The cell biology pipeline

Cell division cycle analysis in T. brucei is typically carried out to characterize the effect of depletion of a protein of interest. Depletion is usually carried out using RNAi directed against the target protein, or through construction of a conditional knockout cell line in which a single ectopic allele is under regulated and inducible expression (Wirtz et al. Reference Wirtz, Leal, Ochatt and Cross1999; LaCount et al. Reference LaCount, Bruse, Hill and Donelson2000; Wang et al. Reference Wang, Morris, Drew and Englund2000; Alibu et al. Reference Alibu, Storm, Haile, Clayton and Horn2005) (Fig. 3). After the generation of the required genetically modified cell line, it is vital that a number of control experiments are carried out and reported in order to verify its genotype and properties. These include confirming the integration of targeting constructs at the intended loci, confirming the expression, expression level and localization of any ectopic protein, and determining the kinetics of expression or depletion of the protein of interest (Fig. 3). Following validation of the cell line, the first step in its analysis is to assay for a growth defect upon modulation of the protein of interest, and select a timepoint or timepoints for further investigation. The choice of timepoint is made based on knowledge of the growth data and the level of protein depletion at that moment. Having identified the timepoint for analysis, the relative numbers of cells in different cell cycle states will be quantified using DNA labelling. Kinetoplast and nuclear DNA are labelled using DNA dyes such as DAPI or Hoechst, and the cells examined using either fluorescence microscopy or flow cytometry. Microscopic analysis of the cells allows the quantification of cells with one or two kinetoplasts (K) and one or two nuclei (N). The ordered progression of the cell division cycle, in which kinetoplast segregation precedes nuclear division, allows the easy partitioning of the cycle into three normal states: 1K1N, 2K1N and 2K2N. Accumulation of cells in one state or the appearance of cells with abnormal K/N numbers (for example, 1K0N, 0K1N and 1K2N) is immediately indicative of a possible cell division cycle effect. This approach has the advantage of decent resolution but is often done manually and is thus very labour intensive. Flow cytometry, while faster and more automated, permits only the sorting of G1/S and G2/M proportions of the cell cycle and the asynchrony of trypanosome populations further limits its resolution. Monitoring the actual duplication of DNA using for example BrdU is typically not considered unless replication of nuclear or kinetoplast material is the main focus of the project (Robinson and Gull, Reference Robinson and Gull1994).

Fig. 3. The cell biology pipeline. A generalized pipeline for the molecular cell biology analysis of a genetically modified cell line is shown. The pipeline is principally written for analysis of an RNAi or conditional knockout (CKO) cell line, but is applicable also to gene deletion and ectopic overexpression cell lines. Not every control is applicable for every type of cell line. Note that this assumes that the localization of the target protein has already been validated.

A much greater problem, common to both approaches, is that the majority of the cell biological events associated with the cell division cycle – replication of the basal body, Golgi, ER exit site, flagellar pocket, flagellar pocket collar – are completed in the interval between the 1K1N and 2K1N states (Fig. 2). Around 70% of the cells in an asynchronous population of T. brucei will be 1K1N, yet this category covers everything from interphase cells, which have just completed cytokinesis, to cells with an almost completely replicated complement of organelles (Archer et al. Reference Archer, Inchaustegui, Queiroz and Clayton2011). Greater consideration of the cell biological states that lie between 1K1N and 2K1N cells is clearly needed.

Next, or coincident with the microscopic analysis of K/N states, is morphological characterization of the cells. At its most basic this involves imaging of live or fixed cells using transmitted light to determine whether there are any obvious abnormal phenotypic effects – partially or wholly detached flagella, enlarged flagellar pockets, impaired motility, and accumulation of cells showing arrested or impaired cytokinesis are a number of the most common examples. Further characterization can be accomplished using indirect immunofluorescence analysis with antibodies specific for various cellular structures or organelles. These assays allow determination of whether the duplication and/or segregation of major cellular features is affected; tabulation of these observations alongside the K/N counts can generally allow a rough estimation of where in the cell division cycle things begin to go awry. An important control is confirmation that the affected cells do indeed have lower levels of the target protein so that morphological changes can be directly correlated with impaired function.

