Comparison of osteoclast differentiation protocols from human induced pluripotent stem cells of different tissue origins

Background: Ever since their discovery, induced pluripotent stem cells (iPSCs) have been extensively differentiated into a large variety of cell types. However, a limited amount of work has been dedicated to differentiating iPSCs into osteoclasts. While several differentiation protocols have been published, it remains unclear which protocols or differentiation methods are preferrable regarding the differentiation of osteoclasts. Methods: In this study we compare the osteoclastogenesis capacity of a peripheral blood mononuclear cell (PBMC)-derived iPSC line to a fibroblast-derived iPSC line in conjunction with either embryoid body-based or monolayer-based differentiation strategies. Both cell lines and differentiation protocols were investigated regarding their ability to generate osteoclasts and their inherent robustness and ease of use. The ability of both cell lines to remain undifferentiated while propagating using a feeder-free system was assessed using alkaline phosphatase staining. This was followed by evaluating mesodermal differentiation and the characterization of hematopoietic progenitor cells using flow cytometry. Finally, osteoclast yield and functionality based on resorptive activity, Cathepsin K and tartrate-resistant acid phosphatase (TRAP) expression were assessed. Results were validated using qRT-PCR throughout the differentiation stages. Results: Embryoid-body based differentiation yielded CD45+, CD14+, CD11b+ subpopulations which in turn differentiated into osteoclasts which demonstrated TRAP positivity, Cathepsin K expression and mineral resorptive capabilities. This was regardless of which iPSC line was used. Monolayer-based differentiation yielded lower quantities of hematopoietic cells that were mostly CD34+ and did not subsequently differentiate into osteoclasts. Conclusions: The outcome of this study demonstrates the successful differentiation of osteoclasts from iPSCs in conjunction with the embryoid-based differentiation method, while the monolayer-based method did not yield osteoclasts. No differences were observed regarding osteoclast differentiation between the PBMC and fibroblast-derived iPSC lines.

Ever since the discovery of resetting somatic cells to an embryonic-like state by Shinya Yamanaka in 2006 (1), induced pluripotent stem cells (iPSCs) have opened up previously inconceivable possibilities in the area of regenerative medicine, disease modeling and drug discovery (2). As was later established, the creation of these cells by transfecting them with the so-called Yamanaka factors (Oct3/4, Sox2, Klf4, c-Myc) allowed for their differentiation into all three germ-layers and subsequently into numerous terminally differentiated cell types (3)(4)(5)(6)(7)(8)(9)(10), thereby proving their pluripotency. As such, iPSCs provide the possibility to create any conceivable cell type to be used in an autologous manner (11,12). Even though challenges and considerations still exist when it comes to clinical translation (11), the possibility for autologous usage could allow for the treatment of currently untreatable diseases (13)(14)(15).
One such scenario of autologous usage of iPSCs consists in the utilization of engineered osteoclasts (OCs) as a strategy to treat bone diseases (16) and disorders involving ectopic calci cations (17)(18)(19).
Our group previously developed engineered murine OCs (iRANK cells) by transfecting OC precursors with a viral construct containing an intracellular receptor activator of nuclear κB (RANK) signaling domain linked to a fusion protein (FKBP12) that provides a binding site for the chemical inducer of dimerization (CID), allowing for a drug-controlled cell differentiation independent of receptor activator of nuclear factor κB ligand (RANKL) and osteoprotegerin (OPG) (17,20). These engineered iRANK cells haven been shown to be effective in applications involving the controlled resorption of heterotopic ossi cations (17,21), necrotic bone (22) and even more potential applications can be envisioned. Additionally, iRANK cells have also been used in disease modeling of diseases that involve OCs (22).
In terms of transitioning from murine cell to human cells, a robust source for human OCs is necessary not only for the creation of human iRANK cells but also other applications that require human OCs (23,24).
While CD34 + peripheral blood mononuclear cells (PBMCs) are still used as the prevalent source for OCs by many (25)(26)(27), the small number that can be harvested from a patient in conjunction with the inability to expand them in vitro (28, 29) limit their potential autologous use. iPSCs on the other hand can overcome the problem of cell number limitation as they can be expanded in vitro inde nitely and thereby allow for upscaling of OC production. Additionally, iPSCs can be created from a wide variety of tissue origins and do not require large amounts of blood to be drawn from the patient.
Several iPSC lines have to date been successfully differentiated into OCs (30,31). However, the entire process of generating human OCs from iPSCs can differ widely (32). Differences may arise starting with the vector for iPSC creation (33)(34)(35) and the protocols used for iPSC expansion (36). In order to generate OCs, iPSCs must undergo mesodermal and hematopoietic differentiation, which is followed by terminal OC differentiation (10,(37)(38)(39). Different approaches for mesodermal and hematopoietic differentiation have been published. One approach involves the creation of a single-cell iPSC suspension which is used to created small, spherical, embryoid body-like structures, which are said to simulate early stages of postimplantation embryonic development (4,40). Another approach uses iPSC colonies in monolayer to start mesodermal induction and hematopoietic differentiation (41,42). Both approaches give rise to cellforming complexes that produce hematopoietic progenitor cells (HPCs) but differ in the cytokines used for differentiation. While several differentiation protocols have been published, it remains unclear which protocols or differentiation methods are preferrable regarding the e cient and robust differentiation of human OCs.
In this study, we compare a PBMC-derived iPSC line to a broblast-derived iPSC line (PBMC-derived vs. broblast-derived) in conjunction with either an embryoid body-based (EB) (31) or a monolayer-based (MB) (37) differentiation protocol. Both cell lines and differentiation protocols were investigated regarding their ability to generate OCs and their inherent robustness and ease of use. First, both cell lines' ability to remain undifferentiated while propagating in a feeder-free system was assessed. This was followed by evaluating mesodermal differentiation and characterization of hematopoietic progenitor cells produced under the differentiation strategy. Finally, OC yield and OC functionality based on resorptive activity, Cathepsin K and TRAP expression were assessed and compared.

