CRISPR-enhanced human adipocyte “browning” as cell therapy for metabolic disease

Obesity and type 2 diabetes (T2D) are associated with poor tissue responses to insulin1,2, disturbances in glucose and lipid fluxes3–5 and comorbidities including steatohepatitis6 and cardiovascular disease7,8. Despite extensive efforts at prevention and treatment9,10, diabetes afflicts over 400 million people worldwide11. Whole body metabolism is regulated by adipose tissue depots12–14, which include both lipid-storing white adipocytes and less abundant “brown” and “brite/beige” adipocytes that express thermogenic uncoupling protein UCP1 and secrete factors favorable to metabolic health15–18. Application of clustered regularly interspaced short palindromic repeats (CRISPR) gene editing19,20 to enhance “browning” of white adipose tissue is an attractive therapeutic approach to T2D. However, the problems of cell-selective delivery, immunogenicity of CRISPR reagents and long term stability of the modified adipocytes are formidable. To overcome these issues, we developed methods that deliver complexes of SpyCas9 protein and sgRNA ex vivo to disrupt the thermogenesis suppressor gene NRIP121,22 with near 100% efficiency in human or mouse adipocytes. NRIP1 gene disruption at discrete loci strongly ablated NRIP1 protein and upregulated expression of UCP1 and beneficial secreted factors, while residual Cas9 protein and sgRNA were rapidly degraded. Implantation of the CRISPR-enhanced human or mouse brown-like adipocytes into high fat diet fed mice decreased adiposity and liver triglycerides while enhancing glucose tolerance compared to mice implanted with unmodified adipocytes. These findings advance a therapeutic strategy to improve metabolic homeostasis through CRISPR-based genetic modification of human adipocytes without exposure of the recipient to immunogenic Cas9 or delivery vectors.

ex vivo of immune cells taken from a human subject to enhance their ability to disrupt malignancies upon infusion back into the same subject. In theory, this strategy should be effective in diseases in which cells with relevant therapeutic potential can be genetically modified to enhance that potential. Here we take advantage of recent discoveries revealing the utility of thermogenic adipocytes to function as major beneficial regulators of whole body metabolism in such metabolic diseases as type 2 diabetes and obesity [15][16][17][18] . Thermogenic adipocytes, denoted as brown 27 , beige 28 or brite 27,29 , are distinct from the more abundant lipid storing white adipocytes not only by their high oxidative capacity and expression of mitochondrial uncoupling protein (UCP1) but also by their secretion of factors that enhance energy metabolism and energy expenditure [15][16][17][18] . Multiple studies have demonstrated that implantation of mouse brown adipose tissue into obese, glucose intolerant mice can improve glucose tolerance and insulin sensitivity [30][31][32] . Recently, human beige adipocytes expanded from small samples of subcutaneous adipose tissue were shown to form robust thermogenic adipose tissue depots upon implantation into immune-compromised obese mice and to lower blood glucose levels 33 . Collectively, these data provide the framework to apply genetic modifications to adipocytes to further improve their therapeutic potential.

