in some samples may well be of therapeutic benefit in patients with sickle cell disease.

Gene Therapy 2001 9, 118-126. DOI: 10.1038/sj/gt/3301610

gene repair; sickle cell disease; chimeric oligonucleotide; microinjection

Introduction

Sickle cell anemia is a hereditary hemoglobin disorder caused by a point mutation (A to T) in the sixth codon of the -globin gene. A major consideration in the treatment of any hemoglobinopathy is the maintenance of the physiological and tightly regulated erythroid-specific expression of the normal globin gene cluster. Two possible approaches can be envisioned for gene therapy treatment of hemoglobinopathies such as sickle cell anemia: either delivering a replacement copy of the relevant globin gene (eg beta or gamma globin) or actually repairing the globin gene. The complex and coordinate regulation of globin gene expression in vivo raises significant difficulties in treating sickle cell anemia by adding a normal copy(s) of the gene using retroviral vectors or adeno-associated viral vectors. Repairing the mutation itself within the -globin gene would maintain the correct genetic material within its normal chromatin environment; in principle, ensuring appropriated genetic regulation and expression in the progeny erythroid cells.

A strategy to take advantage of the efficient and specific DNA repair process in mammalian cells has been previously developed. The concept centers around the use of a synthetic vector which catalyzes nucleotide exchange at a specific site within a gene via endogenous repair pathways.¹ Vectors, known as chimeric RNA/DNA oligonucleotides, were previously demonstrated to mediate 'gene repair' at the -globin locus in ^SS lymphoblastoid cells.² The specificity of the chimeric oligonucleotide-mediated gene repair was evident by the absence of gene conversion in the closely related globin gene.² Independent groups confirmed this approach in the -globin gene³ and at other loci in vitro.^4,5,6,7,8 Significant proof-of-principle experiments have been conducted successfully in vivo in organisms including rat,⁹ mice,^10,11 dog¹² and yeast.¹³ Taken together, these studies not only demonstrate that a variety of genes in various cell types are amenable to targeted gene conversion, but also reveal the inheritable nature of chimeric molecule-mediated gene conversion.^5,9,11

In this study, we examined the efficacy of chimeric oligonucleotide-mediated gene conversion in human CD34⁺ cells and Lin^-CD38^- cells via microinjection. Glass needle-mediated microinjection of primary human blood stem/progenitor cells is a novel approach that allows delivery of transgene and other macromolecules directly into the nucleus without causing loss of progenitor¹⁴ or stem cell activity (unpublished results). In addition, glass needle-mediated microinjection can deliver a controllable number of repair molecules directly to the nuclei of injected cells with minimal cell-to-cell variation in the number of molecules delivered. Finally, microinjection has the potential to co-deliver the repair oligonucleotide with other accessory molecules that will enhance the conversion process. Here, we evaluated microinjection-mediated delivery of chimeric RNA/DNA oligonucleotides to purified CD34⁺ cells, as well as to a more primitive population (Lin^-CD38^-). Conversion from A to T within codon 6 of the -globin gene at a frequency ranging from 1-15% was observed in three out of 13 CD34⁺ samples and 10 out of 44 Lin^-CD38^- samples following microinjection of the chimeric oligonucleotide. Erythroid progeny, derived from human blood Lin^-CD38^- cells in vitro after microinjection, showed sickle sequence conversion at both the DNA and mRNA level.

Results

Microinjection-mediated delivery of oligonucleotides into hematopoietic CD34⁺ and Lin^-CD38^- cells

Genetic correction of rare, primitive blood stem cells is crucial to long-term treatment/cure of blood cell diseases. We therefore focused on microinjection-mediated delivery of oligonucleotides to populations of human umbilical cord blood cells enriched in stem cell content. Initial experiments were performed with the total CD34⁺ population (approximately 0.5-1% of cord blood mononuclear cells; approximately 1 in 2 ´ 10⁴ CD34⁺ cord blood cells successfully engraft in immune deficient NOD/SCID mice¹⁵), previously shown to contain the vast majority of human hematopoietic stem and progenitor cells. An even more primitive population of blood cells, negative both for various lineage-specific antigens and CD38 (Lin^-CD38^-), was also employed in experiments. The Lin^-CD38^- cells, 80-90% of which are CD34⁺, are even more enriched for blood stem cells (both those of the CD34⁺ and CD34^- phenotype; approximately 1 in 600 CD34⁺CD38^- cells engraft NOD/SCID mice¹⁶).

We first determined the post-microinjection viability of CD34⁺ and Lin^-CD38^- cells injected with OGD (Materials and methods, Figure 1) The viability of CD34⁺ and Lin^-CD38^- cells at 2 h after microinjection was 76% ± 23% (mean ± s.d., n = 12) and 65% ± 21% (mean ± s.d., n = 21), respectively. Similar viability was observed in our previous report,¹⁴ which also demonstrated that microinjected CD34⁺ and CD34⁺CD38^- cells retained their proliferative capacity and the ability to generate myeloid and erythroid progeny.

We also demonstrated successful delivery of oligonucleotides to blood cells. A 25 base DNA oligonucleotide (with no homology to -globin gene), conjugated to the Texas Red fluorochrome (uD3T/TR) and OGD were co-injected into Lin^-CD38^- cells. As shown in Figure 2, both OGD Figure 2b and uD3T/TR Figure 2c co-localize in the same injected cells Figure 2d at 2 h after injection. Moreover, microinjection allows for direct delivery of oligonucleotides to the nuclei of Lin^-CD38^- cells (Figure 2c, arrows). The minimal variation of fluorescent intensity among the injected cells indicates that microinjection can deliver a consistent number of oligonucleotides to each cell.

