KRAS^G12D expression in lung-resident myeloid cells promotes pulmonary LCH-like neoplasm sensitive to statin treatment

Langerhans cell histiocytosis (LCH) is a rare disorder characterized by abnormal growth of CD1a⁺S100⁺ histiocytes that share some characteristics with epidermal Langerhans cells (LCs).^1,2  Multiple organs, including the skin, bone, liver, and lung, are affected in LCH either as a multisystem or single-organ disease, and hematopoietic organ involvement has been shown to be a feature of high-risk diseases.¹  Pulmonary involvement occurs either in isolation or as part of multisystem diseases, though >85% cases of pulmonary LCH have been reported as single-organ involvement.³ 

Recent discoveries of somatic mutations in the genes involved in the RAS/RAF/MEK/extracellular signal-regulated kinase (ERK) pathway have unveiled that LCH pathogenesis is driven by neoplastic transformation associated with deregulated ERK pathway activation.^4-9  The BRAF^T1799A mutation, leading to BRAF^V600E oncoprotein expression, is most common in LCH,⁴  whereas other genes in the ERK pathway, including ARAF⁵  and MAP2K1,^6,7  have been also reported as mutational targets. Furthermore, NRAS⁸  and KRAS⁹  mutations, leading to NRAS^Q61K/R and KRAS^G12V/A oncoprotein expression, respectively, have been identified uniquely in pulmonary LCH cases, suggesting organ-specific contribution of oncogenic RAS to the pathogenesis of LCH.

Both LCs and lesional LCH cells express Langerin (also known as CD207), a C-type lectin that contributes to the formation of Birbeck granules in LCs/LCH cells.^10,11  However, Langerin expression is not restricted to LCs: it is also detected in subsets of dendritic cells (DCs), including CD103⁺ DCs.¹²  A previous study utilizing genetically engineered mouse (GEM) models demonstrated that BRAF^V600E expression in CD11c⁺ or Langerin⁺ DC-lineage cells is sufficient to drive LCH-like pathology in mice, proposing that LCH could originate from CD11c⁺ DC precursors or Langerin⁺ mature DCs.¹³  However, it is currently unclear whether Langerin promoter-driven Cre recombinase (Cre) expression used in the model¹³  drives BRAF^V600E-induced neoplastic transformation of LCs or Langerin⁺ DCs.

In contrast to conventional DCs (including Langerin⁺ DCs) developing in a FLT3 (also known as CD135)-dependent manner,¹²  mouse epidermal LCs, defined by their CD11c⁺Langerin⁺CD103⁻F4/80⁺ phenotype,¹²  arise largely from embryonic/fetal precursors, including yolk-sac–derived myeloid progenitors^14,15  and fetal liver monocytes,¹⁶  in a CSF1R (also known as MCSFR or CD115)-dependent manner.^17,18  LC precursors distributed in the embryonic epidermis express CD11c and Langerin immediately after birth,¹⁹  suggesting that BRAF^V600E induction in the GEM models above¹³  could start during the first week of life at the latest. Importantly, the majority of human pulmonary LCH cases are adult onset, in contrast to multisystem LCH that is predominantly diagnosed during childhood.²⁰  Therefore, animal models recapitulating the adult-onset pathology in the lung are preferred in order to study the pathogenesis of pulmonary LCH that could be considerably different from multisystem LCH.

Conditional knockin mice expressing KRAS^G12D following Cre expression have been previously reported and extensively used to study the pathophysiological consequences of oncogenic RAS expression in different tissues including the lung. Although KRAS^G12D expression in the lung following exposure to an adenoviral vector encoding Cre under control of the cytomegalovirus (CMV) promoter (AdCMVCre)²¹  has been previously used to study lung adenocarcinoma, somewhat unexpectedly, we found that this model developed mixed neoplasms with hematopoietic and epithelial components. Our investigations confirm that the hematopoietic pathology is driven by KRAS^G12D expression in specific myeloid population(s) in the lung, recapitulating some aspects of pulmonary LCH. Using this model as a therapeutic platform, we found that the pulmonary LCH-like disease is sensitive to in vivo treatment with atorvastatin, suggesting therapeutic potential of statins for the treatment of RAS-driven pulmonary LCH.

