Long-term Regulation of Genetically Modified Primary Hematopoietic Cells in Dogs

Abstract

We report long-term results from a large animal model of in vivo selection. Nine years ago, we transplanted two dogs (E900 and E958) with autologous marrow CD34⁺ cells that had been transduced with a gammaretrovirus vector encoding a conditionally activatable derivative of the thrombopoietin receptor. Receptor activation through administration of a chemical inducer of dimerization (CID) (AP20187 or AP1903) confers a growth advantage. We previously reported responses to two 30-day intravenous (i.v.) courses of AP20187 administered within the first 8 months post-transplantation. We now report responses to 5-day subcutaneous (s.c.) courses of AP20187 or AP1903 at months 14, 90, and 93 (E900), or month 18 (E958), after transplantation. Long-term monitoring showed no rise in transduced cells in the absence of drug. Retroviral insertion site analysis showed that 4 of 6 (E958) and 5 of 12 (E900) transduced hematopoietic cell clones persisted lifelong. Both dogs were euthanized for reasons unrelated to the gene therapy treatment at 8 years 11 months (E958) and 11 years 1 month (E900) of age. Three clones from E900 remained detectable in each of two secondary recipients, one of which was treated with, and responded to, AP1903. Our results demonstrate the feasibility of safely regulating genetically engineered hematopoietic cells over many years.

Introduction

Hematopoietic stem cells (HSC) are distinguished by their ability to stably engraft, to self-renew, and to differentiate across multiple blood lineages. Because the number of HSCs in an individual far exceeds the number contributing to blood cell production at any given moment, HSCs can be viewed as competing for the opportunity to contribute to blood cell formation. However, the rules governing this competition are poorly understood. The ability to regulate the competitive stance of transplanted cells could have many advantages for gene and cell therapy.^1,2,3

One method for increasing the frequency of genetically modified hematopoietic cells uses drugs that kill cells unless they are protected by a corresponding resistance gene. Although a number of cytotoxic drug/resistance gene combinations have been evaluated, their effects are often short-lived due to the inability of most drugs to kill unmodified HSC.^4,5,6 In these cases, achieving sustained selection through chronic drug administration is not feasible due to side effects. In contrast, the most potent of these methods uses a gene encoding a derivative of the enzyme methylguanine methyltransferase, followed by selection with O⁶-benzylguanine combined with an alkylating drug such as temozolomide or BCNU (reviewed in ref. 2). The potent and sustained selection observed in mice,⁷ dogs,⁸ and nonhuman primates⁹ is attributable to the activity of the temozolomide or BCNU/O⁶-benzylguanine combination against unmodified HSC. However, alkylating drugs can cause serious side effects, and situations exist in which a conditional expansion of genetically modified cells might be preferable to permanently depleting the HSC compartment.³

We and others have developed an alternative method for selection that relies on conditionally regulatable signaling molecules containing portions of cell surface receptors^{10,11,12,13,14,15,16} or kinases¹⁷ fused to binding sites for drugs called chemical inducers of dimerization (CIDs). The activity of these signaling proteins is regulated through self-association that can be controlled using CIDs.^{10,11,12,13,14,15} Most attractive for eventual clinical use are CIDs that interact minimally with endogenous proteins, and domains for CID docking that are unlikely to evoke immune responses. We have used a system that fulfills these criteria to show that a conditionally activated derivative of the thrombopoietin receptor (F36VMpl) can support the expansion of genetically modified hematopoietic cells in vitro^10,11 and in vivo.^12,13,14,17

In a previous report,¹³ we described the hematological responses of two dogs transplanted with autologous CD34⁺ cells that had been transduced with a bicistronic gammaretrovirus vector-encoding green fluorescent protein (GFP) and an engineered, CID-responsive growth factor receptor (F36VMpl), followed by administration of the cognate CID AP20187.¹³ Both dogs received two 30-day intravenous (i.v.) courses of AP20187, and both responded with rises in GFP⁺ red blood cells (RBCs), platelets, and white blood cells (WBCs), the latter comprised mainly of B cells. Here, we report results from life-long monitoring and responses to additional courses of CID. In contrast to the previously described 30-day i.v. regimen, we tested 5-day subcutaneous (s.c.) courses of AP20187 and its closely related analog, AP1903.¹⁸ Tracking studies demonstrate the ability to conditionally regulate the contributions of individual HSC clones to hematopoiesis over the lifetimes of these large animals, with no evidence of treatment-related serious adverse events.

