Genetic dissection of clonal lineage relationships with hydroxytamoxifen liposomes

Targeted genetic dissection of tissues to identify precise cell populations has vast biological and therapeutic applications. Here we develop an approach, through the packaging and delivery of 4-hydroxytamoxifen liposomes (LiTMX), that enables localized induction of CreERT2 recombinase in mice. Our method permits precise, in vivo, tissue-specific clonal analysis with both spatial and temporal control. This technology is effective using mice with both specific and ubiquitous Cre drivers in a variety of tissue types, under conditions of homeostasis and post-injury repair, and is highly efficient for lineage tracing and genetic analysis. This methodology is directly and immediately applicable to the developmental biology, stem cell biology and regenerative medicine, and cancer biology fields.

T he ability to genetically isolate and trace the clonal expansion and fate of individual stem and progenitor cells has wide-ranging applications in biology and medicine. The Cre/lox recombination system permits tissue-specific labeling 1 in conditional transgenic and gene knockout animals 2 . With fusion of the ligand-binding domain of the estrogen receptor (ER) to the Cre recombinase, temporal control over induction can be achieved as recombination only occurs after tamoxifen is administered and binds to the ER 3 . However, spatial control over induction remains limited to the cell and tissue-specific expression of Cre in available driver strains. With systemic tamoxifen administration, all tissues expressing the Cre driver are induced, which can be very broad depending on the driver selected. There is a potential lack of promoter specificity as well as a possible lack of validated promoters in certain cell types. A more refined system with temporal and spatial control over CreER T2 activation is beneficial to analyze complex tissues.
Efforts have been made previously to use optical "uncaging" of CreER T2 activity (using one and two-photon photo-activation) to specifically label cells of interest 4 . However, this technique requires delivery of caged compounds as well as light-mediated activation, which can be invasive and difficult to achieve in certain tissue types. Liposome-encapsulated tamoxifen has previously been explored for topical application in the setting of skin carcinogenesis in order to avoid the systemic anti-estrogen effects 5,6 .
We devised an innovative strategy for the packaging and delivery of 4-hydroxytamoxifen to locally induce CreER T2 recombinase in transgenic mice. This methodology is efficient for cell labeling using a variety of fluorescent labeling constructs in mice with specific or ubiquitous Cre drivers in multiple different tissue types and both at homeostasis and during repair postinjury.
LiTMX application to articular cartilage. Compared with the local application of non-liposomal-packaged 4-hydroxytamoxifen, which results in non-specific cell labeling of the bone marrow compartment when applied to the articular plate in a Gli-CreER T2 ::Rosa26-mTmG mouse model (Fig. 1d, e), topical application of LiTMX application yields precise labeling of cells along the joint surface (Fig. 1f) permitting specific visualization of clonal cell populations. The Gli1 promotor labels bone cells allowing detailed study of bone homeostasis and healing/remodeling after injury. In the mTmG mouse model, all cells express membrane tdTomato (mTomato), but after CreER T2 recombination, induced cells express membrane green fluorescent protein (mGFP) (Fig. 1d), allowing the activity of specific cell types to be traced. We labeled cells with LiTMX at the articular cartilage at the time of limb injury in mice with the same construct, and were able to visualize (Fig. 1g) and quantify (Fig. 1h) the clonal expansion of cells involved in tissue repair. Rainbow mice have been described previously for lineage tracing of individual cells with more precision than prior systems permit 8 . This reporter strain contains a multicolor Cre-dependent reporter construct in the ROSA locus (Rosa26-VT2/GK3). With activation of Cre recombinase, individual cells randomly and permanently express one of four fluorescent protein (FP) colors (cytoplasmic green-eGFP, membrane red-mCherry, membrane yellow-mOrange, and membrane blue-mCerulean) (Fig. 1i), allowing clonal expansion of cells of interest (depending on the Cre driver selected) to be visualized as continuous regions of single colors 8,9 . We induced the articular cartilage with LiTMX in Actin-CreER T2 ::Rosa26-VT2/GK3 mice to achieve focal CreER T2 induction (Fig. 1j). Compared with standard systemic administration of tamoxifen in corn oil via intraperitoneal (IP) injection for CreER T2 recombination, which non-specifically labels the entire bone marrow compartment (Fig. 1k), LiTMX showed targeted labeling of tissue-resident cells along the articular surface (Fig. 1l) permitting accurate lineage tracing.