Animal experiments were performed under UK Home Office License authority. Kras^LSL-G12D, Braf^LSL-V600E, and their recombined alleles were genotyped as described.^22,23  Nasal delivery of adenoviral vectors was performed as described.²³  Atorvastatin (10 mg/kg; Sigma) dissolved in phosphate-buffered saline was orally administrated once a day for 5 weeks, starting at 5 weeks after adenoviral induction. Lung tissues were processed for hematoxylin-and-eosin (H&E) staining and immunohistochemistry (IHC) as described.²²  Antibodies used for IHC are described in the supplemental Methods (available on the Blood Web site).

Paraformaldehyde-fixed, Triton-X–permeabilized cells were stained for Mac2, and blocked for endogenous biotin as described.²⁴  Then, the cells were stained with anti-CD207-biotin antibody (1:100; eBioscience), followed by Alexa Fluor 488–conjugated streptavidin (Invitrogen) staining (1:2000). Images were obtained and processed as described.²⁴  Protein lysates were prepared by solving frozen tissues into 1× sodium dodecyl sulfate sample buffer or cultured cells into NP40 buffer, and analyzed by western blotting as described.^25,26  Cell surface markers were analyzed by flow cytometry as described.²⁴  For in vitro 5-bromo-2′-deoxyuridine (BrdU) incorporation, cells labeled with 10 μM BrdU for 18 hours were analyzed as described.²⁴  For cell sorting, lung tissues digested by 30-minute collagenase/DNase treatment were stained for cell surface markers as described,²⁴  and sorted using FACSAria II (BD Biosciences). Antibodies used are described in the supplemental Methods.

Cell cultures were performed essentially as described.²⁴  Detailed methods are described in the supplemental Methods.

Cells from KRAS^LSL-G12D lungs were plated at 2 × 10⁵ per well to 10 × 10⁵ per well, and infected with adenoviral vectors at a multiplicity of infection equal to (MOI) 100. After 1 week of infection, media containing the virus were removed, and the cultures were maintained in Dulbecco modified Eagle medium/10% fetal calf serum with half media changes every 3 to 5 days without disturbing nonadherent cells. Limiting dilution analysis was performed using ELDA software.²⁷ 

A total of 1 × 10⁵ sorted myeloid cells were infected with adenovirus vectors at MOI = 100 in Dulbecco modified Eagle medium/10% fetal calf serum for 24 hours, and transferred onto fibroblast feeders established by serial passages of wild-type lung cells. On day 7, media containing the virus were removed, and the cultures were maintained as described for limiting dilution culture.

Comparison between any 2 groups was performed by the unpaired Student t test. χ² tests were used to statistically evaluate the differences in limiting dilution cultures.

Nasal inhalation of AdCMVCre into KRAS^LSL-G12D/+ mice has been widely exploited to investigate lung carcinogenic processes. Although AdCMVCre could induce oncogene expression in any type of cell when successfully infected, nasal inhalation of 10⁵ plaque-forming units (pfu) AdCMVCre preferentially promotes epithelial transformation in this model.²¹  To investigate whether higher viral doses could induce neoplastic transformation of other cell types, we administrated 3 × 10⁷ pfu AdCMVCre into KRAS^LSL-G12D/+ mice by nasal inhalation (referred to as AdCre/KRAS^G12D mice hereafter), and AdCre/KRAS^G12D mice exhibiting severe respiratory symptoms at 4 to 6 months following nasal inhalation were analyzed histologically. As expected, a number of solid mass lesions containing epithelial components were observed, but these lesions were frequently intertwined with round cells with a less eosinophilic cytoplasm (Figure 1A). In addition, distal alveolar spaces were occasionally filled with similar round cells, mimicking the pathology of desquamative interstitial pneumonia (DIP) (Figure 1A). The round cells in both solid and DIP-like lesions were positive for S100, a marker for LCH²  (Figure 1A).

