Concept of exBAT
The three steps of exBAT are: 1) harvesting of host WAT; 2) exposure of WAT fragments to browning factors via single-step culture; and 3) re-implantation of converted BAT within subcutaneous WAT (Fig. 1). While only small amounts of BAT can have a significant impact on metabolism48, large amounts of viable WAT can be obtained by plastic surgeons using well-established harvesting techniques (i.e. liposuction)49,50. Moreover, compared to traditional processes that involve sorting and purification of isolated progenitor cells followed by long periods of cell-culture expansion, the exBAT procedure is quick, because a single browning step acts on whole tissue fragments to convert WAT to BAT mass, which is ready for direct implantation.
Development and characterization of ex vivo browning process in mice
We first developed a mouse model to test whether whole WAT fragments could be converted to BAT ex vivo (Fig. 2a). We first excised a small piece (~0.5 mL) of subcutaneous WAT from the left inguinal depot, located along the rear flank of the mouse above the hindlimb. Next, we gently minced the WAT tissue into fragments of approximately 2 to 5 mm in diameter to mimic the size of fragments obtained during fat harvesting procedures in humans and suspended the fragments in either browning media or control media (i.e. basal media without browning factors). Working with whole pieces of tissue is simpler than working with individual cell populations such as adipocyte progenitors, which are frequently isolated from adipose tissue depots and require expansion in 2D culture9,16. We used a cocktail of rosiglitazone (PPARγ agonist), isobutylmethylxanthene (IBMX, phosphodiesterase inhibitor), T3 (thyroid hormone), indomethacin (COX inhibitor), CL316,243 (β3 adrenoreceptor agonist), and vascular endothelial growth factor (VEGF) (see Methods for details). This single-step cocktail was found to induce browning over a duration of 1–3 weeks ex-vivo while maintaining cell viability, compared to multi-step induction which was previously performed for browning of adipocyte progenitors51,52.
We performed live-cell staining on whole tissue fragments cultured in the presence of browning media (Fig. 2b,c). Initially, the WAT fragments displayed cytoplasmic and mitochondrial staining around large lipid droplets, as well as in branching vascular structures (Fig. 2b, left panel). After one week in browning media, tissues displayed higher cell density with smaller and increased number of lipid droplets, consistent with formation of BAT-like tissue (Fig. 2b, middle and right panels). Assessment of cell viability, which could affect the efficacy of the implant in vivo, showed that viability after 3 weeks of ex vivo browning was comparable to that of control media under the same duration (Fig. 2c).
To further test the extent to which we could achieve ex vivo browning from subcutaneous WAT fragments, we performed immunohistochemistry on samples that had been cultured in both browning and control media for 10 days. Samples cultured in BAT media exhibited smaller lipid droplets and significantly higher UCP1 staining (Fig. 2d, p < 0.001), consistent with what would be expected of native BAT.
Autologous re-implantation into mice and persistence of BAT-like phenotype in vivo
Next, we examined whether the BAT-like properties of tissues cultured in browning media would persist weeks after re-implantation. After 3 weeks of ex vivo culture, we subcutaneously implanted approximately 0.2 mL of autologous converted BAT on the right inguinal WAT depot and extracted the tissue again from the implant site after 8 weeks.
After 8 weeks of implantation, tissue fragments cultured in control media during the ex vivo process retained a WAT-like appearance (Fig. 3a). However, we observed that tissues cultured in browning media during the ex vivo period, after 8 weeks of implantation, continued to exhibit a brown color consistent with BAT which contains a high density of iron-rich mitochondria (Fig. 3a). We performed immunostaining for UCP1 (with counterstaining to label lipid droplets and cell nuclei) on both whole-tissue fragments immediately before re-implantation and on extracted tissue after 8 weeks of implantation. As negative and positive controls, native uncultured inguinal WAT and interscapular BAT tissue fragments were also stained (Fig. 3b, left panels). Tissues cultured in control media prior to re-implantation retained a WAT-like appearance (Fig. 3b, middle panels). By comparison, tissues cultured in browning media exhibited high levels of UCP1 signal, numerous small lipid droplets, and a high density of cell nuclei (Fig. 3b, right panels), both in pre-implant whole-tissue fragments and extracted tissues after 8 weeks of implantation.