Results

Two dogs, E900 and E958, were transplanted with transduced autologous CD34-enriched marrow cells after receiving myeloablative conditioning. Demographic information for both dogs (E900 and E958) is provided in Table 1. The bicistronic gammaretroviral vector that was used for these studies makes use of the MSCV LTR promoter, resulting in expression of GFP, and by inference F36VMpl, across all hematopoietic lineages. Both dogs received three courses of AP20187 within the first 18 months after transplantation. The first two courses were administered i.v. at doses of 1 mg/kg twice daily for 30 days, whereas the 3rd course was administered s.c. at a dose of 5 mg/kg/day for 5 days. Responses to the first two courses were reported previously.¹³

Table 1. Demographic information for dogs.

AP20187 treatment produced rises in both the percentages (Figure 1) and absolute numbers (Supplementary Figure S1) of GFP⁺ RBCs, WBCs, and platelets. Among WBC subsets (Supplementary Figure S2), GFP⁺ CD21⁺ B cells exhibited prominent responses to the first two courses of i.v. AP20187 and a lesser response to the s.c. course. GFP⁺ CD3⁺ T cells rose following the first course of AP20187 in both animals. In contrast, GFP⁺ DM5⁺ granulocytes and CD14⁺ monocytes responded only minimally. The effects of CID treatment on GFP⁺ red cells and platelets waned rapidly upon completion of the i.v. courses and more gradually following the s.c. course (Figure 1). The efficacy of the s.c. course appeared to be generally similar to the i.v. courses, despite the lower cumulative dose (25 mg/kg versus 60 mg/kg) and duration (5 days versus 30 days) of the s.c. course.

Figure 1.

Sustained chemical inducer of dimerization (CID) responsiveness in dogs E900 (left) and E958 (right). Percentages of GFP⁺ red blood cells (RBCs, top), white blood cells (WBCs, middle), and platelets (Plts, bottom) are indicated on the y axis. X axis indicates the number of days post-transplantation. Black boxes and descending dashed lines indicate periods of CID treatment. The first two courses of CID used AP20187 at a dose of 1 mg/kg twice daily intravenously (i.v.) for 30 days. The third course of CID used an abbreviated subcutaneous (s.c.) regimen of AP20187 (5 mg/kg/day s.c. × 5 days). Hatched lines across the x axis indicates a >5-year period of monitoring without CID treatment. In E900, GFP⁺ cells remained stable at low levels throughout the >5-year period of monitoring without CID. In contrast, GFP⁺ cells fell to very low levels in E958 during this interval. Administration of CID courses 4 and 5 in E900 (using AP1903 at 5 mg/kg/day s.c. for 5 days) prompted responses that were much more modest albeit qualitatively similar to course 3. ^†Euthanasia due to the development of an idiopathic cardiomyopathy.

Blood count monitoring revealed a reproducible, transient decline in platelet counts concomitant with CID administration (Supplementary Figure S1 and data not shown), as described previously,¹³ falling as low as 59,000 in E900 and 50,000 in E958 (both during course #2). Studies using platelet-enriched plasma from E958 did not reveal evidence of CID-induced platelet aggregation in vitro (data not shown), and thus the basis for this evanescent drop in platelet count is not known. To the contrary, declines in hematocrit during and immediately following CID administration are likely attributable to the daily phlebotomies that occurred during this period. The s.c. route of AP20187 administration was also associated with significant inflammation at the site of the injection, reflected in a marked, transient rise in circulating neutrophils (Supplementary Figure S3). Following completion of AP20187, GFP⁺ cells returned to low baseline levels in all hematopoietic lineages except for B cells and, to a lesser extent, T cells, which remained at persistently higher levels than RBCs, granulocytes and platelets (Figure 1, Supplementary Figure S2).

Long-term monitoring

After completing the s.c. course of AP20187, both dogs were returned to routine housing and monitored without further CID treatment for >6 years (Figure 1 and Supplementary Figures S1–S3). Complete blood counts remained normal throughout this period. A moderate step-up in total platelet counts was recorded in both animals beginning in the fall of 2003 and remained stable thereafter (Supplementary Figure S1 and data not shown), the likely consequence of a change in the automated complete blood count procedure. The frequency of GFP⁺ RBCs, WBCs, and platelets in E900 remained relatively stable throughout the monitoring period. A somewhat different pattern was observed in E958, in which GFP⁺ WBCs remained stable at 1–2%, whereas GFP⁺ RBCs and platelets fell to ~0.1% (Figure 1). The absence of a constitutive rise in GFP⁺ cells over a >6-year period of observation demonstrates the lack of selective advantage in the absence of CID. Both dogs retained stably higher frequencies of GFP⁺ cells among B and T lymphocytes compared to neutrophils, monocytes, RBCs, and platelets (Supplementary Figure S2), suggesting that the earlier courses of CID treatment may have affected long-lived GFP⁺ lymphoid cells.