Because the morphology of the round cells observed in AdCre/KRAS^G12D lungs was suggestive of their hematopoietic origin, we analyzed hematopoietic compartments in the AdCre/KRAS^G12D lungs using flow cytometry. Compared with KRAS^+/+ mice nasally delivered with AdCMVCre (referred to as AdCre/KRAS^WT mice hereafter), the percentage of CD11b^lowCD11c⁺ cells was robustly increased in the AdCre/KRAS^G12D lung (Figure 1B). Interestingly, the CD11b^lowCD11c⁺ population in the AdCre/KRAS^G12D lung included the majority of the cells expressing a macrophage/LC marker F4/80 (Figure 1B blue in the plot), whereas most F4/80⁺ cells in the AdCre/KRAS^WT lung were detected in CD11b^intCD11c^low/− cells (Figure 1B green in the plot). Collectively, the hematopoietic pathology was characterized by aberrant accumulation of CD11b^lowCD11c⁺F4/80⁺ cells.

Because nasal delivery of AdCMVCre into KRAS^LSL-G12D lungs induces epithelial neoplasms,²¹  it is unclear whether the accumulated CD11c⁺ cells could represent KRAS^G12D-transformed lung myeloid cells, or occur as a reactive response of nonneoplastic CD11c⁺ cells (not expressing KRAS^G12D) to epithelial neoplasms. To distinguish these possibilities, we sorted lung CD45⁺ hematopoietic subpopulations according to their CD11b/CD11c profiles, and investigated KRAS^G12D recombination in each subpopulation. KRAS recombination was clearly detected in the CD45⁺CD11b^lowCD11c⁺ population (Figure 2A), indicating that the accumulation of CD11c⁺ cells was driven by their KRAS^G12D expression.

Morphologically, the sorted CD11c population was composed of smaller round cells with a basophilic cytoplasm and larger macrophage-like cells (Figure 2B). Monocytes or neutrophils with typical morphologies were not identified. Because of the F4/80⁺ phenotype and histological findings that are consistent with LC/LCH characteristics (Figure 1A),^{2,12,14,28,29}  we next performed CD207 immunostaining of enriched CD11c⁺ cells combined with a myeloid marker MAC2.³⁰  We found that most cells showed reticular cytoplasmic CD207 staining with plasma membrane MAC2 (Figure 2B), suggesting that the CD11c⁺ cells are committed to LC-lineage differentiation. Consistently, CD207⁺ cell accumulation in the AdCre/KRAS^G12D lung was detected by IHC, and increased CD207 protein expression was confirmed by immunoblotting (Figure 2C).

To investigate the proliferative potential of the CD11c⁺ cells, CD11c-enriched AdCre/KRAS^G12D lung cells were cultured with BrdU. However, BrdU uptake was almost negligible in the CD11c^high cells (Figure 3A), indicating that they were mitotically inactive. Interestingly, some BrdU incorporation was detected in CD11c^dim cells (Figure 3A), but they failed to grow in long-term culture (data not shown). In contrast, when CD11c^high cells were removed by selective adhesion to plastic plates and the resulting “CD11c-depleted” nonadherent cells were cultured, floating cell clusters and endothelial (CD31⁺) networks developed along with fibroblasts and epithelial cells (Figure 3B). The majority of floating cells were CD45⁺CD11b⁺ small round cells that exhibited KRAS^G12D recombination (Figure 3B-C), demonstrating that the floating cells are myeloid cells promoted by KRAS^G12D expression. Even after 7 weeks of culture, they maintained proliferative potential, and expressed surface CD11c and F4/80 (Figure 3C). Notably, CD11c^highF4/80^high cells in this population partially expressed surface CD207, whereas CD11c^lowF4/80^low cells did not (Figure 3C), suggesting that upregulation of CD11c and F4/80 could be associated with LC-lineage commitment.

