Cell Tracking with Lipophilic Membrane Dyes

Optimal staining is a key component for studying tumorigenesis and progression. Learn useful tips and techniques for dye applications, including examples from recent studies.

Abstract

Because lipophilic cell tracking dyes such as PKH26, PKH67, and CellVue^® Claret can be used to label almost any cell, they have enabled cancer biologists to track a wide variety of tumor and immune cell functions in vitro and in vivo. These include: migration and adhesion; proliferation of stem and progenitor cells; differentiation and growth control; mechanisms of antigen presentation (e.g., membrane transfer, phagocytosis); and interactions of effector and regulatory cells with each other and with tumor cells. This article summarizes methods for optimal staining with these dyes and provides examples from the recent literature illustrating their use to study mechanisms involved in tumor progression and anti-tumor immunotherapy.

Methods

General Membrane Labeling

Using Diluent C, the staining vehicle provided in our cell linker kits, general membrane labeling occurs by nearly instant partitioning of the tracking dyes’ long-chain alkyl tails into the membrane bilayer (Figure S1*). The labeling protocol is rapid, simple, and applicable to essentially any membrane-containing cell or particle (Figure S2*), but obtaining bright, uniform, and reproducible labeling requires attention to different variables than when staining with antibodies or other equilibrium binding reagents.^1–3 In particular:

• Final staining intensity depends on cell concentration as well as dye concentration. The amount of dye required for bright, homogeneous labeling increases as total number and/or size of cells to be stained increases (i.e., as total amount of membrane increases). Unfortunately, many authors report only the dye concentration used. Table S1* summarizes non-perturbing staining conditions reported for several membrane dyes, but concentrations found in the literature should always be tested on each laboratory’s cell type(s) of interest. In our experience, a preliminary titration at constant cell concentration is the easiest way to select an optimized dye concentration that gives acceptable post-labeling viability, recovery, and fluorescence intensity as well as unaltered cell function.^2,3
  
  
• Uniform labeling requires rapid, homogeneous mixing of cells and dye. Because cell-to-dye ratio affects staining intensity, it is important that all cells are exposed to the same concentration of dye at the same time. Because dye uptake into the membrane is so rapid, this is more easily achieved by admixing similar volumes of cells and dye (both pre-dispersed in Diluent C) than by trying to rapidly disperse a tiny volume of concentrated ethanolic dye into a large volume of cells in Diluent C (Figure S1*). Dispensing and mixing should be as nearly simultaneous as possible. In theory, it should make no difference whether cells are added to dye or dye to cells, but in our experience new users can more reliably obtain uniform labeling by adding cells to dye.³
  
  
• Prepare 2X dye and 2X cell suspensions immediately prior to staining. Diluent C is an aqueous staining vehicle designed to keep cells viable and membrane dyes in solution for long enough to complete the brief labeling step. It minimizes membrane dye aggregation and maximizes staining efficiency by avoiding salts, while minimizing cell toxicity by maintaining osmotic pressure and avoiding organic solvents. However, lack of physiologic salts can impair cell viability over prolonged periods, and the dye aggregates will gradually form even in the absence of salts.^1–3
  
  
• Use neat fetal bovine serum (FBS) or other system-compatible protein as the “stop” reagent. Since labeling occurs by partitioning rather than equilibrium binding, dye uptake into membrane continues for as long as cells are in contact with dye, eventually compromising membrane integrity and function. Addition of buffered salt solutions or serum-free culture medium stops labeling by causing rapid dye aggregation. However, cellassociated dye aggregates are not efficiently removed by washing and act as reservoirs of unbound dye that can transfer to unlabeled cells present in an assay. Addition of a relatively concentrated protein solution (e.g., FBS; 10–15 mg/mL albumin) is therefore preferred as the “stop” reagent. FBS provides hydrophobic binding sites for excess dye, and its lower salt concentration reduces aggregation.³ In systems where animal-derived proteins must be avoided, recombinant albumin may be an acceptable substitute.

Proliferation Tracking

In addition to being useful for general membrane labeling and cell tracking4, the membrane dyes in Table S2* have also been shown to be useful for proliferation monitoring based on the principle of dye dilution.² In non-dividing cells fluorescence intensity is stable for weeks to months, but in dividing cells it decreases progressively as dye is segregated evenly between daughter cells with each round of division. Using flow cytometry and commercially available software, it is possible to estimate how many divisions each cell has undergone by comparing its intensity to that of the starting population immediately post-labeling (Figure 1) and, from that information, to determine precursor frequencies, extent of expansion to a given stimulus, etc. Counterstaining with phenotypic reagents to identify subsets of interest allows differential proliferation monitoring within complex cell populations, avoiding the need for subset isolation. The availability of membrane dyes emitting from the green to the far red (Table S2*) allows the choice of proliferation tracking dye to be tailored for compatibility with other key markers and probes (e.g., fluorescent proteins or dimly expressed antigens).