LiTMX for clonal analysis in intact skin and healing wounds. We applied LiTMX to the intact, dorsal skin of Rainbow mice, with resulting visualization of clonal cell expansion on 2D analysis ( Supplementary Fig. 1a-b). We next created dorsal, full-thickness skin wounds with splinting of the wound edges using silicone washers in the manner described by Galiano et al., 2004, which mimics human wound healing kinetics 10 , in Actin-CreER T2 ::Rosa26-VT2/GK3 Rainbow mice (Fig. 2a, b). Unlike systemic tamoxifen induction (delivered IP or via oral gavage (OG)), LiTMX only activates the Rainbow construct ( Fig. 2a) locally demonstrated by the absence of Rainbow cells in the peripheral blood (Fig. 2c). Application of LiTMX reliably activates the Rainbow construct with near equal distribution of each of the fluorescent colors across the wound bed (Fig. 2d).
With topical application of LiTMX, in vivo clonal analysis of cutaneous wound healing is visualized on 2D confocal imaging of sectioned specimens at post-operative day (POD) 9 (Fig. 2e), with the size of clones increasing at POD 12 after wounding (Fig. 2f), and further at the time of wound closure, POD 15 (Fig. 2g). The resident cells in the wound appear labeled at POD 9 when LiTMX induction is conducted at the time of wounding, but the systemic inflammatory infiltrate recruited to the wound appears green (GFP+) only indicating that those cells have not been induced (Fig. 2e, left panel, white box inset). Increased incidence of fluorescent cell labeling is seen with increased LiTMX dose (Fig. 2e, right panel).
To visualize the 3D architecture of these wound-responsive cells, we developed a whole-mount technique using the cutaneous wound model (Fig. 3a-c). Histological staining of full-thickness skin wounds from the same model at POD 15, revealed linear migration of cells involved in wound healing (Fig. 3d). Clonal expansion of tissue-resident (non-circulating) wound-responsive cells in the wound bed can be visualized in 3D on Imaris analysis (Fig. 3e). Clones of cells involved in tissue regeneration increase in size over time during the healing process (Fig. 3f). After LiTMX application, robust Rainbow construct activation is particularly apparent along the wound edges, with linear expansion of distinct clones from activated progenitor cells (Fig. 3g-POD 10) engaged in tissue regeneration, seen on whole-mount confocal imaging ( Fig. 3h-POD 14, Fig. 3i-POD 21). Clonality is maintained in the dermis, as well as in the induced epidermis using the Actin-CreER T2 ::Rosa26-VT2/GK3 Rainbow mouse mode, over the long term after 2 months of healing and tissue remodeling (Supplementary Fig. 2h-i).
LiTMX compared with standard techniques. To further validate our methodology, we produced LiTMX at various concentrations (10 mg/ml, 20 mg/ml, and 30 mg/ml) and applied these to the femoral periosteum in Gli-CreER T2 ::Rosa26-mTmG mice. Consistent local induction of the mTmG construct along the periosteal surface resulted, and increased dosage correlated with increased frequency of labeled cells ( Supplementary Fig. 2a-c). Standard techniques for CreER T2 recombinase induction, systemic administration of tamoxifen in corn oil via IP injection, or OG at both low and high doses, yield patchy patterns of cell recombination in complex tissue such as skin and connective tissue ( Supplementary Fig. 2d-g). In contrast, LiTMX application permits more precise cellular labeling of a specific cell lineage of interest ( Fig. 2e-g, Fig. 3g-i).
