Cell Lineage Analyses and Gene Function Studies Using Twin-spot MARCM

Critical Steps within the Protocol

Steps 1.1.1, 1.2.1, 2.3, 2.5, and 3.2 are critical for obtaining good tsMARCM results. Tissue-specific-GAL4 drivers that do not express GAL4 in neural progenitors are preferred for steps 1.1.1 and 1.2.1. Avoid over-crowding the animals grown in the fly-food vials in step 2.3. Because the induction latency, the expression level of FLP upon heat-shock, and the average dividing time of neural progenitors (i.e., NBs and GMCs) are not known, it is best to vary the heat-shock time (10, 20, 30, 40, and 50 min) on different samples of the same genotype and developmental stage in step 2.5. This will allow for the determination of the optimal heat-shock time in order to obtain the best ratio of tsMARCM clones of interest. In addition, be careful during fly brain dissection in step 3.2 in order to obtain intact brain samples; this is crucial for accurately counting neuronal cell numbers. For researchers who are interested in using tsMARCM for studying the function of genes on other chromosome arms (e.g., 2R, 3L, and 3R), tsMARCM-ready flies should be assembled as in step 1.2 by choosing appropriate pairs of transgenes of FRT, UAS-mCD8.GFP.UAS-rCD2i,and UAS-rCD2.RFP.UAS-GFPi, which are indicated in Table 1. We also recommend that the female tsMARCM-ready flies carry hs-FLP on the X chromosome (e.g., hs-FLP^[22],w; UAS-rCD2::RFP,UAS-GFPRNAi,FRT^40A; +; +), which facilitates repeated tsMARCM experiments.

Modifications, Troubleshooting, and Limitations of the Technique

The efficiency and perdurance of RNAi suppressors and reporters are the major determinants for whether neural lineage analyses and gene function studies can be successfully performed in tsMARCM experiments. Generally, the combination of low perdurance of RNAi suppressors and reporters derived from neural progenitors, together with the high expression of RNAi suppressors and reporters in post-mitotic neurons, permits better tsMARCM results. Because the expression of RNAi suppressors and reporters is controlled by GAL4 drivers, researchers should select strong GAL4 drivers that express GAL4 in differentiated neurons but not in neural progenitors. For example, GAL4-OK107 is a strong GAL4 driver that expresses GAL4 in mushroom body (MB) progenitors and their derivatives, MB γ, α'/β', and α/β neurons^11,19. Unfaithful labeling of MB γ neurons has been observed in two clonal patterns (i.e., single-cell associated with single-cell clones and two-cell-GMC associated with multi-cellular-NB clones) when GAL4-OK107 was used in tsMARCM experiments¹¹. Intriguingly, the above issue can be solved by using MB247-GAL4, a different MB GAL4 driver that expresses GAL4 in differentiated MB neurons¹¹.

Not all GAL4 drivers can be used in the tsMARCM system or in other mosaic labeling systems (e.g., MARCM, Q-MARCM, and twin-spot generator)^4,20,21 if the driver is not expressed in the neurons of interest. Therefore, prior to using tsMARCM for neural lineage analyses, GAL4 drivers should be tested for their coverage of the neurons of interest in the dual-expression-control MARCM with a pan-cell driver (e.g., tubulin promoter-LexA::GAD)²². We note that tsMARCM clones sometimes cannot be analyzed unambiguously, especially when a GAL4 driver with a complex neural expression pattern is used (e.g., green lateral horn (LH) neurons prevent clear visualization of the axonal elaboration pattern of the VL2p adPN in Figure 3B and 3C). However, the above issue of ambiguous patterns also occurs in other mosaic labeling systems (e.g., MARCM, Q-MARCM, and twin-spot generator)^4,20,21 because the limitation originates with the GAL4 driver. We also note that, to conduct rescue experiments for the authentication of gene functions, the rescue transgenes should be engineered with the target sequences of the RNAi suppressors in 5' or 3' untranslated regions of the transgenic rescue constructs, or an alternative twin-spot labeling system (e.g., coupled MARCM, a combination of Q-MARCM and conventional MARCM)²¹ should be adopted.

Significance of the Technique with Respect to Existing/Alternative Methods

Besides tsMARCM, twin-spot generator and coupled MARCM would also, at least in theory, allow for the visualization of neurons derived from common neural progenitors in two distinct colors (a comparison of these three mosaic labeling technologies has been presented previously¹⁵). The minimum number of transgenes required for assembling twin-spot generator, the original version of tsMARCM (or the new version of tsMARCM), and coupled MARCM is 4, 8 (or 6), and 9, respectively. However, tsMARCM is the only system that has been used to conduct comprehensive neural lineage analyses^12,15,18,23. Site-specific recombination does occur in non-mitotic cells in the twin-spot generator system, and this leads to the co-expression of the two reporters in the same cell, making it challenging to use this system to study the development of the central nervous system¹⁵. In contrast, mosaic clones obtained in the tsMARCM and coupled MARCM systems are derived from mitotic cells. However, because the coupled MARCM system requires two independent drivers to control the expression of the different reporters in the Q-MARCM and conventional MARCM systems, this raises concerns about the labeling of neurons derived from common neural progenitors if the two independent drivers do not faithfully label the same neurons in the neural lineages of interest¹⁵.

