Cytological characterization of murine bone marrow and spleen hematopoietic compartments for improved assessment of toxicity in preclinical gene marking models
- First Online:
- Cite this article as:
- Yang, M., Büsche, G., Ganser, A. et al. Ann Hematol (2013) 92: 595. doi:10.1007/s00277-012-1655-3
- 203 Views
Gene therapy has proven its potential to cure diseases of the hematopoietic system, but potential adverse reactions related to insertional mutagenesis by integrating gene vectors and chromosomal instability in long-lived repopulating cells have emerged as a major limitation. Preclinical gene therapy in murine models is a powerful model for assessment of gene marking efficiency and adverse reactions. However, changes in the hematologic composition after transplantation with retrovirally modified hematopoietic stem cells have not been well investigated in large cohorts of animals by systematic cytological analyses. In the present study, cytological analyses of bone marrow and spleen were performed in a large cohort (n = 58) of C57BL/6J mice over an extended observation period after gene marking. Interestingly, we observed hematological malignancies in four out of 30 animals transplanted with dLNGFR (truncated form of the human p75 low-affinity nerve growth factor receptor) and tCD34 modified stem/progenitor cells. Our data demonstrate that cytological analysis provides important information for diagnosis of hematological disorders and thus should be included in preclinical studies and performed in each investigated animal. Together with histological analysis, flow cytometric analysis, and other analyses, the quality and predictive value of preclinical gene therapy studies will be improved.
KeywordsPreclinical gene therapy studiesCytological analysisdLNGFRtCD34
Hematopoietic stem cells (HSC) are characterized by their ability to self-renew and differentiate into all cell types of the hematopoietic system . Targeting HSC with integrating gene vectors allows a life-long expression of the therapeutic transgene in all progeny cells, and has been effective for the treatment of monogenetic diseases like X-linked severe combined immunodeficiency (X-SCID) , chronic granulomatous disease (CGD) , Wiskott-Aldrich syndrome (WAS) , adenosine deaminase deficiency (SCID-ADA) , and β-thalassemia . In the first three conditions, murine leukemia virus (MLV) derived gammaretroviral (GV) vector integrations in the vicinity of proto-oncogenes have been the initiating event for pre-malignant clonal dominance and malignant transformation [7, 8]. Importantly, we reported similar observations in murine models prior to these clinical discoveries [9–11]. Long-term side effects in hematopoietic stem and progenitor cells in large animals have also been reported by other groups [12, 13].
Well-designed animal models and multicenter efforts will be required for systematic risk assessment of side effects related to transgene insertion and expression, especially when targeting long-lived stem cells [14, 15]. To develop safer and more effective gene therapy with HSCs, it is important to evaluate genotoxicity and phenotoxicity induced by gene marking in mice and/or large animals (e.g., nonhuman primate) transplanted with gene-modified HSCs. Although preclinical gene therapy in murine models is a powerful model for assessment of gene-marking efficiency and adverse reactions, however, changes in the hematologic composition after transplantation with retrovirally modified HSCs have not been well investigated in large cohorts of animals by systematic cytological analyses. It is worth to point out that the differences between the hematopoietic systems of mice and man must carefully be evaluated to diagnose reactive and neoplastic blood cell disorders with certainty and to improve the predictive value of the animal model [16, 17]. In contrast to human leukemia, the spleen is involved in almost all cases and bone marrow is only variably involved in murine models [18, 19].
