This protocol adapts cell cycle measurements for use in a mass cytometry platform. With the multi-parameter capabilities of mass cytometry, direct measurement of iodine incorporation allows identification of cells in S-phase while intracellular cycling markers enable characterization of each cell cycle state in a range of experimental conditions.
The regulation of cell cycle phase is an important aspect of cellular proliferation and homeostasis. Disruption of the regulatory mechanisms governing the cell cycle is a feature of a number of diseases, including cancer. Study of the cell cycle necessitates the ability to define the number of cells in each portion of cell cycle progression as well as to clearly delineate between each cell cycle phase. The advent of mass cytometry (MCM) provides tremendous potential for high throughput single cell analysis through direct measurements of elemental isotopes, and the development of a method to measure the cell cycle state by MCM further extends the utility of MCM. Here we describe a method that directly measures 5-iodo-2′-deoxyuridine (IdU), similar to 5-bromo-2´-deoxyuridine (BrdU), in an MCM system. Use of this IdU-based MCM provides several advantages. First, IdU is rapidly incorporated into DNA during its synthesis, allowing reliable measurement of cells in the S-phase with incubations as short as 10-15 minutes. Second, IdU is measured without the need for secondary antibodies or the need for DNA degradation. Third, IdU staining can be easily combined with measurement of cyclin B1, phosphorylated retinoblastoma protein (pRb), and phosphorylated histone H3 (pHH3), which collectively provides clear delineation of the five cell cycle phases. Combination of these cell cycle markers with the high number of parameters possible with MCM allow combination with numerous other metrics.
Mass cytometry enables detection of approximately 40 parameters by taking advantage of the high resolution and quantitative nature of mass spectroscopy. Metal-labeled antibodies are used instead of fluorophore conjugated antibodies that allow for a higher number of channels and produce minimal spillover1,2. MCM has advantages and disadvantages in regard to cell cycle analysis in comparison to flow cytometry. One major advantage of MCM is that the large number of parameters enables the simultaneous measurement of cell cycle state across a large number of immunophenotypically distinct T-cell types in highly heterogeneous samples. MCM has been successfully used to measure the cell cycle state during normal hematopoiesis in human bone marrow3 and transgenic murine models of telomerase deficiency4. Analysis of cell cycle state in acute myeloid leukemia (AML) showed that cell cycle correlated to known responses to clinical therapies, providing an in vivo insight into functional characteristics that can inform therapy selections5. A second advantage of mass cytometric cell cycle analysis is the ability to measure a large number of other functional markers that may be correlated with cell cycle state. Recent work has been able to correlate protein and RNA synthesis with cell cycle state through the use of IdU and metal tagged antibodies to BRU and rRNA6. This kind of highly parametric analysis measuring cell cycle state across numerous populations in a continuum of differentiation would be nearly impossible with current flow cytometry technology. The major disadvantage of MCM is the lack of comparable DNA or RNA stains as those used in fluorescent flow cytometry (e.g., DAPI, Hoechst, Pyronin Y, etc.). Fluorescent dyes can give relatively precise measurements of DNA and RNA content, but this precision is only possible due to the changes in the fluorescent properties of these dyes that occur upon intercalation between nucleotide bases. MCM analysis is thus unable to measure DNA or RNA content with similar precision. Instead, mass cytometric cell cycle analysis relies on measurements of proteins related to cell cycle state such as cyclin B1, phosphorylated retinoblastoma protein (pRb), and phosphorylated Histone H3 (pHH3) combined with direct measurement of the iodine atom from IdU incorporation into S-phase cells. These two measurement approaches yield highly similar results during normal cellular proliferation, but can potentially be discordant when cell cycle progression is disrupted.
Measurement of the number of cells in each cell cycle phase is important in understanding normal cell cycle development as well as cell cycle disruption, which is commonly observed in cancers and immunological diseases. MCM provides reliable measurement of extracellular and intracellular factors using metal-tagged antibodies; however, measurement of the S-phase was limited as the iridium-based DNA intercalator was unable to differentiate between 2N and 4N DNA. In order to define cell cycle phases, Behbehani developed a method that utilizes IdU with a mass of 127, which falls within the range of the mass cytometer and allows direct measurement of cells in S-phase3. This direct measurement circumvents the need for secondary antibodies or use of DNA denaturing agents such as acid or DNase. In conjunction with intracellular cycling markers, it allows high resolution of cell cycle distribution in experimental models.