Methods iPSC culture
The OC proliferation and differentiation of two iPSC lines derived from distinct tissue origins were compared. MCND-TENS2, a peripheral blood mononuclear CD34 + /CD38 − cell-derived iPSC line from a healthy donor (37) (received from NIH National Heart Lung Blood Institute, Bethesda, MD, USA; registered at https://hpscreg.eu/cell-line/RTIBDi001-A) was compared to GM28404*B, a broblast-derived iPSC line documented to originate from "apparently healthy individuals" (received from Coriell Institute Cell Repository, Camden, NJ, USA; registered at https://www.cellosaurus.org/CVCL_C0M4). Both cell lines were created using a Sendai virus reprogramming kit (37,43). Upon defrosting, cells were cultured with mTeSR Plus (StemCell Technologies, Vancouver, Canada) on Cultrex basement membrane extract (R&D Systems, Minneapolis, USA) coated 6-well plates and incubated at 37°C and supplied with 5% CO 2 . 10 µM of Rho-associated, coiled-coil containing protein kinase (ROCK) inhibitor Y-276432 was used when defrosting to improve cell survival. Media changes were performed every second day. iPSCs were passaged at 70-80% con uency using 5 U/mL Dispase (StemCell Technologies, Vancouver, Canada).

Mesodermal induction, hematopoietic differentiation, and osteoclast differentiation
In this study a cytokine de ned differentiation protocol published by Rössler et al. (31) was compared to a commercially available differentiation kit (STEMdiff hematopoietic kit, StemCell Technologies, Vancouver, Canada), using the above-mentioned iPSC lines from different tissue origins. Both cell lines underwent mesodermal induction at passage 27.
EB differentiation according to Rössler et al. (31) was performed by creating a single cell suspension which was seeded in a round bottom ultra-low attachment 96-well plate at a cell density of 1.25x10 4 cells in 100 µL of mTeSR Plus supplemented with 50 ng/mL human bone morphogenetic protein 4 (BMP4) (StemCell Technologies, Vancouver, Canada), 50 ng/mL human vascular endothelial growth factor-165 (VEGF165) (StemCell Technologies, Vancouver, Canada), 20 ng/mL human stem cell factor (SCF) (StemCell Technologies, Vancouver, Canada), and 10 µM ROCK inhibitor Y-27632 (StemCell Technologies, Vancouver, Canada). Subsequently, the 96-well plate was centrifuged for 3 min at 100 x g and half medium changes performed after one and two days. Four days after mesodermal induction, embryoid bodies were transferred to a 6-well plate and further differentiated in X-VIVO 15 Medium (Bioscience Lonza, Basel, Switzerland) supplemented with 2 mM Ultraglutamine (Bioscience Lonza, Basel, Switzerland), 55 µM 2-mercaptoethanol (ThermoFisher, Waltham, USA), 25 ng/mL human interleukin 3 (IL-3) (StemCell Technologies, Vancouver, Canada), 100 ng/mL human macrophage colony-stimulating factor (M-CSF) (StemCell Technologies, Vancouver, Canada). A full medium change of 3 mL was performed after 5 days. After 10 days of differentiation, oating suspension cells were harvested and used for further OC differentiation.
MB differentiation using STEMdiff hematopoietic kit (37) (StemCell Technologies, Vancouver, Canada) was performed according to manufacturer's instruction. In short, 16-40 cell aggregates were seeded onto each 12-well plate coated with Cultrex basement membrane extract and treated for 3 days with medium A (containing bFGF, BMP4, VEGFA) (37) and consecutively for 10 days with medium B (containing bFGF, BMP4, VEGFA, SCF, Flt3L, TPO) (37). The oating suspension cells were harvested at the end of the period and used for further OC differentiation.
Following hematopoietic differentiation cells were transferred to 6-well plates at 2 x 10 5 cells/cm 2 and incubated in alpha-MEM (Bioscience Lonza, Basel, Switzerland) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 50 ng/mL human M-CSF for 3 days after which cells were incubated in alpha-MEM supplemented with 10% heat-inactivated FBS, 50 ng/mL human M-CSF and 70 ng/mL human RANKL (StemCell Technologies, Vancouver, Canada) for 7 days. Medium changes were performed every 2-3 days. OC differentiation was nished by a nal treatment using alpha-MEM supplemented with 10% heat-inactivated FBS and 80 ng/mL human RANKL for 2 days.