SpyCas9/sgRNA RNPs for ex vivo gene editing
In order to enhance the therapeutic potential of adipocytes in obesity and diabetes, we initially targeted the mouse Nrip1 gene. Nrip1 had been previously shown to strongly suppress glucose transport, fatty acid oxidation, mitochondrial respiration, uncoupling protein 1 (UCP1) expression as well as secretion of such metabolically beneficial factors including neuroregulin 4 21,22,34 . NRIP1 functions as a transcriptional co-repressor that attenuates activity of multiple nuclear receptors involved in energy metabolism, including estrogen related receptor (ERRa), peroxisome proliferator activated receptor (PPARg) and thyroid hormone receptor (TH) 35 . NRIP1 knockout in white adipocytes upregulates genes that are highly expressed in brown adipocytes, enhancing glucose and fatty acid utilization and generating heat. Nrip1 ablation in mice elicits a lean phenotype under high fat diet conditions, and greatly enhances energy expenditure, glucose tolerance and insulin sensitivity 21 . However, NRIP1 is not an attractive target for conventional pharmacological intervention as it is not an enzyme and has a multiplicity of tissue specific roles such as regulating the estrogen receptor in the reproductive tract 35 .
Targeting NRIP1 selectively within adipocytes represents an ideal approach to capture its therapeutic potential without undesirable side effects.
A key aspect of our strategy in targeting the Nrip1 gene was to employ methods that would ablate its expression in adipocytes but not cause immune responses upon implantation of the cells. CRISPR-based methods based on continuous expression of Cas9/sgRNA to modify adipocytes that function in vivo have been reported 36,37 , but they expose recipients to Cas9 and delivery agents that cause immune responses. Direct administration of Cas9/sgRNA complexes in mice have not been adipocyte-specific and could cause undesirable effects in other tissues 36 . Ribonucleoprotein complexes of SpyCas9/sgRNA are desirable vehicles for such modifications since they are rapidly degraded following DNA disruption 38 . A previous attempt at delivery of such CRISPRbased complexes to adipocytes were suboptimal as efficiencies of delivery of RNPs to these cells was only modest 34 . We overcame these deficiencies by disrupting Nrip1 in mouse preadipocytes with ribonucleoprotein (RNP) complexes of Cas9 and sgRNA by modifying electroporation methods 39 described for other cell types (Extended Data Fig 1), and confirmed Cas9 protein is rapidly degraded following indel formation in preadipocytes(Extended Data Fig. 2). Electroporation conditions were developed to optimize the efficiency of Nrip1 gene targeting in mouse preadipocytes by Cas9/sgRNA RNPs without perturbing their differentiation into adipocytes (Fig. 1a,b and Extended Data Fig. 1).
The efficiencies of indel formation by 7 different sgRNAs against various regions of Nrip1 gene exon 4 ( Fig. 1a) were uniformly sustained in the 90% range in preadipocytes and upon their differentiation into adipocytes (Fig. 1b,c). Indels were quantified by Sanger sequencing data analysis of PCR fragments spanning the upstream and downstream double stranded breaks of the Nrip1 genomic DNA (Fig. 1c,d) with little change in the total Nrip1 mRNAs (Fig 1e). High frequencies of frameshift mutations in Nrip1 by all 7 sgRNAs were found and similar indels were found in the corresponding Nrip1 mRNA species, as exemplified by sgRNA M3 and M4 (Extended Data Fig. 3).
While the mRNA of Nrip1 was equally abundant in all groups, indicating no increased degradation due to disruption (Fig. 1e), surprisingly, not all of the sgRNAs were effective in eliciting loss of the NRIP1 protein (Fig. 1f) . Consistent with these data, thermogenic responses to the various sgRNAs as reflected by elevated expression of UCP1 mRNA ( Fig. 1g) and protein (Fig. 1h) correlated with the loss of native full length NRIP1 protein.
Taken together, these data show that sgRNAs targeting the regions of Nrip1 DNA that encode the N-terminal region of the NRIP1 protein are not effective in eliminating synthesis of functional NRIP1 protein. Most likely, additional transcription or translation start sites beyond these target sites are functional under these conditions. Thus, sgRNAs that are optimal for inducing thermogenic genes must be identified by such screening methods.

Implantation of CRISPR-modfied mouse adipocytes
To test the ability of NRIP1-deficient adipocytes to improve metabolism in mice, large numbers of primary preadipocytes obtained from 2-3 week old mice were electroporated with RNPs consisting of either SpyCas9/non targeting control (NTC) sgRNA or SpyCas9/sgRNA-M6 complexes, and then differentiated into adipocytes and implantated into wild type mice. The implanted mice were kept on normal diet for 6 weeks during the development of adipose tissue depots from the injected adipocytes, then placed on a high fat diet (HFD) regimen to enhance weight gain (Fig. 2a). Adipocytes treated with Cas9/sgRNA-M6 displayed upregulation of Ucp1 and other genes highly expressed in brown adipocytes (e.g., Cidea) prior to transplantation (Extended Data Fig. 4). A transient decrease in overall body weights were detected between mice implanted with RNPs containing the Cas9/sgRNA-M6 versus the Cas9/NTCsgRNA group, but by 6 weeks of HFD no significant difference was observed (Fig. 2b). Nonetheless, implantation of NRIP1KO adipocytes prevented the increase in fasting blood glucose concentration due to HFD that occurs in the control adipocyte-implanted mice (Fig. 2c). Glucose tolerance was also significantly improved by implantation of NRIP1KO adipocytes (Fig 2d,e). The implanted adipose tissue depots retained their elevated expression of UCP1 16 weeks after implantation, at which time they were excised for analysis (Fig. 2f). The livers and inguinal white adipose tissues (iWAT) from the Cas9/sgRNA-M6 group of mice had lower weights (Fig. 2g) and lower iWAT to body weight ratios (Extended Data Fig. 4), revealing a strong systemic effect.
Livers of the NRIP1 deficient adipocyte-implanted mice were less pale (Fig. 2h), were smaller as assessed by lower liver to body weight ratios (Fig. 2i) Fig. 4) as well as liver triglyceride determination (Fig. 2m). The decrease in hepatic lipid accumulation and inflammation in response to implantation of the NRIP1 depleted adipocytes suggests that this therapeutic approach might mitigate these T2D co-morbidities in humans 6 .