This protocol adapts cell cycle measurements from common flow cytometry protocols for MCM. Our methods provide a convenient and simple way to include cell cycle parameters. IdU incorporation of in vitro samples requires only 10 to 15 minutes of incubation at 37 °C, which is shorter than most BrdU staining protocols that recommend incubation times of several hours3,7. IdU and BrdU incorporated samples can be fixed using a proteomic stabilizer and then stored for some time in a -80 °C freezer. This allows large numbers of IdU stained samples to be archived for batch analysis without reduction in sample quality.
1. Preparation of IdU stocks
2. IdU incubation and sample preservation
3. Staining samples for mass cytometry
4. Mass cytometer operation
NOTE: Mass cytometry operation can be machine specific. It is always advisable to check the CyTOF user’s manual before operation. Additionally, there are currently two JoVE articles dealing with machine start up and maintenance9,11.
5. Data analysis
Utilizing HL-60 cells and a human bone marrow aspirate it is possible to show how experimental conditions can affect cell cycle distribution and analysis. First, the gating strategy must be established to demonstrate how the cell cycle phases are derived. In Figure 1 we show the establishment of the singlet gate, which is important in separating cellular debris and doublets, establishing a single cell population. For cell lines the singlet gate is all that is needed to move onto cell cycle analysis (Figure 2a). For human samples immunophenotypic populations typically need to be established prior to cell cycle analysis, since the exact boundaries of each cell cycle gate can vary across different cell types. Once the populations have been established (usually by gating on surface markers that define that population) the cell cycle gates will then need to be established. Figure 2b demonstrates the establishment of the S-phase on the IdU vs CyclinB1 biaxial plot. This plot is also used to establish the boundary of the G2/M-phase gate (Figure 2f). Once the G2/M-phase is established the remainder is the G0/G1-phase gate (Figure 2f). The IdU vs pRb is used to establish the pRb+ cycling population first by establishing a gate on IdU incorporating cells (Figure 2g,i). The pRb+/IdUneg population outside this gate is the G0-phase (Figure 2j). M-phase is established on the IdU vs pHH3 where M-phase cells express high levels of pHH3 and exhibit no IdU incorporation (Figure 2k). In the event that pRb is not included the G0-phase can be replicated using Ki67 in a similar way to the method described above (Figure 3a). If IdU incorporation failed or was not performed it is still possible to determine relative cycling fractions using Ki67 and pRb. By using the Ki67 and pRb biaxial two distinct populations form, a pRb+/Ki67+ double positive and a pRblow/Ki67low population. The double positive population represents cells in cycle, while the low represents cells not in cycle (Figure 3b). Using IdU incorporating cells and pRblow cells with no IdU incorporation we show that the S-phase is primarily in the pRb+/Ki67+ population while the G0-phase is primarily in the pRblow/Ki67low population.
Cell cycle analysis relies on good experimental technique especially during the IdU incubation step. While IdU incorporation is flexible (being applicable in cell culture, bone marrow aspirates, and even murine studies), it necessary to perform IdU incorporation and fixation without disrupting the cell cycle state of experimental interest. IdU labeling and thus downstream cell cycle analysis can significantly be affected by time and temperature as indicated by Figure 4. Cells that remain too long in enclosed vessels or at that might be encountered in sample shipment or sample transport between locations, will have reduced S-phase fraction and not be accurate for cell cycle analysis (Figure 4a). Short time periods though, those under an hour in total, will have normal cell cycle distribution indicating that quick transport may not negatively cell cycle analysis (Figure 4b). Another important modifier is cryopreservation that is routinely used in most laboratories. When examining cell cycle state in cryopreserved cells a long equilibration period may be required before cells return to active cell cycling, which may still not reflect the pre-cyropreservation cell cycle state (Figure 4c).