Enzymatic staining
To assess the degree of spontaneous differentiation during cell expansion, both iPSC lines were stained for alkaline phosphatase (ALP) (Abcam, Cambridge, UK) as a marker for pluripotency each time the cell lines were passaged. Additionally, ALP expression was assessed throughout differentiation. For this, cells were seeded onto 8-chamber slides, xed at 70-80% con uency and stained according to manufacturer's protocol. Tiled full well images were taken using an inverted wide eld microscope (DMI6000, Leica, Wetzlar, Germany).
Tartrate-resistant acid phosphatase (TRAP) staining was performed after M-CSF matured hematopoietic cells were seeded onto a calcium-phosphate 24-well bone resorption assay plate (Cosmo Bio, Tokyo, Japan) in order to assess osteoclastogensis using a TRAP staining kit (Sigma-Aldrich, St. Louis, MO, USA). Staining was performed according to manufacturer's instructions. In short, cells were xed using 4% PFA, washed, staining mix added to the plate and incubated for 20 min at 37°C. OCs were then counterstained with methyl green nuclear stain (Sigma-Aldrich, St. Louis, MO, USA) for 10 minutes at room temperature. Image acquisition was performed using a Leica DMI6000 inverted microscope in phase-contrast mode and tiled full well images were analyzed for cell size and multi-nucleation using ImageJ.
Immunocytochemistry Immunocytochemistry (ICC) was performed after EB or MB induction in order to assess mesodermal differentiation and after termination of RANKL treatment in order to assess OCs.
Following OC differentiation, cells were xed with 4% PFA, permeabilized with Triton X-100 (ThermoFisher, Waltham, USA) for 30 min and blocked with normal goat serum for 60 min. Following permeabilization and blocking, OCs were stained with a primary anti-Cathepsin K antibody (Abcam, Cambridge, UK) which was succeeded by staining with an Alexa 647-conjugated secondary antibody (Abcam, Cambridge, UK) (Supplementary information, Table 1). Cells were then additionally stained with TRITC-conjugated Phalloidin and DAPI nuclear stain. Images were acquired and analyzed as mentioned above.

Cell number and viability of hematopoietic cells
Following hematopoietic differentiation, oating suspension cells were collected and analyzed using a Countess 3 cell counter (ThermoFisher, Waltham, USA). Cell viability and cell number was measured after Trypan blue was added to testing samples.

Mineral resorption assay
In order to determine the mineral resorptive capacity of differentiated OCs, HPCs were plated onto a calcium-phosphate 24-well resorption assay plate and treated as mentioned above. After terminating OC differentiation, OCs were removed using 5% bleach and resorptive area was analyzed by taking tiled full well images as described above. OCs derived from human CD34 + PBMCs (healthy donor, received from Fred Hutch, Seattle) were used as positive control. In addition to image acquisition using the phasecontrast mode, images for resorption area quanti cation were taken using the yellow channel in uorescence mode in order to facilitate differentiation between resorbed and un-resorbed areas in ImageJ.