Clonal analysis in adipose-associated vasculogenic cells. LiTMX application is beneficial for tracing vasculogenic lineages and can be used for long-term clonal analysis. Adipose tissue engaged with vasculature in the inguinal fat pad of Rainbow mice shows highly clonal tissue at 6 months after labeling (Fig. 4a) LiTMX for lineage tracing with tissue-specific Cre drivers. We applied LiTMX to soft tissue in mice possessing more specific Cre drivers, PDGFRα-CreER T2 ::Rosa26-mTmG and αSMA-CreER T2 :: Rosa26-mTmG, which mark dermal fibroblasts and myofibroblasts, respectfully ( Fig. 4c-i). When applied to intact, dorsal skin in PDGFRα-CreER T2 ::Rosa26-mTmG, the contribution of cells expressing this surface marker were highlighted (Fig. 4d). After skin wounding, clonal expansion of this fibrogenic cell populations occurs after 14 days of healing ( Fig. 4e) and 21 days of healing (Fig. 4f). When LiTMX is applied to intact, dorsal skin in αSMA-CreER T2 ::Rosa26-mTmG mice, cells engaged with vasculature were labeled (i.e., pericytes or smooth muscle cells around blood vessels) (Fig. 4g). After wounding, expansion of these cells was seen in the wound stroma at the same time points (Fig. 4h-i).

Discussion
The described technique permits identification of clonal cell populations, which can be traced and imaged as described, isolated via laser-capture micro-dissection, or subjected to fluorescence-activated cell sorting (FACS) for subsequent molecular profiling or in vitro culture. Our methodology is highly relevant to living tissue. A specific advantage of this system includes the effectiveness of local LiTMX application to the tissue surface, which can offer distinct benefits at sites tissue delicacy. Our system has several potential limitations. Primarily, we document successful application of this technique to distinct tissue types, however, we do not know if liposomal 4-hydroxytamoxifen induction of the mTmG, Rainbow, or other constructs of interest would be effective in all areas of the body. In addition, we have carefully determined the optimal dosage and timing of tamoxifen treatment to achieve reliable Rainbow activation for the tissues examined, but again this may not be standard across all tissues and could require additional experimentation. Finally, compared with systemic tamoxifen administration, which can be provided to a mouse in a noninvasive manner (via IP or OG), delivery of LiTMX can necessitate a procedure to expose certain target cells such as the tibia or peritoneum. Compared with systemic tamoxifen administration, however, LiTMX provides the opportunity to induce specific tissues with significantly more precision and efficiency.
In summary, we have successfully generated 4-hydroxytamoxifen liposomal vesicles for targeted induction of CreER T2 -mediated recombination. This technology is efficient and reproducible in a variety of animal tissue types at various concentrations. Spatial and temporal control over Cre driver activation presents a significant advancement providing a tool for multimodal delineation of tissuespecific processes. This methodology is feasible and accurate to conduct cellular lineage tracing in vivo in dynamic tissues with great accuracy and specificity. The models discussed demonstrate a unique opportunity to locally lineage trace cells involved in tissue regeneration. This methodology is highly applicable to developmental biology, and tissue repair and regeneration research, as well as projects involving skeletal, hematologic, and cancer biology.

Methods
Animals. Actin-CreER T2 , αSMA-CreER T2 , PDGFRα-CreER T2 , and Gli1-CreER T2 mouse strains were obtained from Jackson Laboratories, Rosa26-VT2/GK3 Rainbow mice were provided as a gift from the Weissman Laboratory, Stanford University School of Medicine. All animal procedures were carried out under the guidance of the Stanford University Administrative Panel on Laboratory Animal Care (APLAC).