Results and Discussion

Because they are simple to use and allow the fate of tagged cells to be tracked in vitro or in vivo (reviewed in Ref. ⁴), PKH26, PKH67, and CellVue^® Claret are powerful tools for studying how different cell populations in the heterogeneous tumor environment impact disease progression and/or responses to therapy. It is impossible in this short space to discuss all of the ways in which these reagents have been used since PKH26 first became available from us in 1993.

Tracking the Effects of Anti-Tumor Immunotherapy

One of the earliest applications of membrane dyes was as a nonradioactive probe for tracking NK cell killing of tumor targets⁵. As recently reviewed by Zaritskaya,⁶ variations on this theme have expanded exponentially in the last decade. Membrane dye labeling of target cells and/or effector cells combined with multicolor flow cytometry continues to provide detailed insight into signaling pathways, mechanism of killing, and other factors affecting cytotoxic efficacy, both in vitro^7,8 (Figure 2) and in vivo⁹ (Figure S3*). A multicolor cytotoxicity protocol using PKH67 and CellVue Claret in combination with a fixable live-dead probe is found in Ref.³, which also includes a discussion of how trogocytosis (contact-mediated membrane transfer between cells) may impact cytotoxicity assays. In another multicolor application, HoWangYin et al.¹⁰ studied the functional consequences of trogocytic acquisition of a tolerogenic antigen (HLA-G1) by PKH26- or PKH67-labeled monocytes. Using PKH26-labeled M8 melanoma cells expressing EGFP-tagged antigen, they found efficient trogocytosis by both T cells and monocytes. More rapid disappearance of HLA-G1 from the surface of monocytes, however, appeared to limit the ability of those cells to inhibit autologous T cell proliferation and inflammatory cytokine production.

Applications of dye dilution proliferation analysis (Figure 1) have also grown dramatically in the last decade. The PKH dyes and, more recently, CellVue Claret¹¹ have served as fruitful probes for qualitative and quantitative monitoring of antigen-driven cell division in complex populations.^2,3,12 Using multicolor flow cytometry, cells can be counterstained for surface or intracellular markers to define how lineage, activation or differentiation state, cytokine or chemokine expression, antigen-binding ability, etc. correlate with extent of cell division as reflected by dye dilution. In particular, as our understanding of immune effector/regulatory cell interactions has improved and anti-tumor vaccine strategies have grown more complex, the dye dilution proliferation assay has allowed investigators to determine: 1) whether a given vaccination protocol increases the frequency of precursor cells able to respond to a given antigen; and 2) whether/how changes in precursor cell frequency post-immunization correlates with clinical response.¹ In an exciting recent example, Barth et al. used PKH67 dye dilution analysis to determine anti-tumor T cell precursor frequencies in colorectal cancer patients before and after adjuvant therapy with an autologous dendritic cell vaccine.¹³

Although the presence of anti-tumor T cell responses pre-therapy did not correlate with recurrence-free survival, patients who exhibited an increase in CD4/CD8 T cell precursor frequencies 1 week postvaccination exhibited substantially better recurrence-free survival at 5 years than patients who did not show an increase in precursor cell frequency (Figure 3). PKH26 dye dilution proliferation analysis has also been used to investigate changes in adaptive immune responses in prostate cancer patients undergoing androgen deprivation therapy for metastatic disease.¹⁴ Increased proliferative responses to CD3 and CD28 stimulation were found in almost all T cell subpopulations (naïve, effector and memory) within 1 month after initiation of therapy, with the most pronounced increases in the effector and central memory subsets.

Tracking Tumor Stem Cells

A rapidly growing application for PKH26, PKH67, and CellVue^® Claret is to monitor the effects of proliferative behavior in different cell subpopulations on tumor progression and/or responses to chemotherapy or immunotherapy. In particular, the fact that non-dividing or slowly dividing cells remain brightly labeled with membrane intercalating dyes has been used by many investigators to identify and sort low-frequency “label-retaining cells” for further characterization. Again, space precludes a comprehensive summary, but several examples from the recent literature will serve to illustrate the usefulness of this approach.

