Molecular Basis of Fracture Pseudarthrosis Associated with Neurofibromatosis Type 1

Abstract

Neurofibromatosis Type 1 (NF1) is a tumor-predisposition syndrome caused by heterozygous mutations in the NF1 gene. In addition to oncologic manifestations, individuals with NF1 face a heightened risk of developing secondary skeletal complications associated with somatic loss-of-heterozygosity. Long bone dysplasia and post-fracture pseudarthrosis are among the earliest skeletal manifestations observed in children with NF1. Most commonly occurring in the tibia, pseudarthroses are often treated with a combination of resection, fixation, bone grafting, and rhBMP2. Although rates of initial union are reported to be high, the risk of re-fracture and subsequent pseudarthrosis is similarly elevated, and some patients may ultimately require amputation. Basic science research related to NF1 pseudarthrosis is uncovering the cellular pathogenesis of the disease and unlocking new potential treatment paradigms. Here, we review the current understanding of the biological etiology of NF1 pseudarthrosis, informed by molecular studies of patient-derived tissue samples and mechanistic studies utilizing pre-clinical mouse models. Results from these studies are presented within the context of contemporary treatment paradigms. Finally, we introduce novel treatment paradigms currently under investigation in pre-clinical models and discuss how these may potentially translate to improve surgical outcomes in children with NF1 pseudarthrosis.

Key Concepts

• (1)Fracture pseudarthrosis in patients with NF1 is caused by somatic mutations in the NF1 gene.
• (2)Somatic NF1 gene mutations hyperactivate the MAPK signaling pathway, disrupting cellular mechanisms necessary for proper fracture healing.
• (3)Emerging therapies aim to target hyperactive MAPK signaling to promote fracture healing by fostering osteogenic instead of fibrogenic pathways in skeletal progenitor cells.

Introduction

Neurofibromatosis Type 1 (NF1) is an uncommon RASopathy characterized by the occurrence of pleiotropic secondary manifestations affecting diverse organ systems [1]. NF1 is caused by heterozygous loss-of-function mutations in the NF1 gene, which encodes for the neurofibromin protein [[2], [3], [4]]. Mutations can be inherited from affected family members or can occur spontaneously as de novo mutations within the NF1 gene. The neurofibromin protein negatively regulates the activation of the RAS signaling protein, thereby limiting activation of the downstream mitogen-activated protein kinase (MAPK) pathway [5,6]. Therefore, NF1 gene mutations lead to activation of the MAPK pathway, predisposing patients to a variety of secondary manifestations that occur with varying frequencies throughout the NF1 population. Secondary manifestations may include dermal, skeletal, ocular, or oncologic presentations. The clinical diagnostic criteria for NF1 were recently revised to identify affected individuals based on the presence of one (if known family history) or more (if no family history) of these secondary manifestations [7].

Skeletal manifestations in NF1 are collectively common and may include systemic and focal differences compared to age- and sex-matched individuals without NF1. For example, children and adults with NF1 typically have lower bone density and shorter stature compared to their non-NF1 peers [[8], [9], [10], [11]]. Additionally, spine deformities may include dural ectasia and scoliosis, including early-onset scoliosis [12]. Scoliosis in children with NF1 is often classified as either non-dystrophic or dystrophic. Non-dystrophic scoliosis resembles the more common non-NF1 adolescent idiopathic scoliosis, while dystrophic scoliosis presents with a variety of radiographic atypia that contribute to more severe and progressive deformity [13]. Dystrophic scoliosis may also be associated with paraspinal neurofibromas [14]. Finally, plexiform neurofibromas, benign Schwann cell-derived tumors occurring in up to 50% of individuals with NF1 [15], may be associated with skeletal abnormalities, though a causal relationship between the tumor and bone abnormalities is unclear [16].

In the long bones, children with NF1 may present with significant dysplasia, most often affecting the tibia, which is prone to fracture [17]. While tibial dysplasia is rare, its incidence is increased in children with NF1. Treatment of tibial dysplasia focuses on preventing fractures during skeletal growth, often through bracing or, potentially, by correcting mechanical alignment with growth modulation [18,19]. Fractures of the dysplastic tibiae often result in pseudarthrosis at the fracture site, which may be treated with resection, intramedullary or external fixation, grafting, and addition of recombinant human bone morphogenetic protein 2 (rhBMP2), though the clinical efficacy of rhBMP2 remains unclear [[20], [21], [22]]. Interestingly, proximal osteotomies performed during the treatment of the fracture do not result in pseudarthrosis [23], suggesting that biologic factors contributing to pseudarthrosis are localized to the fracture site.