Analysis of gene expression
RNA extraction and isolation throughout the differentiation process for both iPSC lines using either the EB or the MB method was performed using a RNeasy kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. In short, a lysis buffer was added to the cells and lysed mechanically using a 20G needle and syringe. Following RNA extraction RNA was isolated and puri ed using spin-columns. RNA concentration was measured using a NanoDrop (ThermoFisher, Waltham, USA). A total of 250 ng of RNA was used to generate complementary DNA (cDNA) using Omniscript (Qiagen, Hilden, Germany) at 37°C for 1 hour. The cDNA was then used to determine the expression of POU5F1, CSF1R, TNFRSF11A, NFATC1, CA2, MMP9 using a TaqMan quantitative reverse transcription PCR (qRT-PCR) (Supplementary information, Table 2). Gene expression levels were normalized to 18s ribosomal RNA levels and calculated using the ∆∆Ct method to determine fold gene expression throughout the differentiation process.

Statistical analysis
Statistical analysis was carried out using GraphPad Prism 9. Data are shown as means ± SD. Statistical signi cance was assessed using Tukey's multiple comparison post-hoc test unless indicated otherwise in individual experiments. An adjusted p-value of p < 0.05 was considered to be statistically signi cant.

Results
iPSCs remain undifferentiated during propagation and retain ALP expression in centrally located cells throughout differentiation ALP staining was performed for both PBMC and broblast-derived iPSC lines and used as a stem cell marker in order to monitor spontaneous differentiation and con rm stemness of the described iPSC lines prior to mesodermal induction (Fig. 1). Both cell lines showed a consistent ALP expression pattern throughout cell propagation while the cell colony peripheries stained more intensively as has been previously described elsewhere (44) (Fig. 1A, B, G, H). Both cell lines retained a certain degree of ALP expression following mesodermal differentiation under either using the EB differentiation (Fig. 1C, D, I, J) or MB differentiation protocol (Fig. 1E, F, K, L). There is a higher degree of ALP expression can be seen in broblast-derived iPSCs differentiated under the MB protocol (Fig. 1K) in comparison to the PBMCderived iPSCs (Fig. 1E). Additionally, a higher degree of variability in size and distribution was observed in MB cell forming complexes (data not shown). Following hematopoietic differentiation, cell-forming complexes still retained a level of ALP expression in the most centrally located cells within the cell forming complex (arrows in Fig. 1D, F, J, L). Cell-forming complexes of the PBMC-derived iPSCs differentiated under the EB protocol (Fig. 1D) appeared to maintain their structure to a higher degree compared to the broblast-derived, EB cell-forming complexes (Fig. 1J). As expected, an abundance of ALP negative cells was observed following hematopoietic differentiation in both differentiation protocols (empty arrows in Fig. 1D, F, J, L).
EB cell forming complexes display a higher degree of organization compared to MB cell forming complexes Immunocytochemistry in conjunction with confocal microscopy of the cell-forming complexes was performed to assess the differentiation process following mesodermal induction (Fig. 2). As depicted, all cell-forming complexes, originating either from PBMC-derived iPSCs or broblast-derived iPSCs and differentiated either according to EB or MB differentiation protocols expressed markers for all three germ layers. In comparison to the broblast-derived cell line (Fig. 2G-L), the PBMC-derived cell line showed a higher degree of organization based on the expression of the ectodermal, mesodermal, and endodermal transcription factors under both EB ( Fig. 2A-C) and MB differentiation protocols ( Fig. 2D-F). Both cell lines and both differentiation protocols showed a higher expression of the ectodermal marker Otx-2 compared to SOX1 ( Fig. 2A, D, G, J). SOX17, an endodermal marker, was consistently expressed centrally within cellforming complexes differentiated according to the MB protocol in both iPSC lines ( Fig. 2F and L). EB cellforming complexes from PBMC-derived iPSCs expressed the mesodermal marker Brachyury (Fig. 2B) while EB cell-forming complexes originating from the broblast-derived cell line expressed low to no Brachyury (Fig. 2H).

