Summary
We report a novel cause of partial lipodystrophy associated with early B cell factor 2 (EBF2) nonsense variant (EBF2 8:26033143 C>A, c.493G>T, p.E165X) in a patient with an atypical form of partial lipodystrophy. The patient presented with progressive adipose tissue loss and metabolic deterioration at pre-pubertal age. In vitro and in vivo disease modeling demonstrates that the EBF2 variant impairs adipogenesis, causing excess accumulation of undifferentiated CD34+ cells, extracellular matrix proteins, and inflammatory myeloid cells in subcutaneous adipose tissues. Thus, this EBF2 p.E165X variant disrupts adipose tissue structure and function, leading to the development of partial lipodystrophy syndrome.
Main Text
The hallmark of lipodystrophy syndromes (LD) is loss of adipose tissue (AT) and resulting deficiency in function, leading to insulin resistance and metabolic syndrome1. These diseases can present as generalized or partial LD2. The genetic underpinnings of about 30% of generalized LD and about 50% of partial lipodystrophy (PLD) syndromes remain unsolved3. To better define the genetic causes of LD, we initiated genetic screenings of individuals affected by LD using whole-genome sequencing (WGS) and next-generation sequencing panels. With these efforts, we are encountering an increasing number of gene variants of unknown significance (VUS)4; however, defining the biological roles played by these VUS in the pathogenesis of LD remains crucial. Through these efforts, we identified a novel heterozygous nonsense variant of the early B-cell factor 2 (EBF2), encoding a helix-loop-helix transcription factor, in a patient with atypical PLD. Murine Ebf2 plays a crucial role in mesenchymal tissue development, including adipocyte differentiation, brown fat development, and bone metabolism5-10. However, the role of EBF2 in human AT development is unknown. A subset of white adipocytes of human visceral AT expresses EBF2 11, and an EBF2 single nucleotide polymorphism is associated with waist-hip ratio11,12, implicating the potential role of EBF2 in regulating human AT function.
Therefore, there was biological plausibility in investigating the impact of the newly identified nonsense EBF2 variant to determine if it played a causal role in our patient’s presentation. Here, we describe the functional impact of the observed EBF2 nonsense variant on AT development and function through in vitro and in vivo disease modeling experiments.
Methods
(Details and statistical methods are in the supplementary appendix.)
Patient studies
Our team has followed this patient since pre-pubertal age, and she has participated in research studies that allowed clinical phenotyping and tissue biopsies13,14. After informed consent was obtained, an experienced plastic surgeon obtained AT biopsies from the back of the neck, abdomen, and upper thigh regions of the patient.
DNA sequencing
WGS and data processing were performed using the Genomics Platform at the Broad Institute of MIT and Harvard.
In vitro disease modeling
3T3-L1 mouse preadipocytes and primary human preadipocytes were used for in vitro disease modeling. Multiple siRNA oligos and lentiviral shRNA constructs targeting Ebf2 were used for transient and permanent gene silencing. Wild-type and truncated variants of EBF2 were expressed using bicistronic lentiviral constructs to assess the functional difference between the wild-type and the nonsense variant of EBF2. An EBF2 reporter assay was performed to determine EBF2 transcriptional activity. Total RNA was extracted from cells and analyzed using targeted real-time qPCR and bulk RNA sequencing.
In vivo disease modeling
The University of Michigan Transgenic Mouse Core generated a knock-in mouse model by inserting the exact single nucleotide variant, Ebf2 p.E165X, using Crispr-Cas technology. The mice were backcrossed into the C57BL/6J strain for five generations before analysis.