Leukemia diagnosis primarily relies on morphological, cytochemical, and immunophenotypic features of the neoplastic cells to define their lineage and degree of maturation. The Bethesda proposal for diagnosis of murine myeloid neoplasia based on histopathology and immunophenotyping was published 10 years ago . Histopathology allows diagnosis of a wide range of hematopathological disorders including leukemia, lymphoproliferative, and myeloproliferative disease, and it is the method of choice to diagnose/confirm spread of neoplasia to other organs. However, histologic analysis requires a sectioning of cells which may impair the cytological analysis of individual cells. Therefore, cytological examination of complete (not sectioned) bone marrow cells is a valuable additional tool in the (differential) diagnosis of hematologic neoplasms. Analysis of cell morphology remains the clinical gold standard for the diagnosis and follow-up of leukemia and myelodysplastic syndromes (MDS). Cell morphology provides best information pertaining cell maturation, granularity, and percentage distribution of cell populations, while histopathology enables examination of quantity (cellularity) and the topographical distribution of cellular constituents of the marrow, their relationships to marrow mesenchyme, vascularity (sinusoids), and bone trabeculae, and a possible infiltration of adjacent tissues.
Our recent analyses of healthy untreated murine bone marrow and spleen provide important reference ranges of each cell type and for the myeloid/erythroid ratio, which can be used in interpreting data generated from transplanted mice . The major purpose of the present study is to document the change in the hematologic composition of bone marrow and spleen after transplantation with retrovirally modified HSCs. Examination of bone marrow and spleen revealed limited effect following gene marking, unless insertional mutagenesis was induced and/or the transgene itself strongly influenced hematopoietic differentiation. Thus, our study demonstrates that in addition to histopathological examination and immunophenotyping, careful cytological analysis of spleen and bone marrow can improve the diagnosis of hematological alteration of animals and thus the quality of preclinical gene therapy trials.
Animals and retroviral gene marking
C57BL/6J mice (aged from 2–4 months) were kept in the animal laboratories of Hannover Medical School. Animal experiments were approved by the local ethical committees in Hannover and performed according to the national guidelines. Retroviral vectors, vector production, retroviral transductions of murine HSCs, and transplantation have been described elsewhere [10, 11, 21, 22]. SFa11tCD34 and SFa11dLNGFR vectors code for human tCD34 and the cytoplasmically truncated form of the human p75 low-affinity nerve growth factor receptor (dLNGFR), respectively [21, 22]. To investigate safety of the marker gene dLNGFR [11, 23], a cohort of animals were transplanted with dLNGFR modified stem/progenitor cells between 2006 and 2008. To this end, lineage negative cells from C57BL/6J male donors were cultured in serum-free medium (STEMSPANTM SF, StemCell Technologies, Vancouver, Canada) supplemented with mSCF (100 ng/ml, R&D system), mIL-3 (20 ng/ml, PeproTech, NJ), hIL-11 (100 ng/ml, PeproTech, NJ), and hFLT3-ligand (100 ng/ml, PeproTech, NJ) for 2 days at 37 °C in 5 % CO2 before exposure to retroviral particles. Gene marking for the groups dLNGFR and tCD34 after two transduction rounds was very similar (64.2 % vs. 61.5 %). dLNGFR-modified cells were transplanted into ten female primary recipients. Half of them were killed at day 218 after transplantation, and bone marrow cells from primary recipients were separately transplanted into irradiated secondary animals (two to three recipients for each primary animal, total n = 13) with a mean observation time of 258 days (range, 101–309 days). From the other five primary recipients, one died of unknown cause at day 387, one was killed at day 286, and the experiment with the other three animals was terminated at day 427 after transplantation.
