The clinical use of ex vivo expanded HLECs to treat patients with LSCD has preceded a clear scientific understanding of the mechanism of action of this therapy. Specifically, it is unclear whether ex vivo cultured HLEC cell transplants reintegrate into the recipient's LESC niche and continue to function in the long term to maintain a healthy corneal epithelium. An alternative hypothesis is that the temporary restoration of a healthy corneal epithelium by the transplanted cells acts as a biological bandage and promotes regeneration of the recipient's own LESC population. This question could most definitively be answered by tracking transplanted cells in vivo. This has not been performed to date, not least due to the lack of laboratory data on labeling efficiency, duration, toxicity, and effects on transplanted HLEC function. The aim of this study was to determine whether Qdots could be used to label and track ex vivo cultured limbal epithelial cells without impacting on cell viability and function. We have shown that Qdots have potential in this regard and have demonstrated proof of principle for their use in tracking transplanted cells in the cornea. Last, we demonstrated how existing clinical equipment could be modified to enable in vivo detection of Qdot-labeled cells in animal or human studies.
The ability to follow the distribution and migration of biologically active cells in living organisms is crucial for both the development of cell-based therapies and for the elucidation of biological mechanisms in stem cell research. Two key components are required to achieve this: a method of selectively labeling transplanted cells and a technique for detecting and visualizing the transplanted cells in vivo.
There are two distinct approaches to cell labeling: indirect and direct labeling of cells. Indirect labeling methods use a reporter gene that is introduced into a cell and translated into enzymes, receptors, or fluorescent or bioluminescent proteins. If the expression of the reporter gene is stable, labeled cells can be observed over the entire lifetime of cells. Indirect labeling has proved an immensely valuable tool in terms of basic research, but there are significant obstacles to the translation of this technology into clinical practice, not least the ethical and regulatory issues that result from the use of genetically modified cells. Direct cell labeling refers to attachment of exogenous detectable labels (fluorophores, magnetic particles, or radioisotopes) to the cell membrane or their intracellular incorporation using a variety of techniques. Direct labeling is technically straightforward and does not involve genetic modification of the cell. It is therefore a safer and more clinically useful approach and the one we chose to use.
By virtue of its transparency, the cornea is an ideal tissue in which to directly observe transplanted stem cells using a fluorophore label and in vivo microscopy. This combination can achieve a level of resolution that other detection systems, such as magnetic resonance imaging, single-photon emission computed tomography, and positron emission tomography, cannot.
10 When selecting a fluorophore, we set the criteria that it should be nontoxic, photostable (i.e., resistant to photobleaching), and sufficiently bright to generate a signal detectable through living tissues. We chose to use Qdots because they exhibit some important differences as compared with traditional fluorophores, such as organic fluorescent dyes and naturally fluorescent proteins.
Structurally, Qdots are inorganic composites consisting of nanometer-scale (roughly protein-sized) clusters of atoms (containing from a few hundred to a few thousand atoms) of a semiconductor material (cadmium mixed with selenium or tellurium has been the most common for biological applications), which has been coated with an additional semiconductor shell (ZnS) to improve the stability and optical properties of the material. Quantum dots fluoresce in a completely different way than do traditional fluorophores. In Qdots, fluorescence is achieved via the formation of excitons, or Coulomb-correlated electron-hole pairs that are created between the two semiconductor materials. The exciton can be thought of as analogous to the excited state of traditional fluorophores; however, excitons typically have much longer lifetimes (>10 ns, up to ∼200 nanoseconds, versus <5 ns for organic fluorophores).
18 This structural and functional difference engenders two key advantages of Qdots over conventional fluorophores. First, photostability of Qdot nanocrystals is many orders of magnitude greater than that associated with traditional fluorescent molecules and hence photo-induced deterioration of the emitted signal (known as bleaching) is substantially less than that seen with other types of fluorophores. Second, the optical efficiency of Qdots (quantum yield, in general terms the amount of light emitted relative to that absorbed) is many times that observed for other classes of fluorophores. As a result, less illumination is required to achieve a detectable signal from Qdots than from other classes of fluorophores. One more general advantage over classic fluorophores comes from multiplexing. Different color Qdots can be excited with a single wavelength and the narrow emission band of Qdots makes deconvolution straightforward.
We used a commercial Qdot kit that uses a custom cell-penetrating peptide bound to the surface of Qdots to induce their cellular uptake.
19,20 These peptides are a beta-amino acid analogue of Tat 47-57, the portion of the human immunodeficiency virus (HIV) Tat DNA-binding protein that is responsible for the ability of the HIV Tat protein to readily cross the cell membrane.
21 Tat 47-75 analogues have negligible cytotoxicity in HeLa, Chinese hamster ovary (CHO-K1), and human umbilical vein endothelial cells.
22 The corneal toxicity of cell-penetrating peptides has been assessed in a rabbit drug-delivery model and found to be negligible.