Chimeric RNA/DNA oligonucleotide mediated gene conversion in CD34⁺ and Lin^-CD38^- cells

The ability of chimeric RNA/DNA oligonucleotides specific for the sickle cell -globin mutation to accomplish gene conversion in microinjected CD34⁺ and Lin^-CD38^- cells was examined. Cells were isolated from pooled samples of human umbilical cord blood. For these experiments, an 88 base chimeric oligonucleotide (SC2/88MD-3) designed to convert an A to T (normal ^A to sickle ^S) mutation in the -globin gene was employed. Approximately 300 to 500 CD34⁺ cells were temporarily immobilized on RN-coated dishes and injected with SC2/88MD-3. Assuming an approximate microinjected volume of 0.5 femtoliter, 500-2500 copies of oligonucleotide were delivered per cell in the majority of experiments (). After detachment and transfer, injected cells were cultured for 1 week in S media to promote expansion of stem/progenitor cells followed by an additional 3 weeks in SM medium to promote myeloid expansion. Genomic DNA was isolated and used for PCR and restriction digestion analysis. Detection of the normal to sickle conversion is facilitated by specific loss of DdeI (and Bsu36 I) restriction site (Figure 3a). As shown in Figure 3b, three of the four experimental samples (lanes 1-4; these four experimental samples were from the same pool of cord blood cells) showed incomplete DdeI digestion, suggesting gene conversion from normal to sickle -globin sequence. In addition, the failure to completely digest with DdeI has been closely paralleled in Bsu36 I digestion analysis, ie whose recognition site is similarly destroyed by the normal to sickle conversion. Approximate conversion rates of 1 to 10% were determined by comparing the intensity of the incompletely digested 297 bp fragment from the three positive experimental samples with controls (containing ^S/^S SC-1 in a background of ^A/^A Lin^-CD38^- cells at a 0%, 1%, 10%, and 100% frequency; Figure 3b, lanes 5-8). Since the sensitivity of this DdeI analysis is approximately 1% conversion (Figure 3b, lanes 5-8), we are unable to determine whether the absence of an undigested band in one of the four samples (Figure 3b, lane 1) reflects a complete absence of conversion or simply conversion below the 1% rate. Confirmation that the loss of DdeI digestion site reflected the intended A to T -globin gene conversion was provided by sequence analysis of cloned PCR products. One out of 14 total clones analyzed from two positive samples (Figure 3b, lanes 3 and 4) had the specific A to T sickle -globin conversion (7% conversion rate, Figure 4 legend). Thus, the 7% conversion rate obtained by sequencing is consistent with the estimated conversion rate of 5-10% determined by DdeI digestion Figure 3b. None of the five clones from the unconverted sample (Figure 3b lane 1) showed conversion (data not shown).

Similar conversion frequencies were obtained when the SC2/88MD-3 molecule was injected into the more primitive Lin^-CD38^- cells. Following microinjection, cells were expanded either in the culture conditions previously described (S medium SM medium) or in culture conditions designed to favor first proliferation and differentiation of both myeloid and erythroid cells (ME medium) followed by erythroid differentiation (E medium) (Figure 1 and ). The DdeI restriction enzyme analysis for Lin^-CD38^- cells is shown in Figure 3c, lanes 1-8 (experimental samples 1-3, 4-5, and 6-8 represent three separate pools of cord blood cells). As expected, Lin^-CD38^- cells injected with OGD alone (no SC2/88MD-3) showed no conversion (Figure 3c, lane 1). Lin^-CD38^- cells injected with SC2/88MD-3 with or without OGD showed various levels of conversion ranging from undetectable (lane 6), to approximately 1% (lane 3), to 5-10% (lanes 2, 4-5, 7-8). Similar results were obtained from two separate analyses. TA-cloning and sequence analyses were used to confirm normal (A) to sickle (T) -globin gene conversion in two positive samples (Figure 3c, lanes 4 and 5). Three out of 20 clones analyzed from these two samples had the specific A to T conversion (15% conversion rate; Figure 4 legend). TA-cloning and sequence analyses of PCR product from cells injected with OGD alone confirmed none of the 10 clones analyzed showed sickle -globin sequence.

Expression of ^S mRNA by the erythroid progeny of microinjected Lin^-CD38^- cells

To demonstrate that the normal to sickle gene conversion in genomic DNA is reflected in the transcribed -globin mRNA, we cultured SC2/88MD-3 injected Lin^-CD38^- cells under the conditions described above to promote erythroid lineage development - the ultimate cell lineage in which -globin is expressed. Erythroid development in these cultures was confirmed in two ways. Firstly, increased time in erythroid culture conditions yielded significant number of cells with a red appearance - evidence of hemoglobin production (data not shown). Secondly, expression of the erythroid marker glycophorin A on in vitro derived erythroid cells was assessed. As shown in Figure 5, less than 2% of the starting Lin^-CD38^- cells expressed glycophorin A, whereas more than 73% of cultured cells expressed glycophorin A, indicating erythroid development in the culture.

Both DNA and RNA were isolated from erythroid culture cells and used for PCR and RT-PCR analysis, respectively. Figure 6 highlights the PCR/DdeI analysis of both genomic DNA and mRNA for two experimental samples (from two separate pools of cord blood cells) in which Lin^-CD38^- cells were injected with SC2/88MD-3. DdeI digestion of amplified genomic DNA demonstrates that one of the samples (Figure 6b, lane 1) showed conversion at an approximately 10% level (Figure 6b, lanes 3-5 contain 1%, 10%, 100% standards), whereas the other sample showed no detectable conversion (Figure 6b, lane 2). We confirmed that the 10% incomplete DdeI digestion observed in the experimental sample (Figure 6b, lane 1) reflected the intended normal to sickle conversion by the following methods. First, TA-cloning of the 345 bp PCO2/PCO5 DNA PCR product and analysis of 95 individual clones by allele-specific PCR using upstream primers specific for either the normal or sickle -globin allele revealed that 10 of 95 clones contained the targeted normal to sickle DNA sequence conversion, while 85 of 95 clones represented normal -globin sequence (data not shown). Control amplification verified that this technique is accurate: normal Lin^-CD38^- cells only amplify with the normal upstream primer, and SC-1 cells only amplify with the sickle upstream primer. The frequency (10/95 = 10.5%) of conversion is consistent with the 10% frequency deduced from DdeI digestion. We have confirmed the accuracy of the allele-specific PCR analysis by sequencing four 'sickle' clones and two 'normal' clones. Sequencing revealed the expected sickle or normal sequence with no other mutation identified (data not shown).