It is vitally important that such assays are carried out as soon as phenotypic effects become manifest, ideally at or just before the point when the affected population begins to exhibit a growth defect. The seeming lack of some cell cycle checkpoints in T. brucei means that impaired cells almost never arrest at a particular stage – instead, organelle and DNA replication continues unabated in the absence of cell division, leading to the generation of polyploid ‘monster’ cells with multiple organelles and flagella. It is impossible to infer the original defect in such cells and the phenotypes are not informative. Only by determining what are the primary effects of a protein's loss can its likely function be inferred with any degree of confidence, and even then teasing apart direct and indirect effects can be challenging (Hammarton et al. Reference Hammarton, Monnerat and Mottram2007).

Once the morphological characterization is complete and an appropriate timepoint or timepoints for analysis have been determined, functional assays must be utilized (Fig. 3). These allow specific hypotheses concerning the protein's function to be tested. The wide availability of reverse genetics tools facilitating the deletion, overexpression, mutation, and truncation of genes in T. brucei is a distinct advantage of the system.

This morphological and functional characterization is extremely labour intensive and additionally carries a very large element of bias – only candidate structures/organelles are interrogated, so discovering unexpected phenotypic consequences of protein depletion (such as might be revealed in genetic suppressor screens) are not possible. The choice of timepoint is likewise subjective. An additional problem comes in the quantification of such assays. There exists no common standard for the conduct of these experiments, and the number of independent experiments (ideally at least three) is not always disclosed. Quantification is often carried out manually. Detailed understanding of statistics is not widespread in the life sciences, and the misuse of P-values to infer whether a result is real or not is as frequent here as elsewhere. For detailed practical recommendations on the proper processing and presentation of quantitative datasets, the reader is referred to the following articles: Cumming et al. Reference Cumming, Fidler and Vaux2007; Vaux, Reference Vaux2014; Klaus, Reference Klaus2016; Weissgerber et al. Reference Weissgerber, Garovic, Milin-Lazovic, Winham, Obradovic, Trzeciakowski and Milic2016. Undersampling (which compromises interpretation) and oversampling (which reduces efficiency) are also problems.

Cell synchronization in T. brucei

The asynchronous growth of T. brucei populations has long been an impediment for accurate cell division cycle analysis. There appears to be small stochastic variation in the cell division cycle length in individual cells, resulting in a passive asynchronization of cell division cycle progression at a population level. Consequently, determining what are the primary effects of a protein's loss is made extremely difficult by the fact that such loss is initiated simultaneously in cells covering all stages of the cell cycle. Additionally, the depletion kinetics are slow and specific to individual proteins. In both RNAi and conditional knockouts, protein depletion depends on mRNA degradation or transcription cessation, followed by gradual protein loss as the target is turned over according to its intrinsic half-life. Mitotically-regulated proteins will exhibit very fast depletion as they are naturally degraded each cell cycle; most other proteins often exhibit very slow depletion as they are more stable, in general far more stable than their mRNAs. Thus, determining a protein's likely function currently requires consideration of how quickly it is simultaneously being lost from cells at each stage of the cell cycle – a task that is complex in the extreme.

To date, four main approaches have been utilized for synchronization of T. brucei, which will be considered in the following paragraphs. They are:

• • starvation-induced,

• • hydroxyurea-mediated,

• • Vybrant DyeCycle,

• • elutriation.

Starvation-induced synchronization is the simplest of the protocols. It requires growing cells to stationary phase, and then releasing them by dilution into fresh media (Gale et al. Reference Gale, Carter and Parsons1994). The absence of drugs or other manipulation is a plus, although allowing the cells to accumulate in stationary phase may carry its own set of consequences, as noted below. The cells do not achieve full synchrony, with a majority accumulating at G1, but a small number are still in G2/M phases (Gale et al. Reference Gale, Carter and Parsons1994; Archer et al. Reference Archer, Luu, de Queiroz, Brems and Clayton2009, Reference Archer, Inchaustegui, Queiroz and Clayton2011). Upon release, the cells remain semi-synchronous for around 12 h, then resume asynchronous proliferation. Despite its simplicity, the principle drawbacks of this method are the incomplete synchronization achieved, the likelihood that the ‘synchronization’ is actually a recovery phase, and the high degree of variation between replicates (Archer et al. Reference Archer, Inchaustegui, Queiroz and Clayton2011). Starvation is also known to induce the formation of stress granules (Cassola et al. Reference Cassola, De Gaudenzi and Frasch2007; Fritz et al. Reference Fritz, Vanselow, Sauer, Lamer, Goos, Siegel, Subota, Schlosser, Carrington and Kramer2015). The protocol has been almost exclusively applied to procyclic form cells. Use of starvation in bloodstream form cells has not been intensively pursued, although there are indications that some cell arrest occurs (Morgan et al. Reference Morgan, Laufman, Otieno-Omondi and Black1993). However, growing pleomorphic cell lines to stationary phase will induce the formation of stumpy cells, and in monomorphic cell lines a pseudo-stumpy morphology is produced. For the reasons mentioned above, the use of starvation-induced synchronization cannot be recommended.