Responses at later time points

After documenting the lack of a spontaneous rise in GFP⁺ cells during prolonged monitoring absent CID treatment, we wanted to test whether the ability to respond to CID was retained. E958 developed a dilated cardiomyopathy and was euthanized at 8 years, 11 months of age (8 years, 4 months post-transplant). We therefore retreated E900 with CID. Because s.c. administration of AP20187 had been accompanied by significant inflammation at the injection site, we tried the closely related CID AP1903. After confirming that AP20187 and AP1903 retained similar in vitro potencies (Supplementary Figure S4) we found that AP1903 was indeed much better tolerated following s.c. injection. Although this difference in local inflammation might be attributable to structural differences between these closely related CIDs, they might also be due to differences in drug formulation or in the age of the animal at the time that CID was administered.

E900 received two courses of AP1903 (5 mg/kg s.c. for 5 days). To discern even modest hematological responses, we developed a flow cytometric method for assessing GFP⁺ reticulocytes (see Materials and Methods section). AP1903-induced prominent responses in GFP⁺ reticulocytes, which transiently reached frequencies of 50–60% (Figure 2, Supplementary Figure S5). These responses were accompanied by corresponding rises in GFP⁺ RBCs, WBCs, and platelets, albeit to much lower levels than following the earlier courses of AP20187 (Figure 1). Another noteworthy departure from the responses to CID courses 1–3 was the lack of a B-cell response to courses 4 and 5 (Supplementary Figure S2). We do not know whether these differences were attributable to differences between AP1903 and AP20187 or, alternatively, the much older age of the animal during courses 4 and 5.

Figure 2.

Graphical depiction of percentages of green fluorescent protein positive (GFP⁺) reticulocytes (diamonds) and red blood cells (RBCs) (squares) in E900 in response to chemical inducer of dimerization (CID) courses 4 and 5. Y axis indicates percentage of GFP⁺ reticulocytes (left) and red blood cells (right)—note difference in scale. x axis indicates days post-transplantation. Black boxes indicate courses of AP1903 (5 mg/kg subcutaneous (s.c.) daily × 5 days).

Assessing clonal responses

Quantitative PCR of bone marrow cells sampled at different time points confirmed rises in transduced bone marrow cells in the periods during or immediately following CID exposure (Supplementary Figure S6). To track individual clones, we used unfractionated bone marrow cells to sequence 12 unique retrovirus insertion sites (RIS) in E900 and 6 RIS in E958 (Table 2). PCR of single progenitor colonies using RIS-specific primers consistently detected only a single RIS per colony, demonstrating the utility of RIS analysis as a means for clonal tracking (data not shown). Of note, two of the RIS that we identified were located within the vicinity of genes tagged as potential oncogenes in the mouse Retrovirus Tagged Cancer Gene Database (http://rtcgd.abcc.ncifcrf.gov/). One of these genes, DUSP2 (clone #2), encodes a phosphatase that inactivates the ERK1 and ERK2 members of the mitogen-activated protein kinase superfamily associated with cellular proliferation and differentiation. The other of these genes, KDR (clone #9), encodes one of the two receptors for vascular endothelial growth factor, and is the primary mediator vascular endothelial growth factor-induced endothelial cell proliferation and survival. However, as discussed below, neither of these clones exhibited a notable growth advantage throughout the course of our studies.

Table 2. Results of retroviral vector integration site analysis.

We looked for the presence or absence of each RIS in sorted CD34⁺ cells, T cells, B cells, and granulocytes early (<day 642) and late (>day 2,673) post-transplant (Supplementary Table S1). In E900, seven RIS were detected in flow-sorted B cells, T cells, and granulocytes, one was detected in B cells and granulocytes only, and four were detected in B cells only. In E958, three RIS were detected in sorted B cells, T cells, and granulocytes, one RIS was detected in sorted T cells only, and two RIS were detected only in sorted CD34⁺ cells at a very early time point (117 days) post-transplant.

In E900, three clones (#'s 1, 4, and 6), including the most abundant clone (#1), remained CID responsive throughout 7 years of observation (Figure 3). Each of these clones contributed to B cell, T cell, and granulocytic progeny. Consistently, clone #4 could be detected only during, or immediately following, CID administration. Two clones (#'s 3 and 5) were present throughout the observation period, but did not appear to respond to CID (Figure 3). One clone (#2) rose to detectable levels in association with the third round of CID administration, but than remained at a nearly constant level throughout the remaining observation period in the primary recipient. Clone #'s 2 and 5 were present in B cells, T cells, and granulocytes, whereas clone #3 was detected in B cells and granulocytes only. The six remaining clones in E900, including clone #9, showed a pattern of declining abundance over time, and included all of the clones that were detected in B cells only (#'s 7, 8, 10, and 12). In contrast clone #11 (detected in all three lineages) rose to detectable levels during the final course of CID administration.