For analysis of -globin mRNA, primers D and R which span intron 1 of the -globin gene Figure 6awere used to differentiate the cDNA-derived PCR product (353 bp) from the genomic DNA-derived PCR product (482 bp). After RT-PCR and DdeI digestion, the presence of a 216 bp fragment can be used to differentiate sickle mRNA from normal mRNA Figure 6a. RT-PCR yielded the anticipated 353 bp -globin product from cDNA (Figure 6c, lanes 3 and 4) derived from the experimental samples previously shown in Figure 6b, lanes 1 and 2. As shown in Figure 6c lane 1, the sample which demonstrated normal to sickle conversion at the DNA level (Figure 6b, lane 1), also showed conversion at the mRNA level (with an approximate conversion frequency of 2-5%). Similarly, the sample with no detectable conversion at the DNA level (Figure 6b, lane 2) also failed to demonstrate conversion at the RNA level (Figure 6c, lane 2). Direct sequencing of the 216 bp fragment that failed to digest with DdeI confirmed the presence of sequence containing the A to T conversion. We again confirmed the accuracy of the DdeI digestion analysis for the converted sample (Figure 6c, lane 1) by TA-cloning and allele-specific PCR. Analysis of 186 individual RT-PCR clones yielded nine clones with sickle globin sequence for a 4.8% conversion rate. We sequenced both positive and negative clones and confirmed expected sickle or normal -globin mRNA sequence.

Discussion

RNA/DNA chimeric oligonucleotides have been successful in promoting single nucleotide exchange in cell-free extract, cell culture, and animals.^3,17,18 However, the application of this gene repair technology to primitive hematopoietic cells has been hindered by several factors. First, to our knowledge, electroporation or liposome-mediated transfection conditions have not yet been reported which allow for the efficient delivery of DNAs and/or proteins to primary human stem/progenitor cells without significant loss of viability or stem cell function.^19,20 Second, these nonviral methods result in extensive cell-to-cell variation in the number of molecules delivered per cell making consistent delivery of precise quantities of molecules impossible. Glass needle-mediated microinjection was recently developed as a novel gene delivery method for hematopoietic stem/progenitor cell gene modification.¹⁴ In contrast to other gene delivery methods, microinjection allows for the direct delivery of genetic material into the nuclei of hematopoietic stem/progenitor cells, which increases the accessibility of chimeric oligonucleotides to the target genomic DNA.

In this investigation, we tested the efficacy of microinjection-mediated delivery of chimeric oligonucleotides into human hematopoietic CD34⁺ cell and Lin^-CD38^- cells. Delivery of fluorescent molecules (ie Oregon Green-conjugated dextran and Texas Red-conjugated oligonucleotide) demonstrated nuclear localization of injected material in most cells. Furthermore, delivery to both CD34⁺ and LinC^-D38^- populations was achieved with excellent post-injection viability (76% ± 23% and 65% ± 21%, respectively). We previously demonstrated that microinjected CD34⁺ and CD34⁺CD38^- cells exhibited excellent survival rates, retained their ability to proliferate, and produced a similar number of progeny as noninjected cells.¹⁴ In addition, microinjected CD34⁺CD38^- cells generated colony-forming cells capable of producing both erythroid and myeloid progeny - at rates indistinguishable from that of uninjected cells.¹⁴ Since approximately 90% of Lin^-CD38^- cells are actually CD34⁺CD38^- Figure 5a, we fully expect that microinjected Lin^-CD38^- cells retain their biological function. In preliminary experiments, we have recently demonstrated that microinjected cord blood Lin^-CD38^- cells retain their stem cell activity as measured by engraftment in the NOD/SCID immunodeficient mouse assay (data not shown).

Although the ultimate goal is to correct the sickle cell anemia mutation in stem cells obtained from sickle patients, we chose to first demonstrate proof-of-principle by introducing the sickle mutation into normal cord blood cells. In contrast with earlier work, where only CD34⁺ cells were used for evaluating gene conversion,²¹ we evaluated both CD34⁺ cells and the more primitive Lin^-CD38^- cell population for the efficiency of gene repair mediated by chimeric oligonucleotides. The sickle mutation was successfully introduced in both the CD34⁺ and the Lin^-CD38^- populations, indicating that even the more primitive cells can be targeted and repaired by chimeric oligonucleotides. Further, since significant -globin expression does not occur until erythroid differentiation, these results strongly suggest that active transcription of the target gene (ie -globin) in the microinjected hematopoietic cells is not required for efficient conversion. Progeny of the injected CD34⁺ cells and Lin^-CD38^- cells showed -globin gene conversion at the DNA level 3-4 weeks (8-12 cell doublings) after microinjection, demonstrating that the targeted nucleotide change was passed on to subsequent generations. These finding are consistent with previous observation of gene correction mediated by chimeraplasts in other cell types.⁵ This demonstration of the inheritable nature of sickle conversion introduced in Lin^-CD38^- cells represents a significant advance over the earlier report, in which gene conversion was evaluated only at 16 h after as transfection.²¹ In addition, our report presents the first evidence that the ^A to ^S gene conversion is reflected in -globin mRNA transcribed in erythroid progeny. Further analyses of -globin proteins (eg by HPLC or immunostaining with antibodies specific for sickle globin) in in vitro differentiated red blood cells will be required to verify the presence of sickle -globin protein. In addition, our in vitro experimental results cannot tell us whether or not the -globin gene in true stem cells has undergone conversion. For example, even our Lin^-CD38^- cell samples, enriched in human blood stem cell activity, only contain approximately one cell in 600 capable of engrafting immunodeficient NOD/SCID mice.¹⁶ In vivo experiments designed to answer this question (eg reconstituting the bone marrow of a NOD/SCID mouse with modified human blood stem cells) are currently underway in our laboratory. Since the target cells were not absolutely pure for CD34⁺ or Lin^-CD38^- cells at time of injection (due to presence of contaminating cells in the initially isolated cells, as well as some maturation/differentiation of blood stem/progenitor cells during culture before injection, see ), we cannot state unequivocally that it was only the most primitive cells (eg CD34⁺ or Lin^-CD38^- cells) that were genetically converted in these experiments. However, we did not observe significantly higher conversion rates as cells were maintained for longer times in culture before injection. For example, Lin^-CD38^- samples maintained in S medium for 2, 3 or 4 days demonstrated conversion in three of 17 samples (18%), in two of nine samples (22%), or two of eight samples (25%), respectively. If only maturing or mature cells were targets for conversion, we would have expected a more significant trend toward higher conversion rates with longer times in culture.