When the floating cells were replated, they altered their morphology within 48 hours. Half of the cells showed LC-like dendritic morphology whereas one-third had a macrophage-like round/amoeboid morphology (Figure 3D). Plasma membrane CD207 expression and nuclear MAC2 localization were characteristic of LC-like cells and were not observed in the macrophage-like cells (Figure 3D). These in vitro data suggest the presence of proliferative precursors in the AdCre/KRAS^G12D lung that differentiate into CD11c^highF4/80^high LC-lineage cells.

The high AdCMVCre virus dose required for CD11c⁺F4/80⁺ cell accumulation in the AdCre/KRAS^G12D lung suggested that cells initiating the pathology could be relatively rare. This prompted us to estimate the frequency of “myeloid pathology-initiating cells” (MPICs). To this end, we infected unrecombined KRAS^LSL-G12D lung cells in vitro with AdCMVCre, or control vectors expressing β-galactosidase (AdCMVβgal) or Cre under the surfactant protein C promoter (AdSPCCre).³¹  When 2 × 10⁵ AdCMVCre-infected cells were cultured for 4 weeks, 15% of the culture developed cell clusters (Figure 4A-B) in a similar manner to the AdCre/KRAS^G12D lung cultures (Figure 3B). When the plating cell number was increased to 1 × 10⁶, 75% of the culture produced cell clusters, by which the frequency of MPICs was estimated as 1 in 0.83 × 10⁶ lung cells (95% confidence interval, 1 in 1.39 × 10⁶ to 1 in 0.50 × 10⁶) (Figure 4C). Floating cell clusters developing in these cultures showed clear KRAS^G12D recombination (Figure 4D). In contrast, cells infected with control vectors only developed mixed cultures with macrophage-like cells and fibroblasts (Figure 4A-B), indicating that KRAS^G12D expression was essential for the aberrant growth of small round cells forming cell clusters. Half of AdCMVCre-infected cultures plated at 2 × 10⁵ cells per well also developed macrophage/fibroblast mixed cultures (Figure 4A-B), and floating macrophages collected from the cultures showed KRAS^G12D recombination (Figure 4D), suggesting that KRAS^G12D expression in lung macrophages might not be sufficient for initiating the development of cluster-forming cells.

Floating cell clusters were maintained in culture for 7 weeks, and their CD11c⁺F4/80⁺ phenotype was confirmed (Figure 4E). Morphological changes to LC-like cells were also observed upon replating (Figure 4F), but plasma membrane CD207 staining or MAC2 nuclear translocation was rarely detected in this condition. In the in vitro AdCMVCre infection culture, neither endothelial networks nor epithelial cells were developed. Such differences in the nonhematopoietic components observed following in vivo and in vitro KRAS^G12D induction might affect the CD207/MAC2 phenotype of the developed LC-like cells.

To prospectively evaluate MPIC potentials of lung myeloid populations, 4 myeloid subpopulations in uninduced KRAS^LSL-G12D lungs, including CD11b^lowCD11c⁺ (Fr-1), CD11b^intCD11c^dimGr1⁻ (Fr-2), CD11b^intCD11c⁻Gr1^int (Fr-3), and CD11b^highCD11c⁻Gr1^high (Fr-4) cells, were isolated and infected with AdCMVCre (Figure 5A-B). KRAS^G12D recombination was undetectable in Fr-3/4 cells that could not survive in this condition, but AdCMVCre-infected Fr-1/2 cells showed weak KRAS^G12D recombination (Figure 5B). When cultured without feeders, infected Fr-1/2 cells were viable for a couple of days, but eventually died within 10 days without showing any signs of active growth. However, AdCMVCre-infected Fr-2 cells proliferated to form round cell clusters on fibroblast feeders (Figure 5C), suggesting that lung-resident CD11b^intCD11c^dimGr1⁻ cells were likely transformed into MPICs by KRAS^G12D expression. These data also suggest that the poor adenoviral infection efficiency into this population might be the major reason for the rarity of MPICs.