Figure 3.

Heterogeneity among chemical inducer of dimerization (CID) responsive hematopoietic cell clones in dogs E958 (top), E900 (bottom), and H247 (bottom right). Y axis depicts the relative abundance of individual clones over time compared to a cellular gene (KIT). X axis indicates the days post-transplantation, with the hatched lines showing a >5-year interval of observation without CID. Black boxes and descending dashed lines indicate courses of CID treatment as described in the legend for Figure 1. **Clones that were detectable in all three lineages (granulocytes, B cells, and T cells); all other clones were only detected in one or two lineages. Three clones that persisted throughout the life of E900 remained detectable in its secondary recipient, H247, and responded to CID.

In E958 it was possible to determine baseline clonal frequencies prior to CID exposure. Each of three clones that were present in all lineages (#'s 13, 15, and 16) rose from undetectable levels prior to the first course of CID to persist at detectable levels for the remainder of the animal's lifetime. Two of the remaining three clones were detectable only in CD34⁺ cells (#'s 17 and 18) early post-transplant. The sixth clone (#14), found only in T cells, remained barely detectable as late as 7 years, 4 months post-transplant.

The lack of a response among transduced B cells in E900 following courses 4 and 5 of CID administration contrasted the prominent responses to courses 1–3 (Supplementary Figure S2). Notably, the clonal composition of sorted GFP⁺ CD21⁺ cells was very similar early versus late post-transplantation (Supplementary Figure S7). These findings suggest an age-associated loss of B lymphopoietic function within individual HSC clones, rather than succession from lymphoid-biased to myeloid-biased HSCs.¹³ Alternatively, these differences may be attributable to the different CIDs used in courses 1–3 (AP20187) versus courses 4 and 5 (AP1903).

Retention of CID responsiveness in a secondary recipient

An age-associated decline in B lymphopoiesis has been reported previously,¹⁹ and we wanted to test whether the loss of CID responsiveness in the aged B cells of E900 might be reversed in the environment of a younger animal, as has been observed in mice.²⁰ We therefore transplanted mobilized peripheral blood mononuclear cells from E900 into each of two lethally irradiated, unrelated, dog leukocyte antigen-matched recipients. Both recipients engrafted with similar kinetics (Supplementary Figure S8) and both exhibited engraftment of gene-modified cells, demonstrating the transduction and maintenance of gene-modified stem cells in the donor, E900. In both secondary recipients, percentages of GFP⁺ B cells and T cells were higher relative to granulocytes, red cells, or platelets (Supplementary Figures S9 and S10). Treatment of one of the secondary recipients (H247) with a 5-day s.c. course of AP1903 prompted a pronounced rise in GFP⁺ reticulocytes and red cells (Supplementary Figure S10), to nearly the same extent as had occurred in E900. However, CID treatment of H247 did not elicit a rise in GFP⁺ B cells (Supplementary Figure S9).

Clonal analysis demonstrated that the three clones that were most prominent in E900 (#'s 1–3) remained detectable in both of the secondary recipients (Figure 3, bottom right and data not shown). Furthermore in H247, all three clones increased in response to CID (Figure 3, bottom right). The rise in clones #2 and #3 in response to AP1903 in H247 contrasted with their apparent unresponsiveness in the donor, E900.

Lack of apparent long-term toxicity from intermittent CID administration

E900 developed a retroorbital adenocarcinoma and was euthanized at 10 years, 11 months of age, or 8 years, 11 months post-transplantation. Autopsy also revealed a perianal gland tumor and a benign s.c. lipoma. Because constitutive Mpl activation via exon 10 mutations have been associated with myelofibrosis and essential thrombocythemia,²¹ we evaluated the bone marrows and spleens of E900 and E958. The spleens of both animals were normal in size. Bone marrows and blood smears of E900 and E958 at the time of autopsy revealed normal trilineage hematopoiesis, with no evidence of myelofibrosis (Figure 4 and data not shown). Autopsy reports for E900 and E958 are provided in the Supplementary Materials and Methods.

Figure 4.

Absence of myelofibrosis in E900 and E958. Photomicrographs of hematoxylin and eosin stained bone marrow biopsies of E900 (left) and E958 (right) at ×100 and ×250 magnification. Bottom right panel shows ×100 magnification of reticulin stain for E958.

Discussion

Our findings demonstrate the ability to safely regulate the proliferation of genetically modified hematopoietic cells in dogs over a long period of time. All dogs displayed CID-dependent rises in genetically modified red cells and, to a lesser extent, platelets. E900 and E958 exhibited CID-dependent rises in GFP⁺ B cells within the first 18 months post-transplantation, accompanied by more gradual and subtle rises in GFP⁺ T cells during the same time period. In both dogs a clear discrepancy was evident between the responses of red cells and platelets versus B cells and T cells. Whereas GFP⁺ red cells and platelets returned to baseline levels within several weeks following the completion of CID, GFP⁺ B and T cells persisted at significantly higher levels. This disparity was most apparent in E958, where the frequency of GFP⁺ red cells and platelets fell to <0.1% with prolonged observation, whereas GFP⁺ B cells and T cells remained in the range of 5–10%.