Assessment of the gene conversion rate was determined both by DdeI digestion and by analysis of individual TA-clones by allele-specific PCR and sequencing. In order to estimate the conversion rate, the intensity of the band that failed to be digested with DdeI (due to the sickle conversion) was compared with that of a group of standards that contain 0-100% SC-1 cells (^SS) in a background of normal cells (^AA). With this procedure, our sensitivity for detection of conversion was approximately 1%. Twenty-three percent of experiments using either CD34⁺ cells (three out of 13 experiments) or Lin^-CD38^- cells (10 out of 44 experiments) showed conversion at a 1-15% frequency, as determined by DdeI digestion.

Analysis of individual TA-clones generated from amplified -globin gene sequence yielded similar conversion rates as determined by DdeI digestion. We have not yet determined the reason(s) for the observation of significant (ie 1%) repair in some experiments but not others. For example, the presence or absence of conversion did not correlate with parameters such as viability after injection, the percentage of plated cells that were injected, efficiency of nuclear delivery, the concentration of oligonucleotides injected, or whether fresh versus frozen cells were utilized. In addition, it is possible that some conversion did occur in the 'negative' samples but was simply less than our current limits of detection. At this time, we believe that the two most likely causes for variation in conversion rate are heterogeneity from one cord blood sample to another and heterogeneity in the proliferative ability of individual Lin^-CD38^- cells. We found that certain pools of cells were more amenable to gene conversion than others. For example, in CD34⁺ cell experiments, all of the three experiments that showed conversion used the same pool of cord blood cells, whereas no conversion was detected in two other pools of CD34⁺ cells. In Lin^-CD38^- experiments, all the conversion occurred in five out of 11 pools of cord blood cells we examined. Since microinjection allows for direct delivery of molecules to nuclei of cells, this indicates that differences in gene repair from cell sample to sample is not due to differences in efficient delivery of molecules - that for instance would be the case for electroporation or liposome-mediated delivery. Rather, the clustering of repair in only certain cell samples strongly suggests that there are intrinsic differences from cell sample to cell sample with regard to 'repairability'. Cells from different donors may have different levels of repair or homologous recombination proteins, which may affect the ability of chimeric oligonucleotides to induce gene conversion in certain samples. For example, Cole-Strauss et al¹⁸ found that reduced expression of hMSH2 correlated with significantly reduced gene conversion mediated by chimeric oligonucleotides. Future studies will seek to regulate the repair efficiency in cells, both in order to increase the repair frequency in 'permissive' cells, as well as to enable repair in those cell samples currently showing no conversion. We believe that the extreme heterogeneity in the proliferative ability of individual Lin^-CD38^- cells likely contributed to the variability in observed conversion rates. In single cell culture, some Lin^-CD38^- cells give rise to thousands of progeny, whereas others give rise to few progeny (with 10-20% of the individually plated cells giving rise to 80-90% of the total progeny; Davis et al, data not shown). Therefore, if the conversion occurred in highly proliferative Lin^-CD38^- cells, the conversion rate, determined for the culture as a whole, would be much higher than had the conversion occurred in Lin^-CD38^- cells with poor proliferative ability. Since we only injected several hundreds of cells in each experiment, such variation is not unexpected.

We do not yet know what significance to attribute to the difference in conversion rates observed at the DNA (approximately 10% conversion rate) versus mRNA level (approximately 5% conversion rate) for the sample shown in Figure 6b, lane 1. Several factors could contribute to this discrepancy. It is conceivable that unconverted erythroid progenitor cells may have a proliferation advantage over converted erythroid progenitor cells. In addition, since -globin mRNA is only expressed at later stages of erythroid development, early erythroid precursors which have -globin gene conversion at the DNA level would not contribute converted -globin mRNA.

In this study, we achieved conversion rates as high as 10-15% in 11% of experimental samples. This conversion of 10-15% of the -globin alleles represents conversion in a minimum of 10-15% of cells if two alleles converted per cell (from ^AA to ^SS) to a maximum of 20-30% of cells if one allele converted per cell (from ^AA to ^SA). Based on bone marrow transplantation studies, as few as 10% normal peripheral donor blood cells (^AA) will dramatically improve the health of sickle patients.²² Since under normal conditions, cells from sickle cell trait patients (^SA) are functional, transplanted cells modified at only one allele may also have therapeutic benefit. Thus, the level of gene conversion achieved in this study could be of therapeutic benefit in patients with sickle cell disease, provided that the converted cells were maintained at significant frequency in the marrow of transplant recipients. Since the corrected mature red blood cells have prolonged survival advantage over sickle cells, it is likely that the percentage of marrow stem cells requiring correction will be <10%.

