A description of the clinical course of the proband has previously been reported . Briefly, he is a Caucasian male born at 37 weeks of estimated gestational age following labor induction due to maternal hypertension. No dysmorphic features were noted. On day of life 1, he developed a patchy and scaly erythematous rash that intermittently disappeared and recurred, and was urticarial at times. At 1 week of age during evaluation for his rash, he was found to be thrombocytopenic with a platelet count of 36 K/μL. He then developed fever and lethargy and was admitted to the hospital where he underwent a sepsis rule out, skin biopsy, and bone marrow biopsy. The skin biopsy revealed perivascular and interstitial inflammation with eosinophils. The bone marrow biopsy showed a hypercellular marrow with predominantly maturing myeloid elements, a high myeloid:erythroid ratio, and presence of megakaryocytes. Blasts were < 2% and the karyotype was normal 46 (X,Y). No hemophagocytosis was noted. He was discharged home, but about 1 week later, he was re-admitted after developing hematochezia associated with a platelet count of 10 K/μL. Hepatomegaly was noted for the first time, which progressively worsened. He also developed persistent tachypnea and intermittent hypoxia requiring supplemental oxygen. A chest X-ray was unremarkable. It was thought that his tachypnea was due to mechanical causes related to his marked hepatomegaly. Liver biopsy revealed extensive extramedullary hematopoiesis. He subsequently developed splenomegaly. He continued to have a recurrent polymorphic rash as well as intermittent angioedema and nodules on his hands. A ferritin level at 18 days of life was 594 ng/mL, and fibrinogen was 211 mg/dL. Repeat skin biopsy showed inflammatory infiltrates with lymphocytes, histiocytes, and rare neutrophils and eosinophils. He developed anemia and neutropenia in addition to his thrombocytopenia and became both red blood cell and platelet transfusion dependent. He also developed mild eosinophilia with absolute eosinophil counts ranging from 0.52 to 3.2 K/μL. Immunoglobulin levels were all elevated (IgM 202 mg/dL, IgG 1830 mg/dL, IgE 353 IU/mL, and IgA 81 mg/dL) except for IgD. Peripheral lymphocyte subset studies were normal. Repeated bone marrow biopsies revealed a hypercellular bone marrow with megakaryocyte dysplasia and moderate to marked fibrosis that was positive by reticulin and trichrome stains (Fig. 1). The karyotype continued to be normal, and no excess of blasts was found. In vitro GM-CSF hypersensitivity testing to assess for possible juvenile myelomonocytic leukemia was negative. He was diagnosed with idiopathic infantile myelofibrosis. He was treated with a 4-day course of cytosine arabinoside (Ara-C) at 20 mg/m2/day to try to reduce his hepatosplenomegaly, but he did not have a clinical response. He continued to have baseline hypoxia with massive hepatomegaly causing restrictive lung disease. Oxygen saturations were 88% breathing room air, but increased to > 94% with supplemental oxygen. Echocardiography did not show evidence of pulmonary hypertension, and chest CT did not show evidence of parenchymal disease. At 2 months of age, he underwent bone marrow transplantation (BMT) with his 4-year-old female HLA-matched sibling serving as the donor. He was conditioned for BMT with busulfan (1.2 mg/kg/dose every 6 h for 16 doses, with busulfan pharmacokinetics to achieve a range of 600 to 900 nm/mL) followed by cyclophosphamide, 50 mg/kg for 4 days. There was no radiation as part of the formal BMT conditioning regimen. However, he did receive 300 cGy to his spleen prior to BMT conditioning to try to reduce the spleen size because of mechanical interference with his respiration. The patient’s respiratory compromise improved in the first few days following BMT, and he no longer required supplemental oxygen. However, beginning on day 21 following BMT, he developed worsening tachypnea and hypoxia. There were no infiltrates on his chest X-ray. Cardiac catheterization demonstrated severe pulmonary hypertension with baseline right pulmonary arterial pressure of 70 mmHg compared with a systemic pressure of 65 mmHg. He was placed on diuretics and calcium channel blockers, but his respiratory disease worsened and he died on day 32 following BMT. Autopsy revealed severe intimal hyperplasia of the pulmonary arterioles.