EB differentiation gives rise to later stage hematopoietic cells compared to MB differentiation
Following hematopoietic differentiation, the suspension cell population that arose from the cell-forming complexes was harvested and analyzed for quantity and cell viability (Supplemental Fig. 1). More cells were harvested from cell-forming complexes differentiated according to the EB protocol (Supplemental Fig. 1A) compared to the MB protocol independent of the iPSC lines used in this study.
In order to further characterize the harvested oating cell population, ow cytometry was performed using hematopoietic and monocytic markers (Fig. 3). Additionally, cells were stained with an anti-RANK antibody to investigate whether an early difference in RANK expression can account for differences in OC activity between cell lines and differentiation protocols.
Undifferentiated iPSCs, used as negative control ( Fig. 3A and D), displayed a CD34 + sub-population which was larger in the PBMC-derived iPSC line compared to the broblast-derived iPSC line. Additionally, both undifferentiated iPSCs populations yielded a small fraction of a CD265 + sub-population.
EB differentiated cells generated larger CD43 + and CD45 + populations ( Fig. 3B and D) compared to MB differentiated cells while MB differentiation resulted in larger CD34 + populations ( Fig. 3C and F). Additionally, oating hematopoietic cells differentiated according to the EB protocol had a larger population of monocytes (CD14 + and CD11b + cells) for both iPSC lines (Fig. 3B and E). No differences were observed in RANK expression between both differentiation protocols or cell lines.

EB differentiation gives rise to bona de OCs
Following OC differentiation, cells were stained for Cathepsin K (turquoise) and F-actin (red) and counterstained with nuclear DAPI stain (blue) (Fig. 4). Both broblast and PBMC-derived iPSC lines showed multiple large spread-out multinucleated cells when differentiated according to the EB protocol (Fig. 4A, B, E, F). Several large polykaryons with up to 100 nuclei were observed in both groups (solid arrows in Fig. 4A and E), while those derived from the broblast iPSC line appeared to exhibit more Cathepsin K (arrow heads in Fig. 4F) compared to the PBMC-derived cell line (arrow heads in Fig. 4B). Both cell lines had the strongest signals for Cathepsin K in proximity to nuclei clusters. Additionally, mononuclear cells with varying degrees of Cathepsin K expression were interspersed with the large multinucleated cells (empty arrows in Fig. 4A and E).
The relative homogeneity of the two cell lines differentiated according to the EB protocol (Fig. 4A, B, E, F) was contrasted by the distinct appearances of PBMC-and broblast-derived cell lines differentiated according to the MB protocol (Fig. 4C, D, G, H). Similar observations were made between multiple experiments. First, PBMC-derived iPSCs displayed only few multinucleated cells with no more than 5 nuclei per cell (solid arrow in Fig. 4D). Cathepsin K expression was observed in some of the mono and multinucleated cells (arrow heads in Fig. 4D). While some cells showed a spread-out morphology, most cells were mononuclear and had a stellar or spindle-like appearance (empty arrows in Fig. 4D). In contrast to the PBMC-derived cell line, the broblast-derived iPSC line (Fig. 4H) exhibited mononuclear cells with a distinct stellar-like phenotype following OC differentiation. While most of these stellar-like cells appeared to express little to no Cathepsin K (arrow heads in Fig. 4H), a well-formed F-actin cytoskeletal structure was visible following visualization of F-actin with TRITC-conjugated Phalloidin (chevron arrows in Fig. 4H). Small mononuclear cells with varying degrees of Cathepsin K expression were also seen in this group (empty arrows in Fig. 4H). Image quantitation con rmed the signi cant differences between both protocols (Fig. 4I) while demonstrating no signi cant differences in osteoclasts differentiated according to the EB protocol ( Fig. 4J and K) using either the PBMC or broblast-derived cell line.
As presented in Fig. 5, both iPSC lines differentiated according to the EB protocol generated large, multinucleated OCs following OC differentiation (Fig. 5A, B, E, F) as determined by TRAP and methyl green nuclear counterstaining (solid arrows in Fig. 5B and F). TRAP is commonly used to stain OCs, albeit also expressed in leukocytes as further discussed below. Resorption pits are visible in both groups (borders outlined using white dashed lines in Fig. 5A and E). OCs differentiated from the broblastderived cell line under the EB protocol were larger, more numerous and stained more intensively for TRAP compared to the PBMC-derived cell line. Both cell lines differentiated according to the MB protocol (Fig. 5C, D, G, H) show lightly stained TRAP positive mononuclear cells. A lower cell density of mononuclear cells is observed for the PBMC-derived cell line (Fig. 5D) in comparison to the broblastderived cell line (Fig. 5G and H). Analogous to the ndings in Fig. 4G and H, Fig. 5H shows again TRAP − cells with a stellar-like phenotype (empty arrows in Fig. 5H).
Overall, assessment of cell morphology as well as expression of Cathepsin K and TRAP demonstrate the differentiation of human OCs when using the EB protocol.
OC resorptive activity was assessed by quantifying the resorption area on calcium-phosphate coated wells (Fig. 6). Wells with OCs differentiated according to the EB protocol showed clearly visible resorption pits for both PBMC-derived (Fig. 6A) and broblast-derived cell lines (Fig. 6E). However, the quanti ed area of resorption pits created by OCs from the broblast-derived iPSC line was signi cantly higher than from the PBMC-derived cell line (36.9% vs 57.2%, p < 0.05). In contrast, wells with OCs differentiated according to the MB protocol ( Fig. 6C and G) showed almost no visible resorption pits.
To further assess gene expression throughout the differentiation process, qRT-PCR was performed (Fig. 7) using a stemness marker (Fig. 7A), an osteoclast precursor marker (Fig. 7B) and osteoclast markers (Fig. 7C, D, E, F). The expression of POU5F1 (Oct4) was signi cantly reduced throughout the differentiation process in all groups (Fig. 7A). However, it can be noted that iPSCs differentiated according to the EB protocol showed a larger decrease compared to the MB differentiation protocol.
CSF1R (M-CSFR) showed a signi cant increase during transition from the mesodermal to the hematopoietic stage during EB differentiation (Fig. 7B). PBMC-derived iPSCs differentiated according to the MB protocol showed no increase in CSF1R. Ct values for CSF1R within the broblast-derived iPSC line were below the detection threshold. OC markers RANK (Fig. 7C) and NFATC1 (Fig. 7D) showed a signi cant increase in expression in the OC stage compared to the previous hematopoietic stage when differentiated according to the EB protocol. This increase was seen in both iPSC lines. Additionally, markers associated with resorptive activity, CA2 (Fig. 7E) and MMP9 (Fig. 7F) were also signi cantly elevated in the OC stage when differentiated using the EB protocol.