Tamoxifen preparation for systemic induction. Prior to preparing tamoxifen (Sigma-Aldrich), corn oil (Sigma-Aldrich) was sterilized using a 0.22 μM syringe filter (Millipore Sigma). Tamoxifen was dissolved at a concentration of 20 mg/ml, unless otherwise specified, wrapped in foil and placed in a 37°C oven for 24 h. After 24 h, the tamoxifen was vortexed at 30 s intervals to ensure the tamoxifen was completely dissolved in the corn oil. Aliquots of the tamoxifen were stored at −20°C for long-term use.
4-Hydroxytamoxifen preparation for localized induction. 4-Hydroxytamoxifen powder (Sigma-Aldrich) was resuspended in 100% methanol per the manufacturer's recommendations. A dose of 10 μg, unless otherwise specified, was applied to the site of interest topically using a pipette.
Intraperitoneal injection of tamoxifen. Prior to injection, the tamoxifen was thawed and vortexed to ensure that no precipitate was present within the solution. Mice were anesthetized with 2% inhaled isoflurane (Henry Schein, Melville, NY). Using a 1-ml syringe (BD) attached to a 25 gauge needle (McKesson), the tamoxifen was intraperitoneally administered at a dose of 20 μg, unless otherwise specified, for a course of 5 days.
Oral gavage administration of tamoxifen. Prior to injection, the tamoxifen was thawed and vortexed to ensure that no precipitate was present within the solution. Mice were scruffed by hand and held in an upright position. Using a 1-ml syringe (BD) attached to a 20 gauge curved oral gavage needle (McKesson), the tamoxifen was administered at a dose of 20 μg, unless otherwise specified, for a course of 5 days.
Generating liposome vesicles. In a 25-ml round bottom flask, 14 µmol of a 90:10 mol:mol mixture of DMPC (1,2-dimyristoyl-sn-glycero-3-phosphocholine) and cholesterol (Avanti Polar Lipids) was desiccated under a 5 kPa stream of nitrogen gas Fig. 1 Generation of LiTMX for targeted tissue labeling. a Schematic of LiTMX generation and extrusion to 100 nm diameter. b Liposomal stability as determined by DLS. Hydrodynamic diameter after filtered extrusion of gold (Au) control nanoparticles at 10-nm core size (purple) and 100-nm core size (blue), and for liposomes at 100 nm (green) and 200 nm extrusion (orange). Experiment conducted with n = 5 replicates per condition. c Stability of LiTMX at 24 h (green), maintained at 6 (red) and 12 weeks (blue) after production, using UV-Vis spectrophometric detection. Following production of LiTMX at various concentrations (10 μg-purple, 20 μg-orange, 30 μg-black, 40 μg-gray), the supernatants show uniformly minimal tamoxifen content demonstrating near complete uptake by the liposomes. d Cre-driven recombination resulting in GFP expression in the bone marrow compartment. e Local application of non-liposomal 4-hydroxytamoxifen (10 μg) to the femoral periosteum (white dotted lines indicate the superficial periosteum and femur cortical bone border, left panels-white indicates DAPI, right panels-green indicates GFP). f In comparison to (e), with application of LiTMX at the same dose (individual GFP+ cells indicated by yellow arrowheads). g Twenty-one days after limb injury, clonal expansion of labeled cells involved in healing occurs along the bone periosteal surface. h The average size of GFP+ clones from g increases with time post-injury (UI = uninjured, n = 5 replicates per timepoint, error bars indicate standard deviation, SD). for 5 min, followed by vacuum desiccation at 25°C for 3 h. Then 935 µl of 1 × PBS (Fisher BioReagents, 10 × stock solution-diluted to 1 × with DI water) was added to the desiccated lipids, followed by sonication (Branson) in a water bath for 15 s. The lipids were then reconstituted by extruding 35 times through a 100-nm pore-diameter polycarbonate membrane (Sigma-Aldrich) at 32°C.
