Human bone marrow-like ossicles improve primary leukemia cell engraftment in NSG mice

Both hematopoietic stem cell (HSC) and leukemia initiating cell (LIC) survival and propagation require complex interactions between stroma cells, extracellular matrix, growth factors, and cell adhesion molecules in the bone marrow compartment.  This “bone marrow niche” in the highly immunodeficient NSG™ mice (005557) supports human HSC engraftment and multilineage differentiation. HSC-derived human myeloid cells, however, require human-specific factors for optimal expansion, and their absence in mice likely explains why many types of human primary myeloid leukemias are difficult to propagate in NSG™ hosts.  In the July 2016 issue of Nature Medicine, researchers at Stanford University (Palo Alto, CA) and Paracelsus Medical University (Salzburg, Austria) describe a new engraftment model in which a human ossicle – a bone marrow-like organoid - placed into the flank of NSG™ mice acts as a stem cell niche (Reinisch, et al 2016).  The human ossicle enables engraftment of normal human HSCs, as well as both patient-derived myeloid and lymphoid leukemias.

Development of a human bone marrow niche

Human ossicles were initiated by expanding human bone marrow-derived mesenchymal stem cells (MSCs) in culture. The MSCs were harvested and mixed with a solution of extracellular matrix molecules and growth factors, and the mixture was injected subcutaneously into four separate sites in the flank of NSG™ mice (005557).  The mice were treated with human parathyroid hormone for 28 days to promote the growth and development of the BM niche within the ossicles (see figure).

humanized ossicle

Establishment of a humanized ossicle in NSG™ mice creates a bone marrow niche that supports human hematopoietic stem cell (HSCs) and myeloid leukemia engraftment. Human bone marrow-derived mesenchymal stem cells are expanded in culture, mixed with extracellular matrix proteins and growth factors, and engrafted subcutaneously in the flank of NSG™ mice (005557).  After 8-10 weeks, human HSCs or myeloid leukemia progenitors can be injected either intravenously or directly into an ossicle.  Human hematopoietic or leukemia engraftment is monitored by flow cytometry and histology 7-24 weeks later.

The functional development of the ossicles was tested by injecting human cord blood-derived CD34+ HSCs.  The HSCs were injected either directly into a single ossicle or intravenously (i.v.) into the tail vein. The HSCs were allowed to develop in the mice for 7-24 weeks before both ossicles and endogenous bone marrow from the mice were harvested, and the presence of human cells was evaluated by flow cytometry and immunohistochemistry.  When a single ossicle was injected, human CD45+ hematopoietic cells were found not only in the injected ossicle, but also in non-injected ossicles as well as in host bone marrow.  In mice that received the i.v. HSC injections, human CD45+ hematopoietic cells were also detected in both ossicles and bone marrow.  Flow cytometry revealed that the ossicles contained human T and B cells, NK cells, neutrophils, eosinophils, monocytes, mast cells, and dendritic cells.  These cell types were further confirmed by cell sorting and May-Gruenwald Giemsa staining and microscopy.  Multipotent progenitors (MPP), lymphoid-primed multipotent progenitors (L-MPP), granulocyte macrophage progenitors (GMP), and megakaryocyte erythroid progenitors (GMP), also were detected by flow cytometry, as were HSCs (detected by sorting for CD34+ CD38- cells).  When these ossicle-derived HSCs were transplanted into secondary NSG™ recipients, human hematopoietic cells were detected 12 weeks later.  Therefore, human HSCs homed to the ossicle niches, indicating that the ossicles express appropriate adhesion molecules and signaling factors. The engrafted HSCs can differentiate into immature progenitors and phenotypically mature lymphoid and myeloid cells. Taken together, these results demonstrated that the ossicles develop into fully functional bone marrow niches.

Ossicles provide a preferential niche for human leukemia engraftment

Having demonstrated the ossicles’ functionality, the authors next evaluated whether the ossicles would support human leukemia cell engraftment.  Ossicles bearing mice were injected with primary human AML, B-ALL, or T-ALL, either directly into the ossicle or i.v.  Fifteen different human AML samples were tested by direct intra-ossicle injection, and 12 weeks post-injection, myeloid leukemic blasts (CD3-, CD33+) were observed in 13 of the 15 ossicle-bearing NSG™ hosts.  The human AMLs also expanded into the mouse bone marrow.  Four human AML samples were injected i.v., and all four homed to and engrafted the ossicles.  Intra-ossicle injected B-ALL and T-ALL also engrafted.  These results demonstrated that the ossicles support human primary leukemia engraftment and growth.

