Author contributions: R.M.R.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing, E.S.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing, J.-H.L., J.B.L., S.-H.H, and E. Szabo: collection and/or assembly of data and data analysis and interpretation; M.B.: conception and design, financial support, data analysis and interpretation, manuscript writing, and final approval of the manuscript. R.M.R. and E.S. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS November 30, 2011.
Hematopoietic stem cells (HSCs) can regenerate the entire hematopoietic system in vivo, providing the most relevant criteria to measure candidate HSCs derived from human embryonic stem cell (hESC) or induced pluripotent stem cell (hiPSC) sources. Here we show that, unlike primitive hematopoietic cells derived from hESCs, phenotypically identical cells derived from hiPSC are more permissive to graft the bone marrow of xenotransplantation recipients. Despite establishment of bone marrow graft, hiPSC-derived cells fail to demonstrate hematopoietic differentiation in vivo. However, once removed from recipient bone marrow, hiPSC-derived grafts were capable of in vitro multilineage hematopoietic differentiation, indicating that xenograft imparts a restriction to in vivo hematopoietic progression. This failure to regenerate multilineage hematopoiesis in vivo was attributed to the inability to downregulate key microRNAs involved in hematopoiesis. Based on these analyses, our study indicates that hiPSCs provide a beneficial source of pluripotent stem cell-derived hematopoietic cells for transplantation compared with hESCs. Since use of the human–mouse xenograft models prevents detection of putative hiPSC-derived HSCs, we suggest that new preclinical models should be explored to fully evaluate cells generated from hiPSC sources. STEM CELLS 2012; 30:131–139.
Hematopoietic stem cells (HSCs) found in bone marrow (BM), mobilized peripheral blood (M-PB), and umbilical cord blood (UCB) are transplanted in patients to treat hematological disorders, thereby providing the gold standard and functional definition of human HSCs [1–3]. The number of HSCs and progenitors extracted from these sources are often limited in supply and donor to recipient immunological compatibility . Hence, unlike human embryonic stem cells (hESCs), human induced pluripotent stem cell (hiPSC)-derived HSCs would offer an abundant alternate source of transplantable autologous cells capable of accelerating short-term hematopoietic recovery that would reduce patient mortality rates and eliminate graft-versus-host disease .
Both mouse embryonic stem cells (mESCs)  and reprogrammed mouse iPSCs  give rise to chimera mice with functional hematopoietic systems. These observations provide direct evidence that mouse pluripotent stem cells (mPSCs) possess the developmental potential to generate functional HSCs via in utero differentiation. Furthermore, mESCs differentiated in vitro toward the hematopoietic lineage are capable of reconstituting the BM of mice [8–10], suggesting that in vitro differentiation can also generate “HSC-like” cells from mPSC sources. Despite dozens of reports demonstrating primitive hematopoietic potential from hiPSCs sources in vitro, in vivo hematopoietic reconstitution has not been reproduced with the same level of success as mPSC-derived cells [11–13].
By definition, the ability to engraft and reconstitute the hematopoietic system of a conditioned recipient serves as the only functional measure of HSCs . Thus, transplantation of human hematopoietic cells into irradiated immune-deficient mouse recipients is proposed as a reliable in vivo model that approximates to human clinical transplantation procedures  and can therefore be used to identify putative HSCs from hiPSC sources [5, 15]. Accordingly, preclinical validation of hiPSC-derived HSCs requires evidence that these cells are functionally similar to HSCs derived from somatic hematopoietic tissues such as M-PB, adult BM, or UCB that are capable of regenerating hematopoiesis in human–mouse xenotransplantation models.
Here, using a xenotransplantation mouse model capable of measuring the processes that collectively detect HSCs, we provide an in depth in vivo and molecular comparison of the hematopoietic developmental potential of hESCs and fibroblast-derived hiPSCs to bona fide somatic HSC derived from adult sources. Unlike derivatives of hESCs, hiPSC-derived hematopoietic cells engraft, like adult HSCs, but fail to differentiate in vivo and thus do not regenerate hematopoiesis. However, hiPSC-derived graft cells isolated form xenotransplant recipients generate colony forming units (CFUs) in vitro thereby revealing their hematopoietic potential. The molecular mechanism underlying the in vivo block in hematopoiesis of hiPSC derivatives was correlated to specific microRNA (miRNA) misregulation.