Characterizing liposome size. The liposomes were spun down at 30,230 × g (13,000 r.p.m.) for 10 min using a micro benchtop centrifuge. The liposomes were resuspended by vortexing in PBS. Next, using a NanoBrook Omni dynamic light scattering (DLS) instrument (Brookhaven), DLS measurement was performed using a 10 s acquisition time at 37°C (with the laser power adjusted to maintain the intensity between 500,000 and 2,000,000 counts). The radii, size distribution, as well as the poly dispersity index (PDI) of the liposomes were calculated using the regularization algorithm provided by the associated software. We found that the 100-nm extrusion method yielded homogeneous liposomes with a mean diameter of 151.2 ± 1.9 nm, while extrusion through a 200-nm membrane yielded a mean diameter of 242.1 ± 2.07 nm.
Preparing LiTMX. The liposome vesicles were incubated with 4-hydroxytamoxifen (Sigma-Aldrich) in a ratio of 10 mg to 1 ml (10 μg; based on wet mass after centrifugation), 20 mg to 1 ml (20 μg), 30 mg to 1 ml (30 μg), or 40 mg to 1 ml (40 μg) under nitrogen gas at 25°C for 6 h. The supernatant was decanted and the liposomal pellets were resuspended with an equal volume of 1 × PBS and stored under nitrogen gas at 4°C. The loading capacity and efficiency of the 4-hydroxytamoxifen in the liposome vesicles was calculated, as well as the retention profile for 4-hydroxytamoxifen based on the different ratios. Optimal concentration of LiTMX was determined in different tissue compartments of different mice strains prior to performing the experiments. Topical application of LiTMX to unwounded mouse skin. Adult mTmG (Actin-CreER T2 ::Rosa26-mTmG, αSMA-CreER T2 ::Rosa26-mTmG, and PDGFRα-CreER T2 :: Rosa26-mTmG) and Rainbow (Actin-CreER T2 ::Rosa26-VT2/GK3) mice were anesthetized with 2% inhaled Isoflurane (Henry Schein, Melville, NY) and 30 mg/ kg Buprenorphine SR (Reckitt Benckiser Pharmaceuticals Inc., Richmond, VA) was injected subcutaneously for pain control. Hair on the dorsal skin was removed with hair clippers followed by depilatory cream (Nair™, Church & Dwight Co., Ewing, NJ). Silicone, 12-mm diameter washers (Invitrogen) were secured around the perimeter of the regions of interest with cyanoacrylate glue and interrupted 6-0 nylon sutures (Ethilon®, Inc., Somerville, NJ, purchased from eSutures.com, Mokena, IL) to adhere the washers to the skin. LiTMX (20 μg, unless otherwise specified) was delivered topically to each of the 4 quadrants (1 μl per quadrant) at the inner diameter of the ring using a 5 μl Hamilton syringe (Hamilton Company, Reno, NV), for a course of 2 days. After 10 min, a sterile Tegaderm dressing (3M Healthcare, St Paul, MN) was applied to the area to protect the skin within the ring. Animals were euthanized and intact skin was harvested 1 week after induction.
Application to splinted full-thickness excisional wounds. Adult mTmG (Actin-CreER T2 ::Rosa26-mTmG, αSMA-CreER T2 ::Rosa26-mTmG, and PDGFRα-CreER T2 :: Rosa26-mTmG) and Rainbow (Actin-CreER T2 ::Rosa26-VT2/GK3) mice were anesthetized and hair on the dorsal skin was removed as previously described 9 . The skin was prepared for aseptic surgery with alternating alcohol and Betadyne prep (three times each) to the site followed by placement of a sterile drape. A 6-mm fullthickness wound was created extending through the panniculus carnosus on the dorsum of the mice. A silicone washer was secured around the perimeter of the wound and LiTMX (20 μg, unless otherwise specified) was delivered topically to each of 4 quadrants in the wound followed by placement of a sterile dressing, for a course of 2 days. The dressing was changed every other day until the wound was closed (defined as the time at which the wound bed was completely reepithelialized and filled with new tissue). Digital photographs were taken at the time of surgery and every other day at the time of dressing changes until the wound was closed. Animals were euthanized and the wound and surrounding skin was harvested at various time points after induction.