Future Applications or Directions after Mastering the Technique

Using tsMARCM to conduct high-resolution neural lineage analyses
By applying the approach used in Figure 3 to systematically analyze tsMARCM clones of adPNs, 40 types of adPNs, with a total of 82 neurons, were identified in a previous study¹². Among these 40 types of adPNs, 35 were uni-glomerular adPNs (i.e., innervating a single glomerulus in the antennal lobe (AL)), which relay olfactory information to other regions of the Drosophila brain¹² (see Figures 2-4 in Yu et al., 2010). The other five types of adPNs were poly-glomerular adPNs (i.e., innervating multiple glomeruli in the AL), which could integrate olfactory information from the AL and provide an additional level of functional complexity to the olfactory system¹² (see Figures 3E-3H and 4Z in Yu et al., 2010). Embryonic-born adPNs have only one neuron per type, whereas larval-born adPNs are represented by multiple neurons per type, suggesting that there is interplay between specificity and redundancy in the olfactory system¹² (see Figures 2-4 in Yu et al., 2010). Interestingly, neurons produced within certain developmental windows are destined to die, and many types of adPNs are found with a fixed number of neurons in the ALad1 neural lineage, suggesting that the number and types of adPNs for assembling the olfactory system are tightly regulated¹² (see Figures 2H, 4V, and 5 in Yu et al., 2010; also see Figure 3P–3T).

Using tsMARCM to accurately study gene functions
Besides birth-dating adPNs, tsMARCM has also been applied to determine the birth-order of six central complex neurons labeled by GAL4-OK107. These six neurons make up a small subset of neurons in the LALv1 neural lineage that contribute to the circuitry of the Drosophila central complex to control intricate motor patterns^2,11 (see Figure 4a and 4b in Yu et al., 2009). Interestingly, each of these six central complex neurons possesses a distinct neuronal morphology¹¹ (see Figure 4d-4h and 4o in Yu et al., 2009). All of them carry distinct axonal patterns in the lateral accessory lobe (LAL). The first four neurons distribute their dendrites to sub-compartments of the noduli, a sub-structure of the central complex, whereas the remaining two neurons project their dendrites to the ellipsoid body, another sub-structure of the central complex. Specifically, the 3^rd-born central complex neuron projects its dendrites to the ventral noduli, with a compact axonal pattern in the LAL, whereas the 4^th-born central complex neuron distributes its dendrites to the dorsal noduli, with a broad axonal pattern in the LAL¹¹ (see Figure 4f and 4g in Yu et al., 2009).

Beyond comprehensively identifying neurons, the power of tsMARCM would be better exemplified by studying gene functions in neurons of interest, since the tsMARCM system permits better phenotypic analyses by enabling the examination of identical neurons in different animals. For example, tsMARCM detects a single temporal fate change that requires chronologically inappropriate morphogenesis (chinmo, a BTB-zinc finger nuclear protein) in the 3^rd-born central complex neurons¹¹ (see Figure 4i, 4p, and 4w in Yu et al., 2009). If the chinmo mutation had been characterized in the original MARCM system (i.e., single green-labeled neurons, without the labeling of neurons derived from the associated sides of the twin cells as the reference), the chinmo-deficient neuron shown in Figure 4i from Yu et al., 2009 would have been misidentified as the 4^th-born, rather than the 3^rd-born, central complex neuron. This would have resulted in the misinterpretation of chinmo function as controlling axonal guidance rather than specifying temporal fate¹¹ (double-arrow in Figure 4i from Yu et al., 2009). However, by using the magenta multi-cellular-NB sides of tsMARCM clones as the reference (via counting the cell numbers in Figure 4m, 4n, and 4p from Yu et al., 2009), the chinmo-deficient neuron shown in Figure 4i from Yu et al., 2009) was proven to be the 3^rd-born central complex neuron, which therefore led to the correct conclusion that chinmo functions in temporal fate specification for central complex neurons¹¹. Taken together, the above results highlight the important advantages of the tsMARCM system for high-resolution phenotypic analyses, thereby accurately disclosing gene functions in neurons (or other types of cells, if applicable) for future neural developmental studies (or the developmental studies of other tissues).