MATERIALS AND METHODS
Derivation of Reprogrammed iPSCs
Lentiviral vectors used to induce ectopic expression of Oct4, Nanog, Sox2, and Lin28 in adult dermal fibroblasts were provided by J.A. Thomson . Briefly, fibroblasts were transduced with lentiviral gene expressions and maintained on Matrigel-coated plates in Dulbecco's modified Eagle's medium/F12 supplemented with 20% knock-out serum replacement, 12 ng/ml of basic fibroblast growth factor, and 10 ng/ml of insulin-like growth factor 2. hiPSCs were examined for the expression of pluripotent markers including Oct4, cell surface stage-specific antigen 3 (SSEA3), SSEA4, and TRA-1-81 by flow cytometry.
Hematopoietic Development of hiPSCs and Isolation of hiPSC-Derived Primitive Hematopoietic Cells
Hematopoietic differentiation of hESCs and hiPSCs was performed as previously described [17, 18]. Day 18 embryoid bodies (EBs) were dissociated into single cells and then stained with CD34-FITC and CD45-APC (BD Biosciences, San Diego, CA, www.bdbiosciences.com/eu/index.jsp). During hematopoietic EB development, loss of pluripotent markers (Oct4, Nanog, and Sox2) and acquisition of mesodermal lineage markers (MIXL1 and Brachyury) were quantified by quantitative real-time polymerase chain reaction (qRT-PCR). CFU assays were conducted on hESC-derived cells, hiPSC-derived cells, and HSCs (Fig. 2A) following a previously described protocol .
Transplantation of hiPSC-Derived Hematopoietic Cells
To examine hematopoietic repopulation capacity of hiPSCs, dissociated cells from developing EBs were injected (4.0 × 105 cells per mouse) into the femurs of sublethally irradiated (3.65 Gy) either 8–10-week-old NOD.CB17-Prkdcscid/J (NOD-SCID) or NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice (Jackson Laboratories, Bar Harbor, ME, www.jax.org) . For hiPSC-derived cells, 20.8% of cell injected were measured to be CD34+CD45+. As a control, lineage-negative M-PB cells were injected. The injected femur, collateral bones, and spleen were harvested 10 weeks post-transplantation and stained for human marker (human leukocyte antigen A/B/C [HLA-A/B/C]), pan-hematopoietic marker (CD45), myeloid (CD33), B cell (CD19), T cell (CD3), primitive population (CD34), and endothelial/hematopoietic (CD31) markers and analyzed using flow cytometry (FACS Calibur, BD).
Gene Expression and miRNA Profiling
The CD34+CD45+ population (defined as the primitive hematopoietic cell population) was isolated by fluorescence-activated cell sorting (FACS; FACSAria II, BD Biosciences) and used for mRNA and miRNA profiling. Total RNA from purified populations was extracted and amplified as described previously . Amplified-labeled RNA was hybridized to HG-U133Plus v2.0 chip (Affymetrix) by using standard protocols at the Ottawa Health Research Institute (Ottawa, Canada) and Microarray Facility of the Robarts Research Institute (London, Ontario, Canada). Array data were normalized and comparisons were performed using the DNA-Chip Analyzer (dChip) software (Harvard School of Public Health, Boston, MA). dChip software implements invariant set normalization and probe-level model-based expression analysis on multiple arrays and computes the t statistic and the p value based on the t distribution . Batch effect was corrected using dChip software. Hierarchical clustering was performed with dChip where the distance between two genes was defined as 1 − r where r is the Pearson correlation coefficient. Sample clustering in Figure 3D was performed based on expression values of genes expressed in at least 30% of the arrays analyzed and whose variation coefficient (SD/mean) ranged between 125 and 1,000. In addition to new array acquisitions, we used several public arrays for analysis, including HSCs extracted from umbilical fetal cord blood (HSC-FB; array numbers: GSM87705, GSM87706, and GSM87707), HSCs extracted from neonate cord blood (HSC-CB; array numbers: GSM87705, GSM87707, and GSM87709), HSCs resident in the BM (HSC-BM; array