Discussion
Several hematopoietic differentiation protocols have been published. However, it remains unclear which protocol or differentiation method is favorable for osteoclastogenesis. Here, we compare an EB to a MB differentiation protocol using iPSC lines from different tissue origins. We found that EB differentiation yielded bona de osteoclasts independent of tissue origin of the iPSC line used while MB differentiation led to hematopoietic cells of earlier stages which did not subsequently differentiate into OCs.
Differentiation of iPSCs into OCs requires a complex series of events. While the entire differentiation process can be divided into three major steps of mesodermal, hematopoietic and OC differentiation, each step can be further subdivided into steps that are recapitulated in ontogenesis. Hematopoietic differentiation has been shown to require stages of mesodermal induction, hematoendothelial speci cation, vascular arterialization, endothelial-to-hematopoietic transition, and hematopoietic maturation (40,45). Both EB and MB approaches have previously been reported to recapitulate these steps to varying degrees (37,(46)(47)(48)(49)(50)(51)(52).
A prolonged expression of pluripotency markers has been described elsewhere during EB formations (53,54), similar to the retained ALP expression in the center of cell-forming complexes of both EB and MB protocols even after hematopoietic differentiation.
EB differentiation has been reported to simulate early stages of post-implantation development by exhibiting ectodermal, mesodermal, and endodermal germ layers (40). Accordingly, embryoid bodies showed expression patterns for all three germ layers while differences in the degree of their expression and spatial organization existed between the two iPSC lines from different tissue origins. In contrast to the EB differentiation approach, MB differentiation showed similar, yet not as spatially organized expression patterns of markers for the three germ layers.
Cell forming complexes, either EB or MB, have the potential to in uence and alter hematopoietic cell production. This occurs through cell-cell interactions, extracellular matrix (ECM) interactions, and cytokine signaling from cell niches (53,55). The right microenvironment therefore appears to be critical for successful hematopoietic differentiation (53,(56)(57)(58). For example, Sturgeon et al. showed that activation of Wnt-β-catenin during mesodermal speci cation can determine whether cells will undergo a de nitive hematopoietic program (59-61) or a primitive program.
It has been shown that endogenous Wnt-β-catenin signaling is more active in EB (or 3D) cell-forming complexes over MB (or 2D) cell forming complexes (62). It could be hypothesized, that due to the larger relative surface area of a monolayer compared to a sphere, the MB cell-forming complexes cannot maintain their own microenvironment as well as EB complexes and are more susceptible or dependent on the exogenous cytokine concentrations and composition. Differentiation outcomes of embryoid bodies have also been reported to be highly variable based on their size and quality (40), which could also be explained by the same hypothesis.
Despite variability in the size of embryoid bodies, EB differentiation yielded more hematopoietic cells with a higher cell viability compared to MB differentiation. Even though MB differentiation in conjunction with pre-made differentiation media is less work intensive and a signi cant improvement in usability, this method demonstrated more variability from well to well with respect to numbers of iPSC colonies per well, colony size and their distribution within the well.