The purpose of this paper is to review the current state of knowledge regarding the biologic understanding of NF1-associated fracture pseudarthrosis, informed in part by analyses of genetically engineered mouse models (GEMMs) and patient-derived research specimens. Results from these studies are beginning to implicate novel therapies that, when combined with current surgical approaches, may have potential to alter the natural history of this skeletal NF1 manifestation.

Molecular basis of NF1 pseudarthrosis

Neurofibromin limits RAS pathway activation

Neurofibromin, the protein encoded by the NF1 gene, functions within cells to inactivate the oncogenic RAS protein [5,6]. Therefore, like other “RASopathies”, mutations in the NF1 gene lead to increased RAS pathway activation and subsequent clinical sequelae. An increase in the active form of the RAS protein leads to the activation of the MEK-ERK signaling pathway (also known as the MAPK pathway) and the PIK3CA-AKT-MTOR pathway (Fig. 1), among others. Individuals with NF1 harbor a heterozygous mutation in the NF1 gene, leading to a reduction in functional neurofibromin protein and increased RAS activation compared to individuals without NF1. As mentioned above, this is associated with systemic sequelae or increased risk of secondary manifestations.

Figure 1.

Model of RAS pathway hyperactivation in NF1 pseudarthrosis. In the iliac crest, neurofibromin (NF1) inactivates RAS to regulate physiologic levels of MAPK and PIK3CA-AKT-MTOR signaling. In pseudarthrosis, loss of NF1 causes hyperactivation of RAS-regulated pathways, which results in impaired osteogenesis and delayed fracture healing.

Secondary manifestations of NF1 are associated with the spontaneous occurrence of second-hit somatic NF1 gene mutations, which further reduce the availability of functional neurofibromin protein. This results in hyperactivation of the RAS pathway. The molecular and cellular consequences of hyperactive RAS are cell type specific. Largely informed by GEMMs, secondary manifestations of NF1 have been attributed to somatic NF1 gene mutations in specific cell types, such as melanocytes (pigmentation and café-au-lait macules), Schwann cells (neurofibromas), and glial progenitor cells (optic pathway gliomas), for example. Surprisingly, second-hit somatic mutations have also been described in post-mortem non-pathologic tissues from an NF1 subject [24], suggesting that a somatic NF1 gene mutation is necessary but not sufficient to cause detectable pathologies in some tissues or cell types.

Somatic NF1 gene mutations in pseudarthrosis

Like other secondary manifestations in NF1, second-hit NF1 gene mutations were investigated in tissue samples from patients with pseudarthrosis. By testing selected genetic markers around the NF1 gene, results from multiple studies suggested the presence of tissue-specific somatic conversion of heterozygous to homozygous gene mutations (i.e., loss-of-heterozygosity) in NF1 pseudarthrosis [25,26] (Fig. 2). Subsequent studies using massively-parallel next-generation sequencing techniques confirmed the association of somatic NF1 mutations with pseudarthrosis [[27], [28], [29]]. Consistent with the somatic basis of NF1 pseudarthrosis, second-hit NF1 gene mutations were not detected in patient-matched iliac crest samples [28,29]. Likewise, MAPK pathway activation was higher in pseudarthrosis samples compared to patient-matched iliac crest samples [29,30]. These studies demonstrate a shared genetic etiology between pseudarthrosis and other non-skeletal manifestations of NF1.

Figure 2.

Somatic basis of NF1 pseudarthrosis. Patients with NF1 may inherit heterozygous NF1 gene mutation (NF1^+/−). Somatic second-hit mutations (i.e., loss-of-heterozygosity) leads to secondary manifestations, such as pseudarthrosis. Cells harboring somatic mutations (NF1^−/−) are localized to the pseudarthrosis site.