numbers: GSM87729, GSM87731, and GSM87733), human fibroblasts (array numbers: GSM557000, GSM5577001, GSM5577002, and GSM586150), hiPSCs (array numbers: GSM556996, GSM556997, and GSM556998), and hESCs (array numbers: GSM556994 and GSM556995). Total RNA was extracted (Total RNA isolation kit, Norgen, Thorold, Canada, www.norgenbiotek.com), and the miRNA expression profile for each sorted population (Fig. 4A) was analyzed using an Applied Biosystems (Foster City, CA, www.appliedbiosystems.com/absite/us/en/home.html) miRNA TaqMan panel of 365 individual miRNAs. RT reaction and amplification were done following ABI instructions. The amount of RNA was calibrated to the expression of RNU48. Diana software (http://diana.cslab.ece. ntua.gr) was used to identify genes targeted by miRNAs. qRT-PCR validation of selected miRNA was performed using TaqMan primer/probes and normalized to the RNU6B endogenous control using delta Ct method.
PCR analyses were performed to detect human DNA extracted from the BM and spleen of host recipients. High-molecular DNA was isolated using phenol–chloroform, and chromosome-17 α-satellite PCR was performed following an established protocol .
The mean ± SD was calculated for data sets and analyzed using a Student's t test.
Differentiation of hiPSCs and hESCs Generates Primitive Hematopoietic Progenitors
Human iPSCs were generated from human dermal fibroblasts transduced with Oct-4, Nanog, Sox-2, and Lin-28  and gave rise to colonies with hESC-like morphology (Supporting Information Fig. S1a) that expressed several markers of pluripotency (SSEA3, Tra-1-81, Oct4, and SSEA4; Supporting Information Fig. S1b). Pluripotency status of hiPSCs was demonstrated by their ability to form EBs (Fig. 1A) and ability to give rise to all three germ layers in an in vivo teratoma assay (Supporting Information Fig. S1c). Hematopoietic differentiation was induced in EBs from both hESCs and hiPSCs  and resulted in time-dependent downregulation of the pluripotency markers, Oct4, Nanog, and Sox2, and upregulation of mesodermal differentiation markers, MIXL1 and Brachyury consistent with differentiation responses (Fig. 1A). These results indicated that hematopoietic lineage specification of hiPSCs follows a similar trajectory as hESC in response to soluble cytokines. To determine whether hiPSCs follow similar developmental stages during hematopoietic commitment as hESCs [18, 23], the frequency of cells at each stage of hematopoietic specification and emergence was assessed (Fig. 1B). Although, higher levels of hemogenic precursors (CD31+CD34−CD45−) were generated from hiPSC, equivalent levels of primitive hematopoietic cells (CD34+CD45+) and mature blood cells (CD45+CD34−) were generated from hiPSC and hESC sources (Fig. 1C). Similar trends in hematopoietic development were observed on a second, independently reprogrammed hiPSC line (Supporting Information Fig. S1d). These results verify that fibroblasts reprogrammed to a pluripotent state not only follow the same stages of hematopoietic development as hESCs but also differentiate over the same time period with equivalent output of hematopoiesis.
Functionally, the hematopoietic potential of hiPSC-derived primitive hematopoietic cells was assessed in vitro using the CFU assay. This assay provides a quantitative in vitro measure of hematopoietic differentiation and detects robust clonal proliferation into multiple hematopoietic lineages from both HSCs and hematopoietic progenitors . Similarly, both hESC-derived and hiPSC-derived hematopoietic cells were able to respond to in vitro signals for sustaining hematopoiesis as indicated by their capacity to generate CFUs (Fig. 1D). Using CD34 as a marker of primitive hematopoietic cells, hiPSC derivatives yielded equal numbers of CFUs when compared with hESC counterparts but lower than adult HSCs (Fig. 1D). We demonstrate that hiPSCs-derived cells have comparable in vitro hematopoietic capacity to hESC-derived counterparts and HSCs.
