This study aims to understand how neurons from different stem cell lineages wire with each other to form a specific circuit motif. In particular, we wanted to test the hypothesis that Drosophila larval nerve cord circuits are assembled by preferential connectivity between temporal cohorts (Meng and Heckscher, 2021). First, however, we needed to better understand the relationship between temporal cohorts, circuit function, and circuit anatomy. Specifically, we needed to distinguish between graded and sharp wiring transition models (see ‘Introduction). To do so, we need to determine the relationship between neuronal birth order and synaptic connectivity at single neuron resolution.

As a model, we use the NB3-3 lineage in the first abdominal segment (A1) of the nerve cord (Figure 1A). In A1, there is a left–right pair of NB3-3 neuroblasts that are thought to be identical. We will refer to these neuroblasts as NB3-3A1L/R. First, NB3-3A1L/R undergoes a type 1 division to self-renew and generate a GMC (Baumgardt et al., 2014). The GMC divides to produce two neurons of unknown function (Figure 1B; Schmid et al., 1999). Next, NB3-3 begins to divide using a type 0 division pattern to self-renew and generate EL interneurons (Figure 1B; Baumgardt et al., 2014). The birth timing of ELs has been characterized using three different methods (Tsuji et al., 2008; Wreden et al., 2017; Mark et al., 2021). Nonetheless, the precise relationship between morphology and birth order for NB3-3A1L/R progeny remains unknown. Here, we determined the birth order and morphology of every neuron produced by NB3-3A1L/R. Using morphology, we identified matching neurons in the connectome. Using the connectome, we found all upstream synaptic partners. Finally, we analyzed the patterns of wiring using network science approaches.

First, we needed to determine birth order and morphology of EL interneurons at the single-neuron level. To do so, we used the gold standard in the field for lineage tracing, Twin-spot mosaic analysis with a repressible cell marker (ts-MARCM) (Yu et al., 2009b). Briefly, in ts-MARCM, heat shocks are delivered to dividing cells, rendering their progeny competent to express either a UAS-red or -green fluorescent reporter (Figure 1B and C). Reporter expression is driven by a cell-type-specific GAL4 line. Providing heat shocks at various times is used to reconstruct lineage progression. The morphology of singly labeled neurons is used to determine neuron identity (Figure 1E’’’’) and counting the number of alternatively labeled, subsequently born cells determines birth order (Figure 1E’–E’’’). To drive ts-MARCM reporter expression, we used EL-GAL4 that drives in all EL neurons, together with a newly generated genetic cassette that amplifies GAL4 expression (Figure 1D). We find that in segment A1, ELs are produced in the following order: A08x, A08j1, A08j3, A08j2, A08m, A08o, A08c, A08s, A08e3, A08e1, and A08e2 (Figure 2). Notably, all these ELs have been suggested to be NB3-3A1L/R progeny, and our data provides the first confirmation for several (A08o, A08j1-j3) (Mark et al., 2021; Wreden et al., 2017; Heckscher et al., 2015). Because we already knew the first two neurons in the lineage were non-ELs, our data precisely define the birth order for all neurons in NB3-3A1L/R lineage at cellular resolution (Figure 2).

