Fresh human DCIS lesions reproducibly generate in vitro anchorage independent, neoplastic epithelial cells that generate 3-D structures including spheroids and duct-like structures. Neoplastic cells with this phenotype can emerge from a high proportion of replicate DCIS lesion samples from the same patient, and can be serially propagated for at least one year. No anchorage independent cells arose from tissue containing histologically verified normal appearing glands and adipose tissue. The anchorage independent epithelial cells were observed to arise from all grades of DCIS including ADH ().
The cytogenetically abnormal cells contained in fresh human DCIS lesions are potentially malignant by the following criteria: a) generation of neoplasms in NOD SCID mice, b) anchorage independent 3-D structures that increase in size and frequency over time, which can be subcultured for at least one year (), c) abnormal neoplastic-type cytogenetics (, Figures S4
), and d) invasion of autologous stroma in the organ culture (–).
It is unlikely that the neoplastic cell strains propagated from fresh human DCIS ducts are isolates of micro-invasion areas or invasive cancer in the original histopathology.The patient source material was verified histologically to be devoid of microinvasion or invasive cancer before and after DCIS lesion tissue procurement. The verification was by an independent pathologist with no knowledge of the research findings for the individual specimen. The DCIS lesion source material was evaluated by IHC for type IV collagen and was found to contain an intact basement membrane (Figure S8
). If the neoplastic, cytogenetically abnormal cells isolated in this study represented areas of microinvasion, this would be expected to be a rare event. This was not the case. Generation of spheroids and 3-D structures arose spontaneously from multiple, independent human DCIS duct tissue fragments from the same patient and from different patients.
The propagated spheroid forming cells, generated from fresh human DCIS tissue, were cytogenetically abnormal and were associated with xenograft tumor formation. We hypothesize that these cytogenetically abnormal cells pre-exist in the fresh human DCIS duct. This hypothesis was supported by the results of the Illumina 300,000 cytoSNP beadchip comparing the molecular cytogenetics profile of the organoid, the epithelial and cuboidal cultured monolayers, and the spheroids and 3-D structures grown in the same mixed culture. It is well established that the transition from normal to neoplastic growth involves deregulation at various cellular levels including gene-specific mutational events, alterations in signal transduction and growth control pathways, and dysfunctional DNA synthesis and repair mechanisms resulting in the generation of chromosomal abnormalities. At the chromosomal level, it is now accepted that some degree of copy number variation (CNV) and copy-neutral loss-of-heterozygosity (LOH) exists in the normal karyotype 
. However, we have observed in this study that cells with these features can rapidly emerge during the time course of in vitro
culture. The early emergence of chromosomally abnormal cells may be the drivers of the dominant neoplastic phenotype observed during DCIS culture outgrowth.
Transformed cells containing regions of excessive CNV can change the dosage of key regulatory genes and protein output; while de novo regions of LOH can potentially expose harmful recessive alleles. Of note was the strikingly similar loss on chromosome 6 in the spheroid forming cells of three different patients. This region on chromosome 6 narrowly encompasses the SUPT3H gene (protein coding GIFtS:59, GC06M044904, UniProtKB/Swiss-Prot: SUPT3_HUMAN, O75486). Little is a known about the encoded 399 amino acid protein. It is thought to be a transcription factor, participating in the STAGA complex (SPT3-TAF9-GCN5-acetylase) involved in p53 and c-Myc regulation 
. From this study, the genes contained within affected chromosomal regions of gain or loss can be studied in further detail to gain greater insight into the transformation process. Higher density microarrays, gene expression studies, and gene sequencing can be applied in candidate gene approaches using these data as a starting point. Even in the case in which different patients show different patterns of aberration, the same or similar regulatory pathways related to cell survival, oxidative stress, angiogenesis, or autophagy may be involved.
A strong rationale links autophagy to the survival and invasion of pre-malignant breast cancer. The first link is hypoxia and nutrient stress 
. Proliferating ductal epithelial cells accumulating within the breast duct do not have access to the vasculature outside the duct. For this reason, high grade DCIS is associated with central necrosis, and the accumulation of lipofuschin. Autophagy is a pathway activated to promote survival in the face of hypoxic and nutrient stress 
. Consequently the activation of autophagy may divert the hypoxic cells away from apoptosis and thereby support the survival and growth of DCIS neoplastic cells within the lumen 
. The second link is anoikis, the triggering of apoptotic cell death for cells that have been separated from their normal adhesion substratum 
. Normal glandular epithelial cells require attachment to, or association with, the basement membrane extracellular matrix (ECM) for continued survival. During ductal hyperplasia and dysplasia epithelial cells exist within the duct at a substantial distance away from association with the peripheral basement membrane. Moreover, invading carcinoma cells can migrate into the stroma in the absence of a basement membrane anchor 
. Autophagy has been shown to be a key regulator of survival for cells deprived of an anchoring substratum 
, and may play an important role for cell survival in any anchorage independent state. A third link is matrix degradation 
. High grade DCIS, microinvasion, and overt carcinoma invasion are associated with interruptions, remodeling, and enzymatic breakdown of the basement membrane and the stromal ECM 
. Autophagy may facilitate cell movement through areas of degraded matrix by the phagocytic processing of matrix breakdown fragments 
. A fourth link is calcium. Microcalcifications are mammographic indicators of high grade DCIS 
, and calcium phosphate precipitates are potent inducers of autophagy 
. Five out of seven tissues used for ex vivo organoid culture were noted to have microcalcifications.