23 The intracellular target of Tat 47-57 and its analogues is the nucleus.
19,20 In the initial 24 hours after labeling, Qdots formed membrane-less aggregates throughout the cytoplasm, presumably reflecting the fact that they are trafficking to the nucleus. By 1 week after labeling, Qdots were primarily located in the nucleus, as expected. Some cytoplasmic aggregates were still present at this time point and on electron microscopy these were found to be encapsulated by a membrane and associated with the Golgi apparatus. These findings suggest that some metabolism of and possibly exocytosis of Qdots were occurring at this stage. By 2 weeks after labeling, Qdots were located primarily in the nucleus, with few cytoplasmic aggregates, suggesting that by this time point the labeling was stable. There is evidence that mouse embryonic stem cells can both degrade and exocytose the same commercial Qdot system that we have used, resulting in loss of Qdot signal within 48 hours of labeling.
24 In contrast, the same study found that mouse embryonic fibroblasts did not degrade or exocytose Qdots, and that the reduction in signal seen over time in this cell type was due to the diluting effect of cell division. Our results show that Qdots can result in a labeling efficiency of more than 94% immediately after labeling, which gradually reduced to 22% after 14 days as the cells proliferated. This is remarkably similar to the findings in mouse embryonic fibroblasts
24 and we therefore hypothesize that although limited metabolism and exocytosis do occur in HLECs, the primary reason for the reduction in Qdot signal over time is dilution due to cell division.
Another significant limitation of our organ culture model is that we have been unsuccessful in maintaining limbal epithelial cell sheets (both labeled and unlabeled) in culture on stromal explants for longer than 2 weeks. This is because the transplanted epithelium differentiates rapidly between day 14 and 21 post transplantation. Although our model does not permit us to track cells beyond 2 weeks, this does not mean that quantum dot–labeled cells will be undetectable beyond this time point. In vivo studies will be necessary to determine the duration of labeling with certainty. Two-photon imaging has been used to demonstrate that Qdots can be retained within the cornea for up to 26 days in an in vivo mouse model, albeit this was a model of toxicity and not a model of cell tracking.
25
For Qdots to be used clinically, it is essential for their toxicity to be investigated. It was encouraging to find that a high level of labeling efficiency could be achieved and that neither the labeling process nor the ongoing presence of Qdots within cells caused toxicity or promoted differentiation. This is in contrast with one previous study by Kuo et al.
25 in which two-photon microscopic imaging was used to study the distribution of chemically similar (CdSe/ZnS core/shell) Qdots throughout the cornea. Kuo et al.
25 found evidence of cytotoxicity in bovine corneal stromal cells. Cell viability decreased significantly as the Qdot concentration and incubation period increased. The reasons for the discrepancy between this study and our results could be that different cell types or even different species handle Qdots differently. Alternatively, the different concentrations of Qdots used (50 nM versus 10 μg/mL in our study) and exposure periods (24–48 hours versus 60 minutes in our study) reflect the fact that Kuo et al.
25 were seeking to define limits of toxicity, whereas we were using an optimized and validated protocol for cell labeling.
When assessing the effect of Qdots on proliferation and differentiation of HLECs, we did not perform clonal analysis. We used a mixed population of cells that will have included holoclones, meroclones, and possibly some paraclones. We cannot therefore be certain whether these results apply specifically to LESCs (holoclones), transient amplifying cells (meroclones), or both.
16,17
Our organ culture transplantation model showed that we could still detect transplanted cells at 14 days after transplantation. At present, there is a reliable method of directly visualizing transplanted ex vivo cultured HLECs transplanted in human patients. Our techniques could be used to monitor cells for up 2 weeks and perhaps longer, depending on the rate of proliferation of transplanted cells. Although it would be desirable to be able to track cells for their entire lifetime, the technology to do this safely in vivo is not currently available. The ability to track cells even for 1 month after transplantation would lead to significant advances in understanding the biological mechanism of action and posttreatment behavior of such transplants.
Based on the observations that LESCs divide infrequently in vivo and therefore retain bromodeoxyuridine longer than their differentiated progeny
26 and that the rate of proliferation determines the rate of label dilution and therefore duration of labeling, we hypothesize that post transplantation cells that divide rapidly, such as transient amplifying cells, will see the number of Qdots in their cytoplasm rapidly reduce to undetectable levels, whereas cells that divide infrequently, as stem cells do, may retain a high number of Qdots. In our organ culture model of HLEC transplantation, it was observed that there was a marked heterogeneity of Qdot signal with the brightest cells containing most Qdots being found in clusters at the limbus. This suggests that the limbus has an effect on the transplanted cells, which the cornea does not. It also suggests that some of these slow-cycling cells may reintegrate into the limbus and remain in a slow-cycling or quiescent state. From our data it is not possible to confirm that these cells possess a limbal stem cell phenotype but further work will aim to investigate whether this is indeed the case.
In vivo confocal microscopy has been increasingly used to investigate the cornea in health and disease. In particular, the Rostock corneal module for the HRT II has enabled cellular-level investigation of corneal structure and the limbal stem cell niche.
15,27,28 Our results from ex vivo rabbit corneas demonstrate that the Rostock corneal module can be modified to work with the Heidelberg Spectralis HRA-OCT to enable in vivo detection of cells labeled with Qdots. There is undoubtedly room for a more precisely engineered method of combining in vivo confocal microscopy with fluorescence detection, but the data we present demonstrate that this is technically possible.