Although the CD11b^intCD11c^dimGr1⁻ phenotype was consistent with resident/patrolling monocytes,^32,33  CD11b⁺ DCs³⁴  and interstitial macrophages^35,36  could also show similar phenotypes. Therefore, we further analyzed Fr-2 cells for their expression of major histocompatibility complex class II and DC markers.³⁷  They were found to be CD115⁺SSC^low cells negative for DC markers CD103/CD117/CD135, but included class II⁺ subpopulations (Figure 5D), likely composed of professional antigen-presenting cells and/or transitional (inflammatory to resident) monocytes with class II upregulation.³⁸ 

Because BRAF is frequently mutated in human pulmonary LCH,^4,8,9,39-41  we investigated whether nasal delivery of AdCMVCre into the BRAF model also induced LCH-like neoplasms. AdCMVCre (3 × 10⁷ pfu) was nasally delivered into the lung of BRAF^LSL-V600E mice (referred to as AdCre/BRAF^V600E mice hereafter), and lung tissues were histologically analyzed at 4 months following AdCMVCre infection. In contrast to AdCre/KRAS^G12D mice, well-demarcated papillary adenomas accompanied by myeloid cells in the microenvironment developed, especially in adenoma-rich areas (Figure 6A left) as we have previously reported.²⁴  On the other hand, alveolar structures in adenoma-free areas were largely intact without developing DIP-like lesions (Figure 6A right).

Although we previously reported reactive myeloid cell accumulation in the BRAF^V600E model induced by spontaneous activation of CreER,²⁴  the myeloid cells similarly accumulating in the AdCre/BRAF^V600E lung were not formally proven to be nonneoplastic. To address this, we sorted hematopoietic subpopulations from the AdCre/BRAF^V600E lung, and BRAF recombination was examined. Similarly to the AdCre/KRAS^G12D lung (Figure 2A), 3 distinct hematopoietic populations were identified, but the CD11c⁺ population in the AdCre/BRAF^V600E lung showed only subtle recombination of the BRAF^V600E allele (Figure 6B), confirming that CD11c⁺ cells in the BRAF model were largely nonneoplastic. These data demonstrate that BRAF^V600E expression in lung-resident myeloid cells by nasal delivery of AdCMVCre is not sufficient for promoting myeloid neoplasms as seen in the AdCre/KRAS^G12D lung.

Isoprenylation of RAS proteins promotes their association with the plasma membrane where the proteins are activated.⁴²  Consistently, genetic modifications inhibiting RAS prenylation have been reported to ameliorate KRAS-driven myeloproliferative disorders.^43,44  Therefore, we sought to investigate whether chemical inhibition of protein prenylation suppresses myeloid neoplasm developing in the AdCre/KRAS^G12D mice. Because isoprenoid lipids synthesized through the mevalonate pathway are required for RAS prenylation,⁴²  we used atorvastatin, an 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitor that suppresses the mevalonate pathway upstream of isoprenoid lipid production. AdCre/KRAS^G12D lung cultures treated with 3.3 μM atorvastatin showed significant reduction of proliferating CD45⁺ cells compared with vehicle control cultures (Figure 7A). This antiproliferative effect was not associated with differentiation induction because CD11c and F4/80 expression was rather suppressed by atorvastatin (Figure 7B). Our biochemical analysis of CD11c-enriched AdCre/KRAS^G12D lung cells treated with atorvastatin showed decreased AKT, but not ERK, phosphorylation (Figure 7C), suggesting that inhibition of RAS prenylation inhibits RAS interaction with phosphatidylinositol 3-kinase (PI3K), as reported for marrow-derived macrophages.⁴⁵  Of note, poly (ADP-ribose) polymerase 1 (PARP-1) cleavage was also induced by atorvastatin (Figure 7C), indicating that apoptosis of atorvastatin-treated CD11c⁺ cells might contribute to the reduced CD11c levels in atorvastatin cultures (Figure 7B).