Nevertheless, both differentiation approaches gave rise to hematopoietic cells. Even though some variation between iPSC lines could be observed, MB cells showed a larger CD34 + population for both iPSC lines, whereas the EB differentiated cell populations were comprised of large portions of CD43 + and CD45 + cells. Both CD43 and CD45 appear ontogenetically after CD34, and CD45 is associated with progressive myeloid commitment (63-65). Additionally, EB differentiated cells comprised a larger portion of CD11b + and CD14 + monocytes, suggesting an overall further downstream position compared to MB hematopoietic cells. These ndings were consistent with ndings by Rössler et al., who showed similar population sizes of CD14 + and CD11b + cells using EB differentiation, albeit being slightly higher than our populations (31). Ruiz et al. on the other hand did not assess CD14 and CD11b marker expression, showed however a larger CD43 + population compared to the CD45 + population after 10 days of differentiation when differentiated according to the MB protocol (37). This was comparable to our ndings, showing a smaller CD45 + population for the MB protocol compared to the EB protocol. Nevertheless, a prolongation of treatment with the hematopoietic differentiation medium can increase the CD45 + portion (37) when differentiating hematopoietic cells according to the MB protocol using the STEMdiff hematopoietic kit.
While few studies have focused on in depth phenotyping of OC precursors (66) the traditional view depicts a multipotent progenitor (MPP) that is able to give rise to a common myeloid progenitor (CMP) and a common lymphoid progenitor (CLP). The CMP in turn gives rise to the megakaryocyte/erythrocyte progenitor (MEP) and granulocyte/macrophage progenitor (GMP). GMPs differentiate into granulocytes, monocytes, macrophages, dendritic cells, and OCs (66-71). GMPs are capable of giving rise to CD11b + and CD14 + monocytes which are well known to be able to differentiate into OCs (72)(73)(74)(75)(76). However, CD14 − cells (75,77) and a CD11b − CD14 − CD115 + (CD115 ≙ CSFR1/c-FMS) population have also been shown to possess high osteoclastic potential (78). Whether or not CD14 − cells obligatorily pass through a CD14 + stage in order to differentiate into OCs remains to be seen. Recent work has shown that not only GMPs, but also a multilymphoid progenitor (MLP) is capable of giving rise to macrophages and dendritic cells in addition to lymphoid cells, while being devoid of a megakaryocyte/erythroid potential (68, 79, 80). Downstream of the GMP or MLP a CD11b − CD34 + c-KIT + FLT3 + IL3Rα high population was identi ed as a common macrophage, OC and dendritic cell progenitor that was mRNA positive for CSFR1, did however not show any presence of CSFR1 at the cell surface (66). Trajectory analysis of an OC population using single cell RNA sequencing (scRNA-seq) shows a differentiation path beginning at a CD14 + stage, followed by a dendritic cell and macrophage stage before reaching a more active OC type, expressing increased levels of Cathepsin K and V-ATPase subunit D 2 (81). Early stage (c-Kit + , c-Fms + , Mac-1 dull and RANK − ) and late-stage OC precursors (c-Kit − , c-Fms + , Mac-1 + and RANK + ) have been described (82, 83) that are driven towards mature OCs by a wide range of factors (84-90). Ultimately, a mononuclear OC fuses with adjacent mononuclear OCs. This is regulated by proteins such as dendritic cell-speci c transmembrane protein (DC-STAMP), osteoclast stimulatory transmembrane protein (OC-STAMP) and syncytin-B (91)(92)(93), to form a large multinucleated cell. These mature OCs are capable of mineral and bone resorption (94) and show an increase in expression of proteins related to the resorption process such as V-ATPase subunit D 2, carbonic anhydrase 2, Cathepsin K, and MMP9 (69, 81, [95][96][97].
We showed that iPSCs differentiated according to the EB protocol gave rise to large multinucleated cells for both PBMC-and broblast-derived iPSC lines. qRT-PCR results showed a signi cant increase in expression of TNFRSF11A, NFATC1, CA2 and MMP9, supporting their identity as OCs. Finally, TRAP expression, Cathepsin K expression, and mineral resorptive capacity clearly identify the differentiated multinucleated cells as bona de OCs.
iPSCs differentiated using the MB method only gave off a very limited number of multinucleated cells with up to 5 nuclei for the PBMC-derived iPSC line, and no multinucleated, yet stellar-shaped appearing cells for the broblast-derived iPSC cell line. Both iPSC lines differentiated according to the MB protocol did not show mineral resorptive capacities. We hypothesize, that since the MB differentiation process gave rise to ontogenetically earlier HPCs compared to the EB method, MB cells might not have reached the OC stage in the same differentiation time span as compared to cells differentiated using the EB protocol and might have remained in earlier stages or side trajectories such as macrophages or dendritic cells.