Figure 2

Download asset Open asset

Next, we needed to find all NB3-3A1L/R neurons in the larval connectome because this provides high-resolution pictures of neuron morphology, including the locations of dendrites (i.e., regions with postsynaptic densities, Figure 3D, cyan) and axons (i.e., regions with presynaptic active zones, Figure 3D, red). To locate neurons of interest in the connectome we used two criteria: first, lineage-related neurons often have cell bodies that cluster and neurites that form a tight bundle entering the neuropil (i.e., the axon, dendrite and synapse-rich, cell body-free region of the CNS) (Schmid et al., 1999; Mark et al., 2021). We generated a new NB3-3-GAL4 line, which selectively labels NB3-3 (Figure 1—figure supplement 1). Using both NB3-3-GAL4 and EL-GAL4 to drive membrane GFP, we confirmed that the GFP-labeled neurons have clustered soma and bundled neurites (Figure 3A and B). Second, we used ts-MARCM data as ground truth to identify neurons with corresponding morphologies in the connectome (Figures 1–2). With these two criteria, we identified a lineage bundle that contained all the ELs plus two other non-EL neurons (Figure 3C). The non-EL neurons were an undifferentiated neuron, which has never been reported before, and a motor neuron, which is consistent with previous reports (i.e., MN 22/23, Figure 3D, top row; Schmid et al., 1999). Then, with the high-resolution morphology data, we asked to what extent is EL morphology correlated with birth order and temporal cohort borders. We found one morphological group of early-born ELs, which included A08x, A08j1, A08j3, A08j2, and A08m (Figure 3D, second row). The axons of these ELs project to the central brain, with one exception, A08m, which projects to the thorax. Their dendrites are found on both sides of the midline (ipsilateral and contralateral to the soma) with one exception, A08m, which has only contralateral dendrites. All project dendrites ventrally from the main neurite, and of these, two had been previously identified as early-born ELs (A08x and A08m) (Wreden et al., 2017). The second morphological group contained late-born ELs and included A08o, A08c, A08s, A08e3, A08e1, and A08e2 (Figure 3D, third and fourth rows). The axons of these ELs project to the central brain or remain local. Their dendrites are ipsilateral to the soma and project dorsally from the main neurite, with one exception, A08c, which projects dendrites both dorsally and ventrally. All of these ELs, except A08o, had been previously identified as late-born ELs (Wreden et al., 2017). Thus, we find morphological groupings of neurons in the NB3-3A1L/R lineage that are strongly correlated with neuron birth time and that reflect previous grouping of ELs into early-born and late-born temporal cohorts.

Figure 3

Download asset Open asset

Figure 3—source data 1

NB3-3A1L/R cortex neurite length.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig3-data1-v2.xlsx

Download elife-79276-fig3-data1-v2.xlsx

In a previous report, Mark and colleagues used the length of a neurite between the soma and the point at which it enters the neuropil as a proxy for neuronal birth order (Mark et al., 2021). This measure is called ‘cortex neurite length.’ Here, we sought to quantify the relationship between cortex neurite length and neuron birth order because for the first time we had a complete lineage with both precisely defined birth order information and cortex neurite length data. We plotted cortex neurite length for each neuron in the NB3-3A1L/R lineage as well as their average and found a strong positive correlation as measured by Pearson’s correlation (r(11) = 0.88, p<0.0001 and r(11) = 0.85, p=0.002 and r(11) = 0.92, p<0.0001 for A1L, A1R, and average, respectively; Figure 3E, Figure 3—source data 1). However, exact birth order cannot be accurately inferred by neurite length. We conclude that it is possible to roughly, but not precisely, infer birth order of neurons within a lineage using only anatomical data.

Next, to characterize the relationship between neuronal birth order and patterns of synaptic connectivity, we needed to comprehensively identify all the synaptic inputs to NB3-3A1L/R neurons. To do so, we mined the connectome and identified 4944 synaptic inputs on NB3-3A1L/R neurons, which came from 1179 different neurons (or neuronal fragments) (Figure 4—source data 1). We categorized the sources of NB3-3A1L/R synaptic inputs: a majority (61%) came from other nerve cord interneurons, followed by sensory neurons (19%), with the remainder being from difficult to annotate neuron fragments or neurons of the central brain (Figure 4A, Figure 4—source data 1). Because a large majority of inputs onto NB3-3A1L/R neurons come from sensory neurons and nerve cord interneurons, we focused on these in the rest of this article.