In summary, we demonstrated single nucleotide conversion in blood CD34⁺ cells and Lin^-CD38^- cells using chimeric oligonucleotides delivered by microinjection. The introduced sickle mutation is genetically stable, propagates into erythroid progeny, and is transcribed into -globin mRNA. These studies indicate delivery of chimeric oligonucleotides to primitive blood cells via microinjection may be a viable method for treatment of genetic diseases, such as sickle cell anemia.

Material and methods

Synthesis and purification of oligonucleotides

SC2/88MD-3 is an 88-mer chimeric oligonucleotide designed to convert normal A to sickle T in the -globin gene. uD3T/TR is a control 25-mer DNA oligonucleotide (specific for the kanamycin resistance gene) with Texas Red conjugated to the 5' end. Sequences of these two oligonucleotides are shown here with uppercase representing DNA and lowercase

representing 2'-O-methylated RNA. Oligonucleotides were synthesized using phosphoramidites on controlled pore glass supports. After deprotection and detachment from the solid support, each oligonucleotide was gel-purified according to Gamper et al²³ and concentrations were determined spectrophotometrically (33 or 40 g/ml per A₂₆₀ unit of single-stranded or hairpin oligomer).

Isolation and culture of human CD34⁺ and Lin^-CD38^- cells

Mononuclear cells (MNC) were isolated from pools of human umbilical cord blood of normal donors (University of Texas Medical Branch) using Ficoll-Hypaque (Pharmacia Biotech, Uppsala, Sweden) density centrifugation. CD34⁺ cells were immunomagnetically purified from MNC using either the Progenitor or Multisort Kits (Miltenyi Biotec, Auburn, CA, USA). Fluorescence-activated cell sort (FACS) analysis confirmed that 80% to 90% of these cells were CD34⁺. Lin^-CD38^- cells were purified from MNC using negative selection with StemSep system (Stem Cell Technologies, Vancouver, Canada). Lineage markers used for the depletion include CD2, CD3, CD14, CD16, CD19, CD24, CD36, CD38, CD45RA, CD56, CD66b and glycophorin A. FACS analysis of purified Lin^-CD38^- cells showed that more than 95% of cells were CD38^-, and 80-90% of cells were CD34⁺ Figure 5a. For FACS analysis, cells were stained either with CD34 FITC and CD38 CyChrome, or with IgG1 FITC and IgG1 CyChrome (BD Pharmingen, San Diego, CA, USA) as controls, and then analyzed on a FACS Vantage (Becton Dickinson). Cells for microinjection were either freshly isolated or cryopreserved, and cultured in Stem Medium (S medium) for 2 to 5 days before microinjection. S medium contains basal medium (Iscoves' modified Dulbecco's medium without phenol red (IMDM) with 100 g/ml glutamine/penicillin/streptomycin (G/P/S; Gibco BRL)), 1 ´ BIT 9500 (50 mg/ml bovine serum albumin, 50 g/ml bovine pancreatic insulin, 1 mg/ml human transferrin, and IMDM; Stem Cell Technologies) with 40 g/ml low-density lipoprotein (LDL; Sigma), 50 mM HEPES buffer (Gibco BRL), 20 ng/ml each of thrombopoietin (TPO), flt-3 ligand, stem cell factor (SCF), and human interleukin (IL)-6. All cytokines were purchased from PeproTech Inc (Rocky Hill, NJ, USA), unless otherwise stated. CD34⁺, CD34⁺CD38^-, and Lin^-CD38^- cells generally retained their primitive phenotypic profiles after culture in S medium for at least 2 days. Even 4 days of culturing Lin^-CD38^- cells in S medium still yielded a population that was 60% for CD34⁺CD38^- cells. After microinjection, cells were detached and transferred to 48-well plates. For DNA assay, microinjected cells were cultured in S medium for 1 week and then cultured in Stem-Myeloid Medium (SM medium; S medium supplemented with 20 ng/ml each of IL-3, granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF)). For assay of both DNA and RNA, microinjected cells were cultured in myeloid-erythroid medium (ME medium; basal medium supplemented with 40 g/ml LDL, 50 ng/ml SCF, 10 ng/ml IL-6, 20 ng/ml IL-3, 10 ng/ml macrophage colony-stimulating factor (M-CSF), 21 ng/ml G-CSF and 3 units/ml recombinant human erythropoietin (EPO; Stem Cell Technologies)) for 4 days, then cultured in erythroid medium (E medium; basal medium supplemented with 0.001 ng/ml GM-CSF, 0.01 unit/ml IL-3, 10 units/ml EPO) for 3 weeks.