Both EB or MB differentiated cells displayed TRAP + mononuclear cells that might be presumed to be mononuclear OCs or OC precursors. However, TRAP expression is not speci c to OCs but also present in immune cells such as macrophages and dendritic cells (DCs) (98-100). Isoform 5b has been found to be predominantly secreted by OCs (101) which could be used to further de ne mononuclear cells. However, single cell methodology such as scRNA-seq might offer the best possibility to get an overall picture of sub-populations and differentiation trajectories within the different groups (102).
Differences based on tissue origin of iPSC lines were observed regarding cell growth as broblast-derived iPSCs showed higher proliferation rates during propagation (data not shown). Fibroblast-derived iPSCs also gave off more hematopoietic cells compared with PBMC-derived iPSCs. Nevertheless, both cell lines could be successfully differentiated into OCs independent of their tissue origin. This is in line with research showing iPSC differentiation potential to be largely independent of cell type origin (103,104), while minor differences have been attributed to epigenetic memory persisting even after reprogramming to iPSCs (105)(106)(107). On average, PBMC-derived iPSCs gave off 2289 OCs per embryoid body, while broblast-derived iPSCs gave off 5505 OCs per embryoid body.
Limitations of this study consist mainly in the use of a commercially available differentiation kit for MB differentiation, as cytokine composition and concentration are not fully disclosed and thus, ndings in this study may be attributed in part to differing cytokines and not just on the 3D shape of EB cell-forming complexes over the 2D one in MB cell-forming complexes. Nevertheless, commercially available kits have its place and also require comparison and validation of established methods.
Regarding reproducibility, one additional limitation consists in the usage of FBS for terminal OC differentiation which can result in batch-to-batch variability (108) as our lab has noted difference in osteoclastogenesis when using exosome-free FBS over standard/untreated FBS (unpublished data). In view of clinical translation, a fully de ned differentiation process without the use of FBS for terminal OC differentiation will be required (109).
Finally, the scope of this study did not reveal the exact subtype of OCs that was differentiated. Many different subtypes of OCs such as chrondroclasts, odontoclasts, septoclasts, vascular-associated OCs have been described by other authors (110)(111)(112)(113). Single cell transcriptomics could offer also in this regard a deeper insight into differentiation trajectories and cell subtypes.

Conclusion
The outcome of this study demonstrates the successful differentiation of OCs from iPSCs in conjunction with an EB differentiation method. In contrast, a MB differentiation method that used a commercially available hematopoietic differentiation kit did not yield OCs. Presence of bona de OCs was validated using osteoclast marker expression and determining the mineral resorptive activity. The differentiation process was continually evaluated following mesodermal induction, hematopoietic differentiation and after terminal OC differentiation for both EB and MB approaches using iPSCs from two distinct tissue origins. No differences were observed regarding OC differentiation between a PBMC-and a broblastderived iPSC line used in this study.

Availability of data and materials
The datasets supporting the conclusions of this article are available in the Dryad repository, [unique persistent identi er and hyperlink to dataset(s) in http:// format will be provided upon acceptance as Dryad requires a manuscript number which is generated upon submission].

Competing interests
The authors declare no competing interests.

Funding
National Institutes of Health grant R35 HL139602-01. We also acknowledge NIH S10 grant S10 An abundance of ALP negative cells can be observed following hematopoietic differentiation (empty arrows). Scale bar = 500 µm.     showed either no signi cant increase or an initial Ct value below the detection threshold. Osteoclast