Figure 4

Download asset Open asset

Figure 4—source data 1

All synaptic input neuron information onto NB3-3A1L/R.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig4-data1-v2.xlsx

Download elife-79276-fig4-data1-v2.xlsx

Figure 4—source data 2

Sensory input onto NB3-3A1L/R neurons.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig4-data2-v2.xlsx

Download elife-79276-fig4-data2-v2.xlsx

Figure 4—source data 3

Summary of NB3-3A1L/R input neurons based on number of input.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig4-data3-v2.xlsx

Download elife-79276-fig4-data3-v2.xlsx

Figure 4—source data 4

NB3-3A1L/R input from L/R paired neurons.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig4-data4-v2.xlsx

Download elife-79276-fig4-data4-v2.xlsx

A major functional difference between early-born ELs and late-born ELs is their response to sensory stimuli (i.e., early-born ELs respond to vibration, but late-born ELs do not) (Wreden et al., 2017). First, we characterized the relationship between birth order and sensory neuron input. In the NB3-3A1L/R lineage, the earliest born, non-EL neurons get no input from sensory neurons (Figure 4B, Figure 4—source data 2). All early-born ELs receive input from chordotonal sensory neurons (CHOs), which agrees with previous findings (Wreden et al., 2017) and characterizes the sensory input onto A08j1-3 for the first time (Figure 4B). All late-born ELs get input from proprioceptive and other sensory neurons, but not CHOs (Figure 4B). This agrees with previous finding (Heckscher et al., 2015) and newly shows that A08o gets a low amount of proprioceptive input (Figure 4B). Thus, we conclude that based on input from sensory neurons, there are two sharp transitions in NB3-3A1L/R lineage progression, from non-EL neurons to early-born ELs and from early-born ELs to late-born ELs.

Next, we asked if there were similarly sharp transitions in NB3-3A1L/R lineage progression when looking at inputs from nerve cord interneurons. However, first, we wanted to understand how many synaptic contacts each interneuron made onto neurons in the NB3-3A1L/R. This is because neurons contributing few synapses may represent biological noise (promiscuous wiring) or technical noise (annotation errors). In our dataset, the number of contacts from one interneuron onto a NB3-3A1L/R neuron varied from 1 to 28 synapses. To examine the full distribution of interneuron inputs onto neurons inNB3-3A1L/R, we generated a histogram with number of neurons on the Y-axis and number of synapses onto a NB3-3A1L/R on the X-axis (Figure 4C, Figure 4—source data 3). A majority (79%) of interneurons contributed only one synapse onto an NB3-3A1L/R neuron (Figure 4C). To avoid analyzing potentially noisy data, we focused on the subset of nerve cord interneurons for which left–right (L/R) pairs formed a minimum of two synapses onto a NB3-3A1 neuron on one side and four synapses onto the corresponding NB3-3A1 on the other side (see ‘Materials and methods’ for details, Figure 4—source data 4). We identified 198 ‘L/R paired’ interneurons, which together provided 1990 synapses onto NB3-3A1L/R neurons. From these L/R paired interneurons, undifferentiated neuron gets no synaptic input, and MN22/23 gets inputs that largely do not synapse onto ELs (Figure 4D). For early-born ELs, many L/R paired interneurons synapse onto multiple neurons within the temporal cohort, but not other neurons in the lineage (Figure 4D). Similarly, for late-born ELs, L/R paired interneurons synapse onto multiple neurons within the temporal cohort, but not other neurons in the lineage (Figure 4D). We conclude that based on L/R paired interneuron input, NB3-3A1L/R lineage progression undergoes sharp transitions in patterns of input connectivity. Further, these transitions are correlated with temporal cohort borders.

The qualitative analysis above suggested that early-born and late-born ELs have distinct connectivity patterns. To quantitatively test this suggestion, we analyzed all input data, not selected subsets, using a network science-based approach, specifically distance analysis. Briefly, we treated inputs to a neuron as a vector that contained the number of synaptic contacts. The order of inputs was the same for every neuron. For every pair of neurons, we calculated the Euclidean distance between the two vectors, which produced a metric for the difference in connectivity between two neurons. We call this the ‘real’ value. For the distance of the real value to be zero, not only would the input partners need to completely overlap, but the number of synapses contributed by each of these partners would need to be identical. To provide an estimate of how the same set of neurons could be wired by chance, we computationally permutated the input vector for each neuron independently, preserving number of inputs, but shuffling input identity. We did this 100 times for each neuron to produce a ‘shuffled’ dataset. For shuffled datasets, Euclidian distances were computed. Real values were normalized to shuffled values to generate a z-score that allowed us to determine statistical significance (see ‘Materials and methods’). In examining the distribution of z-scores, early-born ELs have significantly similar patterns compared to other neurons in their temporal cohort, but not compared to other neurons in the lineage (Figure 5A, Figure 5—source data 1). Late-born ELs have significantly similar patterns compared to other neurons in their cohort, but not compared to other neurons in the lineage (Figure 5A). Importantly, in general, early-born and late-born ELs have significantly different connectivity from each other (Figure 5C, Figure 5—source data 2). This provides statistical support for the idea that early-born ELs and late-born ELs have sharp transitions in patterns of connectivity beyond what is expected by chance. Further, these sharp transitions in connectivity correlate with previously characterized circuit function and temporal cohort borders.