iPSC-Derived Cells Graft the BM but Do Not Reconstitute Human Hematopoiesis
Although hESC-derived primitive hematopoietic cells (hESCphc) and hiPSC-derived PHCs (hiPSCphc) show similar phenotypic expression to adult HSCs (CD34+CD45+), surface marker expression and CFU capacity does not provide a surrogate for in vivo reconstitution. To functionally examine the hematopoietic reconstitution potential of differentiated hESC- and hiPSC-derived cells, we intrafemorally injected each cell type into irradiated NOD.CB17-Prkdcscid/J (NOD-SCID)  or NSG  recipients and directly compared the level of reconstitution to that achieved by adult HSCs (Fig. 2A, Supporting Information Fig. S2). Intrafemoral injection was selected as the mode of delivery to localize cell delivery to the BM and minimize any deviation in cell migration or variability in engraftment. In doing so, transplanted cells are isolated within this defined in vivo microenvironment. Importantly, hiPSC graft cells are compared with control samples based on adult HSCs, which are delivered in an identical manner thereby normalizing any possible influences from cell migration/engraftment variability. NSG mice provide a more permissive environment for HSC and progenitor engraftment and have therefore been adopted for reconstitution studies  while NOD-SCID recipients exclusively detect putative HSCs [14, 26]. Flow cytometry was used for quantification of human engraftment levels due to its ability to discriminate between live and dead cells by way of 7-aminoactinomycin D exclusion and size complexity (side scattered and forward scattered) gating. Human engraftment was not detected in the injected femurs, contralateral bones, or spleens of NOD-SCID recipients injected with either hiPSC- or hESC-derived cells whereas adult HSCs readily engrafted these recipients (Supporting Information Fig. S2). In contrast, when transplanted in NSG recipients, human hiPSC-derived cells could be detected by flow cytometry analysis in the injected femurs at a level in excess of the 0.1% threshold defining graft formation similar to mice transplanted with adult HSCs, whereas hESC-derived cells were below this threshold (Fig. 2A). However, hiPSC-derived cells were not detected in the contralateral bones or spleen of NSG recipients, while HSCs were detected in these tissues (Fig. 2A). Human graft formation of iPSC-derived cells was further validated by detection of a human specific genomic DNA sequence (chromosome 17) in the injected femur, but similar to the flow cytometer analysis, this marker was not detected in the spleen and contralateral bones of recipients (Fig. 2B, Supporting Information Fig. S2b). These data indicate that, unlike hESC-derived cells, hiPSC-derived hematopoietic cells are able to graft and survive in the BM of recipient NSG mice for 10 weeks post-transplantation but are unable to leave the delivery site and/or enter the circulatory system to migrate to other sites to establish hematopoiesis. Therefore, transplanted iPSCphc lack a key property associated with bona fide adult HSC reconstitution processes.
The phenotypic properties of iPSC-derived graft cells recovered at 10 weeks post-transplantation were specifically analyzed for hematopoietic, endothelial, and primitive hematopoietic markers and compared with the cellular phenotype of engrafted adult HSCs (Fig. 2C). Despite their ability to form and sustain a graft for 10 weeks post-transplantation, grafted human hiPSCs-derived cells neither express the pan-hematopoietic marker CD45 (Fig. 2D) nor did they express the mature hematopoietic markers of myeloid (CD33), B-cell (CD19) and T-cell (CD3) lineages (Supporting Information Fig. S3), therefore indicating a lack of hematopoietic reconstitution of the BM. Instead, they expressed a primitive hematopoietic marker CD34 and a common endothelial and hematopoietic lineage marker CD31 (Fig. 2E). When this phenotype is compared with engrafted HSCs, hiPSC-derived graft cells are shown to express the primitive hematopoietic marker CD34 at significantly higher levels relative to HSC grafts while expressing equal levels of CD31 (Fig. 2E). Indeed, it is documented that engrafted HSCs display low levels of CD34+ cells . The complete lack of a discrete HLA+CD45+ in hiPSC-derived graft cells indicates these cells are blocked in a primitive state.