One cohort (n = 8) with tCD34 was done in parallel with dLNGFR as a control. Similarly to the dLNGFR group, half of the primary recipients were killed at day 219 after transplantation, and bone marrow cells from primary recipients were separately transplanted into irradiated secondary animals (one to two recipients for each primary animal, total n = 7) with a mean observation time of 287 days (range, 224–309 days). The other four primary animals were observed with a mean observation time of 408 days (range, 388–428 days) before termination of the study or until death. Mice transplanted with EGFP (n = 10) modified stem/progenitor cells were collected from other experiments with similar gene marking efficiency [24, 25]. The mean observation time was 205 days (range, 163–300 days). Animals were visited >2 times/week. At the end point analysis, blood smear, cytospins of spleen, bone marrow, thymus (in majority of transplanted animals), and tumor, and histopathology of spleen, bone marrow (sternum and tibia), liver, thymus, lung, kidney, and tumor, if any, were accomplished for each animal. Cytospins from single-cell suspensions were made by centrifugation (800 rpm, 10 min) on microscope slides using a Thermo Shandon Cytospin 4 centrifuge (Thermo Scientific, Waltham, MA). A complete blood count was performed with an automated animal blood counter (Scil Vet abc, Scil animal care company GmbH, Viernheim, Germany). All cytospins were stained with the Pappenheim staining method. Perls Prussian blue reaction was also done in selected cases.
Cytological and histological examination
Only animals, for which cytospins of bone marrow and spleen, histology, and blood smear were available, were enrolled into this study, with exception for a few cases in which blood smears were lacking. Thus, three mice from the dLNGFR group were excluded due to lack of cytospins. Morphology was assessed by examination with a ×100 objective, and a 500-cell differential count for bone marrow and spleen was performed and compared with the reference data of myelogram and splenogram established from healthy controls in our laboratory . Histological analyses were also performed in all analyzed animals. A 100-cell differential count for peripheral blood was also performed in selected animals. Diagnoses of hematopoietic neoplasms were made according to the Bethesda proposals by the hematopathology subcommittee of the Mouse Models of Human Cancers Consortium [20, 26].
Our protocol for ligation-mediated PCR (LM-PCR) followed published conditions . Genomic DNA was harvested using the QIAamp blood kit (Qiagen, Hilden, Germany). Single-cell suspensions of mouse organs prepared at necropsy were collected by centrifugation.
Differences between groups were analyzed by Wilcoxon test. A two-tailed P value <0.05 was considered to indicate statistical significance.
Cytological analysis of mice after gene marking (tCD34, LNGFR, and EGFP)
Since use of dLNGFR as a surface marker in gene therapy clinical trials is controversial [23, 27], it is important to investigate the potential risks associated with this and any such molecule for gene marking. To this end, we conducted a new cohort of animals transplanted with dLNGFR modified stem/progenitor cells with very long observation period (the longest time for some primary animals together with their secondary recipients was 527 days). Expression of dLNGFR, tCD34, and EGFP in spleen of primary mice at the time of sacrifice was 1.7–82.3 % (mean 34.4 %), 12.8–94.1 % (mean 57.7 %), and 1.3–76.0 % (mean 25.3 %), respectively. Primary and secondary mice with <1 % transgene expression were analyzed and summarized in a separate group (see chapter below).
Myelogram of the mouse #597
% of total number
Total of myeloid cells
Total of erythroid cells
Other cells (megakaryocyte etc)
Ratio myeloid/erythroid (M/E)
Splenogram of the mouse #597
% of total number
Total of myeloid cells
Total of erythroid cells
Other cells (megakaryocyte etc)
In animal #520, cytological analysis revealed a normal myelogram, but strongly increased erythroid cells (21.3 %) in the spleen. The presence of splenomegaly (weight of spleen 268 mg) and increased neutrophils, monocytes, and megathrombocytes in PB (Fig. 2c) indicated a myeloproliferation neoplasm. A final diagnosis of myeloproliferation (genetic) was made after histological examination showed typical megakaryocytic alteration in spleen (Fig. 2d) and infiltration of megakaryocytes in the liver (data not shown). Moreover, there were strongly increased Sca1+/c-Kit+ progenitor cells both in bone marrow and spleen flow cytometric analysis. Taken together, these three cases illustrate how important cytological analyses are for proper diagnosis of hematological disorders. Combining cytological analysis with histological and flow cytometric analysis and clinical parameters can improve diagnosis of hematological malignancies in mice after gene marking. Moreover, in the EGFP group (n = 10) we observed one animal (#486) with precursor B-cell lymphoblastic leukemia as previously described .