Microinjection of CD34⁺ and Lin^-CD38^- cells

Experimental design is shown in Figure 1. Briefly, 35-mm dishes were coated overnight at 4°C with 50 g/ml fibronectin fragment CH-296 (Retronectin (RN); TaKaRa Biomedicals, Panvera, Madison, WI, USA) in phosphate-buffered saline (PBS) and washed with IMDM containing glutamine/penicillin/streptomycin. Three hundred to 2000 cells were added to cloning rings and attached to the plates for 45 min at 37°C and 2 ml of S medium added to each dish for microinjection. Injection needles were pulled from 10-cm borosilicate capillaries with a 1.2-mm outer/0.94-mm inner diameter using a Flaming/Brown Micropipette Puller Model P-97 (Sutter Instrument, Novato, CA, USA) and had a range of 0.22 to 0.3 outer tip diameter, as determined by scanning electron microscopy. Semi-automatic injections were performed with Eppendorf micromanipulator 5171 and transjector 5246 together with an Olympus IX70 microscope equipped with a Fryer A-50 temperature-controlled stage (set at 37^oC). Thirty to 700 cells (6% to 85% of the attached cells) were successfully injected. A successfully injected cell was defined as a cell that remained alive, intact and attached to the plate after injection. In order to monitor both flow of material and delivery to cells, as well as to assay the viability of cells afater injection, we delivered Oregon Green Dextran (OGD, 10 000 Da; Molecular Probes, Eugene, OR, USA; 0.1 to 1 mg/ml) either alone or together with the chimeric oligonucleotide (SC2/88MD-3, generally 1000 to 5000 copies/fl, but some experiments with 10000-20000 copies/fl) or Texas-Red conjugated oligonucleotide (uD3T/TR, 0.1 mg/ml). Viability (defined as number of fluorescent cells/number of successfully injected cells ´ 100) was assessed 2 h after injection. Alternatively, chimeric molecules were delivered to cells alone. The estimated number of oligonucleotide copies delivered per cell (500-2500 in most experiments) was derived from the approximate volume delivered per cell (0.5 fl; based on the estimated volume expansion of cells during microinjection) and the oligonucleotide concentration. The actual number of molecules delivered per nucleus in a given experiment, utilizing a fixed oligonucleotide concentration, likely varied by a factor of 2-3, determined by the exact volume delivered and the accuracy of nuclear delivery. Following injection, cells were detached from dishes by adding a mixture of FN CS-1 fragment (0.42 mg/ml), H-Arg-Gly-Asp-Ser-OH (1.0 mg/ml) and Phenylac-Leu Asp-Phe-D-Pro-NH2 (1.0 mg/ml; Bachem BioScience, Torrance, CA, USA) and incubating at 37°C for 15 min. Detached cells were transferred to 48-well plates for subsequent culture.

Analysis of genomic DNA for sequence conversion

For genomic DNA analysis, microinjected cells were cultured for 4 weeks. Over this time period, cell numbers grew from the original 300 to 2000 cells to 1-4 ´ 10⁶ cells, representing about 11 population doublings. Genomic DNA was isolated from 1 ´ 10⁶ cells using either QIAamp DNA mini-kit (Qiagen, Valencia, CA, USA) or Proteinase K digestion. SC-1 cells (ATCC#CRL-8756; lymphoblasts homozygous for the sickle cell allele (^SS)²⁴) were mixed with normal Lin^-CD38^- cells at different ratios (1% SC-1, 10% SC-1, 100% SC-1). Genomic DNA isolated from 1 ´ 10⁶ total mixed cells was used as positive control for sickle sequence and used to determine sensitivity of our genomic DNA analysis method. Primers PCO2 (5'TCCTAAGCCAGTGCCAGAAGAG3') and PCO5 (5'CTATTGGTCTCCTTAAACCTG3') were used to amplify a 345 bp length -globin sequence from genomic DNA using AmpliTaq gold polymerase (PE Biosystems, Foster City, CA, USA). After an initial incubation of 10 min 30 s at 95°C to activate AmpliTaq gold polymerase, the program was run for 35 cycles of 95°C, 1 min; 52°C, 1 min; and 72°C, 1 min. Eighteen l of PCR product were digested with 10 units of DdeI (CTNAG) or Bsu36 I (CCTNAGG) (New England Biolabs, Beverly, MA, USA) in a 20 l reaction at 37°C for 2 h and then run on an agarose gel to differentiate normal versus sickle -globin sequence. DdeI digestion of PCR product amplified from normal -globin sequence gives rise to 180 bp, 117 bp, 45 bp, and 3 bp fragments, whereas that from sickle -globin sequence gives rise to 297 bp, 45 bp, and 3 bp fragments. Similarly, Bsu36 I digestion can distinguish normal (228 bp and 117 bp fragments) from sickle -globin sequence (345 bp fragment). For TA cloning, the PCR product was cleaned using QIAquick PCR Purification Kit (Qiagen) and then cloned using TA Cloning Kit (Invitrogen, Carlsbad, CA, USA). Sequence analysis was done with forward primer PCO2 using an ABI Prism kit on an ABI310 capillary sequencer (PE Biosystems).

In vitro erythropoiesis derived from Lin^-CD38^- cells

We modified the published procedure of Malik et al,²⁵ first culturing cells in ME medium for 4 days and then culturing in E medium for 3 weeks. Erythropoiesis was evident by both glycophorin A expression and the presence of red color (signifying the presence of hemoglobin) in the cultured cells. For FACS analysis, both starting Lin^-CD38^- cells and erythroid cultured cells were stained with either CD34 FITC, glycophorin A PE, or isotype controls (IgG1 FITC and IgG_2b PE; BD PharMingen), and then analyzed on a FACScan flow cytometer (Becton Dickinson).

Analysis of genomic DNA and RNA in erythroid culture

After 3 weeks in E medium, both genomic DNA and RNA were isolated from erythroid culture cells. Genomic DNA was isolated and analyzed as described above. Total RNA was isolated from 1 ´ 10⁶ cells using RNeasy mini-kit (Qiagen). cDNA was synthesized using ProSTAR First-Strand RT-PCR kit (Stratagene) with oligo dT priming as per the manufacturer's instruction. Five microliters of cDNA was used to amplify -globin sequence using primers D (5'CTGACACAACTGTGTTCAC3') and R (5'TGAAGTTCTCAGGATCCAC3') and Taq polymerase (Sigma, St Louis, MO, USA). PCR program was 35 cycles of 91°C, 1 min; 54°C, 1 min; and 72°C, 2 min. RT-PCR product was digested with DdeI and electrophoresed through a 2.5% agarose gel.

In one experiment, the 216 bp fragment resulting from incomplete DdeI digestion was used for direct sequencing. The fragment was purified from the agarose gel using QIAquick Gel Extraction Kit (Qiagen) and then sequenced with primer D.