MAPK-associated dysregulation in NF1 pseudarthrosis

MAPK hyperactivation in NF1 leads to cell type-specific changes in gene expression, which in turn alter cellular functions. For pseudarthroses, multiple studies hypothesized that MAPK hyperactivation impairs the differentiation of skeletal progenitor cells into osteo-lineage cell types required for proper fracture healing. These studies were initially informed by analyses of GEMMs (discussed below). To investigate alterations associated with the loss of NF1 that contribute to pseudarthroses, recent studies have utilized single-cell sequencing to dissect the molecular dysregulation specifically in somatically mutated pseudarthrosis-derived cells. Gene expression studies identified increased expression of the EREG gene, which encodes epiregulin, as an expression biomarker of NF1-deficient pseudarthrosis-derived cell [28,31]. Comparing gene expression between EREG^HIGH and EREG^LOW cells demonstrated significant differences in the expression of genes involved in skeletal development and fracture healing [28]. Despite detecting changes in gene expression, no significant differences were observed in the early stages of osteogenic differentiation between EREG^HIGH and EREG^LOW cells in laboratory experiments. However, single-cell expression analysis implicated defects in matrix mineralization within the somatic EREG^HIGH cell population. Consistent with the shared genetic etiology of different secondary manifestations, gene expression signatures previously described in NF1-associated plexiform neurofibromas were also evident in NF1-deficient EREG^HIGH pseudarthrosis-derived primary cell [28].

More recently, single-cell sequencing of patient pseudarthrosis- and iliac crest-derived cells revealed expression signatures consistent with increased MAPK pathway activation, which is thought to be due to the presence of NF1-deficient cells from the pseudarthrosis [29]. Compared to iliac crest cells, pseudarthrosis-derived cells demonstrated increased expression of genes associated with fibrosis, suggesting these cells are primed toward a fibrotic rather than osteogenic fate. To test this, iliac crest- and pseudarthrosis-derived cells from patients were transplanted within the fracture site of immunocompromised mice. While iliac crest-derived cells contributed to a cartilage callus after fracture, consistent with a proper post-fracture healing response, transplantation of pseudarthrosis-derived cells resulted in significant fibrosis and impaired healing after fracture [29]. Taken together, studies utilizing patient-derived cells and tissues reproducibly demonstrate impairments in the regulation of osteo-lineage gene expression associated with NF1-deficient MAPK pathway hyperactivation.

Pre-clinical evidence implicating Nf1 in skeletal development and fracture healing

Pre-clinical studies, primarily utilizing genetically engineered mouse strains, facilitate mechanistic investigations addressing key molecular and cellular questions in the field. For example, delineation of the somatic cell of origin and the associated molecular dysregulation leading to dysplasia and pseudarthrosis remains incomplete. Complete loss of the Nf1 gene in mice (Nf1^−/−) is incompatible with survival, and these knock-out mice display multiple congenital defects leading to embryonic lethality [32]. Loss of a single gene copy (Nf1^+/−), however, increases the risk of spontaneous tumor formation in aged mice [33], although mice fail to display other manifestations evident in the human condition. To circumvent the lethality inherent in Nf1^−/− mice, “Nf1 floxed” (Nf1^flox) mice were engineered [34], wherein two loxP sequences were inserted into the gene, which would then facilitate deletion of the intervening sequence (and hence gene inactivation) within cells expressing the gene encoding the Cre recombinase enzyme. The “Cre/Lox” system enables the selective deletion of the Nf1 gene in specific cells by crossing Nf1^flox mice with mice engineered to express the Cre gene in targeted cell types. In the presence of Cre expression, the Nf1^flox gene is deleted, but the Nf1 gene is expressed at normal levels (and encodes for a normal neurofibromin protein) in cells that do not express the Cre gene. The Cre/Lox system has been further developed to enable temporal control of Cre expression in mice or to induce Cre expression using viral vectors. Likewise, numerous mouse strains have been developed to express Cre in specific skeletal cell populations [35], though it is essential to recognize that Cre may also be expressed in other extraskeletal tissues, which may limit their usefulness in specific experimental contexts [36]. Here, we highlight how pre-clinical mouse studies have shaped the field's understanding of skeletal disease in NF1.

Nf1 is essential for endochondral bone formation in mice

Early studies sought to broadly delete the Nf1 gene in skeletal tissues. Deletion of Nf1 in the embryonic limb mesenchyme in Prrx1-cre;Nf1^flox mice resulted in growth plate disturbances, joint abnormalities, impairment in long bone mineralization, and dysplasia [37]. To more broadly evaluate the necessity of Nf1 in the skeleton, Nf1 was deleted in cartilage tissues throughout the axial skeleton in Col2a1-cre;Nf1^flox mice [38]. The skeletal manifestations present in these mice were more severe compared to Prrx1-cre;Nf1^flox mice. Col2a1-cre;Nf1^flox mice were significantly runted with impairments in endochondral bone formation affecting the spine and leading to scoliotic deformity. In the long bones, bone formation was significantly impaired in both the trabecular and cortical compartments, a finding also observed in the vertebrae. Survival was limited, with most mice not surviving to weaning age (∼1 month).