Primary human samples are often composites of multiple different cell types, these different cell types can have different sensitivities to processing leading to different cell cycle gating. In two bone marrow aspirates that were IdU labeled immediately, stored for 30 minutes before IdU labeling, or cryogenically stored after Ficoll separation there are differences between each sample and population (Figure 5a,b). Two immunophenotypic populations were examined for differences in IdU incorporation; T-cells (CD45high/CD3high) and monoblasts (CD33+, HLADR+, CD11blow, CD14neg). With the correct combination of surface marks, it is possible to examine further immunophenotypic populations. In marrow #2 there was a noticeable T-cell activation effect after 30 minutes storage that was not seen in the monoblasts from the same patient (Figure 5b). Like cultured cells there were noticeable changes in IdU labeling after cryogenic storage that was also dependent on population. Marrow #1 had reduction in the T-cell population but increase in monoblasts IdU labeled fractions when compared to baseline (Figure 5a,b), Marrow #2 showed reduction in both T-cells and monoblasts when compared to baseline (Figure 5a,b). Frozen cells then require a notable incubation period before returning to normal cell cycle state and this can influence studies that rely on modifying cell cycle state or cell cycle state as a metric of drug or experimental effect.
Another benefit of MCM is the ability to discriminate cells in cell cycle arrest or that have abnormal cell cycle distribution. While DNA dyes commonly used in flow cytometry are able to discriminate between 2N and 4N DNA content, they are very bright, which can greatly complicate measuring other parameters from that laser. IdU, however, only takes one mass channel and has minimal spill over allowing for other markers to be used in cell cycle determination. MOLM13 cells that were irradiated show decreased IdU incorporation and decrease in M-phase when compared to control cells (Figure 6). Disruption of the normal cell cycle checkpoints might alter the apparent cell cycle state by MCM. Looking at pH2AX and cPARP populations in the non-irradiated cells the pH2AXlow and cPARPlow population shows normal cell cycle distribution while cells expressing higher levels of pH2AX or cPARP localize mainly in the G0/G1-phase which is expected (Figure 6a). In the irradiated cells the pH2AXlow and cPARPlow, however, the cells are almost entirely localized in the G0-phase, while the pH2AXhigh and cPARPlow cells show a cell cycle arrest phenotype with IdU incorporation and localization to the G0/G1-phase and G2-phase with an absence of M-phase. The pH2AXhigh and cPARPhigh cells also show cells incorporating some IdU and localizing to the G0/G1-phase indicative of radiation damage (Figure 6b).
Figure 1: Establishing the singlet gates using 191-Ir by Event Length and also Gaussian parameters, Residual and Offset.
The differences in T-cell (CD45+/CD3+) and S-phase (IdU+) between an ungated sample (a), an event length vs 191-Ir singlet gate (b), or a singlet gate combined with Gaussian parameters, residual and offset (c). Singlet gating removes debris, doublets, and beads shown in the loss of the pRbhigh population on the right corner of the biaxial. This singlet gate can be further optimized by including Gaussian parameters such as residual and offset, removing more debris. Please click here to view a larger version of this figure.
Figure 2: The gating schema for establishing cell cycle gates for G0, G1, S, G2, and M phases using IdU, CyclinB1, pRb, and pHH3.
The singlet gate is established to remove doublets and debris (a). The S-phase must be established (b), once the S-phase is established the IdU+ population can be used to establish the G2/M-phase boundary (c,d). The establishment of the G2/M-phase boundary establishes the bounds of the G0/G1-phase population (f). The pRb+ and G0-phase population are established on the IdU vs pRb biaxial. The IdU+ cells (h) are used to establish the boundary for the pRb+ population (i). The boundary of the pRb+ population establishes the boundary of the G0-phase population (j). The M-phase is established on pHH3+ cells that are IdU– (k). Please click here to view a larger version of this figure.
Figure 3: Establishing cell cycle gates without the use of pRb or without the use of IdU incorporation.