Results
Case Report
The patient’s complex clinical presentation has been reported previously14 13. She is a young woman with PLD who first presented with progressive AT loss affecting hips and legs (Fig. 1A,B). No lipodystrophy phenotype was observed in her relatives. One of her parent, diagnosed with atypical lupus, heart disease, and type 2 diabetes, died prematurely at a young age; therefore, we do not have access to the genetic material. The patient also displayed scoliosis, hand contracture, and hypogonadotropic hypogonadism13. During follow-up, she developed progressive liver enzyme elevations and proteinuria and underwent liver and kidney biopsies. Her liver showed hepatocyte ballooning with excess lipid accumulation and fibrosis; her kidney showed fibrotic extracellular matrix (ECM) accumulation in the glomerulus and interstitial space (Supplement-Fig. 1), which was called “Alport-like pathology” by the case pathologist. A Clinical Laboratory Improvement Amendments (CLIA) certified lipodystrophy gene panel (AGPAT2, AKT2, BSCL2, CAV1, CIDEC, LMNA, PLIN1, PPARG, PTRF, TBC1D4 and ZMPSTE24)13 did not detect any known variants. Subsequently, we performed WGS on samples from the proband, mother, and two siblings. We identified a heterozygous nonsense gene variant of zinc-knuckle DNA-binding domain containing transcription factor, EBF2, which we confirmed with targeted Sanger sequencing (Fig. 1C). The heterozygous nonsense variant of EBF2 (NM_022659.4) in exon 6, leads to the premature termination of EBF2 at amino acid position 165 (EBF2 8:26033143 C>A, c.493G>T, p.E165X).
Biopsies of subcutaneous adipose tissues at three different fat depots (neck, abdomen, hip) showed increased ECM accumulation and inflammation (Fig. 1D), suggesting that disruption of tissue integrity may underlie this patient’s metabolic dysfunction.
EBF2 p.E165X variant impairs adipocyte differentiation and function
We knocked down the endogenous Ebf2 gene in 3T3-L1 preadipocytes using two siRNA oligonucleotides (Supplemental-Fig. 2A-B). Ebf2 silencing significantly impaired adipogenesis, as shown with reduced lipid accumulation and gene expression of adipocyte genes, Pparg and Fabp4 (Supplemental-Fig. 2C). To perform a rescue experiment to assess the adipogenic potential of EBF2, we used five different lentiviral shRNA constructs to permanently knockdown Ebf2 expression in 3T3-L1 cells and selected a clone (clone-4) that showed the specific suppression of Ebf2 relative to Ebf1 (Supplemental-Fig. 2D). When we reconstituted Ebf2-silenced 3T3-L1 cells with EBF2 constructs using lentiviral gene transfer, full-length EBF2 (EBF2-full) restored robust adipogenesis with enhanced expression of Pparg and Fabp4 (Fig. 2A-C); however, the restoration of adipogenesis was not observed with the nonsense EBF2 variant (p.E165X, EBF2-mut). Adipogenic potential conferred by the lentiviral EBF2 gene paralleled the transcriptional activity of EBF2 as determined by reporter assays (Fig. 2D). When unmodified 3T3-L1 cells (which maintain endogenous Ebf2 expression) (Fig. 2E) and human preadipocytes (which express EBF2) (Supplemental-Fig. 3A) were transduced with the EBF2-mut, these cells displayed significantly impaired adipogenesis, as shown by decreased lipid droplet content and reduced expression of PPARG and FABP4. These results suggest a loss of function and a potentially dominant-negative effect exerted by the truncated EBF2 nonsense variant (Fig. 2F-H and Supplemental-Fig. 3B-E). We then performed bulk RNA sequencing of human preadipocytes, transduced with the lentivirus constructs of control, EBF2-full, and EBF2-mut, and stimulated with adipogenic cocktails (Fig. 2I). The gene transfer of EBF2-full and EBF2-mut led to differential gene expression in the ECM and cytokine-cytokine receptor interactions pathways (Fig. 2J-K). The EBF2-mut repressed the expression of COL1A1, COL1A2, COL4A1, THBS1, and TNXB while increasing the expression of LAMB3, SPP1, LAMA1, ITGA10, ITGA5 (Fig. 2J). The EBF2-mut also increased the expression several cytokines and growth factors, including IL1A, IL1B, CXCL2, CCL5, IL24, and TGFB2 (Fig. 2K). Together, these results reveal that the nonsense EBF2 variant not only acts as a loss-of-function variant but partly exerts a dominant negative effect on adipogenesis and gene expression in the pathways of ECM remodeling and cytokine-cytokine receptor interaction. Consistent with these in vitro findings, the biopsied AT from our patient showed dysmorphic adipocytes surrounded by abundant ECM proteins and eosinophilic amorphous structures with excess myeloid cell infiltration (Fig. 1D).