Hydroxyurea-mediated synchronization has probably been the most popular of the techniques to date. Hydroxyurea inhibits ribonucleotide reductase, one of the enzymes involved in the synthesis of dNTPs (Hofer et al. Reference Hofer, Schmidt, Gräslund and Thelander1997). The drop in cellular dNTP levels inhibits DNA replication, causing cells initiating or undergoing S-phase to accumulate. Early work using hydroxyurea on procyclic and bloodstream from T. brucei was discouraging however, with population growth being inhibited but DNA synthesis apparently not (Brun, Reference Brun1980; Mutomba and Wang, Reference Mutomba and Wang1996). This approach was revisited by the Englund laboratory in 2008, and a protocol for successful synchronization was reported (Chowdhury et al. Reference Chowdhury, Zhao and Englund2008). The protocol was rapidly adapted for bloodstream form cells, with similar results (Forsythe et al. Reference Forsythe, McCulloch and Hammarton2009). The exact mechanism remains unclear, as kinetoplast S-phase is completed while nuclear division is strongly inhibited, leading to an accumulation of 2K1N cells. Synchronization is thus achieved by slowing progression through the cell cycle, rather than arresting it. Once released from the block by hydroxyurea washout, the cells synchronously progress through the remainder of the cell cycle, before gradually becoming asynchronous over the course of the subsequent cycle. The hydroxyurea protocol is simple and appears to be nontoxic; however, its principle drawback is the relatively late stage at which cell cycle slowdown occurs – 2K1N. Most organelle and cellular structure replication has already occurred at this stage, which greatly limits the utility of the protocol for studying earlier events in the cell division cycle (Fig. 2). The relatively short time period in which the cells remain synchronized limits the opportunity for analysis. The use of chemical treatments generally to synchronize cells has been discouraged (Shedden and Cooper, Reference Shedden and Cooper2002). The principle objection to chemical (or starvation)-induced synchronization is that it is often not clear that true synchronization has been achieved. Furthermore, the treatment may itself alter gene expression patterns.

Vybrant DyeCycle protocols are not true synchronization methods but instead produce selective enrichment of different cell cycle states. They involve the use of commercial DNA dyes (Thermo Fisher Scientific) coupled with fluorescence-activated cell sorting of the labelled cells. Vybrant DyeCycle Orange has been successfully used in procyclic cells (Siegel et al. Reference Siegel, Kawahara, Degrasse, Janzen, Horn and Cross2008a , Reference Siegel, Hekstra and Cross b ). Vybrant DyeCycle Violet was subsequently used in bloodstream form cells (Kabani et al. Reference Kabani, Waterfall and Matthews2010). Despite their simplicity and applicability to both procyclic and bloodstream form cells, the protocols are yet to be widely-adopted by the trypanosome research community. One key drawback is low yield, which reduces the number of cells available for downstream analysis. Arguably more important though is the low resolution afforded by fluorescence-activated cell sorting of T. brucei. In T. brucei, nuclear S-phase initiates before segregation of the duplicated kinetoplast DNA (Woodward and Gull, Reference Woodward and Gull1990; Siegel et al. Reference Siegel, Hekstra and Cross2008b ; Kabani et al. Reference Kabani, Waterfall and Matthews2010). Thus, selection of G1 cells will yield a mixed population of ‘true’ 1K1N cells in which kinetoplast duplication has not yet begun, and late 1K1N cells in which kinetoplast duplication is almost complete. As noted earlier, replication of many cytoplasmic organelles will already have occurred within this window (Fig. 2). Similarly, selection of G2/M cells yields a mixed population of late 1K1N cells, 2K1N cells and 2K2N cells. However, probably the principle objection is that the effect on the cells after labelling with the dye and sorting is unclear, with bloodstream form cells showing a significant delay in growth rates after the procedure (Kabani et al. Reference Kabani, Waterfall and Matthews2010). The effect of the dye on cell division cycle progression in procyclic cells post-sorting has not been reported.