Our in vivo analysis indicates that in contrast to previous studies with hESC-derived hematopoietic cells [11, 12], which display microchimerism at levels below the 0.1% threshold [27, 28], hiPSC derivatives can sustain a graft following intrafemoral delivery into immune-deficient NSG mice at levels above 1%. However, these iPSC-derived graft cells are blocked in primitive phenotype and are unable to undergo complete hematopoietic maturation. The retention of hiPSC-derived graft cells in the BM of NSG mice affords us the opportunity to systematically relate the in vivo properties of hiPSC-derived cells to the underlying molecular basis for failure to reconstitute the hematopoietic system in vivo.
Explanted hiPSC-Derived Graft Cells Retain Hematopoietic Capacity In Vitro
Prior to transplantation, hESC derivatives, hiPSC derivatives, and adult HSCs were shown to generate CFUs (Fig. 3A, 3B) but only hiPSC derivatives and adult HSCs survived 10 weeks in the mouse BM (Fig. 1A). We also assessed engraftment levels at 2–3 weeks and found undetectable levels of hiPSC graft cells (data not shown). Although incapable of reconstituting the BM, when iPSC-derived graft cells were recovered from xenotransplanted mice and cultured in methocellulose conditions optimized for human cell growth (Fig. 3A), they were able to generate hematopoietic CFUs (specifically, granulocytic, monocytic, and granulomonocytic) at approximately one-third the CFU potential of engrafted adult HSCs (Fig. 3C, Supporting Information Fig. S4) although similar CFU equivalents were transplanted in mice (Fig. 3B). The ability to generate CFUs indicates that these graft cells retain hematopoietic capacity for the duration of the transplant study but were unable to respond to the in vivo hematopoietic maturation signals present in the mouse BM. This dormancy in the early stages of hematopoietic development appears to be a property unique to hiPSC-derived cells because HSCs are able to respond to the same signals of the mouse BM microenvironment and initiate and sustain hematopoiesis. Accordingly, we sought to determine why hiPSCs-derived cells failed to respond to the conductive in vivo signals of the mouse BM.
Gene Expression Profile of iPSCphc Is Distinct from Putative Adult HSCs
We hypothesized that the inability for hiPSC-derived graft cells to respond to in vivo signals could be explained by a differential molecular state compared with adult HSCs fully competent for reconstitution. We have previously demonstrated that the primitive hematopoietic fraction differentiated from hiPSCs is solely responsible for in vitro hematopoietic development  and thus selected this population phenotype (i.e., CD34+CD45+) for molecular interrogation and direct comparison with adult HSCs. Consequently, gene expression profiles of purified hESCphc and iPSCphc were compared with HSC-FB and HSC-CB or HSC-BM in an effort to understand how differences in gene expression could account for the inability of hiPSC-derived graft cells to sustain hematopoiesis in vivo. Hierarchical clustering revealed that the global genetic states of hESCphc and iPSCphc were several orders removed from somatic HSCs (HSC-FB, HSC-CB, and HSC-BM) (Fig. 3D, Supporting Information Fig. S5a, Supporting Information Table S1).
miRNAs Misexpression Correlates with a Block in iPSCphc In Vivo Hematopoiesis
The numerous differences in gene expression between human PSC (hiPSC)-derived hematopoietic cells and adult HSCs (Fig. 3D, Supporting Information Fig. S5a) may be attributed to the absence of key regulators able to control complex gene networks associated with in vivo self-renewal and differentiation [29, 30]. One category of regulators important to hiPSCs is miRNAs [31, 32], where a single miRNA possesses potent suppressor activity capable of affecting hundreds of genes . Based on this property and the ability to effect early cell fate decisions of stem cells and differentiation programs [34, 35], we investigated whether miRNAs were implicated in the inability of iPSCphc to achieve full in vivo hematopoietic reconstitution.