Translation to human adipocytes
To translate these CRISPR-based methods to human adipocytes, adipocyte progenitors were obtrained from small samples of excised subcutaneous adipose tissue as previously described 33 . Electroporation conditions were tested to optimize efficiency of indel formation using various sgRNAs directed against regions of the NRIP1 exon 4 at locations roughly similar to those we targeted in the mouse genomic DNA (compare Fig. 1a to Fig.   3a). Efficiencies of NRIP1 gene disruption were observed with several sgRNAs in the 90% range (Fig. 3b), with indel distributions very similar in preadipocytes and adipocytes ( Fig. 3c). Electroporated, NRIP1 deficient human preadipocytes could be readily differentiated to adipocytes without apparent disruption following indel formation (Fig. 3d).
NRIP1 mRNA was equally abundant in all conditions (Fig. 3e). UCP1 expression increased by up to 100 fold in several experiments in Cas9/sgRNA treated cells, but only with certain sgRNAs (Fig. 3f,g), similar to our results with mouse adipocytes. Functional NRIP1 protein can still be expressed despite high efficiency indel formation in the N-terminal region. NRIP1 disruption in combination with the adenylate cyclase activator forskolin synergistically increased UCP1 expression (Fig. 3h).
To test the efficacy of NRIP1-depleted human adipocytes to provide metabolic benefits in obese glucose intolerant mice, we utilized immune-compromised NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ (NSG) mice that lack T cells, B cells and natural killer cells to accept human cell implants in the protocol depicted (Fig. 4a,b). NRIP1 gene disruption was about 80% with the sgRNA-H5 (Fig 4c) and circulating human adiponectin (Fig. 4d) was the same from NTC vs sgRNA-H5 groups, indicating similar levels of adipose tissue formation. NRIP1KO adipocytes exhibited upregulation of UCP1 levels similar to Fig. 3 (not illustrated). and the resulting NRIP1KO adipose tissue implants harvested from the mice 13 weeks later retained the enhanced UCP1 expression (not shown).
While no body weight difference between groups was detected on normal diet, a highly significant decrease in weight gain on the HFD was observed in mice implanted with NRIP1 disrupted adipocytes (Fig. 4e). Importantly, mice with control human adipocyte implants displayed significantly decreased glucose tolerance 3 weeks after starting a HFD while animals with NRIP1-depleted adipocyte implants did not ( Fig. 4f-i). The difference in glucose tolerance between the two groups at the end of the study was highly significant (Fig. 4h,i). Relative liver to body weight ratios (Fig 4j) and liver triglycerides ( Fig.   4k) were also decreased when implants were performed with NRIP1 deficient adipocytes compared to control adipocytes.
Taken together, the data presented here show that CRISPR-based RNPs can enhance "browning" of adipocytes ex vivo at high efficiency without the use of expression vectors to improve metabolic parameters in two mouse models of obesity. Although CRISPRbased upregulation of UCP1 alone in implanted adipocytes can improve metabolism in mice 37 , targeting NRIP1 has the advantage of upregulating expression of many genes that have favorable metabolic effects in addition to UCP1. The specific approach presented here has several additional advantages, including the rapid electroporation procedure with minimal loss of viability, the fact that NRIP1 deletion does not diminish adipose differentiation, and the lack of immunogenic Cas9/sgRNA complexes in implanted cells. Also, the brief exposure of cells to Cas9/sgRNA that we document here (Extended Fig. 2) reduces the potential for off target effects that are produced by long term expression of these reagents. Since UCP1 expression in NRIP1KO adipocytes does not reach the level of mouse BAT, our approach can be improved by further enhancing adipocyte browning through disrupting combinations of targets in addition to NRIP1. Indeed we find that simultaneous delivery of multiple sgRNAs into preadipocytes each yield high efficiencies of indel formation (not shown), offering the potential for disrupting multiple thermogenic suppressor genes to achieve greater therapeutic potential. Also the use of high fidelity nucleases 40,41 can further maximize cell viability and functionality. Further experiments to define the dose of implanted cells and preferential recipient sites will also help optimize this technique as a step towards testing in larger animals and advancement towards clinical trials.