Importantly, while the total myeloid cells, total erythroid cells, and lymphoid cells were in the normal range in the remaining mice from the EGFP and tCD34 groups when compared to healthy controls , the remaining 15 out of 17 mice from the dLNGFR group had significantly reduced lymphoid cells (mean 17.7 vs. 21.0 %, p < 0.05) in the bone marrow (also trend in spleen, 74.5 vs. 86.9 %, p = 0.99), suggesting a potential altered differentiation of lymphoid cells by dLNGFR. Interestingly, myelograms and splenograms from all groups were generally in agreement with immunophenotypic analyses (e.g., 33.4 % monocytes in splenogram in Table 2 vs. 35.8 % monocytes by FACS analysis in Fig. S3 for mouse #597), with exception of the relatively low percentage of erythroid cells (represented by Ter119 positive cells) and lower blast count in some cases detected by flow cytometric analyses (data not shown). This is consistent with the accepted view that flow cytometric determination of blast count should not be used as a substitute for morphological evaluation .
It should be noted that most cases with hematological malignancies in the dLNGFR and tCD34 groups occurred in secondary recipients (i.e., #589, #597, and #602), supporting our previous observations that serial transplantation stresses hematopoiesis and increases the risk of genotoxic side effects by retroviral vectors [11, 30, 31]. Thus, secondary transplantation should be performed for preclinical study in mouse models.
Impaired lymphomyeloid differentiation in all animals with high expression of dLNGFR
Insertional mutagenesis may have contributed to malignancies
While dLNGFR expression in vivo in this study was very similar as in our earlier study , tCD34 expression was much higher than our previous studies (57.7 vs. 35.4 %) . The higher gene marking efficiency in the tCD34 group and consequently higher risk of insertional mutagenesis may have contributed to development of hematological malignancies in the present study. We therefore performed LM-PCR in all five diseased (i.e., #486, #520, #589, #597, and #602) and selected animals with >20 % transgene expression. Insertion sites recovered are presented in Table S3. Interestingly, all diseased mice (with exception for #602) had two common integration sites (CISs), while the analyzed healthy mice carried maximal only one CIS . Importantly, some CISs recovered from diseased mice are known to play important role in myeloid and lymphoid neoplasms, e.g., Zeb2 , Notch1 , and Hmgb1 . These data strongly suggested that insertional mutagenesis may have contributed to development of malignancies, although the contribution of transgenes or other unidentified factors cannot be excluded. For unknown reason, there was no insertion into the MDS-associated gene Evi-1 in the present study, although insertions into Evi-1 frequently occurred in our previously published studies  and Evi-1 represents one of the most frequent insertion sites listed in the retroviral tagged cancer gene database .
Myelogram and splenogram of mice with <1 % transgene expression
We also performed cytological analyses of bone marrow and spleen in 18 mice without transgene expression (defined as <1 % transgene expression of tCD34, dLNGFR, EGFP, and other transgenes by flow cytometric analysis) from dLNGFR/tCD34 cohort and several other experiments involving other transgenes between 2006 and 2010. Three (#591, #593, and #600) out of five animals after secondary transplantation (only from dLNGFR and tCD34) developed follicular B-cell lymphoma (diagnosed by histological analysis) and precursor T-cell lymphoblastic leukemia (n = 2). The diseases probably resulted from very poor engraftment of donor cells after transplantation, which is consistent with our previous observation and reports from other groups that in vivo expansion of the endogenous T- or B-cell compartment stimulated by radiation and serial bone marrow transplantation induces T- or B-cell (mainly T-cell) leukemia in mice [31, 36, 37]. However, the remaining 13 animals (mean observation time of 264 days; range, 71–343 days), which did not undergo serial transplantation, showed only reduced lymphoid cells in bone marrow (11.9 vs. 21.0 %, p < 0.001) and spleen (80.1 vs. 86.9 %, p < 0.05) and increased erythroid cells (9.7 vs. 5.7 %, p < 0.05) in the spleen when compared with non-irradiated healthy controls, suggesting that irradiation per se has no major effect on hematopoiesis in mice after transplantation.