Figure 5

Download asset Open asset

Figure 5—source data 1

z-scores for similarity.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig5-data1-v2.xlsx

Download elife-79276-fig5-data1-v2.xlsx

Figure 5—source data 2

z-scores for differences.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig5-data2-v2.xlsx

Download elife-79276-fig5-data2-v2.xlsx

Figure 5—source data 3

Similarity z-scores for sensory only input.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig5-data3-v2.xlsx

Download elife-79276-fig5-data3-v2.xlsx

Figure 5—source data 4

Similarity z-scores for interneuron only input.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig5-data4-v2.xlsx

Download elife-79276-fig5-data4-v2.xlsx

It has been suggested that within a lineage temporal cohorts could be copies of each other, implying that although the specific input partners differ, the pattern of connectivity repeats (Wreden et al., 2017). Distance analysis allowed us to asses this idea by helping to visualize connectivity patterns between early-born versus late-born ELs (Figure 5A). To quantify patterns of connectivity we measured two features—similarity among left–right pairs of neurons and similarity among following pairs. An example of similarity among left–right pairs is A08j3 A1L and A08j3 A1R, both of which have significantly similar connectivity compared to A08x A1L (Figure 5A). In contrast, A08c A1L, but not A08c A1R, has significantly similar connectivity compared to A08o A1L (Figure 5A). In general, early-born ELs have significant similarities between left–right pairs of neurons (Figure 5B, open purple circles), whereas late-born ELs did not ( Figure 5B, open magenta circles). An example of similarity among following pairs is that A08c A1L has significantly similar connectivity compared to A08s A1L, which is adjacent to A08c in birth order (Figure 5A). Both early-born and late-born ELs have similarities among following pairs (Figure 5B, filled circles). The idea that neurons within a temporal cohort are more similar to neurons next born in the sequence shows that within a temporal cohort there are graded transitions in connectivity patterns. Additionally, early-born ELs have similarities with many neurons within their temporal cohort, whereas late-born neurons only tend to be similar to neurons that were born next in the sequence (Figure 5A). We conclude that temporal cohorts are not merely copies of each other and the computations performed by early-born and late-born ELs may differ.

Finally, we used distance analysis to ask what drives the clustering of EL interneurons into two cohorts—connectivity with sensory neurons, interneurons, or both? We repeated the distance analysis, dropping out either sensory neuron or interneuron inputs. For drop out of interneurons, only sensory neuron connections remain (Figure 5D, Figure 5—source data 3). In this case, there were still significant similarities in inputs specifically between neurons of a temporal cohort. However, for early-born ELs, the total number of neuron pairs with significant similarities was increased, whereas for late-born ELs, it was reduced. This suggests that interneuron input diversifies the early-born cohort, but unifies the late-born cohort. Data for drop out of sensory neurons is consistent with this idea (Figure 5E, Figure 5—source data 4). This may hint at differing developmental strategies for temporal cohort assembly.

In summary, we find that the NB3-3A1L/R lineage undergoes two sharp transitions during lineage progression—from non-EL neurons to early-born ELs and from early-born ELs to late-born ELs. Sharp transitions are seen with respect to morphology and patterns of input connectivity, supporting a sharp transitions model. Further, sharp transitions are correlated with previously defined early-born and late-born ELs temporal cohort borders (Wreden et al., 2017). Ultimately, these data bring into alignment the concepts of temporal cohorts, circuit function, and circuit anatomy at single-neuron resolution.

