Establishment of Clonal MIN-O Transplant Lines for Molecular Imaging via Lentiviral Transduction & In Vitro Culture

As the field of molecular imaging evolves and increasingly is asked to fill the discovery and validation space between basic science and clinical applications, careful consideration should be given to the models in which studies are conducted. The MIN-O mouse model series is an established in vivo model of human mammary precancer ductal carcinoma in situ with progression to invasive carcinoma. This series of transplant lines is propagated in vivo and experiments utilizing this model can be completed in non-engineered immune intact FVB/n wild type mice thereby modeling the tumor microenvironment with biological relevance superior to traditional tumor cell xenografts. Unfortunately, the same qualities that make this and many other transplant lines more biologically relevant than standard cell lines for molecular imaging studies present a significant obstacle as somatic genetic re-engineering modifications common to many imaging applications can be technically challenging. Here, we describe a protocol for the efficient lentiviral transduction of cell slurries derived from precancerous MIN-O lesions, in vitro culture of “MIN-O-spheres” derived from single cell clones, and the subsequent transplantation of these spheres to produce transduced sublines suitable for optical imaging applications. These lines retain the physiologic and pathologic properties, including multilineage differentiation, and complex microanatomic interaction with the host stroma characteristic of the MIN-O model. We also present the in vivo imaging and immunohistochemical analysis of serial transplantation of one such subline and detail the progressive multifocal loss of the transgene in successive generations.


Introduction
Ductal Carcinoma in situ (DCIS) is the term given to those cancers which demonstrate marked atypia, but are non-invasive. While DCIS without associated invasive breast cancer (IBC) is rarely detected as a palpable lump, it is estimated to account for approximately 20-30% of those cancers found in mammography screening programs [1,2]. It can be expected that as mammography screening in developing countries becomes more common, so too will the diagnosis of DCIS. While DCIS itself is not generally life-threatening, it is associated with an increased risk of IBC. Conventional wisdom regarding the progression of breast cancer from atypical ductal hyperplasia to DCIS to IBC has recently come into question [3]. However, studies examining the relationship between DCIS, IBC and the molecular pathways that link them, while most definitely needed to enhance treatment of patients showing DCIS, remain underrepresented in the literature.
In order to address this deficiency, we have developed and extensively characterized a mouse mammary intraepithelial neoplasia (MIN) that accurately models human DCIS and progresses to invasive carcinoma. The mouse MIN and the outgrowths derived from their transplantation (MIN-O) satisfy the classic operational definition of premalignancy [4,5]. Each of the six MIN-O lines that we developed meets the following transplantation criteria: grows in gland-cleared fat pad (orthotopic); does not grow in the subcutis (ectopic); does not senesce over many generations of transplantation; and consistently transforms to a phenotype characterized by an ability to grow in the subcutis (ectopic). The biological transplantability and reproducibility of MIN-O permit accurate experiment-based in vivo/ex vivo correlations [6,7,8] that are not possible in existing xenograft models or in mouse models based on whole mammary glands. These transplantable precancerous MIN-O lines are clonal and have unique morphological and biological characteristics modeling human tissue heterogeneity. Their biological predictability has allowed detailed biological and molecular analyses of transitions from normal to premalignant to malignant and to metastatic states [6,8,9,10,11]. MIN outgrowths (MIN-O) are easily expandable and amenable to experimental manipulation and the in vivo mammary fat pad provides an ideal microenvironment in which to study the complexities of tumor progression. In addition to transplanting and expanding small pieces of MIN tissue for outgrowth, we have shown that MIN-O can arise in a glandcleared fat pad from a transplanted multicellular sphere or ''MIN-O-sphere'' derived in culture from a single cell previously isolated from MIN-O donor tissue [12]. To summarize, the MIN-O model, in contrast to more commonly used xenograft and cell line based models, allows experiments to be carried out in immune intact mice with the transition from DCIS to IBC occurring in a microenvironment that more closely recapitulates normal disease progression.
The factors which make the MIN-O model such an attractive model of DCIS unfortunately present challenges in the area of transgene expression. As such, there has been limited utility of this model within the field of molecular imaging where expression of imaging reporter genes can be exploited to track cells and study regulation of gene expression in vivo. In vivo imaging studies with the MIN-O model have to this point been limited to 2-deoxy-2-[ 18 F]fluoro-D-glucose ( 18 FDG) studies that track progression based on glucose uptake [13,14]. In this manuscript, we present a protocol for the efficient lentiviral transduction and in vitro culture of ''MIN-O-spheres'' and the successful transplantation of these spheres which ultimately results in propagation of transgenic MIN-O sublines that retain the in vivo properties of this valuable model. The primary goal of this work was to incorporate imaging reporter genes into the MIN-O sublines, thus expanding the reach of MIN-O based molecular imaging studies into areas which uncover the mechanisms and molecular events associated with the progression of DCIS to IBC; however, in a broader sense it is a method that can be employed to rapidly establish clonal transgenic lines derived from a wide range of primary cells.