Using quantitative PCR arrays that detect 365 human-specific miRNAs, the differences in levels of miRNA expression between each stage of cellular development, from precursor to progeny, were compared to levels expressed in HSCs or HSCs recovered from reconstituted xenotransplantations (engrafted HSCs). These differences in miRNA expression between relevant populations were represented as heat maps. In addition, miRNAs, which were undetected in precursors and then detected in progeny, were demarked as “ON” or vice versa as “OFF.” This method of annotation permitted tracking of individual miRNAs at each stage of cell development (Fig. 3E, Supporting Information Fig. S6). Noteworthy is the transition from hiPSC to differentiated primitive hematopoietic phenotype that is marked by a downregulation, to mostly undetectable levels, of numerous miRNAs (Fig. 3E, Supporting Information Fig. S6).
We sought to identify which miRNAs were implicated in hematopoietic reconstitution by characterizing the differential miRNA expression profile between HSCs and engrafted HSCs. Of the nine miRNAs expressed in HSCs (Supporting Information Table S2a), eight were downregulated (specifically, miR-19b, 146a, 149, 191, 223, 324-3p, 490, and 585). An additional 11 miRNAs were turned ON in engrafted HSCs (Supporting Information Table S2b). Due to their prevalence in the early stages of reconstitution, we decided to further interrogate using qPCR only the six miRNAs which were downregulated during the transition from HSC to engrafted HSC (Fig. 3F) but excluded those miRNA detected in hiPSCphc. Complementary sequence analysis was used to identify the genes targeted by these specific miRNAs (Supporting Information Table S3). The levels of these candidate miRNAs were assessed in various hiPSC-derived populations, including: (a) hiPSCphc; (b) hiPSC-derived graft cells recovered from xenotransplantations; and (c) CFU generated from hiPSC-derived cells. The detection of human specific miRNA in hiPSC graft cells and CFUs confirmed the human origin of these populations (Fig. 3G). We found that in contrast to HSC reconstitution, hiPSC-derived graft cells do not downregulate several candidate miRNAs (Fig. 3G). For example, mir-149 and mir-490 are not downregulated when comparing hiPSCphc versus iPSC-derived graft cells in contrast to the reduction of these same miRNAs during the transition of adult HSC to engrafted adult HSC (Fig. 3F). Instead, higher levels of miRNAs were detected relative to their precursor, hiPSCphc, indicating that hiPSC graft cells respond to signals of the BM in an inverse response to that observed during HSC reconstitution.
To assess whether the block in in vivo differentiation of hiPSC-derived graft cells was due their inability to downregulate the candidate miRNAs while in the BM, we measured the miRNA levels of CFUs (which display in vitro multilineage differentiation) generated from the same iPSC-derived precursor population and found a similar downregulation in miRNAs for mir-149, 324-3p, and 490 as seen during HSC reconstitution, while mir-26a and mir-146a were slightly upregulated (Fig. 3G). However, mir-26a and 146a were highly expressed in hiPSC-derived cells both in the mouse BM and in CFUs compared with adult HSCs (Fig. 3G). Furthermore, of the six candidate miRNAs tested, mir-585 was the only candidate that was not detected in the hiPSC-derived precursor or progeny populations assayed, even though this miRNA is highly expressed in HSCs and is downregulated during engraftment. Due to its association as a negative regulator of miRNA biogenesis , we assessed whether the reprogramming factor, Lin28, in hiPSC could affect miRNA regulation. Total Lin28 was determined to be silenced in hiPSC versus hESC and thus excluded as the cause of miRNA misregulation during differentiation (Supporting Information Fig. S1e). The inability of hiPSC-derived cells to downregulate key miRNAs in vivo suggests that the inhibitory function of these miRNAs is likely preserved and can ultimately block in vivo hematopoietic regeneration (Fig. 4).
We show that differentiation of hiPSCs and hESCs generates primitive hematopoietic progenitors. hiPSC-derived cells graft the bone marrow but do not reconstitute human hematopoiesis in xenotransplantation models. However, explanted hiPSC-derived graft cells retain hematopoietic capacity in vitro. Gene expression profile of iPSCphc is distinct from putative adult HSCs. MicroRNA misexpression correlates with a block in iPSCphc in vivo hematopoiesis.