Loss of Nf1 alters osteogenic differentiation of skeletal progenitor cells

Using cultured cells isolated from skeletal tissues of Prrx1-cre;Nf1^flox or Col2a1-cre;Nf1^flox mice, loss of Nf1 resulted in hyperactivation of the MAPK pathway and defects in osteoblast differentiation and mineralization [[37], [38], [39]]. These in vitro results prompted the generation and characterization of GEMMs targeting specific subsets of committed skeletal progenitor cell populations. For example, postnatal deletion of Nf1 in osteoprogenitor cells in Osx-cre;Nf1^flox mice resulted in significant postnatal growth restriction, reduced trabecular bone formation, increased cortical bone porosity, and impaired bone mineralization, all of which contribute to reduced biomechanical bone strength [[39], [40], [41]]. Tibia fracture healing was also impaired in Osx-cre;Nf1^flox mice. Similar to Osx-cre;Nf1^flox mice, targeting bone marrow skeletal progenitor cells in adult LepR-cre;Nf1^flox/- mice demonstrated that Nf1 is required for the maintenance of the skeleton in older mice, though fracture healing was not evaluated in this model [28]. Again, in vitro culture of Nf1-deficient mouse osteoprogenitor cells from Osx-cre;Nf1^flox or LepR-cre;Nf1^flox/- mice demonstrated significant impairment in osteogenic differentiation and mineralization compared to cells from control mice.

Loss of periosteal Nf1 alters skeletal development and fracture healing

With the implication of a periosteal basis for NF1-associated pseudarthrosis [42], multiple pre-clinical periosteal Nf1 models have been developed. One model aimed to broadly delete the Nf1 gene throughout the periosteum by targeting cells expressing the Periostin gene (Postn-cre;Nf1^flox mice). Loss of Nf1 in Postn-cre;Nf1^flox mice increases risk for neurofibromas at later age (Periostin is expressed in nerve tissues) [43]; however, skeletal development and fracture healing were normal [44]. To more accurately mimic human patients who harbor heterozygous NF1 mutations and somatic mutations in the periosteum, Postn-cre;Nf1^flox were crossed to the Nf1^+/− line to generate mice in which all cells are heterozygous for a Nf1 mutation while Periostin-expressing cells (i.e., bone periosteum and nerve cells) are deficient in Nf1 (i.e., loss-of-heterozygosity). Such Postn-cre;Nf1^flox/- mice developed skeletal deficiencies and delayed fracture healing [44,45]. Spatial transcriptomic analyses demonstrated significantly impaired endochondral bone healing and BMP signaling in a Postn-cre;Nf1^flox/- fracture [45].

Multiple studies have identified distinct skeletal progenitor cell populations within the mouse bone periosteum that contribute to bone regeneration after fracture [29,[46], [47], [48], [49]]. These studies often characterize progenitor cell populations based on a) their expression of a specific “marker” gene, and b) their contribution to one or multiple cell types required for fracture healing (i.e., chondrocytes, osteoblasts). As would be expected, genetic perturbation of these cell populations may disrupt fracture healing. While these progenitor cells can be distinguished from other periosteal cell populations by the expression of a marker gene, other extra-periosteal tissues may also harbor cell populations expressing the same marker gene. For example, expression of the gene encoding Cathepsin K (Ctsk) marks a periosteal progenitor cell population in embryonic and early postnatal mice; however, Ctsk is also robustly expressed in osteoclasts, and genetic perturbation experiments may affect both cell type [48]. Other periosteal progenitor cell populations involved in fracture healing include those expressing Gli1 [49] or Prss56 [29].

Although the impact of Nf1 loss in Gli1⁺ or Ctsk⁺ periosteal progenitor cells on fracture healing remains unclear, loss of Nf1 in Prss56-expressing periosteal progenitor cells resulted in post-fracture pseudarthrosis. Consistent with hyperactive Ras signaling following Nf1 loss, fracture healing was partially rescued when mice were treated with the MEK inhibitor selumetinib, and the addition of a SHP2 inhibitor further enhanced fracture healing [29].