The data presented herein, strongly support the conclusion that autophagy plays a necessary role in the DCIS cell malignant phenotype:
a) autophagy is up-regulated in the in vivo DCIS lesion as shown by immunohistochemistry; b) autophagy is up regulated in the cultured DCIS spheroids and 3-D structures, as shown by immunohistochemistry and immunofluorescence; c) autophagy is up-regulated in mouse xenografts as shown by immunohistochemistry; d) autophagy signal proteins are up-regulated in cultured DCIS spheroids, with validation after long term culture, by reverse phase protein microarray measurement; e) disruption of autophagy by chloroquine phosphate (CQ) completely abrogated xenograft tumor formation; f) CQ completely blocked growth and invasion of DCIS cells on autologous stroma; g) CQ markedly suppressed DCIS spheroid formation and outgrowth in culture for independent experiments; and h) CQ completely eliminated cytogenetically abnormal cells from the DCIS cultured cells. Based on these established mechanistic roles, autophagy constitutes a novel target for treating DCIS and arresting DCIS transition to overt invasion.
Autophagy is an established strategy for a cell to avoid apoptosis in the face of oxidative, hypoxic and nutrient deprivation stress. Our results indicate that autophagy is a requirement for the survival and the malignant phenotype (tumorigenicity and invasion) of cytogenetically abnormal cultured human DCIS cells. It is possible that the observed genetic abnormalities directly or indirectly induced the up-regulation of autophagy and thereby promoted survival of the DCIS cells. It is also possible that autophagy was not driven directly by the genetic changes. Instead the malignant precursor cells up-regulated autophagy as a necessary means to generate anchorage independent 3-D structures such as spheroids and pseudo ducts. In either case, the end result is the same: Autophagy is required for the observed phenotype.
Chloroquine phosphate, which suppressed or abolished the anchorage independent, cytogenetically abnormal anchorage independent neoplastic DCIS cells, is an orally administered small molecule inhibitor which blocks the autophagy pathway by accumulating in autophagosomes and inhibiting autophagosomal formation/function 
. Chloroquine has been shown to suppress N-methyl-N-nitrosurea induced mouse breast carcinomagenesis 
, enhances the effectiveness of tyrosine kinase inhibitor treatment of primary CML stem cells 
, and has been proposed as a potential means to enhance the effectiveness of tamoxifen in vitro
in tamoxifen resistant breast carcinoma cells by blocking autophagy dependent cell survival 
. Moreover, chloroquine has been proposed as a chemoprevention therapy for Myc induced lymphomagenesis because it induces lysosomal stress that causes a p53 dependent cell death that does not require caspase mediated apoptosis 
These data provide additional evidence that the local ductal tissue microenvironment influences the progression of in situ carcinomas. Breast stromal cells 
, endothelial cells, and myoepithelial cells 
have been implicated in the transition from in situ to invasive carcinoma. By IHC, Caveolin-1 was found to be down regulated in stroma of DCIS lesions with a greater incidence of subsequent invasive cancer 
. Jedeszko et al 
reported that stromal fibroblasts contributed to the regulation of invasion by the production of hepatocyte growth factor, using a co-culture model with MCF10 and SUM102 cell lines. The present study identifies and characterizes malignant precursor epithelial cells in fresh human DCIS lesions, and therefore goes beyond model systems using established cell lines such as MCF10.
Even though the cytogenetically abnormal DCIS epithelial cells we have derived appear to have the characteristics of progenitor cells by RPMA analysis ( and Table S2
), our method of generating these cells is different than cell sorting methods used for what are defined as “cancer stem cells”. In the present study we purposely did not assume that the DCIS neoplastic cells would be positive for CD44 or any other surface marker. We did not enzymatically dissociate the DCIS lesions and then conduct flow sorting as has been done for studies of breast carcinoma 
. Instead, we allowed the spheroid and 3-D ductal structures to grow spontaneously from epithelial cells that emerged from exposed open ends of the DCIS duct in organ culture. Thus, the cytogenetically abnormal () progenitor-type cells we have identified are self-selected by anchorage independence in organ culture. While there is broad evidence supporting the existence and role of stem cells both in normal mammary gland development and tumorigenesis 
, we cannot formally classify these cells to be, or not to be, “cancer stem cells”. This classification is irrelevant to the fact that these propagated DCIS lesion derived cells are cytogenetically abnormal and have malignant characteristics.
In conclusion, fresh human DCIS organoid culture reveals cytogenetically abnormal, neoplastic, anchorage independent cells that require autophagy for survival. Invasive neoplastic cells may pre-exist within the human breast DCIS duct but are apparently held in check by the ductal niche, and can be coaxed to emerge in organ culture when the duct is cut open. Blockade of the autophagy pathway abolishes the propagation, invasion, and 3-D growth of the DCIS neoplastic cells. The reproducible derivation of cytogenetically abnormal cells with fresh human DCIS lesions provides a new ex vivo model to study the biology of DCIS. Moreover, the yield, and properties, of progenitor-like cells from samples of human DCIS lesions following in vivo treatment of DCIS can be used to screen the effectiveness of an in vivo therapy. Finally, autophagy constitutes a new treatment target for DCIS.