Our study reveals that misexpression of miRNAs in hiPSC-derived leads to deficiencies in in vivo hematopoietic regenerative capacity. Above all other requirements, putative HSCs derived from hiPSC sources must be able to reconstitute the hematopoietic system of host recipients similar to adult HSC function. Unfortunately, limited successes have been demonstrated with the candidate approaches used to date [11–13, 15, 37] and even less has been achieved in correlating the in vivo biological barriers of hematopoietic reconstitution to the molecular features that could account for these shortcomings. Instead of continuing to use candidate methods to generate putative HSCs from hESCs or hiPSCs, here we attempted to elucidate the in vivo biological barriers faced by hiPSC-derived hematopoietic cells.
We demonstrate that hiPSC-derived cells are superior to hESC-derived counterparts due to their ability to graft in the BM of xenotransplantation recipients but fall short of complete hematopoietic regeneration which is a feat currently only achievable with putative human somatic HSCs. The basis of hiPSC-derived cell survival for 10 weeks in the mouse BM requires further investigation. Their ability to generate hematopoietic CFUs following explantation suggests a primitive and quiescent hematopoietic nature. Indeed, adult HSCs are not dissimilar in this functional role. HSCs are able to reside in a stable quiescence stage for extended periods, only to be activated when stimulated . There is also the possibility that hiPSC graft cells proliferate in vivo but are short-lived thereby showing limited increase in cell number. Whatever the mechanism of hiPSC graft cell survival, the property of graft formation affords us the opportunity to use this model to probe the molecular mechanisms preventing hiPSC xenotransplantation hematopoietic reconstitution.
Unlike hiPSC derivatives, hESC-derived cells do not generate an in vivo graft (Fig. 2). We therefore have no cell product which we can molecularly interrogate or explant to generate CFUs and thus cannot demonstrate a block in hematopoietic regeneration. It may well be that failure of hESC derivatives to graft may be due to miRNA inhibition but we cannot exclude multiple other factors that may be causative or investigate this idea given the absence of cells found in vivo. As there is a growing body of evidence suggesting that iPSCs are different from hESC [32, 39], our results present another functional difference between these distinct human pluripotent sources.
We cannot exclude the possibility that hiPSC-graft populations arises from a nonhematopoietic cell origin. It is however unlikely that other metastable cell types, including cells with hemangioblastic properties, could contaminate the mouse BM for 10 weeks, as in vivo conditions of the BM are selective for hematopoietic cell maintenance and differentiation and are thus unlikely to support long-term survival of rare and metastable cell fates such as hemangioblasts or other types of cells capable of giving rise to hematopoietic progenitors (CFU) upon explant. Although it has been documented that vascular endothelial cells derived from hESCs are able to engraft in xenotransplantation studies , endothelial precursors or progenitors by definition are unable to generate hematopoietic CFUs. Therefore, even if the graft population comprises an endothelial precursor/progenitor population, our finding that a fraction of these graft cells give rise to hematopoietic CFUs suggests that a graft cell with hematopoietic potential is able to survive in vivo and be activated to differentiate ex vivo.
We demonstrate that when hiPSC-derived graft cells are recovered from xenotransplants they are able to proceed through in vitro hematopoietic multilineage differentiation indicating that, although poised to differentiate, these cells are not stimulated to fully mature in the host BM. Indeed, the phenotype of hiPSC-derived graft cells (i.e., CD34+CD31+CD45−) is similar to CD45negPECAM-1, Flk-1 and VE-cadherin positive (PFV) cells shown to possess both hematopoietic and endothelial cell fate potential . When subjected to hematopoietic cytokines in vitro, CD45negPFV cells develop a hematopoietic fate yielding hematopoietic multilineage differentiation. Similarly, hiPSC-derived graft cells can respond to these same hematopoieitc cytokines yet are unable to develop in the mouse BM. The inability of hiPSC-derived graft cells to respond to differentiation signals of the mouse BM is likely due to misexpression of molecular regulators of hematopoiesis correlated with multilineage differentiation.