Ethics Statement
All animal studies were completed in accordance with National Institutes of Health and Institutional Animal Care and Use Committee (IACUC) guidelines. The protocol was approved by the IACUC Committee at the University of California, Davis, under animal protocol #15927.

Emerald GFP Expression Plasmid
The lentiviral plasmid backbone used for mammalian delivery in this study is a modified form of pLenti6/TR (Invitrogen). Briefly, the sequence containing the cytomegalovirus (CMV) immediate early promoter, b-Globin Intron II and tetracycline receptor was excised and replaced with a customized multiple cloning sequence to allow for subcloning of various promoter/ reporter combinations. The EF1a promoter was PCR amplified and subcloned into the pLenti6 vector using MluI & SalI restriction sites to create pLenti6/EF1a. Next, emerald green fluorescent protein (EmGFP; 487 nm/509 nm) was PCR amplified from pcDNA6.2/EmGFP-miR (Invitrogen) and subcloned into pLenti6/EF1a using CpoI restriction sites to create pLenti6/ EF1a/EmGFP.

Preparation of Concentrated Lentiviral Particles
293T cells were obtained from ATCC and cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin/L-Glutamine (complete DMEM) and maintained at 37uC in the presence of 5% CO 2 in air. 293T cells were seeded to 15 cm culture dishes and, at approximately 75% confluency, were cotransfected with pLenti6/EF1a/EmGFP and the pLP1, pLP2 and pVSV-G series of viral packaging plasmids (Invitrogen) in a 1:1:1:1 mass ratio. Transfected cells were cultured 48-hours (with a single medium change at 24-hours post-transfection) and supernatant containing lentiviral particles was harvested and filtered through a 0.45 mM acrodisc (PALL Corporation, 4614) to remove cell debris and residual 293T cells. Lentiviral particles were then concentrated 100X by centrifugation in 100 kDa Amicon Ultra-15 centrifugal filter units (Millipore, UFC910024) and used immediately.

Excision of Precancer Lesion and Preparation of Single Cell Slurry
Excision of all lesions and preparation of single cell slurries were completed as described previously [12]. Briefly, animals were anesthetized with nembitol (60 mg/kg) and opened ventrally to allow for manual removal of precancer lesions. Fatty tissue was trimmed away and the lesion was washed twice with phosphate buffered saline (PBS) and manually minced. The minced lesion was incubated overnight at room temperature in DMEM/F12 media supplemented with BSA fraction V (2%v/v), 5 mg/mL insulin, 0.5 mg/mL hydrocortisone and 3 mg/mL collagenase with constant rocking. The following day, digested cells were subjected to a series of washes, brief trypsinization (0.25 mg/mL trypsin in the presence of 5 mL DNaseI) and were finally resuspended in mammary epithelial growth medium [MEGM, mammary epithelial basal medium supplemented with B27 supplement (5%v/v), epidermal growth factor (20 ng/mL), basic fibroblast growth factor (20 ng/mL) and Heparin (100 mg/mL)] at a concentration of 3.75610 5 cells/mL.

Lentiviral Transduction
In a 1.5 mL microcentrifuge tube, 100 mL of the cell slurry (3.75610 4 cells) were mixed with 150 mL concentrated lentivirus. This mix was supplemented with 6 mg/mL polybrene (Chemcon, TR-1003-G) and either centrifuged (30 m/25006g/37uC) or rocked gently for 30 minutes at RT. Cells were then chilled on ice for at least five minutes.

Plating and Culture of Cells on MatriGel Layer
During transduction of isolated cells, 200 mL of a 2:1 mix of Growth Factor Reduced MatriGel (BD Pharmingen) and MEGM was applied to wells of a 24-well plate which was then incubated at 37uC for 30 minutes to allow for polymerization of the gel matrix. Following polymerization, 3.5 mL MEGM was mixed with the cell slurry and this mix was applied to the 24-well plate at 1 mL (1610 4 cells) per well. Medium was refreshed weekly throughout the period of in vitro culture.