Elutriation is a mechanical process in which cells are sorted on the basis of size and density using a counterflow mechanism. Centrifugal forces are used to sediment cells towards the outer edge of an elutriation chamber, while an inward flow of buffer solution produces a counterflow drag force that selectively sorts cells within the population towards the centre of the chamber. By modulating the centrifugal force and buffer flow, different sizes of particles can be eluted from the chamber according to their individual sedimentation properties. A ‘double-cut’ elutriation (DCE) protocol was recently developed in a landmark paper from the Clayton laboratory (Archer et al. Reference Archer, Inchaustegui, Queiroz and Clayton2011). This protocol involves first sorting an asynchronous population to obtain the largest set of cells – i.e. cells at a relatively late stage of the cell division cycle and soon to undergo cytokinesis. These cells are then cultured for around 1 h, and subsequently subjected to a second round of elutriation. The brief incubation period allows time for the cells to undergo cytokinesis, and the second round of elutriation allows selection of the smallest cells, namely those in early G1 phase. The cells remain synchronized for around one cell cycle, and progress through the replication process at a normal rate. A crucial advantage of this protocol is that it allows analysis of the earliest events in the cell division cycle, unlike the hydroxyurea protocol. It also requires no chemical treatment, and although the approximate yield from the original population input is only 5% it can be scaled up easily. To date, the protocol has been applied in two elegant studies of the mitotic kinase TbPLK (Lozano-Núñez et al. Reference Lozano-Núñez, Ikeda, Sauer and de Graffenried2013; McAllaster et al. Reference McAllaster, Ikeda, Lozano-Núñez, Anrather, Unterwurzacher, Gossenreiter, Perry, Crickley, Mercadante, Vaughan and de Graffenried2015). In the first, the DCE protocol was combined with the use of an analogue-sensitive TbPLK to provide rapid and specific inhibition of kinase activity at different stages of the cell cycle, and consequent dissection of the requirements for TbPLK at those time points. The follow-up work combined the DCE and analogue-sensitive TbPLK approach with proteomics in order to identify putative binding partners and substrates of the enzyme. The DCE protocol clearly represents the best of the available protocols for cell synchronization in T. brucei, and the only one that is really compatible with biochemical and proteomic analysis. However, its main drawback is that to date it has been used solely on procyclic cells and it remains unclear whether it can be used for bloodstream form ones (Archer et al. Reference Archer, Inchaustegui, Queiroz and Clayton2011). The smaller size of bloodstream form cells relative to procyclic ones may impair their effective sorting in the procedure.

In summary, of the available protocols for cell synchronization in T. brucei described to date, none represents a truly optimal solution. Only DCE is able to produce synchronized populations of cells at early G1 phase, which progress through the cell cycle without an abnormal delay, and at present this technique has been optimized for procyclic cells only. None of the protocols is able to yield a population of cells that remain synchronized for longer than one cell division cycle. This latter fact already makes synchronization somewhat refractory to RNAi or conditional knockout approaches, as the depletion kinetics are too slow. Moreover, the rapid desynchronization of the artificially-synchronized cells is consistent with an intrinsic variation in the timing of cell cycle initiation and/or progression between individual cells in a clonal population. This stochasticity in the timings must have a biological, and possibly also an adaptive, reason. This begs the question as to whether artificial synchronization should be performed at all, despite its utility.

Electron microscopy-based techniques (tomography, serial block face)

Two electron microscopy-based techniques – electron tomography and serial block-face scanning electron microscopy (SBF-SEM) – both present excellent solutions to the global analysis problem (Gluenz et al. Reference Gluenz, Wheeler, Hughes and Vaughan2015). Both techniques carry the enormous advantages of very high resolution and unbiased acquisition of data from all contrast-stained cellular structures and organelles simultaneously. Dual-axis electron tomography utilizes thick sections that are tilted in 1°–2° increments through a total path of around 120°–140°. Rotating the section through 90° allows the subsequent collection of data along an orthogonal axis, and the datasets can then be combined and aligned computationally to generate a three-dimensional image of the sampled cellular volume. SBF-SEM involves mounting an entire embedded sample within a SEM equipped with a diamond knife (Denk and Horstmann, Reference Denk and Horstmann2004). The face of the block is imaged, and then sliced off with the knife. Sequential iteration of this process results in data acquisition through a large sample volume composed of many (hundreds or thousands) ultrathin sections. The two techniques are compatible in that sample preparation is the same. Separately and together, they have already resulted in a number of seminal papers on trypanosome morphology and ultrastructure, with the Vaughan laboratory's discovery of a cell surface groove guiding flagellum biosynthesis in bloodstream form cells a particular highlight (Hughes et al. Reference Hughes, Towers, Starborg, Gull and Vaughan2013).