Acknowledgememts:
We wish to thank members of the Czech and Corvera laboratories for helpful discussions and Kerri Miller for excellent assistance in preparing the manuscript. We thank the University of Massachusetts Morphology Core Facility for assistance in the histological preparations, stains and analysis.

Animals and Diets
All animal work was approved by the University of Massachusetts Medical School Institutional Animal Care Use Committee with adherence to the laws of the United States and regulations of the Department of Agriculture. Mice were housed at 20-22 °C on a 12hour light/12-hour dark cycle with ad libitum access to food and water. C57BL/6J male mice were purchased from Jackson Laboratory for implant studies. C57BL/6J (Jackson Laboratory) male mice were bred for primary preadipocyte cultures. Briefly, 10-week old male mice arrived and were allowed to acclimate for a week prior to any procedures. Mice were implanted with edited primary mouse adipocytes at 11 weeks of age by anesthetizing prior to the implantation procedure using an anesthesia vaporizer chamber with a continuous flow 500 cc/minute of 02 with isoflurane 3% for induction and 1.5% for maintenance. After the cell injections, animals are allowed to wake up and were placed back in clean cages. Mice were maintained on a chow diet for the first 6 weeks, followed by a 60 kcal% high fat diet (Research Diets, D12492i) for the remainder of the experiment from 6-16 weeks post implant. Glucose tolerance tests were performed after 16-hour overnight fasting with intraperitoneal injection of 1g/kg D(+) glucose. Insulin tolerance tests were performed with 0.75IU/kg after 6-hour daytime fasting. Male NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ (denoted as NSG) mice were kindly donated by Taconic Biosciences, Inc. NSG mice were implanted with edited primary human adipocytes at 11 weeks of age. Mice were maintained on a chow diet for the first 10 weeks, followed by placing them at thermoneutral with a 60 kcal% high fat diet (Research Diets, D12492i) for the remainder of the experiment from 10 to 15 weeks post implant. Housing under thermoneutrality was achieved by placing the NSG mice at 30°C on a 12-hour light/12hour dark cycle. Glucose tolerance tests with NSG mice were performed after a 16-hour fast with intraperitoneal injection 2 g/kg D(+) glucose. Whole blood was drawn and placed in EDTA-containing tubes from living mice with submandibular vein punctures under anesthesia as described above and in the end of the study with cardiac puncture. Plasma was extracted with centrifugation of whole blood for 15 minutes at 300 rcf at 4 o C.

Human Subjects
Abdominal subcutaneous adipose tissue was obtained from discarded tissue following panniculectomy. All subjects consented to the use of tissue and all procedures were approved by the University of Massachusetts Institutional Review Board.

Primary Mouse Preadipocyte Isolation, Culture and Differentiation to Primary
Adipocytes 2 to 3 week old C57BL/6J male mice were euthanized and inguinal fat tissue was harvested (including lymph node) and placed in HBSS buffer (Gibco #14025) plus 3% (w/v) bovine serum albumin (BSA) (American Bioanalytical). The protocol was carried out as described previously 34 with the following modifications; cells were incubated in 2 mg/mL collagenase (Sigma #C6885) in HBSS BSA 3% (w/v) for 20 minutes to digest the tissue. Cells were cultured to sub-confluence in complete media containing DMEM/F12 media (Gibco #11330) , 1% (v/v) Penicillin/streptomycin, 10% (v/v) Fetal bovine serum (FBS) (Atlanta Biologicals #S11550), 100 μg/mL Normocin (Invivogen #Ant-nr-1) at which time they were transfected with RNPs by electroporation and re-plated. Αdipocyte differentiation was induced in the edited cells 24 hours post confluence previously described 34 . Cells grown post differentiation induction were cultured in complete media.