To BMP or not to BMP? That is the question

The challenge of achieving a stable and durable union for NF1 pseudarthroses has prompted the use of additional off-label therapies to promote healing. While the use of Zoledronic acid (ZA) in combination with surgical cross-union has been previously reported in children with NF1 pseudarthrosis [50], outcomes were similar to those of cross-union without ZA [51], and ZA is not widely used in children. In contrast, rhBMP2 delivered via collagen sponge is frequently used as an off-label device to promote union. The use of rhBMP2 to improve healing rates in children with NF1 pseudarthroses has not been demonstrated in a clinical trial [21]. However, multiple retrospective studies related to its use in pseudarthroses have been published, including those involving children with and without NF1 treated with various surgical techniques [[52], [53], [54], [55], [56], [57]]. A recent systematic analysis evaluated multiple factors associated with primary union and refracture for pseudarthrosis, including patients with and without NF1 [58]. Interestingly, this study found an inverse association between union and the use of BMP in patients with NF1. Taken together, the field lacks consensus on whether rhBMP2 provides a significant benefit in promoting healing in the setting of NF1 pseudarthroses.

While BMP is known to promote bone formation, in the setting of NF1, multiple factors may contribute to reduced osteogenic responses. Chief among these is the hyperactivation of MAPK signaling associated with loss of NF1, which has been demonstrated in patient fracture-derived primary cells compared to cells from the iliac crest [29,30,45]. Consistent with this, spatial transcriptomic analysis demonstrated increased MAPK activation and blunted BMP pathway activation associated with persistent fibro-cartilagenous tissue in a patient's pseudarthrosis fracture [45]. These results align with molecular evidence suggesting that fracture-derived primary cells are primed for a fibrogenic rather than osteogenic fate [29]. Taken together, analyses of patient-derived primary cells suggest inherent deficiencies in the response of NF1-deficient cells to native and/or exogenous (i.e., rhBMP2) osteogenic cues. Therefore, modulation of the underlying dysregulation (MAPK, PIK3CA-AKT-MTOR, etc.) may alter the molecular priming of NF1-deficient cells toward a more osteogenic trajectory, thereby promoting healing.

In support of this concept, MEK inhibition partially rescued fracture healing in a novel mouse model where Nf1 was removed from Prss56-expressing periosteal progenitor cell [29]. Although improvements in the rate of union were modest, it remains unclear whether the addition of rhBMP2, as used in patients, may further increase union rates in this model. While inhibiting MAPK hyperactivation with a MEK inhibitor may reverse dysregulation associated with the loss of NF1 in cells at the fracture site, this may not be sufficient to activate the intrinsic molecular pathways responsible for fracture healing absent an extrinsic osteoanabolic cue, such as rhBMP2. Consistent with this, MEK inhibitor alone or rhBMP2 alone failed to rescue osteogenic differentiation of Nf1-deficient osteochondroprogenitor cells from Col2a1-cre;Nf1^flox mice; however, the combination of a MEK inhibitor with rhBMP2 significantly rescued differentiation and mineralization in vitro [39]. Taken together, current research into the therapeutic potential of targeted therapies is likely to implicate their use in combination with existing treatment paradigms that include internal or external fixation, iliac crest grafts, and rhBMP2. Emerging research, as discussed here, aims to improve fracture healing, reduce the risk of re-fracture, and decrease the incidence of amputation in children with NF1 pseudarthrosis.

Author contributions

Aysha Khalid: Conceptualization, Writing – original draft, Writing – review & editing. Nandina Paria: Conceptualization, Writing – original draft, Writing – review & editing. Jonathan J. Rios: Conceptualization, Writing – original draft, Writing – review & editing.

Consent for publication

The author(s) declare that no patient consent was necessary as no images or identifying information are included in the article.

Funding

The authors acknowledge funding from Scottish Rite for Children. The funder had no role in this manuscript but provided salary support for the authors.

Declaration of competing interests

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Jonathan J Rios reports a relationship with US Department of Defense that includes: funding grants. Aysha Khalid reports a relationship with US Department of Defense that includes: funding grants. Nandina Paria reports a relationship with US Department of Defense that includes: funding grants. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.