Plating and Culture of Cells within MatriGel Layer
MatriGel (500 mL) was gently mixed with transduced cells and aliquots were applied to either 24-well plates (200 mL; 1610 4 cells) or 96-well thin wall PCR plates (10 mL; ,500 cells). The plates were then incubated at 37uC for 30 minutes to allow for polymerization of the gel matrix. Following polymerization, each well was topped with MEGM (1 mL and 50 mL for the 24-and 96-well plates, respectively) and returned to the incubator. Medium was refreshed weekly throughout the period of in vitro culture.

Identification and Transplantation of Transduced Spheres
Individual spheres were counted and assayed for EmGFP expression using an IX81 inverted fluorescent microscope (Olympus) and images were acquired using a DP72 12.8 MP digital camera (Olympus). Spheres from 24-well plates were identified and transplanted manually with forceps. Spheres from 96-well plates were selected for transplant based both on transgene expression and the presence of only a single sphere per well. This allowed for transplantation of the entire well via syringe following gentle mixing of the entire contents of the well (MEGM and MatriGel). All transplants were made to pre-cleared 4 th mammary fat pads.

In vivo Imaging
At selected intervals, animals were anesthetized with 2% isoflurane and depilatory cream was applied to the general region surrounding the MIN-O lesion. Depilatory cream was removed and animals were placed into the Maestro 2 (Caliper Life Sciences) fluorescent imaging system. For each animal, a monochrome image was first acquired and then a second fluorescent image ''cube'' was acquired using a blue excitation filter (430-480 nm) and image acquisitions running from 500 nm to 720 nm at 10 nm intervals. This image cube was spectrally unmixed using the accompanying Maestro 2 software. Images presented here are comprised of the unmixed composite image overlaid on the corresponding monochrome image using ImageJ software [15].

Quantitative Real Time PCR
EmGFP negative and positive regions were identified by IHC within excised lesions and isolated via laser capture. Genomic DNA was purified from isolated tissue using the DNAeasy purification kit (Qiagen) and subjected to quantitative real time PCR (qPCR) analysis on an iCycler IQ5 (BioRad) using the following primer pairs: EmGFP Mean Ct values were obtained and subjected to PFAFFL scoring [16] with GAPDH serving as the reference gene.

Transduction and in vitro Culture of Isolated Single Cells
In order to optimize the in vitro portion of our protocol, two methods of culture were utilized. In the first method, spheres were cultured on a layer of MatriGel that was allowed to polymerize in 24-well plates prior to addition of the transduced cell slurry. In the second method, transduced cells were added to MatriGel prior to plating which allowed for sphere culture within the MatriGel layer. In vitro culture of spheres using the second method was completed in 24-well plates and 96-well thin wall PCR plates with similar results relative to sphere formation and transduction efficiency.
Approximately four days following plating, early sphere development was evident regardless of culture method. Also at this time, the presence or absence of EmGFP was readily detected in transduced spheres. Although transduction efficiency was rarely less than 40-50%, this figure varied significantly over the course of our studies with a maximum of ,80% achieved with no significant differences observed between samples subjected to ''spinduction'' or gentle rocking at RT. Approximately one in 650 isolated single cells formed a sphere in vitro when embedded within the MatriGel layer. This is a marked improvement when compared to culture on top of MatriGel which yielded one sphere per approximately 3000 cells. In addition to the increase in sphere formation, sphere morphology was also heavily influenced by culture conditions. Although culture both on top of and within the MatriGel layer yielded multiple sphere types, the relative abundance of various forms was highly dependent upon culture conditions. Following culture on MatriGel, most spheres (.90%) maintained spherical integrity and had what appeared to be a solid core (Figure 1a). However, spheres cultured within MatriGel most frequently displayed a hollow morphology with single sites of irregularity (Figure 1b & 1c). The solid core morphology was never observed in spheres cultured within MatriGel. Additionally, sphere morphology appears to predict growth in vivo. The frequency of lesions following transplantation of hollow spheres is typically between 40-80% with variance likely due to the technical challenge of transplantation. Conversely, spheres displaying the solid core morphology have not produced lesions in vivo following transplantation (n = 20).

Transplantation of Single Spheres into FVB Mice
The results presented in the remainder of this manuscript will focus on spheres cultured within the MatriGel layer in 96-well PCR plates. In our experience, this method is the most efficient and least technically challenging of the three methods we have employed. As mentioned previously, in vitro culture on MatriGel has not been a successful method of generating spheres suitable for transplantation and in vivo propagation. Although we have successfully transplanted spheres cultured within MatriGel in 24well plates, that method has proven to be highly inefficient due to the technical difficulty of excising only a single sphere and correctly transplanting it into the pre-cleared fat pad.