The principle drawbacks of both methods are the extremely demanding technical requirements in terms of both hardware and training. The hardware is expensive and specialized, which necessitates a considerable degree of institutional investment; sample preparation times – which will impact the quality of the data – are also relatively long. The need for advanced training and a period for the acquisition of tacit knowledge and experience mean that both techniques are off-limits to non-specialists. Tomography has a fairly small sampling volume, and the generation of a large number of independent datasets is very time-consuming.

Fluorescence microscopy approaches

Fluorescence microscopy approaches are almost a counterpoint to the electron microscopy-based ones. The principle advantages of fluorescence microscopy are its short preparation time, and lower (although still not insignificant) technical costs. A key advantage relative to electron microscopy is the ability to observe live cells – chemical fixation can introduce artefacts, and the dehydration and plastic embedding steps required for many electron microscopy protocols causes cells to shrink. This makes the analysis of fixed cells (unavoidable for electron microscopy) ultimately less accurate for some morphological parameters, particularly cell length. Recent advances in super-resolution imaging offer the opportunity to close the gap on electron microscopy in terms of resolution, although these techniques are not for casual users. The main disadvantages of fluorescence microscopy approaches stem from its lower resolution compared with electron microscopy, and its requirement for the use of labels (either dyes, protein tags or antibody conjugates), which result in the selective illumination of one or only a small number of cellular structures. This unavoidably introduces an element of bias to the experiment design. In addition, its easy availability has resulted in a proliferation of protocols with consequently little standardization across the field and published data that are not always in agreement. A recent example are the two reported localizations for the phospholipid phosphatidylinositol-(4,5)-bisphosphate (Demmel et al. Reference Demmel, Schmidt, Lucast, Havlicek, Zankel, Koestler, Reithofer, de Camilli and Warren2014; Cestari and Stuart, Reference Cestari and Stuart2015). One paper reported a predominant localization at the flagellar pocket membrane; the other a diffuse localization throughout the plasma membrane and flagellar membrane. Publication of experiment metadata is not standard, and there are few enforced community standards for crucial procedures such as colocalization.

Both fluorescence and electron microscopy-based techniques allow a focus on cellular organelles and structures rather than individual proteins. While delineating the structure and function of individual proteins remains the ultimate goal of the reductionist approach to cell biology, proteins are generally not visible unless present in large complexes or macromolecular arrays. The higher visibility of organelles makes them a more appropriate focus, particularly in the context of analysing cell division cycle progression and phenotypes associated with it. It is worth noting too that correlated light and electron microscopy (CLEM) approaches potentially offer the best of both worlds.

In terms of scientific progression, an additional burden is the fact that work is currently carried out in multiple labs spread across disparate geographical locations. This results in an overall low efficiency in terms of rate of progress, and would benefit from ongoing studies being better able to bootstrap onto the conclusions of prior ones – more community-level organization, in other words. In this respect, the relatively small size and high interactivity of the T. brucei research network represents a distinct though relatively untapped advantage. Recent and ongoing community-level projects at different scales such as TriTrypDB and the TrypTag initiative offer encouraging signs that this nascent potential is beginning to be realized (Aslett et al. Reference Aslett, Aurrecoechea, Berriman, Brestelli, Brunk, Carrington, Depledge, Fischer, Gajria, Gao, Gardner, Gingle, Grant, Harb, Heiges, Hertz-Fowler, Houston, Innamorato, Iodice, Kissinger, Kraemer, Li, Logan, Miller, Mitra, Myler, Nayak, Pennington, Phan, Pinney, Ramasamy, Rogers, Roos, Ross, Sivam, Smith, Srinivasamoorthy, Stoeckert, Subramanian, Thibodeau, Tivey, Treatman, Velarde and Wang2010; Dean et al., Reference Dean, Sunter and Wheeler2016; TrypTag, 2016).

Concluding remarks

In this review, the existing methods for cell division cycle analysis and synchronization in T. brucei have been summarized and evaluated. Techniques for global cell analysis have been compared, and the promise offered by automated and high-throughput computational methods highlighted. The adoption of ISS as a new tool for cell division cycle analysis has been advocated. The advent of single-cell transcriptomics (and potentially even proteomics) is a parallel development that will also enable the analysis of asynchronous populations. By taking advantage of ongoing advances in computational biology, this perhaps offers an unprecedented opportunity for hearing trypanosome bodies talk.