For the rest of this study, we focus on early-born ELs as a model to understand circuit assembly. Early-born ELs and late-born ELs contribute to different circuits (Wreden et al., 2017). In comparison to late-born ELs, the anatomical circuit motif in which early-born ELs are embedded is poorly characterized (Heckscher et al., 2015; Mark et al., 2021). Stem cell lineage-to-circuit relationships can differ depending on the type of circuit generated (Xu et al., 2014). And so, we needed to characterize the early-born EL circuit.

To understand the circuit to which early-born ELs contribute, we started by grouping neurons that synapse onto early-born ELs in A1L/R into classes based on morphology. We analyzed 1945 synapses formed on early-born ELs by 331 neurons (sensory neurons and nerve cord interneurons) (Supplementary file 2, Figure 6—source data 1). We grouped neurons into classes based on morphology. We calculated the total number of neurons within a class and the percentage of total synaptic input onto early-born ELs provided by neurons in each class. 6% of input came from 21 Basin interneurons (Figure 6A and E). Basins are a previously described class of excitatory neurons that have ipsilateral axons and dendrites (Ohyama et al., 2015). 12% of input are from 20 A08 interneurons (Figure 6B and E). In experiments described below, we find that A08 interneurons are early-born ELs from segments other than A1. 17% of synaptic inputs come from 32 Ladder interneurons (Figure 6C and E). Ladders are a previously described group of unpaired inhibitory neurons with cell bodies in the midline and left–right symmetrical arbors (Jovanic et al., 2016). 31% of synaptic inputs come from 72 CHOs (Figure 6D and E), which respond to vibration (Wreden et al., 2017; Ohyama et al., 2013). The remaining 34% of inputs come from 186 other interneurons, which fall into a wide variety of morphological classes (Figure 6E). In summary, in these four morphological classes, 145 neurons provide a majority of the total synaptic input (66%) onto early-born ELs in segments A1L/R.

Figure 6

Download asset Open asset

Figure 6—source data 1

Basin, EL, Ladder, and CHO input onto early-born ELs in A1L/R.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig6-data1-v2.xlsx

Download elife-79276-fig6-data1-v2.xlsx

Figure 6—source data 2

Distribution of input onto early-born ELs in A1L/R.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig6-data2-v2.xlsx

Download elife-79276-fig6-data2-v2.xlsx

Not only do neurons in the four classes (CHO, Ladder, Basin, EL) provide a majority of total synaptic input onto early-born ELs, but on a per neuron basis, these neurons tend to make more synapses onto early-born ELs. This is important from a functional perspective because neurons that provide multiple synaptic contacts onto a downstream are thought to be more effective in driving downstream neuronal activity. To investigate this quantitatively, we grouped CHOs, Ladders, Basins, and ELs into one group and all other neurons into a second group. We generated histograms for each group, binning by number of synapses onto early-born ELs, and counting the total number of neurons in each bin (Figure 6F, Figure 6—source data 2). Comparing these two histograms shows a significant difference in the distribution of synapse number (Kolmogorov–Smirnov, p<0.0001). For example, 36% of all neurons in the group containing CHOs, Ladders, Basins, and ELs, provide 10 or more synapses onto early-born ELs. For other neurons, it is 5%. Put another way, of all the neurons that provide 10 or more synapses onto early -born ELs in A1, >80% are CHOs, Ladders, Basins, or early-born ELs. We conclude that, in general, individual neurons in this group are more highly connected with early-born ELs in A1 compared to other neurons, and neurons in this group are also more likely to be important early-born EL activity.