First Generation MIN-O-D/EmGFP Transplant Line
Following 14 days of in vitro culture, the entire contents of five wells (MatriGel and MEBM; ,40 mL), each identified to have a single EmGFP positive sphere, were individually transplanted via syringe into pre-cleared 4 th mammary fat pads of three FVB mice (bilateral transplants in two mice and a single transplant in the third). By four weeks post-transplant, all five transplants produced lesions; however, the EmGFP signal generated by these lesions was highly variable and in some instances not detectable in vivo. Of the five lesions, one in particular displayed a high level of EmGFP signaling in vivo (Figures 2a & 2b) and was selected for in vivo passage and maintenance. In addition to EmGFP expression, preservation of the characteristics of the MIN-O model was vital. In order to examine this, a portion of the excised lesion was reserved for histologic analysis and immunohistochemistry (IHC). The histology shows MINO morphology characterized by a leading edge of growth into the fat pad stroma, differentiation with basal and luminal epithelial layers in cords and ductules surrounded by remodeled host stroma and increased fat pad vascularity. Progression to central areas of dedifferentiation and proliferation associated with the invasive carcinoma phenotype was also observed. In sum, the morphology and progression was identical to the parent precancer lesion (Figures 3a-d).

In vivo Passage of MIN-O/EmGFP
As the objective of this study is the maintenance of an EmGFP positive subline in vivo, a portion of the lesion was disaggregated and a cell slurry was prepared using the remaining portion of the lesion and injected into the pre-cleared fat pad of another FVB mouse. This cycle of excision, dissociation and injection was repeated three times to generate second, third and fourth generation mice. As with the first generation, a portion of each lesion was reserved for IHC analysis.

Second Generation MIN-O-D/EmGFP
At 28 days post-injection of cells isolated from the first generation lesion, the resulting second generation lesion was imaged in vivo. As expected, the lesion retained EmGFP expression which was readily detectable (Figure 4a). However, IHC analysis showed that, unlike the first generation lesion, there were small regions of EmGFP negative cells (Figure 4b). These regions included areas in which EmGFP expression was nearly completely lost ( Figure 4c) and heterogeneous areas in which EmGFP positive and negative cells were interspersed (Figure 4d).

Third and Fourth Generation MIN-O-D/EmGFP
As in both earlier passages, the third and fourth generation MIN-O-D/EmGFP lesions were easily detected with in vivo fluorescent imaging (Figures 5a & 5b). Consistent with the IHC results from the second generation lesion, however, a pattern of progressive EmGFP loss was evident in the third and fourth generation lesions (Figures 5c through 5f). Again, this loss presented as regions of complete loss and regions of partial loss. This culminated in massive loss of EmGFP expression in the fourth generation lesion.

Generational Loss of EmGFP Expression
In order to determine whether loss of EmGFP expression was due to silencing or genomic loss of the lentivirally delivered EmGFP expression cassette, genomic DNA from EmGFP positive and negative regions (isolated via laser capture) was subjected to qPCR analysis using primers for EmGFP and PyVmT. As the MIN-O transplant lines were originally derived from FVB/ PyVmT mice, PyVmT is the only known genomic difference between MIN-O-derived tissue and tissue from the otherwise genetically identical FVB host mouse. qPCR analysis of PyVmT in this assay thus confirms whether EmGFP negative tissue is MIN- O-derived or recruited from the host. Results of qPCR showed that, while there is no significant difference in the abundance of PyVmT in EmGFP positive and negative regions, sequence encoding EmGFP is nearly absent (0.18% that of EmGFP positive tissue) in EmGFP negative tissue (Figure 6e). Two conclusions can be drawn from these data. First, tissue from both EmGFP positive and negative areas is predominantly MIN-O in origin (as indicated by the similar prevalence of PyVmT in isolated samples). Second, the dominant mode of observed EmGFP loss is expulsion of the lentivirally delivered expression cassette from the genome rather than gene silencing (as indicated by the absence of EmGFP coding sequence in the genomic DNA of EmGFP negative tissue).