Genetically modified HSCs contribute disproportionately to some lineages over others in situations where the favored lineages accrue a selective advantage. Notable examples include the much higher frequencies of genetically modified T cells (relative to other cell types) for adenosine deaminase deficiency²² and X-linked SCID.²³ In contrast, gene marking tends to occur evenly across lineages in the absence of selection, as was observed in a recent gene therapy trial for X-linked adrenoleukodystrophy.²⁴ These observations support the interpretation that the stably higher frequencies of genetically modified B and T cells observed in this study are a lasting consequence of the initial courses of CID administration.

We observed clear differences in the hematological responses of E900 to CID treatment over time. These differences were most notable between CID treatments given before versus following the 6-year observation period. Most noteworthy was a decline in the magnitude of the CID response. Furthermore B cells, which responded dramatically to the initial courses of CID, failed to respond at later times. Although we do not know the reason for these differences, potential explanations include differences in E900's age at the various times of CID administration, differences in the number of prior CID exposures, or differences between the two CIDs. Responses to the first two courses of AP20187, administered i.v. at a dose of 1 mg/kg twice daily for 30 days, were virtually indistinguishable. However, this method of administration consumed a large amount of drug per course (600 mg for a 10 kg animal) and proved difficult logistically, motivating our search for methods that required less drug and were more readily administered. A 5-day s.c. course of AP20187 administration (5 mg/kg/day) proved equally effective (Figure 1) and consumed less drug (250 mg for a 10 kg animal), but caused prohibitive s.c. inflammation (tenderness, warmth, and swelling) at sites of injection. Thus, when we resumed CID treatment, 6 years later, we tested s.c. AP1903. Although this regimen was much better tolerated, it also generated a more muted response than the earlier i.v. or s.c. courses of AP20187, and did not induce a B cell response. A head-to-head comparison showed that AP1903 was at least as potent as AP20187 in stimulating F36VMpl-expressing Ba/F3 cells to proliferate in vitro (Supplementary Figure S4), suggesting that the relatively attenuated response and lack of B cell response were more likely due to other factors. One possibility relates to E900's advanced age by the time AP1903 was administered.²⁵

In contrast to the persistent elevations in genetically modified B and T cells that occurred in E900 and E958 following the first three courses of CID administration, CID-induced rises in genetically modified red cells and platelets were reproducible yet transient. These observations suggest that in the case of F36VMpl, achieving persistent elevations in RBCs and platelets will require ongoing exposure to CID. Although this could pose a limitation in the context of inherited blood disorders, where permanent correction of the hematopoietic compartment is desired, it might constitute an advantage in settings where permanent changes to the kinetics of hematopoiesis are unnecessary. One example might be for stimulating erythropoiesis in contexts where the exogenous administration of erythropoietin is undesirable.³

Other than for the aforementioned soft tissue inflammation, we observed only minor side effects from CID treatment, characterized most notably by a moderate, reversible decline in platelet count during the period of CID administration. The basis for this transient thrombocytopenia is uncertain, because platelet aggregation was not observed in response to CID exposure in vitro. In addition, we observed no long-term toxicities that were attributable to ectopic F36VMpl expression or repeated CID exposures. Whereas we did not observe insertional genotoxicity in our study, this phenomenon is well described in the context of the type of gammaretroviral vector used here.²⁶ Mpl exon 10 mutations occur in a significant fraction of patients with idiopathic myelofibrosis and essential thrombocythemia.²¹ Autopsies revealed normal spleen sizes and no morphological evidence of myelofibrosis either in the marrow or in peripheral blood (Supplementary Materials and Methods). Dilated cardiomyopathy, which occurred in E958, is one of the most prevalent acquired heart diseases of dogs (http://www.merckvetmanual.com/mvm/index.jsp?cfile=htm/bc/121611.htm). Similarly, tumors of the nose and paranasal sinuses account for 1–2% of all canine tumors, occurring at a mean age of 10.5 years, with adenocarcinomas being most common (http://www.merckvetmanual.com/mvm/index.jsp?cfile=htm/bc/121611.htm). Complete autopsy reports from both animals are provided in Supplementary Materials and Methods. As such, we believe that these events were not study-related.