PCR and RT-PCR products from experiments that showed conversion by DdeI digestion were used for TA-cloning, allele-specific PCR, and sequencing. TA-cloning of PCR product was performed as described previously. Allele-specific PCR utilized two sets of primers that specifically amplified either normal or sickle -globin gene. Upstream primers SC9A (5'ACCATGGTGCACCTGACTCCTCA3'; normal -globin) or SC9S (5'ACCATGGTGCACCTGACTCCTCT3'; sickle -globin) were used together with downstream primer SC4 (5'ACCGATCCTGAGACTTCCACACT3') to distinguish between normal and sickle -globin. PCR program was run for 35 cycles of 95°C 1 min, 60°C 37 s, and 72°C 1 min. Genomic DNA of Lin^-CD38^- cells (^AA) and SC-1 cells (^SS) were used as controls for allele-specific PCR. Allele-specific PCR showed that SC9A and SC4 only amplified -globin gene from normal Lin^-CD38^- cells, whereas SC9S and SC4 only amplified sickle -globin gene from SC-1 cells. For TA-clones generated from RT-PCR product, primer R was used as downstream primer in allele-specific PCR. Sequence analysis of individual clones was performed as described above using primer D.

Acknowledgements

The work was funded in part by NIH grant R01HL58583 (to EBK). We thank David B Brown, ND Victor Carsrud, Judith Yannariello-Brown, Aqing Yao and Tamara V Tsulaia for their excellent scientific and technical guidance, Dr Nicole L Prokopishyn and Barbara L Chow for their critical review of this manuscript. We also thank Hetal Parekh-Olmedo and Alyson Cole-Strauss for their technical assistance during the early stages of the work.

1 Kmiec E.B. Targeted gene repair (editorial). Gene Therapy 1999; 6: 1, 1. MEDLINE

2 Cole-Strauss A.et al. . Correction of the mutation responsible for sickle cell anemia by an RNA-DNA oligonucleotide. Science 1996; 273: 1386, 1386. MEDLINE

3 Kren B.T., Cole-Strauss A., Kmiec E.B., Steer C.J. Targeted nucleotide exchange in the alkaline phosphatase gene of HuH-7 cells mediated by a chimeric RNA/DNA oligonucleotide. Hepatology 1997; 25: 1462, 1462. MEDLINE

4 Kren B.T., Bandyopadhyay P., Steer C.J. In vivo site-directed mutagenesis of the factor IX gene by chimeric RNA/DNA oligonucleotides. Nat Med 1998; 4: 285, 285. MEDLINE

5 Alexeev V., Yoon K. Stable and inheritable changes in genotype and phenotype of albino melanocytes induced by an RNA-DNA oligonucleotide. Nat Biotechnol 1998; 16: 1343, 1343. Article MEDLINE

6 Lai L.W., Lien Y.H. Homologous recombination based gene therapy. Exp Nephrol 1999; 7: 11, 11. MEDLINE

7 Beetham P.R.et al. . A tool for functional plant genomics: chimeric RNA/DNA oligonucleotides cause in vivo gene-specific mutations. Proc Natl Acad Sci USA 1999; 96: 8774, 8774. Article MEDLINE

8 Zhu T.et al. . Targeted manipulation of maize genes in vivo using chimeric RNA/DNA oligonucleotides. Proc Natl Acad Sci USA 1999; 96: 8768, 8768. Article MEDLINE

9 Kren B.T.et al. . Correction of the UDP-glucuronosyltransferase gene defect in the gunn rat model of Crigler-Najjar syndrome type I with a chimeric oligonucleotide. Proc Natl Acad Sci USA 1999; 96: 10349, 10349. Article MEDLINE

10 Rando T.A., Disatnik M.H., Zhou L.Z. Rescue of dystrophin expression in mdx mouse muscle by RNA/DNA oligonucleotides. Proc Natl Acad Sci USA 2000; 97: 5363, 5363. Article MEDLINE

11 Alexeev V.et al. . Localized in vivo genotypic and phenotypic correction of the albino mutation in skin by RNA-DNA oligonucleotide. Nat Biotechnol 2000; 18: 43, 43. Article MEDLINE

12 Bartlett R.J.et al. . In vivo targeted repair of a point mutation in the canine dystrophin gene by a chimeric RNA/DNA oligonucleotide. Nat Biotechnol 2000; 18: 615, 615. Article MEDLINE

13 Rice M.C., Bruner M., Czymmek K., Kmiec E.B. In vitro and in vivo nucleotide exchange directed by chimeric RNA/DNA oligonucleotides in Saccharomyces cerevisae. Mol Microbiol 2001; 40: 857, 857. Article MEDLINE

14 Davis B.R.et al. . Glass needle-mediated microinjection of macromolecules and transgenes into primary human blood stem/progenitor cells. Blood 2000; 95: 437, 437. MEDLINE

15 Xu R., Reems J.A. Umbilical cord blood progeny cells that retain a CD34+ phenotype after ex vivo expansion have less engraftment potential than unexpanded CD34+ cells. Transfusion 2001; 41: 213, 213. Article MEDLINE

16 Bhatia M.et al. . Purification of primitive human hematopoietic cells capable of repopulating immune-deficient mice. Proc Natl Acad Sci USA 1997; 94: 5320, 5320. Article MEDLINE

17 Bandyopadhyay P.et al. . Nucleotide exchange in genomic DNA of rat hepatocytes using RNA/DNA oligonucleotides. Targeted delivery of liposomes and polyethyleneimine to the asialoglycoprotein receptor. J Biol Chem 1999; 274: 10163, 10163. Article MEDLINE

18 Cole-Strauss A.et al. . Targeted gene repair directed by the chimeric RNA/DNA oligonucleotide in a mammalian cell-free extract. Nucleic Acids Res. 1999; 27: 1323, 1323. MEDLINE

19 Harrison G.S.et al. . Optimization of gene transfer using cationic lipids in cell lines and primary human CD4+ and CD34+ hematopoietic cells. Biotechniques 1995; 19: 816, 816. MEDLINE

20 Van Tendeloo V.F.et al. . High-level transgene expression in primary human T lymphocytes and adult bone marrow CD34+ cells via electroporation-mediated gene delivery. Gene Therapy 2000; 7: 1431, 1431. MEDLINE