Primary Human Preadipocyte Isolation, Culture and Differentiation to Primary Adipocytes
Explants from human abdominal subcutaneous adipose tissue from individuals undergoing panniculectomy surgery were embedded in Matrigel and cultured in as previously described 33,42 . Human adipocyte progenitors were transfected with RNPs by electroporation and plated at a density greater than 70% confluence to allow for expansion. Cells were grown to confluence then adipogenic differentiation media was added to induce adipogenesis 33,42 . On day 10 post differentiation, cells were harvested for implantation in NSG mice by treating with 0.5 mg/mL collagenase in 1x Trypsin to detach from culture plates.

Transfection of Primary Preadipocytes (Mouse and Human) with RNPs
For ribonucleoprotein (RNP) transfection, we used the Neon Transfection System 100 μL Kit (ThermoFisher, #MPK10096) and we prepared a mix consisting unless otherwise specified of sgRNA 40 pmol (Synthego or IDT DNA) purified SpyCas9 protein 30 pmol (PNA Bio, #CP02 or 3xNLS-SpCas9 43 (prepared by the Scot Wolfe laboratory) in Buffer R provided in the Neon Transfection System Kit. The cells were resuspended in Resuspension Buffer R for a final number of 0.5-6 x10 6 cells per electroporation. For delivering the RNP complex into primary pre-adipocytes the electroporation parameters used were voltage 1350 V, width of pulse 30 ms; number of pulses 1 unless otherwise specified. The electroporated cells were placed in complete media immediately following transfection, expanded, grown to confluence and differentiated into mature adipocytes for downstream applications. We found these methods improved the viability of preadipocytes and adipocytes and their ability to differentiate over methods reported while our manuscript was in preparation 44 .

Implantation of Primary Mouse and Human Adipocytes
Primary mouse and human mature adipocytes on day 6 and 10 post differentiation respectively were washed twice with 1xPBS. 0.5 mg/mL collagenase in 1 x trypsin was used to dissociate the cells from the plate. The detached cells are pelleted at 300 rcf for 10 minutes at room temperature. The cells were washed with 1xPBS, pelleted, and the PBS was removed. Cell pellets were kept on ice for a brief time until implantation. Each mouse adipocyte pellet deriving from 1 x 150 mm fully confluent plate was mixed with matrigel (Corning® Matrigel® Growth Factor Reduced Basement Membrane Matrix, Phenol Red-free, LDEV-free # 356231) up to a total volume of 500 μL on ice and the cell and matrigel suspension (500 ± 20 μL) was drawn into a 1 mL tuberculin syringe without the needle. The cell and Matrigel mixture was injected into the anesthetized mouse recipient with a 20 G needle by tenting the subcutaneous subscapular area, inserting the needle into the tented space and injecting at a slow but continuous rate to avoid cell rupture and solidification of the matrigel. The injection site was pinched gently for 1 minute to allow the implant to solidify, followed by withdrawing the needle with a twisting motion. Each C57BL/6J mouse recipient was injected with 2x150mm plates of fully confluent murine adipocytes split into two bilateral injections in the subscapular area. Each NSG mouse recipient received 1 x 150 mm plate split into two 500 μL bilateral subcutaneous injections in the dorsal area as described above.

DNA Harvest from cells and tissue
At two distinct time-points, 72 hours following transfection and after primary adipocyte differentiation between day 6-10 post differentiation, genomic DNA was isolated from the transfected cells using DNA QuickExtract™ Buffer (Lucigen) in adherence to the manufacturer's instructions.

Indel analysis by TIDE and ICE
Genomic DNA was PCR amplified for downstream analysis using locus specific primers designed with MacVector 17.0 and purchased from IDT DNA and Genewiz, spanning the region 800 bp around the expected DSB. For the PCR, Kappa 2x Hot start HiFi mix was used and PCR products were purified using the QIAgen DNA purification kit following the manufacturer's instructions, and submitted to Genewiz for Sanger Sequencing. Sanger sequencing trace data were analyzed with TIDE and ICE webtools (http://shinyapps.datacurators.nl/tide/, https://ice.synthego.com/#/) that decipher the composition of indels created at the sites of DSBs 45,46 .