Knock-in mouse model phenocopies pleiotropic effects of the nonsense EBF2 variant on tissue structure and function
Based on these data, we hypothesize that the nonsense EBF2 variant underlies the fibrotic and inflammatory AT damage observed in this patient. To test the hypothesis, we generated a knock-in mouse model (Fig. 3A). While we had no difficulty obtaining heterozygous (HET) mice from HET to C57BL/6J mice breeding, we could rarely obtain homozygous (HM) knock-in mice, raising concerns about perinatal lethality. A surviving female HM mouse lacked perigonadal WAT and showed rudimentary subcutaneous and inguinal WAT with dysmorphic adipocytes and extensive accumulation of eosinophilic fibrillar structures (Fig. 3B), which phenocopies the subcutaneous adipose tissues of the patient (HUM for human) displaying the limited number of adipocytes surrounded by excess ECM proteins (Fig. 3C). HET mice similarly demonstrated fibrotic tissue alteration in subcutaneous AT (Fig. 3B). Imaging cytometry (CyTOF) confirmed the deposition of type 1 collagen and the increased number of CD34+ cells and macrophages (F4/80+ cells in mice and CD68+ cells in the patient) (Fig. 3D). CyTOF unveiled the infiltration of neutrophils (Ly6G+ cells in mice and CD15+ cells in humans) and monocytes (CD11b+ cells in mice and CD11c+ cells in humans) in HM and HUM tissues, respectively (Supplemental-Fig. 4A). Immunostaining further validated these findings, showing the reduced number of PLIN1+ adipocytes and increased deposition of type 1 collagen, fibronectin, and elastin (Fig. 3E) coupled with an increased number of smooth muscle cell actin (SMA)+ cells, underscoring fibrotic phenotype of AT in the patient and KI mouse model (Supplemental-Fig. 4B).
Discussion
In this report, we present a case of PLD caused by a nonsense variant of EBF2 for the first time, adding a novel disease mechanism underlying a subtype of this condition. Although our case is a singleton, the in vitro and in vivo disease modeling provide evidence to establish the causality between the EBF2 variant and the disease presentation. Our work also highlights the importance of deep tissue phenotyping in understanding the impact of observed variants in patients with PLD whose disease etiology remains elusive. In this case, the profound disruption in tissue architecture of the patient and striking resemblance of the knock-in model were crucial to drawing the causal link.
Our interest in this variant was spurred by the previous links of EBF2 to adipogenesis, as highlighted in the introduction. Moreover, EBF2 had emerged as a marker of a specific subset of visceral adipocytes in human AT single-cell studies11. The in vivo data was intriguing in demonstrating the profound impact of the variant in cultured cells, including human adipocytes, which supported the previous reports of knock-down experiments in 3T3-L1 cells5. However, our data also suggested a dominant negative effect of this variant inhibiting rescue conferred by full-length variant in cultured cells.
In addition to the phenotypic changes observed in cultured cells, the altered gene programs were in pathways that regulated ECM integrity and local inflammation. These findings supported the observations from the AT of our patient and the KI model in which we observed excess ECM accumulation and inflammatory myeloid cell infiltration in proximity to CD34+ cells. CD34+CD31- mesenchymal stem cells differentiate into adipocytes during AT development and in obesity15,16. The increased number of CD34+ cells juxtaposed with excess ECM accumulation suggests the critical role played by EBF2 in regulating AT character and ECM homeostasis. EBF2 is a crucial transcription factor determining the fate of aging fibroblasts and osteoblasts 6,17. While we focused on adipogenesis in vitro and AT structure in vivo, the EBF2 nonsense variant may exert tissue-specific fibrosis and inflammation in various tissues where EBF2 is expressed. Indeed, previous studies of Ebf1 and Ebf2 knockout mice showed impaired development of kidneys and bones, respectively 18-20.