After identifying these four classes of inputs, we next asked how they were anatomically arranged at the level of circuit motif. We consider early-born ELs to be the circuit outputs because they are excitatory and project to the central brain (or, for A08m, thorax) (Figure 7A–D, Figure 7—source data 1). Input to the circuit comes from CHOs, which provide direct excitatory input onto early-born ELs (yellow arrows, Figure 7A–D). CHOs also provide indirect input excitatory and inhibitory onto early-born EL in A1 via Ladders, Basins, and other early-born ELs (Figure 7A–D, orange, blue, and purple arrows, respectively). Thus, anatomically, early-born ELs in A1 are imbedded in a feed-forward circuit motif.

Figure 7

Download asset Open asset

Figure 7—source data 1

Breakdown of Basin, EL, Ladder, and CHO input onto individual early-born ELs.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig7-data1-v2.xlsx

Download elife-79276-fig7-data1-v2.xlsx

Because early-born ELs get direct excitatory and indirect inhibitory input from CHOs, this suggests that early-born ELs could be activated upon initial chordotonal stimulation and inactivated with delay. In previous calcium imaging experiments, we used a slow, presynaptic calcium sensor and found that early-born ELs respond to vibration (Wreden et al., 2017). Here, to get a more precise understanding of dynamics of EL activity, we monitored stimulus encoding using a cytoplasmic calcium sensor with faster signal decay. In response to vibration, which activates CHOs (Wreden et al., 2017), EL activity initially peaks and then the signal rapidly declines, although the vibrational stimulus remains (Figure 7E–G). These data support the idea that functionally at least a subset of ELs encode stimulus onset.

In summary, we find that all early-born ELs are embedded in a feed-forward circuit motif and that at least a subset of ELs encode the onset of vibrational stimuli.

After characterizing the circuit containing early-born ELs, we were in the position to test the hypothesis that this feed-forward circuit is assembled by preferential connectivity between distinct temporal cohorts (Meng and Heckscher, 2021). To test this, we needed to know to what extent could Basin, Ladder, and A08 interneurons be considered temporal cohorts. To be a temporal cohort, interneurons must be lineage-related and born within a tight time window. We used a combination of light microscopy and connectomics to address this question.

There are four pairs of Basin interneurons per segment, Basins 1–4 (Ohyama et al., 2015; Jovanic et al., 2016). Basins have been suggested to be a temporal cohort from NB3-5 (Wreden et al., 2017). NB3-5 is one of the first neuroblasts to form in Drosophila embryos (Broadus et al., 1995). It is one of the largest embryonic lineages, reportedly producing up to 36 neurons, a subset of which are found at larval stages (Figure 8A; Monedero Cobeta et al., 2017; Moris-Sanz et al., 2014). The first-born neurons express the CCAP neuropeptide, and many NB3-5 neurons die during embryogenesis (Monedero Cobeta et al., 2017; Moris-Sanz et al., 2014). Otherwise, the cell types produced by NB3-5 are poorly understood.

Figure 8

Download asset Open asset

First, we asked if Basins are likely to be a lineage-related set. To do so, we used ts-MARCM and detected ts-MARCM recombination events using Basin-GAL4, which is expressed in all Basin interneurons (Ohyama et al., 2015). First, we induced ts-MARCM clones using brief heat shocks at early times in development. This produced one or two clones per CNS (e.g., Figure 8B) consistent with the idea that recombination levels were low. We counted the number of Basin neurons in each clone and found that most often all four Basins were labeled (Figure 8B’ and K). This is consistent with the idea that recombination occurred in a single dividing neuroblast and that neuroblast produces all Basin interneurons. Next, we used ts-MARCM to determine the relative birth timing of Basins within the lineage. In this set of experiments, we provided heat shocks at a variety of developmental time points and scored the resulting progeny. We find that Basins are born in the following order: Basins 2, 1, 3, 4 (Figure 8A–J). Additionally, based on the types of clones produced at different times (see ‘Materials and methods’ for details), our data suggest Basins are born within a middle-to-late window during NB3-5 lineage progression (Figure 8K). We conclude that Basins are a lineage-related set of neurons born within a tight time window, consistent with the idea that Basins are a temporal cohort.