21 Xiang Y.et al. . Targeted gene conversion in a mammalian CD34+-enriched cell population using a chimeric RNA/DNA oligonucleotide. J Mol Med 1997; 75: 829, 829. MEDLINE

22 Walters M.C.et al. . Bone marrow transplantation for sickle cell disease. N Engl J Med 1996; 335: 369, 369. MEDLINE

23 Gamper H.B.et al. . A plausible mechanism for gene correction by chimeric oligonucleotides. Biochemistry 2000; 39: 5808, 5808. Article MEDLINE

24 Saiki R.K.et al. . Enzymatic amplification of beta-globin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia. Science 1985; 230: 1350, 1350. MEDLINE

25 Malik P.et al. . An in vitro model of human red blood cell production from hematopoietic progenitor cells. Blood 1998; 91: 2664, 2664. MEDLINE

Figure 1 Experimental design. Three hundred to 500 CD34⁺ or 300-2000 Lin^-CD38^- cells were attached to a retronectin-coated dish. Thirty to 700 cells were successfully injected with repair oligonuclotides and/or Oregon Green dextran. After microinjection, cells were detached and transferred for subsequent culture. Cells were either cultured in the myeloid expansion media or erythoid expansion media for subsequent DNA or mRNA analysis, respectively.

Figure 2 Photographs of injected Lin^-CD38^- cells. Lin^-CD38^- cells were attached to RN-coated dishes and injected with both Texas Red-conjugated repair oligonucleotide uD3T/TR and OGD. Two hours after microinjection, cell viability and location of fluorescent material were determined by microscopy. (a) Phase contrast: cells attached to the RN-coated dish; (b) UV light with a blue filter: injected cells contained OGD; (c) UV light with a green filter: uD3T/TR was maintained and concentrated in the nuclei of injected cells (arrows); (d) an overlay of computer images showing co-localization of OGD and uD3T/TR in the same injected cells.

Figure 3 SC2 directed conversion from ^A to ^S in CD34^- and Lin^-CD38^- cells. (a) -Globin gene structure and relative positions of primers PCO2 () and PCO5 (¬) and chimeric oligonucleotide SC2/88MD-3 are indicated. The position of the sickle mutation (^*) and DdeI digestion sites (´) are shown. PCR amplification of the beta-globin gene with primers PCO2 and PCO5 gives rise to a 345 bp product. Normal and sickle beta-globin sequence can be differentiated based upon the DdeI digestion pattern of the PCR product. Conversion from normal to sickle beta-globin sequence induced by the SC2 oligonucleotides results in a loss of a DdeI restriction site, which will give rise to a 297 bp fragment in the DdeI digestion pattern. PCR and DdeI digestion are shown in panels (b and c). (b) M, 100 bp DNA marker; lanes 1 and 3, CD34^- cells injected with SC2/88MD-3 alone; lanes 2 and 4, CD34^- cells injected with SC2/88MD-3 and OGD; lane 5, Lin^-CD38^- cells; lane 6, 1% SC-1 cells, lane 7, 10% SC-1 cells; lane 8, 100% SC-1 cells. (c) M, 100 bp DNA marker; lane 1, Lin^-CD38^- cells injected with OGD; lanes 2 and 3, Lin^-CD38^- cells injected with SC2/88MD-3 and OGD; lanes 4-8, Lin^-CD38^- cells injected with SC2/88MD-3; lane 9, Lin^-CD38^- cells; lane 10, 1% SC-1 cells; lane 11, 10% SC-1 cells; lane 12, 100% SC-1 cells.

Figure 4 Sequences from TA-cloning products. PCR product generated from CD34^- or Lin^-CD38^- cells injected with SC2/88MD-3 were cloned and sequenced. ^S and ^A sequences from individual clones are shown with an arrow highlighting the base target for change. One out of 14 clones from CD34^- genomic DNA showed sickle (^S) conversion. Three out of 20 clones from Lin^-CD38^- genomic DNA showed sickle conversion.

Figure 5 Erythroid lineage development in vitro. Lin^-CD38^- cells were cultured in ME medium for 4 days and then cultured in E medium for 3 weeks. The starting Lin^-CD38^- cells were stained with anti-CD34 FITC and anti-CD38 CyChrome, or isotype control antibodies. Cells before and after culture were stained with anti-CD34 FITC and anti-glycophorin A PE, or isotype control antibodies. (a) CD34 and CD38 expression on starting Lin^-CD38^- cells. (b) CD34 and glycophorin A expression on starting Lin^-CD38^- cells. (c) CD34 and glycophorin A expression on cultured cells.

Figure 6 Gene conversion in erythroid progeny. (a) -Globin genomic DNA and mRNA structure and relative positions of primers (D and R) are shown. The position of the sickle mutation (^*) and DdeI digestion sites (´) are indicated. RT-PCR amplification of beta-globin gene mRNA sequence with primers D and R gives rise to a 353 bp fragment. Since normal to sickle -globin sequence results in a loss of a DdeI restriction site, DdeI digestion of the RT-PCR product was used to differentiate the normal versus sickle -globin mRNA sequence. DdeI digestion of the RT-PCR product from sickle beta-globin mRNA gives rise to a 216 bp fragment. (b) PCO2 and PCO5 primers were used to amplify -globin from genomic DNA. The PCR product was digested with DdeI. Failure of DdeI digestion due to the presence of sickle globin gene mutation gives rise to a 297 bp fragment. M, 100 bp marker; lanes 1 and 2, Lin^-CD38^- cells injected with SC2/88MD-3; lane 3, 1% SC-1 cells; lane 4, 10% SC-1 cells; lane 5, 100% SC-1 cells. (c) RT-PCR and DdeI digestion: M, 100 bp marker; lanes 1 and 2, Lin^-CD38^- cells injected with SC2/88MD-3; lanes 3 and 4, Lin^-CD38^- cells injected with SC2/88MD-3, RT-PCR product, no digestion.