Our findings suggest that dysregulated ECM remodeling is coupled with aberrant cytokine expression, at least in the cultured cells, leading to activation of inflammation, which was observed in patient and KI model-derived tissues. We believe this progressive inflammation and dysregulated ECM remodeling may have played a major role in our patient’s metabolic deterioration. Indeed, the patient showed chronic insulin resistance and metabolic dysfunction-associated steatohepatitis (MASH) and later developed autoimmune diabetes and hepatitis13,14, suggesting the pathological coupling of lipodystrophy syndrome with immune dysregulation.
Although groundbreaking as a new cause of lipodystrophy, our study is limited by the observation of this rare variant in a single patient and our focus mainly on AT. However, EBF2 may have different important roles in other tissues, including bone formation and kidney podocyte differentiation5,6,17. EBF activity is also important for muscle development21,22. Our findings reveal more SMA+ cells and elastin in HM and HUM AT, indicating a need for further studies to clarify EBF2’s role in myofibroblast activity. Previously, disease links of EBF2 to Kallmann syndrome and ventral hernia have been reported23,24. Our patient had clinical findings of scoliosis, hand contractures, hypogonadotropic hypogonadism, disruption of the abdominal wall and renal mesangial expansion, and ECM accumulation, as described previously. The spectrum of clinical findings may all be consistent with EBF2 dysfunction, and future work is needed to define the precise pathophysiological abnormalities in other tissues.
In conclusion, our study identified a role for EBF2 disruption in partial lipodystrophy for the first time.
We propose that this variant leads to aborted adipogenesis, accumulation of CD34+ cells, aberrant ECM deposition, and increased inflammatory myeloid cells in the AT. Though only speculative at this stage, the alteration in EBF2 may result in a novel pathophysiological mechanism that alters the integrity and function of other tissues.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Authors contributions
MF-F conducted experiments, analyzed data, collected patient data, organized figures, and drafted the manuscript. DG performed experiments. AMR contributed to experiments and participated in result discussions. LP analyzed genome sequencing data. MO’L, HLR, and MSU offered variant interpretation guidance and reviewed manuscript drafts. AN contributed to patient data collection. PS participated in result discussions and provided experimental guidance. EAO conceived the project, assembled the study team, formed all aspects of the collaborations, discussed and gathered data, designed, supervised and oversaw patient studies, provided clinical care to the patient, secured funding, critically reviewed data analyses, and contributed to manuscript writing. THC planned laboratory experiments, curated data, performed final analysis, and contributed to manuscript writing. All authors critically reviewed and approved the manuscript. EAO and THC jointly take responsibility for the integrity of the data.
Acknowledgments
We are grateful to our patient and her mother for providing us with an important research question to pursue and for inspiring our translational pipeline. Clinical phenotyping was made possible by NIH grant 5R03DK074488 (Oral EA) and by Lipodystrophy Research Fund (Oral EA) established at the University of Michigan Medical School through philanthropic support (by Sopha Family, Baker family, Rosenblum family and White Point Foundation of Turkey). Infrastructure support has been enabled by NIH grants 2P30DK089503, 5P30DK020572, and UL1TR002240, as well as support from Caswell Diabetes Institute of Michigan Medicine. Sequencing and analysis were provided by the Broad Institute of MIT and Harvard Center for Mendelian Genomics (Broad CMG) and were funded by the NIH grant K23DK114551, National Human Genome Research Institute grants UM1HG008900 (with additional support from the National Eye Institute, and the National Heart, Lung and Blood Institute) and R01HG009141, and in part by Chan Zuckerberg Initiative grant DAF2019-199278 (https://doi.org/10.37921/236582yuakxy), an advised fund of Silicon Valley Community Foundation (funder DOI 10.13039/100014989). We thank The Vector Core of the University of Michigan (Thomas Lanigan and Roland Hilgart) for designing and preparing lentivirus constructs. We thank the Transgenic Animal Model Core of the University of Michigan (Thomas L. Saunders, Zachary T. Freeman, Elizabeth Hughes, Wanda Filipiak, Galina Gabrilina, and Honglai Zhang) for the design and production of the Ebf2 Glu165X transgenic mice.