The inability to eradicate upper aerodigestive squamous cell carcinomas using conventional chemotherapy has been hypothesized to be due at least in part to the emergence of a mesenchymal-like subpopulation during malignancy and the intrinsic drug resistance of this subpopulation. After labeling SCC9 squamous carcinoma cells with PKH67 and culturing for 9 days in vitro, Basu et al.¹⁵ sorted the brightest 10% (low-turnover) and the dimmest 10% (high-turnover) and compared their E-cadherin expression. Consistent with the hypothesized mechanism, the low-turnover subpopulation contained fivefold more mesenchymal-like (E-cadherin low) cells than the highturnover subpopulation. Sorting based on E-cadherin expression confirmed the reduced proliferative status of the mesenchymal-like subpopulation (decreased proportion of Ki-67 expressing cells) and also found increased resistance to paclitaxel in vitro. Schubert et al.¹⁶ used a related approach to evaluate the leukemia-initiating capacity of subpopulations isolated from the bone marrow of AML patients based on CD34 expression, aldehyde dehydrogenase (ALDH) activity, or PKH26 label retention (brightest 5–10% of cells after 5–7 day culture in vitro). Despite considerable variability among patients, they found that “selection of CD34⁺, ALDH^bright, and/or PKH^bright cells correlates with AML cell survival and expansion in the NO D/SCID mouse.” In the hands of these investigators, efficiency of selection for long-term surviving cells was comparable for the ALDH and CD34 methods and slightly, but not significantly, lower for the PKH26 label retaining method.

The ability to sort quiescent or slowly cycling cells based on longterm retention of membrane-intercalating dyes has also enabled identification and functional characterization of cells that contribute to post-therapeutic tumor dormancy and eventual relapse. In one example, investigators isolated PKH26- or PKH67-labeled A4 ovarian tumor cells 3–6 weeks after subcutaneous implantation in NOD/SCID mice, selecting for either PKH^hi, PKH^neg, or PKH^lo populations (intensities equivalent to pre-injection levels, unstained cells, or intermediate levels, respectively; Figure 4). Upon further characterization, the non-proliferating PKH^hi subset exhibited many of the characteristics expected for ovarian cancer stem cells (Figure 4). This observation appeared generalizable since similar characteristics were found for malignant ascites and secondary peritoneal tumors derived from PKHlabeled A4 cells and also for 3-week-old C6 (rat-glioma derived) tumors grown in NOD/SCID mice. Treatment with two cycles of paclitaxel led to enrichment of both PKH^hi (quiescent) and PKH^lo (arrested or slowly cycling) subsets, leading the authors to conclude that “tumor-derived [cancer stem cells] and aneuploid populations contribute to drug resistance and tumor dormancy in cancer progression.¹⁷” Work by investigators in Milan has successfully used long-term retention of bright PKH26 labeling in combination with in vitro mammosphere culture to isolate rare normal mammary stem cells (MSC) and breast cancer stem cells (CSC) and to study the growth properties of both at the single cell level.^18,19 For normal human MSC, cell divisions were predominantly asymmetric with only one daughter cell going on to divide, as reflected by subsequent dye dilution, while the other remained quiescent, as reflected by retention of bright PKH26 labeling (Figure 5). In contrast, both tumorigenic Erb-B2 transformed murine cells¹⁸ and patient-derived CSC¹⁹ exhibited greater numbers of symmetric divisions in which both daughter cells went on to divide. Mammospheres derived from poorly differentiated (Grade 3) breast tumors contained greater numbers of PKH26^bright cancer stem cells than mammospheres derived from more welldifferentiated (Grade 1) tumors. Excitingly, Pece et al.¹⁹ found that the transcriptional profile associated with the PKH26^bright cells (hNMSC signature) was able to predict biological and molecular features of breast cancers. Although this work does not directly address the origin of the cancer-initiating stem cells, it does suggest a model for tumor progression in which transforming event(s) alter the frequency with which stem cells skip asymmetric self-renewing divisions, affecting the final number of cancer stem cells present within the tumor and its biological and clinical characteristics.

Conclusion

The examples provided here illustrate a few of the many ways that tracking cells, membranes, or proliferation with the membrane intercalating dyes PKH26, PKH67, and CellVue^® Claret have been used to study immune cell responses to tumors, tumor cell responses to immune cells, and growth characteristics affecting tumor dormancy and responsiveness to therapy. Other recent publications have applied these reagents to the study of endothelial cell precursors²⁰, quiescent hematopoietic precursors²¹, transfected immune cells^22,23, antigen presenting cells²⁴, regulatory T cells²³, and more. The ability to label almost any cell type, combined with ease of labeling and ability to select among different fluorescence emissions for compatibility with other reagents in multicolor studies, suggests that these reagents will continue to help expand our understanding of tumor biology and anti-tumor immunotherapy.