Drawing the G0-phase can also be done using Ki-67 following the same gating strategy that was used with pRb, if pRb is not included in the experiment (a). If IdU incorporation failed or was not performed it possible to still recover the relative cell cycling fractions through the use of Ki67 and pRb. Ki67 and pRb double positive expression is correlative to cells in cycle as evidenced by demonstrating the IdU+ cells are found primarily in the double positive population (b). The Ki67 and pRb low population correlates with the G0-phase or not cycling population demonstrated by pRblow/IdUneg cells being found in the Ki67 and pRb low population. This method cannot discriminate individual cell cycle phases but can still be used to determine relative cycling fractions in experimental conditions. Please click here to view a larger version of this figure.
Figure 4: Representative Figures of the effect of different storage conditions on the cell cycle distribution of HL-60 cells.
Cells were incubated for an hour followed by an hour rest at the stated temperature conditions in sealed tubes (a). There are noticeable effects on cell cycle distribution when compared to the control. HL60 cells were kept in sealed tubes at room temperature for 30 minutes, a situation that may occur in a clinical setting, before IdU incorporation (b). Sealed tubes kept at room temperature for 30 minutes did not show appreciable cell cycle differences. The effect of cryogenic storage was investigated on cell cycle in HL60 cells, where a sample was taken before cryogenic storage and a sample taken after an hour rest following a week in cryogenic storage (c). One hour post thaw the cell cycle distribution is affected and cell cycle distribution does not return to normal until approximately a week following thaw. Please click here to view a larger version of this figure.
Figure 5: Processing can have an effect on patient samples and representative images are shown between two different patients showing the (a) T-cells (CD45high/CD3high) and (b) Monoblasts (CD33+, HLADR+, CD11blow, CD14neg).
Between marrow one and marrow two it is clear that in marrow 2 there was a T-cell activation effect during the 30 minutes rest whereas marrow one had no such effect. Marrow one showed possible activation effect in the monoblasts population after 30 minutes while marrow two did not. In both marrows, however, it was clear that cryogenic storage impacted cell cycle distribution regardless of cell type. Please click here to view a larger version of this figure.
Figure 6: MOLM13 cells were either left as controls or irradiated using an X-ray irradiator at 10Gy.
MOLM13 cells show the cell cycle distribution in four different populations of pH2AX and cPARP expression. Control cells show minimal pH2AX and cPARP staining, with normal cell cycle characteristics being shown in the pH2AXlow and cPARPlow population (a). While the irradiated cells show an abnormal cell cycle distribution with the majority of cycling cells being located in the pH2AXhigh and cPARPhigh, indicating cell cycle disruption (b). The non-damaged, pH2AXlow and cPARPlow, cells show a lack of cycling characteristics are primarily found in the G0-phase. Without these markers these cells would appear as 4N and 2N cells in normal flow cytometry possibly confounding downstream cell cycle analysis. Please click here to view a larger version of this figure.
The examples presented here demonstrate how to use an MCM platform to analyze cell cycle distribution. It has also been demonstrated that cell cycle analysis is sensitive to experimental conditions such as time and temperature, which is an important consideration researchers must take when considering MCM for their cell cycle analysis14. Samples left in storage for a short period of time, no longer than an hour, will have IdU incorporation comparable to their normal state. Samples in a closed system for long periods of time, approximately 2 hours, will have reduced IdU incorporation, however, the relative cycling and non-cycling fractions will not change allowing coarse cell cycle analysis. Cryogenic storage and subsequent thawing are disruptive to normal cell cycle distribution for a significant period of time. Cryogenic storage has been noted previously to disrupt protein and RNA distribution but has only recently been shown to disrupt cell cycle14,15,16,17. Taken together these indicate that if long storage times are anticipated or cryogenic preservation of samples it would be better to stain the cells with IdU before storage or cryogenic storage, fix the cells and store them until analysis can be done. As cell cycle analysis by MCM does not require live cells this would enable researchers to bank precious samples and perform accurate downstream cell cycle analysis.