The in vitro effects of atorvastatin prompted us to target the myeloid neoplasm in vivo. We thus treated AdCre/KRAS^G12D mice with atorvastatin for 5 weeks. In drug-treated animals, the DIP-like pathology was robustly improved, though some mass lesions, mainly composed of epithelial components, remained in the periphery of the lung (Figure 7D). Quantitative analysis using flow cytometry confirmed up to 70% reduction of CD11c⁺F4/80⁺ cells by atorvastatin, but no significant reduction was observed in other cells (Figure 7E). These data suggest the potential of statins as chemotherapeutic agents against RAS-mutated pulmonary LCH.

Due to its distinct clinical characteristics, pulmonary LCH requires management/therapeutic strategies different from multisystem LCH.⁴⁶  To better understand the pathogenesis of pulmonary LCH, and broaden therapeutic options for patients struggling with advanced diseases, development of animal models recapitulating the human disease is essential. In the present study, we found that adenoviral induction of KRAS^G12D expression in lung-resident myeloid cells promotes pulmonary LCH-like neoplasm in mice, and that the pathology can be ameliorated by in vivo administration of atorvastatin, a cholesterol-lowering drug widely used in the clinic. These data shed light on the potential cellular origin and genetic/molecular pathogenesis of pulmonary LCH, and provide a rationale to explore the therapeutic potential of statins against subsets of pulmonary LCH.

As far as we know, this is the first demonstration that nasal delivery of AdCMVCre to the KRAS^LSL-G12D mouse lung can induce hematopoietic neoplasms. Our data show that nasal delivery of AdCMVCre at relatively high doses is required for inducing myeloid neoplasms, likely due to relatively few cells in the lung initiating neoplastic myeloid growth (Figure 4). However, the lung pathology induced by high-dose AdCMVCre is complex, with hematopoietic phenotypes coexisting with alveolar and bronchiolar neoplasms. These factors may have collectively contributed to the fact that KRAS^G12D-induced neoplastic transformation of lung-resident myeloid cells in this model has not been previously appreciated.

To our surprise, the majority of pathogenic CD11c^high cells in the AdCre/KRAS^G12D lung were found to be mitotically inactive (Figure 3A), though this is consistent with a previous study showing nonproliferative lesional cells in human pulmonary LCH.⁴⁷  Instead, proliferating CD45⁺ cells expressing variable levels of CD11c developed in culture, suggesting that proliferating precursors for CD11c^high cells can exist in the AdCre/KRAS^G12D lung. Proliferating CD45⁺ cells form cell clusters (Figure 3B) in a similar way to immature LCs developing from human CD34⁺ hematopoietic stem/progenitor cells (HSPCs) cultured with transforming growth factor β1 (TGFβ1).⁴⁸  The CD11c-depleted culture was composed of not only hematopoietic, but also epithelial, fibroblastic, and endothelial cells (Figure 3B) and our preliminary analysis detected TGFβ1 in the culture media (T.K. and C.P., unpublished data). Further investigations to identify the cellular source(s) for the secreted TGFβ1 are currently under way.

Although our data suggest that CD11b^intCD11c^dimGr1⁻ monocytes are the major cellular targets of KRAS^G12D-induced transformation in our model, the in vitro infection efficiency of AdCMVCre into this population was very low (Figure 5B). This could be due to the low-level expression of αv integrins that mediate adenovirus internalization.⁴⁹  Granulocyte-macrophage colony-stimulating factor (GMCSF) stimulation of human monocytes reportedly induces αv integrin expression, leading to improved adenoviral gene delivery,⁴⁹  whereas alveolar epithelial cells are the major GMCSF-producing cells in mouse lungs.⁵⁰  Alveolar epithelium-derived GMCSF may induce αv integrin expression in vivo in lung-resident monocytes, making them more susceptible to adenoviral infection.