The proband’s younger brother (Fig. 2a, II-3) was born at 35 2/7 weeks of gestation by cesarean section due to antenatally detected hepatomegaly and reduced fetal movements. He was blue and floppy at birth (SaO2 in the 50% range with poor chest movement) and was intubated. Apgar scores were 3 at 1 min, 6 at 5 min, and 9 at 10 min. He was also noted to have a purpuric rash, hepatosplenomegaly, anemia (initial hematocrit 34%), reticulocytosis (initial reticulocyte count 9.8%), thrombocytopenia (initial platelet count 11 K/μL), leukoerythroblastosis (Fig. 1), and an elevated C-reactive protein level (73.5 mg/dL). A bone marrow aspirate performed on day of life 1 demonstrated moderate erythrocytic dysplasia. The karyotype was normal showing 46(X,Y). No hemophagocytosis was observed. Flow cytometric analysis revealed < 3% myeloblasts. A bone marrow core biopsy was not successful. He was weaned to continuous positive airway pressure (CPAP) by day of life 4, but continued to require supplemental oxygen. There was no evidence of pulmonary hypertension by echocardiography. It was thought that his severe hepatomegaly was causing restrictive lung disease. He was given supportive care while awaiting bone marrow transplantation. He remained platelet and red blood cell transfusion dependent, and continued to have elevated C-reactive protein levels (38.7–96.7 mg/dL). At about 4 weeks of life, he developed labored breathing and tachycardia, followed by bradycardia and decreased respiratory rate. He was re-intubated and diagnosed with presumed septic shock. Chest X-rays showed total atelectasis of the left lower lobe with bronchiectasis with the lobe. There was marked hyperexpansion of the left upper lobe and heavy markings of the right lung. Chest CT was initially concerning for possible congenital lobar emphysema, but this was subsequently ruled out. Direct laryngoscopy and bronchoscopy showed bronchomalacia below the right mainstem bronchus. He subsequently developed multi-organ failure and died at about 2.5 months of age. Autopsy revealed a hypercellular vertebral bone marrow comprised exclusively of early myeloid cells (Fig. 1g), which are positive for myeloperoxidase (MPO) via immunohistochemistry (Fig. 1h) and negative for CD3, CD43, CD34, and CD117. Mild fibrosis could be highlighted by reticulin stain (Fig. 1i), and patchy involvement was noted of the spleen, bilateral kidneys (Fig. 1j), and lymph nodes (Fig. 1k). There was diffuse trichome-positive fibrosis of the liver. The lungs were bilaterally consolidated with left upper lobe hemorrhage and necrotizing pneumonia (Fig. 1l) present in the left upper, left lower, and right lower lobes. There was no evidence of pulmonary hypertension. No fibrosis was noted. Lung cultures showed moderate growth of Pseudomonas aeruginosa.
The proband’s 3 sisters (II-1, II-4, II-5) and both parents (I-1, I-2) are all healthy. Sanger sequencing of peripheral blood mononuclear DNA from the proband and his affected younger brother failed to identify mutations in the coding sequences or intron/exon boundaries of GATA1, JAK2, or MPL.
Given the observed inheritance pattern, we initially searched for X-linked or autosomal recessive causes of this phenotype using WES data generated from all family members (Fig. 2a). This analysis was unrevealing for any putative causal mutations showing appropriate segregation. However, analysis of potential de novo variants revealed a single recurrent event in both affected children, Chr1:22417990:C>T (hg19 coordinates), resulting in the R186C mutation in CDC42 (Fig. 2b). Given that such recurrent de novo mutations can be associated with low-level parental germline mosaicism  and that the father had shown a single read with this variant in WES, we performed deep sequencing to > 100,000-fold of the mutated region of CDC42 in blood-derived genomic DNA from all family members and unrelated controls and observed an increased level of mosaicism from 0.07 to 0.15% in controls to 0.59% in the unaffected father (I-1; Fig. 2c). This finding suggests paternal germline mosaicism as the source for the recurrent mutations noted in the two affected children. This mutation was located in an evolutionarily conserved C-terminal polybasic region of CDC42 (Fig. 2d) and was recently described in patients with a distinct disorder characterized by immune dysregulation [14, 22].