Cell cycle analysis by MCM is a robust system capable of deep interrogation of experimental models. Due to MCM high parameter count, approximately 40-50 mass channels, cell cycle analysis can be compounded with other intracellular or extracellular markers bypassing the need for sorting of cells based on immunophenotype which may cause significant cell cycle effects to be lost. The high parameter nature of MCM lends itself to examining effects in high-dimensional mapping applications such as SPADE and viSNE. While SPADE and viSNE are typically used to define immunophenotypic populations which can then be examined for cell cycle changes it would also be possible to map on cell cycle markers. Depending on experimental conditions mapping cell cycle markers in a high-dimensional space can show cell cycle correlation with drug effect or what immunophenotypic populations may be localizing to each cycle state3,5. While MCM may be limited by the lack of DNA binding fluorescent dyes this is compensated by direct IdU incorporation during S-phase cells and intracellular cell cycle proteins can be used to determine cell cycle state. These cell cycle proteins can also help discriminate between stages of cell cycle arrest that would otherwise appear as 3N or 4N in traditional flow using DNA binding dyes. Such highly parametric systems are not without disadvantages however, and it is sensitive to disruptions in cell cycle. We have shown that long storage times and cryogenic storage can significantly affect cell cycle distribution. This is especially important when trying to rationalize experimental effects for drugs that may affect cell cycle distribution. Treatment of drugs designed to affect cell cycle on frozen primary samples may give false data on cell cycle effect when used immediately after thawing from cryogenic storage. MCM is a versatile technology for cell cycle analysis that can be applied to a number of experimental models and is especially suited to deep profiling of heterogeneous systems. As with other highly parametric methods it is necessary to have a carefully designed experiment with appropriate considerations for how processing and experimental effects will affect cell cycle analysis.
The authors have nothing to disclose.
The authors would like to thank the efforts of Palak Sekhri, Hussam Alkhalaileh, Hsiaochi Chang, and Justin Lyeberger for their experimental support. This work was supported by the Pelotonia Fellowship Program. Any opinions, findings, and conclusions expressed in this material are those of the author(s) and do not necessarily reflect those of the Pelotonia Fellowship Program."
Bovine Serum Albumin (BSA) | Sigma | A3059 | Component of CSM |
Centrifuge | Thermo Scientific | 75-217-420 | Sample centrifugation |
Cleaved-PARP (D214) | BD Biosciences | F21-852 | Identification of apoptotic cells |
Cyclin B1 | BD Biosciences | GNS-1 | G2 Resolution |
Dimethylsulfoxide (DMSO) | Sigma | D2650 | Cryopreservative |
EQ Four Element Calibration Beads | Fluidigm | 201078 | Internal metal standard for CyTOF performance |
FACS Tube w/ mesh strainer | Corning | 08-771-23 | Cell strainer to remove clumps/debris before CyTOF run |
Fetal Bovine Serum (FBS) | VWR | 97068-085 | Cell culture growth supplement |
Helios | Fluidigm | CyTOF System/Platform | |
Heparin | Sigma | H3393 | Staining additive to prevent non-specific staining |
IdU (5-Iodo-2′-deoxyuridine) | Sigma | I7125 | Incorporates in S-phase |
Ki-67 | eBiosciences | SolA15 | Confirmation of G0/G1 |
MaxPar Multi Label Kit | Fluidigm | 201300 | Metal labeling kit, attaches metals to antibodies |
Microplate Shaker | Thermo Scientific | 88880023 | Mixing samples during staining |
Paraformaldehyde (PFA) | Electron Microscopy Services | 15710 | Fixative |
pentamethylcyclopentadienyl-Ir(III)-dipyridophenazine | Fluidigm | 201192 | Cell identification during CyTOF acquisition |
p-H2AX (S139) | Millipore | JBW301 | Detection of DNA damage |
p-HH3 (S28) | Biolegend | HTA28 | M-phase Resolution |
Phosphate Buffered Saline (PBS) | Gibco | 14190-144 | Wash solution for cell culture and component of fixative solution |
p-Rb (S807/811) | BD Biosciences | J112906 | G0/G1 Resolution |
Proteomic Stabilizer | SmartTube Inc | PROT1 | Sample fixative |
RPMI 1640 | Gibco | 21870-076 | Cell culture growth medium |
Sodium Azide | Acros Organics | AC447810250 | Component of CSM/Antibody buffer, biocide |