Discussion
The protocol presented here allows for the development of clonally derived sublines of the MIN-O transplant lines incorporating imaging reporter genes. Prior to this work, noninvasive imaging of this series of lines was limited to studies with administered contrast agents (radioactive or fluorescent) due to an inability to make the type of genomic modifications commonly employed in cell line and xenograft studies. This advance in technology makes possible more sophisticated studies designed to integrate the potential of non-invasive imaging and the biological relevance of MIN-O to elucidate the events involved in progression of DCIS to invasive carcinoma. In addition to the fluorescent reporter construct used here as proof of principle, lentiviral transduction can be readily exploited to deliver a more complex payload such as gene knockdown and overexpression cassettes and/or allele replacement as well as promoter-specific and pathway dependent reporters. The fact that novel sublines are clonal in lineage also allows for greater consistency and clarity of results relative to studies utilizing ''mixed pools'' of transduced cells. In addition to the universal advantages inherent in clonal versus mixed pool lines regardless of the cell origin or method of development, we have found that one of the primary advantages of this protocol is the low number of cells required. In the MIN-O model presented here, less than 50,000 cells are required for the plating of a 96-well PCR plate. A single plate yields approximately 12-18 transplantable wells; that is, wells in which only a single fluorescent sphere developed. This was particularly significant within the context of the MIN-O model due to the inability to allow lesions to progress beyond the precancer stage and the resulting decrease in the yield of primary cells per mouse. Given the difficulty and cost associated with culture of many other types of primary cells, we expect this feature of our protocol to be attractive to researchers in need of similarly engineered lines.
Interestingly, sphere morphology is highly predictive of sphere growth in vivo. This suggests that there are different types of progenitor cells in the MIN-O, with different biologic potential for differentiation and growth in vivo. It is well known that many cell types from the mouse mammary gland show divergent differentiation and growth depending on the culture conditions. In most examples, however, the ''on top'' versus ''in'' MatriGel comparisons are equal [17]. In our experiments here, we found that culture within MatriGel favored the transplantable hollow phenotype over the non-transplantable solid core phenotype. While the precise differences between the two culture methods that contribute to differences in sphere morphology are not known, the critical issue within the context of this study is that culture within MatriGel improves the efficiency of developing transplantable MIN-O spheres. Further evaluation of the factors responsible for this increase in efficiency may be important to cancer stem cell differentiation and biology.
Not surprisingly, the primary drawback to serial transplant of genetically engineered cells is the gradual loss of transgene expression with increasing generations. Loss of transgene expression regardless of the mode (i.e., methylation, deletion, loss of heterozygosity, etc.) has been widely accepted as a pitfall of genetically engineered cells. The rate of loss of expression can be highly dependent upon selective pressures (immunogenicity, changes in proliferative rates, etc.) if a given expression cassette and expressed proteins confer negative selection. Even if there is no significant negative selection, transgenes can still be removed or silenced spontanesously. This difficulty is overcome in vitro by the maintenance of selective pressure via supplementation of growth media with antibiotics for which resistance is genetically linked to transgene expression. Unfortunately, this approach is not easily achieved within the context of the MIN-O model as the main advantage of this transplant line is that, with the exception of PyVmT and the transgene, MIN-O cells are genetically identical to the host mouse. Fortunately, the utility of the fluorescent reporter during in vitro screening can be extended to allow for fluorescence-activated cell sorting (FACS) of cells isolated from precancer lesions in order to enrich for cells in which the transgene remains expressed. In our experience in unrelated projects, we have found that FACS has little or no negative effect on cell viability and lesion formation. Sorting can in theory be completed immediately prior to injection of study cohort(s) to ensure uniform transgene expression as our observations have shown high stability of expression throughout the period of a single generation. Given  that the rate of loss of expression will be dependent upon the type of genetic material and the toxicity or immunogenicity of expressed proteins, potential loss of expression should be considered prior to in vivo propagation. As a related side note, it is also worth noting that we achieved stable EmGFP expression through a single generation only when the elongation factor 1a (EF1a) promoter was used to drive its expression. We completed several iterations of this protocol using lentiviral particles generated with the pLenti6/CMV/EmGFP plasmid and, although similar results were observed in vitro with respect to sphere morphology, transduction efficiency and transgene expression, outgrowths following transplantation were consistently EmGFPnegative indicating ''silencing'' of the CMV promoter as has been reported extensively [18,19,20].
As our understanding of the role microenvironment plays in both normal and atypical development increases, so too will the emphasis placed on the development and utilization of models such as MIN-O which more closely recapitulate the natural processes being queried. As described here, the period from excised unmodified tissue to in vivo imaging of genetically modified lesions which are clonal in lineage is approximately 6 weeks. The transduction efficiency of primary cells, notorious for their difficulty to transfect or transduce, is only slightly lower than what is typically achieved in robust cell lines in vitro. Because of both of these factors we believe this method of genomic modification of primary cells to be a powerful tool for researchers which will expand the use of complex animal models into areas such as in vivo molecular imaging which have been previously largely restricted to studies employing xenografts and cell lines.