CDC42 is a member of the well-studied Rho family of small GTPases and, through interacting proteins known as effectors, regulates signaling pathways that control diverse cellular functions including cell cycle progression, cell migration, and cytoskeletal dynamics . Rho GTPases have been shown to play critical roles in hematopoietic cells, including hematopoietic stem/progenitor cells, as well as cells of the myeloid and lymphoid lineages . In addition, Rho GTPases, particularly RHOA, RHOH, and the GTPase most closely related to CDC42, RAC, have been implicated in malignant transformation. CDC42 is best studied for its role in regulating and enabling signaling pathways to intersect with cytoskeletal activity. Importantly, CDC42 has been shown to have a critical role in HSCs and other hematopoietic progenitors through the study of mutant mice [52, 53]. Conditional deletion of Cdc42 in the mouse hematopoietic system led to a rapidly fatal myeloproliferative disorder . Finally, a different variant in CDC42 (Y64C) in a human patient has been associated with myelofibrosis observed in adulthood . Therefore, the observed mutation appeared to be potentially relevant to hematopoiesis, although the human phenotype described in this kindred is distinct from what has been reported in animal models and other human patients previously.
We initially examined the location of the mutation in CDC42 in a structure involving the complex of CDC42 with its critical guanine nucleotide exchange factor, RhoGDI (encoded by the ARHGDIA gene in humans; Fig. 3a) . The R186 residue is located near the geranylgeranyl moiety and is predicted to alter several key interactions with RhoGDI (Fig. 3a), as has recently been demonstrated through functional analyses . Consistent with this, introduction of either wild type or the R186C mutant CDC42 into 3T3 cells using lentiviral transduction led to abnormal cytoskeletal structure and cell morphology. When transduced cells were plated on fibronectin, we found that the mutant, but not control or WT, CDC42-transduced cells inhibited the formation of filipodia, which depend upon CDC42 activity (Fig. 3b) . After short-term plating of 3T3 cells onto fibronectin for 45 min, we noted that significantly more control or WT CDC42-transduced cells showed a flattened morphology compared with the CDC42 mutant-transduced cells that demonstrated a predominantly rounded morphology (Fig. 3 c, d). These results suggested that the mutant was acting in a dominant-negative or potentially neomorphic manner. Additionally, the phosphorylation of PAK1, a known CDC42 effector protein, was reduced in the mutant-transduced 3T3 cells plated on fibronectin compared with WT CDC42-transduced cells (Fig. 3e).
While these studies suggested dominant-negative or neomorphic effects and deregulated activity of the mutant CDC42, we wanted to gain more precise insights into the function of this mutant protein in hematopoietic cells. Since no hematopoietic cells were available from the patients, we utilized CD34+ HSPCs from healthy individuals. These cells were transduced using lentiviral vectors expressing either the WT or the R186C mutant form of CDC42. Two days after infection and immediately after flow cytometric sorting of the transduced cells, we assessed cell migration in a CXCL12/SDF-1 gradient using a transwell migration assay (Fig. 3f). While we observed no change in the expression of CXCR4 (CD184), the receptor for CXCL12/SDF-1 (Fig. 3 g, h), we noted a > 2-fold reduction in chemokine-directed migration of cells (Fig. 3f). We next assessed cell cycle progression in vitro using BrdU staining in the HSPCs transduced with either the control, WT, or mutant CDC42 in the presence of multiple cytokines including thrombopoietin, stem cell factor, interleukin 3 (IL-3), and the chemokine, CXCL12. Under these conditions, we noted no difference in cell cycle progression (Fig. 3i), suggesting that the observed phenotypes in the patients rely upon disruption of key interactions in the context of an intact niche, as has been observed in conditional knockout mice .
As this CDC42 residue is highly conserved (Fig. 2d), and hematopoiesis in Drosophila melanogaster has many shared features with mammalian hematopoiesis, including critical interactions between hematopoietic cells and their microenvironment; we further explored this mutation in vivo in Drosophila . Rho GTPases have previously described important functions in the fly hematopoietic system, where Rho1 and its downstream effector Wash have key roles in the migration of embryonic hemocytes, while Cdc42 has been shown to be critical for maintaining appropriate cell polarity during hemocyte migration towards wounds [43, 46].
