There are two major ways of forming multinucleate giant cells i.e. acytokinetic cell division and cell fusion [1]. Proliferating neoplastic cells, e.g. Reed-Sternberg cells in Hodgkin’s lymphoma or multinucleated tumor cells in soft tissue, e.g. myxofibro- and osteosarcomas are resulted from incomplete cell division as a result of their cytoskeleton vulnerability e.g. of the contractile ring [2]. On the contrary, inflammatory multinucleated giant cells (GC), such as osteoclast-type giant cells, Langhans-type granuloma giant cells and foreign body giant cells are formed by fusion of cells of the monocyte-macrophage lineage [3].

By testing the expression of cell cycle phase progression associated markers in the mononuclear cells of GCTB, we recently showed that cases with elevated post-G1-phase cell fraction, indicating accelerated cell cycle progression, may predict shorter progression free survival (PFS) [4]. We also recognized that GC nuclei may show diverse proportion of immunoreactions not only for the earlier detected cyclin D and p21^WAF1 [5, 6], and cell cycle control proteins, but also for some cell cycle licensing and promoter markers, which had not been noticed before, despite GCs are considered to be of reactive, osteoclastic phenotype [7]. Therefore, here we studied the expression of a comprehensive set of cell cycle regulatory proteins to see if GCs in GCTB are still show replicative activity and if it has a clinico-pathological relevance.

GCTB is an osteolytic, locally destructive bone lesion, which, besides GCs, is made up mainly of mononuclear monocytic cells which act as precursors for GCs, and of neoplastic stromal cells (Figure 1). The proliferating, neoplastic stromal cells, generally carrying H3F3A G34W mutation [8], are the major drivers of osteoclastogenesis and pathological bone resorption [9, 10]. They produce canonical (RANKL/M-CSF) and non-canonical (e.g., LIGHT, TNFα, IL-6 or vascular endothelial growth factor–VEGF) growth factors and hypoxia inducible factors 1α and 2α [14], which can either directly or through autocrine activation promote osteoclastogenesis and osteolysis [11, 12]. We have shown earlier that besides the replication activity of neoplastic stromal cells, their elevated epidermal growth factor receptor (EGFR) signaling and deregulated gap junction connexin43 expression and channel functions, can contribute to GCTB progression, mediated by GCs [4, 13, 14].

In the present study, we tested the expression of cell cycle regulatory proteins in GC nuclei, including 3 clones for the general proliferation marker Ki67; the DNA replication licensing factors MCM2 and MCM6; the G1-S phase marker cyclin D1 and its complexing partner CDK4/6; the early (CDK2 and cyclin A) and late (topoisomerase 2–TOP2) post-G1 phase markers; and the G2-M-phase markers (aurora kinase A–AURKA, and phospho-histone-H3–pHH3) [15]. Furthermore, the DNA replication inhibitor geminin, the cyclin dependent kinase inhibitors p15^INK4b, p16^INK4a, p21^WAF1 and p27^KIP1; as well as the oncosuppressor retinoblastoma and p53 [16], and the unpaired cyclin G1 [17, 18] proteins were also examined. We aimed to profile cell cycle activity in GCs and see if it is different between primary and recurrent GCTB.

In GCTB, GCs need continuous supply of precursors to fuse, progressively form and resorb bone. They are under the influence of growth factors including receptor activator for NFκB ligand (RANKL) and macrophage colony stimulating factor (M-CSF) and their substitutes e.g. interleukins 6, 11 and 8, and TNFα; as well as VEGF, placental growth factor (PlGF), hepatocyte growth factor (HGF) and FLt-3 ligand, respectively, some of which can induce proliferation [11, 20, 21]. Earlier studies by others showed the widespread expression of cyclin D1 [5, 6, 22–24] cyclin D3 [5] and p21^WAF1 [6], less p16 and scarce Ki67 [22] in GCs. Still, these data were insufficient to declare replication activity in GCs. In our present work, in addition to confirming previous observations, we revealed further cell cycle promoters, which can verify [23] an early cell cycle course in GCs. However, the effect of the generally detected cell cycle licensing MCM6, and the cell cycle promoters CDK4 and cyclin E, are likely neutralized in GCs by the upregulation CDK inhibitors p15^INK4b, p16^INK4a, p27^KIP1 and the p53 induced p21^WAF1, consistent with a cell cycle arrest at the late G1 phase. The potential emergence and links among the markers of this study during the cell cycle are drafted in Figure 6.