In the present study, we investigated changes in the hematologic composition after transplantation with retrovirally modified hematopoietic stem cells in a large cohort of animals by systematic cytological analyses. Our previous study of the bone marrow and spleen of healthy non-transplanted animals provides reference ranges of each cell type and for the myeloid/erythroid ratio , which has been very useful for investigating hematological changes after gene marking experiments in the present study. Totally, we observed hematological malignancies in four out of 30 animals transplanted with dLNGFR and tCD34 modified stem/progenitor cells. Our data demonstrate that cytological analysis provides important information for the diagnosis of hematological disorders and thus should be included in preclinical studies. Moreover, our data strongly support that the spleen is an essential organ for the diagnosis of hematological neoplasm . In animal #520 from our study, the myleoproliferative alteration was much more evident in the spleen (Fig. 2d) than in bone marrow (data not shown). In at least seven other animals from other cohorts not presented in this study, we observed myeloid leukemia exhibiting strong leukemic infiltration in the spleen, while the bone marrow remained almost normal. Interestingly, we also observed one animal from another study with myeloid leukemia showing 87.5 % blasts in the bone marrow and 6.4 % in the spleen, which is far above the normal range (<1 % blasts in healthy spleen ).
It is essential to investigate the potential risks associated with the use of transgene before setting up a clinical gene therapy trial. However, in most cases, it is difficult to make a clear decision whether overexpression of the transgene analyzed is safe or not (e.g., multidrug resistance 1 [30, 38, 39] and IL-2 receptor gamma chain ), unless the transgene dramatically impairs differentiation of hematopoietic cells and increases leukemia risk, e.g., HOXB4 [13, 40]. Thus, long-term observation in different preclinical models should be performed and reproduced in independent laboratories. We for the first time observed hematological malignancies in mice after gene marking with a retroviral vector encoding dLNGFR in 2002 . A single vector integration activating the Evi-1 (ecotropic viral integration site-1) proto-oncogene produced preleukemic lesions which progressed to acute monocytic leukemia, possibly due to cooperation with dLNGFR-dependent signal alterations. One focus of the present study was to investigate the toxicity of dLNGFR overexpression in long-term hematopoietic stem cells in a new cohort of animals. By systematic cytological analysis in combination with histological analysis and other approaches, we identified two animals (#589 and #597) with hematological malignancies apparently associated with dLNGFR expression. In the present study, leukemia risk due to gene marking was very similar among the dLNGFR, tCD34, and EGFP cohorts. However, there are several other reasons that dLNGFR still might not be safe for stem cell marking: (1) The remaining 15 animals of the dLNGFR group which did not develop malignant disease, displayed significantly reduced lymphoid cells in bone marrow. Moreover, flow cytometric analysis demonstrated altered hematological differentiation in peripheral blood in all animals with high expression of dLNGFR (e.g., #535, Fig. 3). (2) Myeloid malignancies, i.e., monocytic leukemia and MDS, induced by dLNGFR, were founded both in the present study and our earlier study . This strongly suggests that the transgene dLNGFR contributed to malignant transformation, as both malignancies rarely occurred following gene marking in our previous studies with exception of the MPL study , even never seen in tCD34 and EGFP cohorts so far. We are however aware that insertional mutagenesis and/or other unknown mutations may also have played an important role. It is worth to point out that NGF/TRK signaling may be important for survival of monocytes and macrophages . (3) TRKA activation by nerve growth factor (NGF) plays an essential role for expansion of murine HSCs (presented by Dr. C. Eaves in Dresden July 17th, 2012). As the truncated form of the LNGFR displays a much higher ability to interact functionally with the TRK receptors including TRKA , thus expression of dLNGFR in HSCs may potentiate masked autocrine loops as well as responses to limited amounts of exogenously provided neurotrophins including NGF, and might subsequently result in aberrant expansion of HSC.