In summary, we have demonstrated the feasibility of safely bringing hematopoiesis under long-term pharmacological control. While limitations in animal numbers and CID supply prevent definitive conclusions regarding issues such as risk for leukemogenesis or HSC exhaustion, our experience does suggest the feasibility of an approach that is highly amenable to optimization by improving gene transfer rates, drug pharmacokinetics, and by employing transgene cassettes that enable tissue-specific expression. Derivations of this approach will allow hematopoiesis to be regulated in ways that are not achievable using conventional growth factors, and may be particularly beneficial for circumventing growth factor—mediated off-target effects.³

Materials and Methods

Methods not described previously¹³ are provided here.

CID administration. AP20187¹³ and AP1903^18,27 were provided by ARIAD Pharmaceuticals (www.ariad.com/regulationkits). AP20187 was a powder and was formulated in 5% solutol at a concentration of 10 mg/mL and stored at 4°C. AP1903 was provided in sterile 2 ml ampules (5 mg/ml) from which the drug was administered directly. Both AP20187 and AP1903 were administered s.c. at a dose of 5 mg/kg/day for five consecutive days.

Assessing CID-responsive reticulocytes. SYTO 62 (Life Technologies, Carlsbad, CA), was used to stain reticulocytes because the peak emission wavelength of thiazole orange (TO) overlaps with GFP. We validated the use of SYTO 62 (SYTO) for reticulocyte analysis by staining normal dog RBCs with TO and SYTO simultaneously. 44.0 ± 1.4% of TO⁺ cells stained with SYTO 62, whereas fewer than 0.1% of TO-negative RBCs were SYTO 62-positive (data not shown). Five milliliter EDTA-anticoagulated peripheral blood was added to 1 ml 0.9% NaCl containing 10 nmol/l SYTO 62, and incubated at room temperature for 20 minutes. Samples were washed once with 0.9% NaCl and analyzed within 30 minutes.

Identifying RIS and quantification of individual clones using quantitative PCR. RIS determinations were performed using linear amplification-mediated-PCR as described previously.^28,29 The closest gene was defined as the Refseq gene with the transcription start site closest to the retroviral integration site.²⁸ Primers to detect individual RIS were designed to amplify unique junction sequences between the dog genome and provirus sequence (Supplementary Table S2). Genomic DNA was purified from bone marrow cells and then subjected to triplicate quantitative PCR to determine the abundance of individual RIS using the ABI HT7900 with ABI SYBR master mix. All RQ values were normalized to genome and provirus copy number, KIT and MSCV, respectively). RQ values that were undetectable or exceeded a 4% coefficient of variance were regarded as falling below the threshold of detection.

Allogeneic secondary transplantation. Peripheral blood stem cells from E900 were mobilized using a combination of recombinant canine G-CSF (rcG-CSF) 5 µg/kg s.c. twice daily (10 µg/kg/day) for 5 days and a single dose of AMD3100 (4 mg/kg) on the day of collection. On the day of collection a dual-lumen venous catheter was placed followed by COBE apheresis, yielding 350–400 ml over 3–4 hours. Unrelated, dog leukocyte antigen-matched recipient animals were treated with cyclosporine, 15 mg/kg orally twice daily beginning 3 days before transplant until day 88 post-transplant for H247 and day 100 post-transplant for H266, and mycophenolate mofetil, 10 mg/kg s.c. twice daily beginning 1 hour after the stem cell infusion and lasting until day 27. Recipients were conditioned with 920 cGy total body irradiation at a dose rate of 7 cGy/min and were treated with standard supportive care protocols for antibiotic, fluid, and transfusion support. On day 64, H266 was found to have moderately elevated liver function tests (bilirubin 3.4 g/dl, ALT 366 U/l, GGT 68 U/l, alkaline phosphatase 812 U/l) and was treated for graft versus host disease with resumption of mycophenolate mofetil (5 mg/kg s.c. twice daily) and prednisone (1 mg/kg orally twice daily for 3 days followed by a taper after 12 days).