In order to explore this specific mutation in Drosophila, we made transgenic flies which express the CDC42 R186C mutation using the inducible UAS-Gal4 system in Pxn-Gal4/UAS-GFP transgenic flies to express this mutant CDC42 specifically in Drosophila hemocytes that also expressed GFP [6, 37]. We explored hemocyte function in wild type flies containing the Gal4 driver only (control), flies overexpressing wild type Cdc42 (Cdc42 WT), and flies overexpressing the Cdc42 R186C mutation (Cdc42 mutant) and found that in contrast to control and Cdc42 WT flies, Cdc42 mutant flies had reduced protrusions in migrating hemocytes (Fig. 4a–c). In order to quantify this defect, we measured the percentage of the hemocyte cell body with protrusions and found a profound loss of protrusions in the Cdc42 mutant compared with control and Cdc42 WT hemocytes (p < 0.0001) (Fig. 4d). Given the defective migration observed with the CDC42 mutant in human hematopoietic cells, we wondered whether Drosophila hemocytes may also fail to migrate appropriately from head to tail in response to cues supplied by their microenvironment. During fly embryogenesis, hemocytes are formed from the head mesoderm, while the head and tail are adjacent. A subset of hemocytes migrate to the tail from the head, dependent on the Rho GTPase, RhoL, before the tail retracts during morphogenesis to the posterior . Once tail retraction is complete and in a chemotactic response to the PVR ligands, Pvf2 and Pvf3, head hemocytes migrate towards the posterior and tail hemocytes migrate anteriorly along the ventral midline towards each other and subsequently distribute throughout the entire embryo [9, 51]. This stereotypic process starts in stage 10 embryos and completes in ~ 2 h in normal embryos. We found that this migration process was severely disrupted in embryos expressing the Cdc42 mutant, but not in control or Cdc42 WT embryos (Fig. 4e–g, n = 20 replicates per condition). Despite this defect, hemocytes in all conditions were able to receive cues to migrate from head to tail and vice versa as hemocytes migrate from both the head and tail and can be seen along the ventral midline (Fig. 4g, 90 min). However, their capacity for migration was greatly diminished as Cdc42 mutant-expressing hemocytes showed ~ 2-fold reduction of migration speed, as compared with the other conditions (p < 0.0001, n = 5 hemocytes per embryo, 20 embryos per condition). Thus, the CDC42 R186C mutation in flies also functions in a dominant-negative or neomorphic manner, impacts cell protrusions, and impairs the ability of hematopoietic cells to migrate in response to tissue microenvironmental cues, leading to an overall defective hematopoietic system.
Relationship to Primary Myelofibrosis
Primary myelofibrosis is a condition associated with somatic mutations in JAK2 or other factors that activate the thrombopoietin signaling pathway [15, 49]. Given the previously characterized role of Rho GTPases in hematopoietic cell localization within the marrow microenvironment [8, 31] and the disruption of this process observed in PMF , we hypothesized that the severe infantile myelofibrosis phenotype that results from the CDC42 R186C mutation may provide insights into the more common acquired PMF cases. CDC42 was broadly expressed in normal human hematopoiesis across the full hematopoietic hierarchy, including in HSCs and other early hematopoietic progenitors (Fig. 5a). To examine whether CDC42 mRNA expression may be altered in PMF patients, we compared gene expression in peripheral blood-derived CD34+ HSPCs from 42 PMF cases with 16 healthy donors . CDC42 was among the most downregulated genes in PMF compared with normal donors with expression being ~ 5-fold less across all the samples (Fig. 5b, c). To validate these observations, we assessed bone marrow sections from 10 individuals with PMF and 7 individuals with normocellular bone marrows. We opted to examine megakaryocytes, given that they are easily distinguished morphologically and highly express CDC42 (Fig. 5a). We observed reduced CDC42 protein expression using immunohistochemistry in the megakaryocytes across the PMF patients, bolstering and extending our results on CDC42 dysfunction occurring more commonly in PMF (Fig. 5d, e, f).