Our data strongly highlight the need for a combination of histological and cytological analyses to diagnose hematopoietic malignancies in murine models. One limitation of cytospin analysis is the relatively low amount of megakaryocytes observed in cytospins. This is an artifact, probably due to the low cell number used for cytospins (1–2 × 105 cells/cytospin). For instance, #520 had strong extramedullar hematopoiesis with increased numbers of megakaryocytes in the spleen by histology, but few megakaryocytes were observed in the cytospin preparation.
Why did we observe hematological malignant transformation more often than most other scientists in the field of gene therapy? After our publication of the first case of leukemia induced by gene marking , this question has been raised by many of our colleagues. Animals can develop hematopoietic neoplasms with latencies ranging between several weeks to up to 2 years. However, many mouse studies are conducted with (1) observation periods rarely exceeding 6–12 months, and without serial transplantation in many studies, (2) observation of mice not always being performed continuously, particularly with lapses on weekends and holidays, and (3) the spleen not being analyzed for malignant transformation. Thus in many studies, it is likely that animals become lost, and some are not able to be properly analyzed and reported as death of unknown reasons or even not mentioned in the publication . Our animals are monitored very regularly, and in some cases visited daily. Another important point is the careful morphological analysis of animals in our studies, which has unfortunately not been performed or completed in many studies from other groups. These analyses are very time consuming and require much experience. The reference data of myelograms and splenograms established from healthy controls in our laboratory can be used to improve the quality of preclinical gene marking studies in murine models .
In summary, we performed cytological characterization of murine bone marrow and spleen after transplantation with retrovirally modified hematopoietic stem cells in the largest cohort reported so far and over a long observation period. Our data demonstrate that cytological analysis provides very important information for diagnosis of hematological disorders and thus should be included in preclinical studies and performed in each investigated animal. Together with histological analysis, flow cytometric analysis and other analyses, the quality and meaningfulness of preclinical gene therapy studies will be improved. Moreover, our data suggest that analysis of spleen is just as important as bone marrow for diagnosis of hematological malignancies. Finally, we believe that a better appreciation of the morphology of the murine blood cells will improve the quality of in vivo preclinical gene therapy, leukemogenesis, and hematopoiesis studies.
This study was supported by the Deutsche Forschungsgemeinschaft (DFG, Li 1608/2-1 and BA1837/ 7-2) and the Deutsche Krebshilfe (grant: 108245). We are very grateful to Prof. Dr. Christopher Baum for his support; Prof. Dr. Kenji Kamino for histopathological examination of animals at the beginning of study; Dr. Mathias Rhein for providing samples; Dr. Olga Kustikova, Teng Cheong Ha, and Kezhi Huang for help with LM-PCR; Rena-Mareike Struß, Jessica Wenzl, Cindy Elfers, Thomas Neumann, Ellen Neumann, Christine Garen, and Mareike Knackstedt for technical assistance; and Dr. Michael Morgan for critical reading of this paper. We also thank Dr. Scott C. Kogan (University of California, San Francisco) for kind discussions about diagnosis of myeloid neoplasms, Jörg Frühauf and Hans Grundke for the irradiation of animals, and Stefanie Ernst and Dr. Michael Schneider (Institute of Biometrics, MHH) for help with statistical analysis. This study is part of the COST Action BM0801 (EuGESMA).
Authorship and conflict of interest statements
MY performed research, collected, analyzed and interpreted data, and wrote the manuscript. GB performed histological analysis. ZL designed and performed research, collected, analyzed and interpreted data, and wrote the manuscript. AG performed research, analyzed and interpreted data, and wrote the manuscript. The authors do not declare any competing financial interests.