BRAF is frequently mutated in human pulmonary LCH (24%-63%)^4,8,9,39-41  with much higher prevalence than KRAS (8%),⁹  though NRAS mutations could be more common than KRAS.⁸  But unexpectedly, a pulmonary LCH-like neoplasm was not observed in our AdCre/BRAF model (Figure 6). This may be explained by different cell populations initiating the BRAF-induced pathology. Our data suggest CD11b^intCD11c^dimGr1⁻ monocytes as the major cell type initiating the KRAS^G12D-driven pathology, whereas common DC progenitors (CDPs) and Langerin⁺ DCs have been proposed to drive BRAF^V600E-induced neoplasms.¹³  Nasally delivered AdCMVCre may not effectively infect lung-resident CDPs or Langerin⁺ DCs, resulting in the failure to develop LCH-like neoplasms. BRAF^V600E-induced MEK/ERK activation may be sufficient for transforming CDPs and Langerin⁺ DCs, but KRAS^G12D-induced PI3K activation may be needed to transform CD11b^intCD11c^dimGr1⁻ monocytes, as suggested by our atorvastatin data (Figure 7C).

Using our model as a platform for preclinical evaluation, we identified antitumor effects of atorvastatin against KRAS^G12D-driven pulmonary LCH-like neoplasm (Figure 7). Antigrowth effects of statins on tumor cells are thought to be attributable, at least in part, to inhibition of RAS family small GTPases.⁵¹  Our biochemical data (Figure 7C) suggest that atorvastatin inhibits prenyl-dependent association between KRAS and PI3K as reported in simvastatin-treated human peripheral blood mononuclear cells.⁴⁵  Intriguingly, PIK3CA/BRAF comutations have been identified in Erdheim-Chester disease, a histiocytic disorder closely related to LCH,⁵²  suggesting that cooperative activation of the MEK/ERK and PI3K pathways could contribute to the pathogenesis of histiocytic neoplasms. Although cancer-associated PIK3CA mutations are rare in LCH,⁵³  we previously reported contribution of autocrine PI3K activation to the aberrant growth of BRAF^V600E-expressing mouse HSPCs.²⁵  Autocrine PI3K activation may also contribute to the pathogenesis of human multisystem LCH, in which marrow CD34⁺ HSPCs reportedly harbor BRAF mutations.¹³  In this regard, PI3K-targeted therapeutics may be beneficial for treating refractory LCH, not restricted to RAS-mutated pulmonary diseases but including BRAF-mutated multisystem diseases.

We exploited an autochthonous GEM model to investigate how KRAS^G12D expression in the lung affects lung-resident myeloid cells in a spatiotemporally controlled manner, leading to a unique opportunity to model pulmonary LCH. However, the clinical relevance of our preclinical study has yet to be fully explored in the clinical setting, though recent discoveries of RAS mutations in human pulmonary LCH are consistent with our findings. Future investigations of pulmonary LCH cases with regard to correlation between statin use and clinical outcomes/RAS mutation status would provide valuable information. Notably, a recent report of a phase 2 clinical trial of pan-AKT inhibitor afuresertib for LCH patients, including 7 evaluable cases with pulmonary involvement, demonstrated that AKT inhibition exerts limited efficacy against pulmonary LCH patients.⁵⁴  Because AKT inhibition induces feedback activation of PI3K,⁵⁵  which could signal through AKT-independent downstream pathways,⁵⁶  AKT may not be an optimal target to suppress the PI3K pathway. Direct inhibition of PI3K isoform(s), rather than AKT inhibitors, would be warranted in future clinical trials for pulmonary LCH patients.

The authors are indebted to the Department of Biomedical Services and the Advanced Imaging Facilities within Core Biotechnology Services at Leicester. The authors are also grateful to Karen Brown and Jennifer Higgins for assistance with cell-sorting experiments.

This work was supported by a Cancer Research UK programme grant (Ref. C1362/A13083). C.P. was also supported by a Royal Society-Wolfson Research Merit Award.

Contribution: T.K. conceived and designed the entire study, performed experiments, analyzed data, and prepared figures; S.G. contributed to animal management and tissue preparation; and T.K. and C.P. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Tamihiro Kamata, Department of Cancer Studies, University of Leicester, Lancaster Rd, Leicester LE1 9HN, United Kingdom; e-mail: tk83@le.ac.uk.