RNA Isolation
Transfected cells were harvested for RNA between day 6-10, post-differentiation depending on the experiment by removing media and washing once with 1xPBS, and adding Trizol reagent to lyse the cells. The protocol for RNA isolation was performed according to manufacturer's instruction with the following modifications; 1μl of Glycol blue (Invitrogen #AM9516) was added to the isopropanol to precipitate the RNA and was either stored overnight at -20°C or placed on dry ice for 2 hours. The isolated RNA was resuspended in RNase free water, then treated with recombinant DNaseI (DNA-free DNA removal kit, Ambion) according to the manufacturer's instructions. RNA concentrations were determined by Nanodrop 2000.

RNA Isolation of Pulverized Tissue/Tissue Piece
Tissue was isolated from the mice and frozen in liquid N2. For RNA isolation, tissue was pulverized in liquid N2, or a piece approx. 100 mg in size was put in a 2 mL tube with screw cap and 1mL of Trizol. Tissue was placed in the Qiagen TissueLyser and homogenized for 3 cycles of 3 minutes at 30Hz. The Trizol and tissue lysate were placed in a new tube, and centrifuged for 10 minutes at 4°C to separate any lipid from the homogenate. Once the homogenate is separated from the lipid, the remaining isolation is carried out according to manufacturer's instructions.
RT-PCR 0.5-1 μg of RNA was used in 20 μL reaction with Bio-Rad iScript cDNA kit according to manufacturer's protocol to synthesize cDNA. cDNA was diluted by adding 80 μL of water to the reaction and 5 μL of cDNA template was used for RT-PCR with Bio-Rad Sybr Green Mix and gene specific primers for a final concentration of 0.3 μM primers. Expression of genes was determined by comparing gene expression levels of target gene compared to housekeeping gene 36B4 and RPL4 for murine and human samples respectively. mRNA expression was analyzed with the ΔΔCT method.

Protein Isolation
Cells grown in culture dishes were washed once with 1 x PBS at room temperature, followed by adding boiling 2% SDS (w/v) with 1 X HALT protease inhibitors and scraping to lyse the cells. Tissue pieces were prepared for western blots by homogenizing a piece approximately 100 mg in Radioimmunoprecipitation Assay (RIPA) buffer with 1x HALT protease inhibitors in the Qiagen TissueLyser and homogenized for 3 cycles of 3 minutes at 30Hz. Tissue and cell lysates prepared with 2% SDS (w/v) buffer or RIPA buffer were sonicated at 60% amplitude with a probe sonicator tip for 30 seconds at room temperature. In figure S2, mouse cells were lysed as described above at different timepoints after transfection and for timepoint 0 hours, after the electroporation the transfection mix consisting of cells and RNPs in Buffer R was centrifugated at 300 rcf. The cell pellet was lysed as described above while the supernatant (sup) was also collected for use as positive control (30pmols of SpyCas9) in the western blot. Protein concentration determination of the tissue and cell lysates was performed using a bicinchoninic acid kit (BCA Protein Assay Kit, Pierce).
Cell lysates used in immunoprecipitation reaction were prepared in non-denaturing NP-40 buffer (20mM Tris HCl pH 8.0, 137mM NaCl, 1% (v/v) Nonident P-40 (NP_40), 2mM EDT) containing 1X HALT protease inhibitors by washing once with 1 x PBS, adding NP-40 buffer and scraping, followed by a 4 °C incubation for 30 minutes to 1 hour with gentle agitation. Cell lysates were centrifuged at 4 °C for 10 minutes at 16,100 rcf and the infranate was collected and used. Protein concentrations were determined on the lysates using Pierce BCA Kit. Protein samples were prepared for running on 7.5-12 % SDS-PAGE mini gels at a final concentration of 1mg/mL protein, 1x Laemmli loading buffer (BioRad) with 2.5 %(v/v) b-Mercaptoethanol, followed by placing in a heat block at 95 °C for 10 minutes.

Triglyceride Assay
For the liver triglyceride assay, we used the Triglyceride Colorimetric assay kit (Cayman Chemical, #10010303). The lysate was prepared by mixing 50 mg of pulverized whole liver with 1.5 mL of the NP-40 lysis buffer and homogenized in the Qiagen TissueLyser with 3 cycles 3 minutes at 30 Hz. The assay ran according to manufacturer instructions with a sample dilution of 1:5.