Using 3D confocal imaging and segmentation of image stacks through whole-mount stained tissues (Supplementary Fig. 1 and Videos 1–4), we quantified UCP1 immunostaining intensity, UCP1 volume fraction, and lipid volume fraction before and after re-implantation (Fig. 3c). Compared to tissues in the control media, tissues cultured in the browning media exhibited significantly higher UCP1 intensity, significantly higher UCP1 volume fraction, and significantly lower lipid volume fraction, both before and after implantation (Fig. 3c). These trends were consistent with phenotypes of the converted BAT in both pre- and post-implant states, and consistent with what we observed for native BAT vs. WAT phenotypes.
Finally, within the whole-tissue fragment, we also observed functional blood vessels as indicated by red blood cell-filled vessels that were visible by eye in the grafted tissue (Fig. 3d). The intact vascular structures within the engrafted converted exBAT were also visible under epifluorescence and brightfield microscopy after 8 weeks of re-implantation (Fig. 3e).
Successful conversion of human subcutaneous WAT into exBAT
We next sought to investigate whether human subcutaneous WAT could also be converted to BAT with our single-step ex vivo browning method. For these studies, we collected excess subcutaneous WAT samples via fat harvesting from the abdominal region of patients who underwent autologous fat grafting procedures (see Materials and Methods for details). Tissues were cultured in the same media and with the same browning factors as previously described for mouse tissues. After 3 weeks of ex vivo culture, we observed that human WAT tissue fragments cultured in browning media developed significant UCP1 expression and smaller lipid droplets (Fig. 4a, top panel, Supplementary Fig. 2d–f). Tissues cultured in control media retained a WAT-like phenotype, similar to native human WAT that was not cultured (Fig. 4a, middle and bottom panels). Immunohistochemistry demonstrated significantly increased UCP1 intensity, UCP1 volume fraction, and significantly decreased lipid volume fraction in exBAT cultured in browning compared to control conditions, comparable with the expression in native tissues (Supplementary Fig. 3). We assessed mRNA expression via qPCR for a variety of genes known to be differentially expressed in native BAT relative to WAT. In the converted exBAT, UCP1 mRNA levels were preferentially upregulated, while DIO2, leptin, and PRDM16 mRNA levels were downregulated (Fig. 4b), concordant with previous reports of human tissues in vitro53. Accordingly, UCP1 protein content (Fig. 4c), and citrate synthase activity (Fig. 4d), a measure of mitochondrial metabolic activity, were significantly increased under browning conditions, indicating that human exBAT possess not only cellular and molecular but also functional characteristics of thermogenically active fat (p < 0.0001).
Allogeneic transplantation of exBAT in mice with diet-induced obesity
Finally, we performed preliminary metabolic phenotyping of exBAT transplantation in a diet-induced obesity (DIO) mouse model to mimic the potential target population in humans. Here, we used allogeneic transplantation, performed in previous DIO mouse models54,55,56,57,58, to minimize variations in stress and weight caused by the multiple surgeries of autologous re-implantation (Fig. 5a). To mimic the envisioned clinical procedure in humans, we developed a minimally invasive delivery method, injecting the converted browned tissue via a 19-gauge (1.1 mm × 25 mm) needle into the subcutaneous, lower dorsal region of recipient mice. For donor tissue, we used visceral WAT from epididymal regions of age- and sex-matched DIO mice which is elevated in obese human subjects25 and exhibits soft mechanical properties (compared to subcutaneous WAT) which made it suitable for needle injection; although epididymal WAT had previously not been as well demonstrated to convert to BAT as inguinal WAT, we found it to brown effectively using our single-step ex vivo method (Fig. 5b, Supplementary Fig. 4). As a control, we also injected visceral WAT cultured in control media, denoted “control WAT”. We injected ~0.2 mL of cultured tissue fragments, which was the yield from one donor fat pad; in this study, we tried just a single dosage to perform a preliminary assessment of the in vivo effect before a full-scale trial of therapeutic dosages and conditions in a future trial.
Throughout 17 weeks post-implantation, mice injected with control WAT and the converted BAT both exhibited weight losses after being moved to the metabolic chamber and after cold exposure, as expected. Mice receiving converted BAT exhibited lower body weight than those injected with control WAT throughout the entire 17 weeks in terms of mean weights, but no statistical significance was observed (Fig. 5c, Supplementary Fig. 5a). EchoMRI measurements of percent fat mass of the animals (Fig. 5d) were taken prior to entry into metabolic chambers, after 1 week in the chambers at room temperature, and after 1 more week in the chambers at 8 °C, and approached significant difference at 15 weeks post-implant (Fig. 5d, p = 0.056). Differences in VO2 and heat expenditure in mice 1 week following implantation were non-significant (Supplementary Fig. 5b,c). Changes in dosage or cell source could be tested in future experiments.