Indeed, we show considerable deviation in the global gene expression between hiPSCphc and adult HSCs and argue that miRNAs, with their property to suppress multiple genes, are capable of altering gene networks. We show that several candidate miRNAs, in their potential capacity as molecular suppressors of hematopoiesis, were downregulated during HSC reconstitution. We associated these miRNAs as critical in hematopoiesis due to their prevalence in the early stages prior to reconstitution and continued expression, albeit lower levels, following xenotransplantation. We also noted that differentiation of hiPSC to primitive hematopoietic cells is marked by a downregulation of numerous miRNAs and suggest this feature to be a hallmark of hematopoietic differentiation, irrespective of the origin of the stem cell (i.e., adult or pluripotent).
This property of miRNA downregulation during differentiation was not replicated in hiPSC-derived graft cells. Instead, higher levels of expression were observed for all of the detected miRNAs in the graft cells shown to be blocked from in vivo hematopoiesis. The inability of hiPSC-derived cells to downregulate key miRNAs in vivo suggests that the inhibitory function of these miRNAs is likely preserved and can ultimately block in vivo hematopoietic regeneration. Indeed, retroviral-mediated overexpression of one of these miRNAs, mir-146a, in mouse HSCs has been shown to impair BM reconstitution . Furthermore, when hiPSC-derived cells were stimulated in vitro to undergo hematopoietic maturity and generate CFUs, several candidate miRNAs were downregulated. When taken together, we surmise that the block in in vivo reconstitution is due the inability of hiPSC-derived graft cells to downregulate key suppressors of gene expression in response to, or lack of, signals in the mouse BM. This relationship can be directly examined by transduction of multiple miRNA sponges in purified hiPSC-derived hematopoietic cells. Such experiments are not without considerable technical obstacles; although, for example, the low transduction frequency for a single miRNA sponge, let alone multiple sponges for miRNAs which may be acting in concert to inhibit reconstitution, coupled with the inability to control the kinetics of miRNA expression must be considered in the context of cell viability required to achieve functional engraftment in NSG mice . Our findings add new insight into the blockage of hiPSC-derived hematopoietic reconstitution and provide the impetus to tackle the technical hurdles. It is also feasible that other xenotransplantation models, which more closely resemble human BM, may be conducive to hiPSC-derived hematopoietic reconstitution.
Hematopoiesis is a complex process that is controlled, to some degree, by miRNA regulation. Considerable genetic variability has been demonstrated between iPSC and hESC lines and this variability extends to the miRNA level [32, 39]. If hematopoietic regenerative medicine is desired, variability between patients and iPSC clones must be taken into consideration. We define a shortlist of six miRNAs critical in adult HSC reconstitution and assert that miRNA variability of iPSC be standardized to this shortlist.
Investigators should focus their attention in engineering hiPSC-derived cells that can respond to recipient BM signals and downregulate key miRNAs following engraftment. In addition, new preclinical models should be explored to fully evaluate cells generated from hiPSC sources. For example, iPSC generation from non-human primates followed by autologous transplantation of primate iPSC-derived hematopoietic cells is one avenue to address whether murine xenotransplantations are reliable hematopoietic models  and has been instrumental to HSC and gene therapy-related studies [43–46]. We propose this will also be the case for cell replacement therapies using non-murine PSCs.
M.B. holds the Canada Research Chair in human stem cell biology. This work was supported by a grant from CIHR and CCSRI to M.B. R.M.R is supported by TTF fellowship (CCSRI), and E.S. is supported by a Ministry of Research and Innovation fellowship. Xabier Aguirre and Robin M. Hallet are thanked for their assistance with microRNA-mRNA analysis, Tracy Wynder, Monica Graham, Aline Fiebig, Marilyne Levadoux-Martin, and Lisa Gallagher for technical support (mice), and Kausalia Vijayaragavan and Marc Bossé for technical support in first attempts at establishing iPSC protocols in the Bhatia Lab; Tamra E. Werbowetski-Ogilvie is thanked for reviewing the manuscript.
DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
The authors indicate no potential conflicts of interest.