Human Adiponectin
Human adiponectin was measured in the plasma of NSG mice was measured using a human-specific adiponectin ELISA from Invitrogen (KHP0041).

Histology
Approximately 0.5 cm 2 of the implant tissue and two 0.5 cm 2 liver pieces from two different lobes per recipient were randomly selected and fixed, followed by processing at the UMass Medical School Morphology Core. Photos of the tissues were taken with an LEICA DM 2500 LED inverted microscope at indicated magnification. Fiji/ImageJ was used to quantify lipid content in H&E images. 4 images per section, 2 sections per liver, were projected into a single montage. The montage was converted from RGB to 8 bit, contrast enhanced, thresholded and binarized. The processed montage was reconverted into individual images and lipid droplets quantified for each image using the particle analysis function (number, size, % of area covered).
Membranes were washed with TBST prior to secondary antibody incubations. HRPconjugated secondary antibodies were diluted with 5% BSA (w/v) in TBST at 1:5,000-10,000 for 45 minutes at room temperature with constant shaking. Membranes were washed in TBST, followed by incubating with Perkin Elmer Western Lightning Enhance ECL. The Bio-Rad Chemi-Doc XRS was used to image the chemiluminescence and quantifications were performed using the system software, or Image J. Immunoprecipitation was performed with NP-40/Halt protein lysates. Briefly, 250 μg of protein lysates were pre-cleared using 50:50 Protein-A Sepharose/NP-40 buffer/1x HALT protease inhibitors for 2 hours at 4°C with end over end mixing. After 2 hours, the lysate/Protein A Sepharose was centrifuged for 5 seconds to pellet the Sepharose, and the lysate was transferred to a new tube. 5 μg of Antibody (Rabbit Non-Immune IgG, Millipore #12-370, or Rabbit anti-Nrip1, Abcam #Ab42126) was added to the lysates and they were incubated overnight at 4 °C with end over end mixing. Antibody/antigen was pulled down by adding 50:50 Protein A Sepharose/NP-40 buffer/1x HALT protease inhibitors for an additional 2 hours at 4°C with end over end mixing. Protein/Antibody/Protein A Sepharose complexes were washed by centrifuging briefly, removing the supernatant and washing the pellet with NP-40 buffer containing protease inhibitors. The captured proteins were eluted from the Sepharose by adding 40 μL of 1xLaemmli buffer containing 2.5% (v/v) β-Mercaptoethanol, vortexing the Sepharose mixture, followed by boiling at 95°C for 10 minutes. All eluted proteins were run on the gel, transferred to nitrocellulose, and immunoblotted as described above.

Plasmid construction
The pCS2-Dest plasmid with CMV promoter expressing SpyCas9 (Addgene # 69220), and sgRNA expressing plasmid (Addgene #52628) were a gift from Dr. Scot Wolfe lab. To clone NRIP1 targeting and non-targeting sgRNAs, oligo spacers with BfuAI overhangs (purchased from IDT) were annealed and cloned into the BfuAIdigested sgRNA plasmid. Lonza pmaxGFP LOT 2-00096 was used to test transfection of these plasmids in various concentrations to determine the efficient dosage range (0.5-1.5 μg) and the electroporation conditions (1350 V, 30ms, 1 pulse) for the delivery and GFP expression was evaluated with EVOS FL fluorescent microscope (Thermo Fisher Scientific).

Statistical analysis
All comparisons are between two groups and student unpaired two-tailed T-Test was performed for the p values. In data that did not follow Gaussian distribution, standardization preceded the statistical analysis. * p < 0.05, ** p < 0.01, *** p < 0.001.

Additional online tools and Software
For the mapping of exons on the Nrip1 gene we used IGV_2.5.3. For the Design of sgRNAs we used a combination of the Broad Institute sgRNA designer, CHOPCHOP and the online sgRNA checkers by Synthego and IDT. For the design of genomic DNA primers we used MacVector 17.0. For the alignment of the Sanger Sequencing traces and the human and mouse coding region we used SnapGene Viewer 5.1.6 and NCBI nucleotide blast. For the design of RT-PCR primers we used Primer Bank (https://pga.mgh.harvard.edu/primerbank/). For the prediction of off-target editing sites,