SUPPLEMENTARY MATERIAL Figure S1. Concomitant increases in GFP⁺ and total blood cell numbers. The absolute numbers of GFP⁺ (closed symbols) and total (open symbols) cells are shown for red blood cells / hematocrit (top panel), white blood cells (WBC − middle panel), and platelets (lower panel) for dog E900. The GFP⁺ hematocrit was determined by multiplying the fraction of GFP⁺ red blood cells with the hematocrit, and thus assumes that the size of GFP⁺ and GFP- red cells is similar. Courses of CID are indicated by the black boxes and dashed lines. See legend of Figure 1 for details. Figure S2. Uneven effects of F36VMpl signaling across white blood cell subsets. Long-term monitoring of GFP⁺ B cells (CD21⁺ − top panels), T cells (CD3⁺ − upper middle panels), granulocytes (DM5⁺ − lower middle panels), and monocytes (CD14⁺ − bottom panels) in dogs E900 (left) and E958 (right). Black boxes above each graph indicate CID courses. Note that CD21⁺ and CD3⁺ cells retained stably elevated GFP⁺ fractions following the first 3 courses of CID compared to DM5⁺ and CD14⁺ cells. Note also the contrast between the prominent response among CD21⁺ cells in both dogs during the first 600 days post transplant versus the lack of response in E900 following the 4^th and 5^th courses of CID. Figure S3. Absolute numbers of neutrophils (top), lymphocytes (middle) and monocytes (bottom) over the lifetime of dog E900. Black boxes and descending dashed lines indicate courses of CID treatment as described in the legend for Figure 1. Figure S4. Dose equivalency of AP20187 and AP1903 in vitro. Cell proliferation (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide, MTT) assays for transduced Ba/F3 cell clones expressing F36VMpl were performed as described previously (30). Proliferative responses are shown in panels A-C, scattergrams indicating frequencies of transduced (GFP-positive) cells are shown in panels D-F. Cells were cultured in the absence of IL3 overnight, and then IL3-containing WEHI-conditioned medium (panels A and D), AP20187 (panels B, E) or AP1903 (panels C,F) were added at the concentrations indicated. Figure S5. Example of flow cytometric analysis for GFP⁺ reticulocytes. Scattergrams of peripheral blood red cells from E900 at days 0, 5, 9 12, 14 and 18 after the start of CID course 4. X axis indicates GFP, Y axis indicates SYTO uptake, a dye used to identify reticulocytes. Left upper panels show controls using peripheral blood from a normal dog (left upper panel) and from E900 without SYTO staining (left lower panel). Note that GFP expression is greater among CID responsive RBCs compared to RBCs generated during basal erythropoiesis. Figure S6. Tracking provirus copy numbers over time. Quantitative PCR demonstrates CID induced rises in provirus containing cells in E900 (Panel A) and E958 (Panel B). Y axis indicates relative quantification as compared to the endogenous KIT locus. Figure S7. Contribution of individual clones to white blood cell subsets. The relative fraction of individual clones in FACS-purified white blood cell subsets are shown for animal E900 at early (day 567) and late (day 2802) time points post-transplantation. Purified cell samples included GFP⁺ DM5⁺ neutrophils, GFP⁺ CD21⁺ B cells, GFP⁺ CD3⁺ T cells, and for the later time point, GFP⁺ bone marrow cells. The frequency of individual clones were determined by quantitative PCR of DNA from these purified cells using primers specific for individual retroviral vector integration sites, and are reported as a fraction of the total provirus present as determined by quantitative PCR for total MSCV provirus. See Supplemental Tables 1 and 2 for more details. Figure S8. Engraftment of secondary recipients. WBC and platelet engraftment kinetics are shown for the two DLA-matched allogeneic transplant recipients from E900. Black boxes denote treatment periods for cyclosporine, mycophenolate and AP1903. Figure S9. Lack of response in white blood cell subsets for secondary recipients. Post-transplant monitoring of GFP⁺ B cells (CD21⁺ − top panels), T cells (CD3⁺ − upper middle panels), monocytes (CD14⁺ − lower middle panels), and neutrophils (DM5⁺ − bottom panels), in the two DLA-matched allogeneic transplant recipients from E900. Black boxes denote treatment periods for cyclosporine, mycophenolate and AP1903. Note that CD21⁺ and CD3⁺ cells retained the stably elevated GFP⁺ fractions present in the donor compared to DM5⁺ and CD14⁺ cells. Note also the lack of response of CD21⁺ B cells to CID course 6. Figure S10. Response of platelets and red blood cells in secondary recipient. Graphical depiction of percentages of GFP positive reticulocytes (diamonds) and RBCs (squares) in DLA-matched allogeneic transplant recipient H247 in response to CID course 6. Y axis indicates percentage of GFP⁺ reticulocytes (left) and red blood cells (right) – note difference in scale. X axis indicates days post-transplantation. Black boxes indicate courses of cyclosporine, mycophenolate and AP1903 (5 mg/kg SC daily x 5 days). Table S1. Presence of individual proviral clones in cell subsets. Table S2. PCR primes for individual retroviral vector integration sites used to track individual transduced clones. Supplementary Materials and Methods.

Supplementary Material

Figure S1.

Concomitant increases in GFP⁺ and total blood cell numbers. The absolute numbers of GFP⁺ (closed symbols) and total (open symbols) cells are shown for red blood cells / hematocrit (top panel), white blood cells (WBC − middle panel), and platelets (lower panel) for dog E900. The GFP⁺ hematocrit was determined by multiplying the fraction of GFP⁺ red blood cells with the hematocrit, and thus assumes that the size of GFP⁺ and GFP- red cells is similar. Courses of CID are indicated by the black boxes and dashed lines. See legend of Figure 1 for details.