The general proliferation marker Ki67, which has been involved in heterochromatin organization, the inhibition of p21-mediated G1/S-phase checkpoint activation and in the formation of mitotic perichromosomal protein sheet, can be detected throughout the replication cycle [25]. So far, only one study reported a weak Ki67 Mib1 staining in <5% of GCs in only 3 out of 29 (10%) GCTB samples. Of the 3 Ki67 clones we investigated in GCs, both Mib1 and B56 showed a low ratio of nuclear positivity (<10%) while SP6 demonstrated a wide range with >75% median. The specificity of the moderate intensity SP6 reaction in GCs was validated by the clear negativity of some adjacent mononuclear cells. Though Ki67 expression peaks at G2/M-phases, the protein is only gradually eliminated from newly divided cells, therefore, its amount may also reflect the time the cell spent in quiescence [26]. Accordingly, the markedly higher ratio of positive GC nuclei in 1-Rec vs. primary cases, and its association with shorter PFS in our explorative survival analysis may also indicate accelerated dynamics of GC formation in the advanced cases. The cytoplasmic Mib1 staining in GCs shown also by others [27] is likely to be related to its metabolic elimination by the ubiquitin proteasome system [28]. The differential occurrence of the Ki67 clones can be partly related to the slower degradation of the epitope region recognized by SP6 compared to the others. However, further signs in this study suggested that this may also be linked with an initial replication activity.

Members of the MCM2-7 helicase complex, which can also be detected all over the cycle, unwind double-stranded DNA and allow the controlled licensing of DNA for duplication [16]. However, while MCM6 complexed with MCM4 and MCM7 is involved in relaxing DNA to single strands, the MCM2 subunit (MCM3, and MCM5 too), has potential inhibitory role on this function [29]. Our widespread detection of MCM6 with moderate intensity but only occasional occurrence of MCM2 in GC nuclei, may reflect the initiation of DNA unwinding in GCs.

Cyclin-dependent serine/threonine kinases complex with their regulatory subunit cyclins to phosphorylate retinoblastoma and promote the transition of different phases during cell cycle progression [30]. Here, we revealed the widespread emergence not only of the previously detected cyclin D1 [5, 6], but also its partner CDK4 in the earliest G1-S phase promoter complex. This complex not only can inactivate retinoblastoma but also support the activation of the next G1-S-phase promoter cyclin E-CDK2 complex through reducing mitochondrial metabolism to prevent cyclin E degradation [31]. In agreement with this, we regularly detected cyclin E in GCs, however, only very rarely found its complexing partner CDK2 in GC nuclei, indicating a late G1-phase arrested cell cycle. At the same time, cyclin D1-CDK4/6 complex can also support cell growth e.g. through mTORC1 activation [32], and high cyclin D1 levels can assist in metabolic substrate utilization toward mitochondrial amino acid production [31], which propose a role for cyclin D1 also in GC growth and differentiation. Furthermore, cyclin D1 overexpression in GCs may also be required for multinucleation [5], as it enhanced the number of nuclei e.g. in cardiomyocytes [33] and was detected primarily in giant trophoblasts rather than in diploid ones [34].

In agreement with earlier studies, in GCs we also revealed the widespread expression of p21^WAF1, the universal CDK inhibitor [35]. However, despite p21^WAF1 is considered to block cyclin D1-CDK4/6 complex and CDK2 activities, paradoxically it may also be important for CDK4/6 complex assembly [36] and for the nuclear export of cyclin D1 [37, 38]. These might explain our frequent detection of CDK4 but not the CDK2, and the high levels of the CDK4 partner cyclin D1 in GC nuclei. P21^WAF1 is an important effector of the cell cycle control functions of p53 [39]. Before us, p53 expression had been noticed only incidentally in one publication [40]. Here, we detected p53 in the majority of GC nuclei using 2 different antibody clones (DO7 and BP53–12 respectively) to give, for the first time, clear evidence that GCs upregulate this important “guardian” of the genome. Cytoskeletal stress, which may occur during multinucleation in GCs, can induce p53 upregulation and activate p21^WAF1 for blocking S-phase entry at G1-checkpoint [39]. The p53-dependent G1 arrest of multinucleated tetraploid cells has been previously described [41] and p53 activity with the contribution of p21^WAF1 can also drive this process even further to result in cellular senescence [42]. We also noticed the general upregulation of p27^KIP1 in GCs, which in normal cells is primarily linked to controlling the cell cycle through inhibiting both cyclin E-CDK2 and cyclin D1-CDK4/6 activities [43]. Both p21^WAF1 and p27^KIP1 play essential roles also in GC functions as shown by the osteopetrotic phenotype in mice with deleted genes encoding these proteins [44]. In GCTB, arresting of the potential cell cycle activities and inducing senescence are required for GC maturation involving the production of key proteases cathepsin K and MMP-9 for pathological bone resorption [45].