Application to the inguinal fat pad. Rainbow (Actin-CreER T2 ::Rosa26-VT2/GK3) mice were placed under anesthesia and analgesia as described and placed in the supine position. A small incision was made in the lower abdominal wall and extended through the peritoneum, the inguinal fat pad was accessed and 2 μl LiTMX (20 μg, unless otherwise specified) was delivered topically to the fat pad bilaterally, single dose only. The peritoneum was closed with running 4-0 monocryl and horizontal mattress suture using 6-0 nylon was used to close the skin. Application to long bone. Gli1-CreER T2 ::Rosa26-mTmG and Rainbow (Actin-CreER T2 ::Rosa26-VT2/GK3) mice were placed under anesthesia and analgesia as described and placed in the supine position. A longitudinal incision was made in the skin parallel to the long bone, subcutaneous fat and muscles were divided parallel to muscle fibers. Dose of 10 μg, unless otherwise specified, of 4-hydroxytamoxifen or LiTMX was applied to the surface of the periosteum using a 2.0-μl Hamilton Modified Microliter Syringe (Sigma-Aldrich) at a recorded distance from the tibial tuberosity, taking care not to damage the periosteum, single dose only. Skin was subsequently closed and the mice monitored postoperatively according to protocol. After a 1-week period, the mice received skeletal injuries as specified.
Results of LiTMX assays were tested and confirmed in a minimum of three independent experiments. Application to uninjured articular cartilage of the knee. Rainbow (Actin-CreER T2 ::Rosa26-VT2/GK3) mice were anesthetized, a 3-mm longitudinal incision over the distal patella extending to the proximal tibial plateau was made. The joint capsule, medial to the patellar tendon was incised. A blunt dissection of the fat pad over the intercondylar area was made, exposing the medial meniscus. LiTMX (20 μg, unless otherwise specified) was topically applied to the medial condyle of the tibial plateau using a 5-μl Hamilton syringe, for a single dose only. The joint capsule was closed with a continuous 8-0 Vicryl® suture with a tapered needle and the subcutaneous layer with 7-0 Vicryl® suture with a cutting needle. The skin was closed the with interrupted 6-0 nylon sutures. All experiments were conducted with n = 3 mice per condition (2 wounds per mouse where applicable). All animal procedures were carried out under the guidance of the Stanford University Administrative Panel on Laboratory Animal Care (APLAC).
Tissue processing. For fixation, the samples were placed in 2% paraformaldehyde for 12 h at 4°C, followed by 3 times 1× PBS washes for 30 min each. At this point, cartilage and bone specimens were decalcified in EDTA for 2 weeks. For 2D sectioning, the samples were then soaked in 30% sucrose in 1× PBS at 4°C for 24 h. The samples were then incubated in Tissue Tek O.C.T for 24 h at 4°C followed by freezing on dry ice. The frozen blocks were mounted on a MicroM HM550 cryostat, sectioned at 10 microns thick and transferred to Superfrost/Plus adhesive slides. For 3D imaging, after fixation, whole tissues specimens were mounted on glass slides in Southern Biotech mounting media.
Imaging. Using a Leica WLL TCS SP8 Confocal Laser Scanning Microscope, we imaged the 10 micron cryosections and whole tissue mounts with ×20, ×40, and ×63 oil objectives. Precise excitation and hybrid detection of mCerulean, eGFP, mOrange, and mCherry was captured and documented. Z-stacked confocal images were rendered into 3-dimensions for further analysis.
Data analysis. ImageJ software was used for analysis of imaging data. Flowjo software was used for FACS analysis of the FACS data. Quantitative analysis was conducted using Graph Pad Prism 6.
Data availability. All relevant data are available from the authors.