Figure S2.

Uneven effects of F36VMpl signaling across white blood cell subsets. Long-term monitoring of GFP⁺ B cells (CD21⁺ − top panels), T cells (CD3⁺ − upper middle panels), granulocytes (DM5⁺ − lower middle panels), and monocytes (CD14⁺ − bottom panels) in dogs E900 (left) and E958 (right). Black boxes above each graph indicate CID courses. Note that CD21⁺ and CD3⁺ cells retained stably elevated GFP⁺ fractions following the first 3 courses of CID compared to DM5⁺ and CD14⁺ cells. Note also the contrast between the prominent response among CD21⁺ cells in both dogs during the first 600 days post transplant versus the lack of response in E900 following the 4^th and 5^th courses of CID.

Figure S3.

Absolute numbers of neutrophils (top), lymphocytes (middle) and monocytes (bottom) over the lifetime of dog E900. Black boxes and descending dashed lines indicate courses of CID treatment as described in the legend for Figure 1.

Figure S4.

Dose equivalency of AP20187 and AP1903 in vitro. Cell proliferation (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide, MTT) assays for transduced Ba/F3 cell clones expressing F36VMpl were performed as described previously (30). Proliferative responses are shown in panels A-C, scattergrams indicating frequencies of transduced (GFP-positive) cells are shown in panels D-F. Cells were cultured in the absence of IL3 overnight, and then IL3-containing WEHI-conditioned medium (panels A and D), AP20187 (panels B, E) or AP1903 (panels C,F) were added at the concentrations indicated.

Figure S5.

Example of flow cytometric analysis for GFP⁺ reticulocytes. Scattergrams of peripheral blood red cells from E900 at days 0, 5, 9 12, 14 and 18 after the start of CID course 4. X axis indicates GFP, Y axis indicates SYTO uptake, a dye used to identify reticulocytes. Left upper panels show controls using peripheral blood from a normal dog (left upper panel) and from E900 without SYTO staining (left lower panel). Note that GFP expression is greater among CID responsive RBCs compared to RBCs generated during basal erythropoiesis.

Figure S6.

Tracking provirus copy numbers over time. Quantitative PCR demonstrates CID induced rises in provirus containing cells in E900 (Panel A) and E958 (Panel B). Y axis indicates relative quantification as compared to the endogenous KIT locus.

Figure S7.

Contribution of individual clones to white blood cell subsets. The relative fraction of individual clones in FACS-purified white blood cell subsets are shown for animal E900 at early (day 567) and late (day 2802) time points post-transplantation. Purified cell samples included GFP⁺ DM5⁺ neutrophils, GFP⁺ CD21⁺ B cells, GFP⁺ CD3⁺ T cells, and for the later time point, GFP⁺ bone marrow cells. The frequency of individual clones were determined by quantitative PCR of DNA from these purified cells using primers specific for individual retroviral vector integration sites, and are reported as a fraction of the total provirus present as determined by quantitative PCR for total MSCV provirus. See Supplemental Tables 1 and 2 for more details.

Figure S8.

Engraftment of secondary recipients. WBC and platelet engraftment kinetics are shown for the two DLA-matched allogeneic transplant recipients from E900. Black boxes denote treatment periods for cyclosporine, mycophenolate and AP1903.

Figure S9.

Lack of response in white blood cell subsets for secondary recipients. Post-transplant monitoring of GFP⁺ B cells (CD21⁺ − top panels), T cells (CD3⁺ − upper middle panels), monocytes (CD14⁺ − lower middle panels), and neutrophils (DM5⁺ − bottom panels), in the two DLA-matched allogeneic transplant recipients from E900. Black boxes denote treatment periods for cyclosporine, mycophenolate and AP1903. Note that CD21⁺ and CD3⁺ cells retained the stably elevated GFP⁺ fractions present in the donor compared to DM5⁺ and CD14⁺ cells. Note also the lack of response of CD21⁺ B cells to CID course 6.

Figure S10.

Response of platelets and red blood cells in secondary recipient. Graphical depiction of percentages of GFP positive reticulocytes (diamonds) and RBCs (squares) in DLA-matched allogeneic transplant recipient H247 in response to CID course 6. Y axis indicates percentage of GFP⁺ reticulocytes (left) and red blood cells (right) – note difference in scale. X axis indicates days post-transplantation. Black boxes indicate courses of cyclosporine, mycophenolate and AP1903 (5 mg/kg SC daily x 5 days).

Table S1.

Presence of individual proviral clones in cell subsets.

Table S2.

PCR primes for individual retroviral vector integration sites used to track individual transduced clones.