As a p53 target, cyclin G1 has been involved in both supporting cell cycle arrest and in driving the S-G2-M-transition [17]. As a supposed oncogene cyclin G1 may activate the MDM2 oncoprotein by recruiting Ser/Thr protein phosphatase 2A (PP2A), which dephosphorylates MDM2 to inhibit and degrade p53 [17, 18, 46]. We detected cyclin G1 in most GC nuclei with a non-significant trend toward higher values in primary vs. 1-Rec GCTB cases. This may rather support its role in cell cycle arrest than as a promoter of replication in GCs which are reactive, non-malignant cells. However, it is also possible, that cyclin G1 upregulation is a rebound effect to control p53 overexpression and prevent apoptosis induction [17, 18, 46]. The permanently elevated cyclin D1 levels in functionally active GCs, but its disappearance from oversized (>40 nuclei), aged GCs may also be consistent with the protective function of cyclin D1 [17, 46].

Besides p21^WAF1 and p27^KIP1, the upregulation of other CDK inhibitors, particularly targeting CDK4 and CDK6 proteins including p15^INK4b and p16^INK4a, further support the cell cycle related activity in GCs [47]. Both p15^INK4b and p16^INK4a are also likely to contribute to G1-phase cell cycle arrest [48, 49], while p16^INK4a may also be linked to cellular senescence induction supporting either the full functional differentiation or the aged-cell decay in GCs [50]. Cell cycle arrest by CDK inhibitors was also reflected by retinoblastoma upregulation in GCs. In line with these findings, practically none of the S-G2-M phase markers including cyclin A, the later complexing partner of CDK2, the cell cycle repressor geminin, the G2-M phase transition associated cyclin B, or the M-phase related AURKA and pHH3 were detected within GCs [16]. This pattern is consistent with the full functional maturation of GCs. The S-G2 phase marker positive mononuclear cells which had no discernible cytoplasmic boarders with GCs, were likely to be monocytes close to or within the fusion process (Figures 4A,B). In support of this, a subpopulation of CD14+/CD33 + monocytes was shown to proliferate in response to M-CSF and form pre-osteoclasts when primed with RANKL [51]. Then pre-osteoclasts loose CD33 expression and fuse with more monocytic cells [52] and the proliferating monocytes form significantly more GCs than the rest of the monocytic pool [51].

Despite silenced, the early signs of replication proved here, may reflect the generation dynamics and age-related activity of GCs. In vitro data confirm an inverse correlation between GC size and resorption activity at mild acidic conditions [53]. This is in line with data showing elevated recurrence potential in GCTB cases, which dominantly contain GCs of <15 nuclei compared to those carrying larger ones [18]. In the present study, the average size and nuclear density of GCs also revealed an inverse trend with the clinicoradiological grade of GCTB. Furthermore, the cyclin D1 and MCM6 reactions in GCs were obviously stronger and more frequent in small sized (∼<15 nuclei) than in larger GCs, particularly when nuclear numbers were ∼>40. These are also in agreement with our earlier finding of the significantly lower average size of GCs in recurrent GCTB cases where the growth related EGFR protein level was elevated in the mononuclear stromal cells, the major drivers of GC formation and activity [13]. All these support the view that smaller sized GCs are the younger, dynamically forming populations, which may show more signs of early replication than the aged, functionally less active oversized GCs (∼>40 nuclei).

Dissecting the replication cycle into major cells fractions through the immunohistochemical detection of nuclear proteins which regulate or control different phases of replication, may allow accurate assessment of cell proliferation dynamics in situ. Using a marker set, selected carefully from the list we described here, can serve cell cycle analysis for diagnostic, prognostic and predictive purposes in any pathological process particularly in cancer.

Though multinucleated GCs in GCTB are thought to be of reactive phenotype formed by fusion of cells of the monocyte macrophage lineage, they had been occasionally shown to express cell proliferation related markers. By using a comprehensive marker set, here we revealed, for the first time in GCs, the general upregulation of cell cycle promoting markers MCM6, CDK4 and cyclin E, indicating primary DNA unwinding and G1-S-phase promoting activities. We also confirmed the earlier published widespread expression of cyclin D1, which all, unequivocally demonstrated an early replication activity in GCs. This, however, was silenced by the widespread expression of CDK inhibitors p15^INK4b, p16^INK4a, p27^KIP1 and p53 induced p21^WAF1 resulting in cell cycle arrest at the G1 checkpoint, confirmed by the missing production of post-G1-phase faction markers. The complex interplay among these elements under the influence of external growth and differentiation factors is required for the functional maturation and bone resorbing activity of GCs in GCTB.