Although these initial experiments clearly demonstrated leukemia cell engraftment in both ossicles and mouse bone marrow (BM), the question remained whether the ossicles provide a preferential niche for the support of these cells.  This question was addressed by selecting AML samples that previously showed very poor engraftment in non-ossicle bearing NSGTM mice following i.v. injection. The engraftment kinetics of these samples was compared following three different injection protocols; i.v. injection into non-ossicle-bearing mice, i.v. injection into ossicle-bearing mice, and direct single intra-ossicle injection.  As expected, no AML cells were detected in non-ossicle-bearing mouse BM at 8 or 12-13 weeks post-injection, and AML cells comprised only 0.5% of total cells at 18-24 weeks.  AML cell distributions in ossicle-bearing mice that received i.v. AML cell injections were as follows: BMnegative ossiclelow(at 8 weeks); BMnegative ossiclehigh (at 12-13 weeks); BMlow ossiclehigh (at 18-24 weeks).  In the mice that received direct, single intra-ossicle injections, the non-injected ossicles were populated with human AML at earlier time points and at higher percentages than the mouse BM.  These results show that the human ossicles contain a niche that provides key components for the preferential engraftment, growth, and expansion of human leukemias that is not provided by mouse BM.

Honing in on the leukemia initiating cells

The superior engraftment supported by the ossicles led the investigators to evaluate the frequency with which leukemia initiating cells (LICs) are found in a given sample using a limited dilution engraftment assay. Identifying the phenotype and genotype of LICs and the frequencies at which they are present in AML samples could be important for understanding disease pathogenesis and, in turn, could contribute to developing new anti-leukemia therapies.  Two patient-derived AML samples were selected and injected into ossicles at sequentially lower doses.  The same samples and doses also were introduced into mice without ossicles by intrafemoral injection.  AML sample SU028 populated all ossicles when as few as 100 cells were injected directly into a single ossicle, but the same dose failed to engraft in the intrafemoral injected, ossicle-free group.  Similar results were found for AML sample SU048, although this sample required 500 cells to achieve engraftment.  Reports from earlier studies estimated that AML LICS were present in approximately 1 in 10,000 cells.  The ossicle-bearing mice revealed that SU028 and SU048 have LIC frequencies of 1:20 and 1:271, respectively.  Therefore, a niche more capable of supporting these neoplasms improves LIC detection sensitivity.

The authors next evaluated whether ossicles could support engraftment of acute promyelocytic leukemia (APL) and reveal the stage in hematopoietic cell differentiation when LICs develop.  APL previously showed little to no engraftment in NSG™ mice.  Following intra-ossicle injection, several APL samples showed high ossicle engraftment but relatively low mouse BM engraftment.  CD33+ myeloid cells within the ossicles were identified by in situ hybridization (FISH) to have the characteristic APL oncogenic translocation, “promyelocytic leukemia-retinoic acid receptor-a” (PML-RARA). May-Gruenwald Giemsa staining confirmed that characteristic azurophilic granules were present in cells resembling immature promyelocytes. Next, CD34+CD38-CD45RA- cells were isolated from primary APL samples. This phenotype identifies a hematopoietic fraction from normal healthy BM that contains primitive HSCs and multipotent progenitors (MPP) capable of differentiating into both myeloid and lymphoid cells.  Engraftment of these APL-derived cells was compared to that of bulk APL cells following intra-ossicle injection.  The APL-derived HSCs and MPPs engrafted and developed into normal CD19+ B-cells and CD33+ myeloid cells that were negative for PML-RARA.  The bulk APL-treated ossicles showed both lymphoid and myeloid cell engraftment at 6-8 weeks that became 100% myeloid cells containing the PML-RARA mutation by 24 weeks.  Collectively, these data strongly suggest that APLs arise downstream from both the HSC and MPP developmental stages in a more lineage-committed progenitor.

Myelofibrosis (MF) is another myeloproliferative neoplasm that engrafts immunodeficient mice poorly.  Unlike APL, MF LICs are thought differentiate from HSCs.  CD34+ cells were sorted from primary patient MF samples and were injected into ossicles.  As few as 50,000 cells allowed engraftment of primarily CD33+ myeloid cells that contained the Janus kinase 2 oncogenic gene variant, JAK2V617F.    The MF samples were then sorted into CD34+CD38low HSC-containing and CD34+CD38+ HSC-negative fraction, and each fraction was injected into ossicles.  The CD34+CD38+ fraction failed to engraft, but the CD34+CD38low/- fraction showed robust engraftment.  MF cells fractionated into MPPs or common myeloid progenitors also failed to engraft.  These results validate that MF LICs arise from primitive HSCs.

In summary, the Stanford and Paracelsus researchers have developed a unique new approach to create a humanized bone marrow niche that is capable of supporting the engraftment, differentiation, and growth of human HSCs and multiple, primary leukemias.  Further, human HSC-derived immune cells preferentially colonize the human organoid ossicles over the mouse host’s endogenous bone marrow.  The improved immune cell support provided by the ossicles enables more accurate determination of LIC frequency and temporal dissection of LIC development. Finally, the model described in Reinisch et al. could represent a significant step forward in developing new treatments for human leukemias.


Reinisch, et al., 2016.  A humanized bone marrow ossicle xenotransplantation model enables improved engraftment of healthy and leukemic human hematopoietic cells.  Nat Med.  July 22(7):812-21. PMID:27213817