Next, we wanted to directly test the idea that Basins are progeny of NB3-5. To do so, we crossed a fly line expressing membrane-GFP in all Basins (Figure 9A) to three different fly lines, each expressing nuclear localized (nls) RFP in the NB3-5 lineage neurons. In all cases, we observed co-expression of Basin and NB3-5 lineage markers (Figure 9B–D’). We conclude that Basins are progeny of NB3-5.

Figure 9

Download asset Open asset

Figure 9—source data 1

NB3-5 input neuron information.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig9-data1-v2.xlsx

Download elife-79276-fig9-data1-v2.xlsx

Figure 9—source data 2

Similarity z-score for NB3-5 A1R.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig9-data2-v2.xlsx

Download elife-79276-fig9-data2-v2.xlsx

Figure 9—source data 3

Similarity z-score for NB3-5 A1L.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig9-data3-v2.xlsx

Download elife-79276-fig9-data3-v2.xlsx

Figure 9—source data 4

NB3-5 cortex neurite length.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig9-data4-v2.xlsx

Download elife-79276-fig9-data4-v2.xlsx

Basin neurons had already been identified in the connectome (Ohyama et al., 2015). And so, next, we used connectome analysis as an independent means of characterizing the development of the Basins. First, we observe Basin cell bodies are clustered and their neurites form a bundle both in light microscopy images and in the connectome (Figure 9A and E). This provides further support for the idea that Basins are a lineage-related set of neurons. Second, we assayed Basin birth order using cortex neurite length as a proxy. To do so, in connectome, we identified all neurons in the lineage bundle that contained Basin neurons (Figure 9E, yellow) because these neurons are likely to be part of the NB3-5 lineage. Then, we measured and plotted the cortex neurite lengths of putative NB3-5 neurons in the hemilineage that included Basins. In this set, Basins possess cortex neurite lengths that are among the longest (Figure 9H). These data are consistent with our ts-MARCM data, which suggest that Basins are middle-to-late born. Therefore, by two independent means we find that Basins are a lineage-related set of neurons born within a tight time window. This provides strong evidence that Basins are a temporal cohort.

In the first section of this article, we found that between neurons in the NB3-3A1L/R lineage there are sharp transitions in input connectivity patterns correlated with temporal cohort boundaries. However, it remained unclear if this was true for early-born and late-born temporal cohorts only, or if this represented a more general ‘rule.’ Because we identified Basins as a temporal cohort, we next examine the relationship between patterns of connectivity and Basin temporal cohort borders. As described in the first section of this article, using distance analysis, we quantified the significant similarities in neuronal input, this time analyzing neurons in the Basin hemilineage (Supplementary file 3). We find that Basins have statistically significant similarities in input connectivity with themselves compared to most other NB3-5 neurons (Figure 9G and H, Figure 9—source data 1–3). Thus, there are sharp transitions in connectivity patterns correlated with Basin temporal cohort borders (see ‘Discussion’ for more).

Ladders have been suggested to be progeny of midline neuroblast (MNB) (Babski et al., 2019). MNB is the only unpaired neuroblast (Wheeler et al., 2009). It produces up to three ventral unpaired midline motor neurons before generating up to six GABAergic unpaired midline interneurons (Figure 10A; Schmid et al., 1999). In the A1 segment of the connectome, there are six unpaired Ladder interneurons with neurites that enter the neuropil in a single tight bundle (Figure 10B). We confirmed Ladders are progeny of MNB by crossing a fly line expressing membrane-GFP in one Ladder to three different fly lines, each expressing nls RFP in the MNB lineage. In all cases, Ladder and MNB lineage markers are found in the same cell (Figure 10C–E). In addition, we used distance analysis to evaluate input connectivity between Ladders in the A1 segment (Supplementary file 3) and find statistically significant similarities (Figure 10F, Figure 10—source data 1–2). We conclude that Ladders are a late-born temporal cohort from MNB. Like other temporal cohorts, we find that the Ladder temporal cohort border correlates with sharp transitions in synaptic connectivity.

Figure 10

Download asset Open asset

Figure 10—source data 1

Ladder input neuron information.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig10-data1-v2.xlsx

Download elife-79276-fig10-data1-v2.xlsx

Figure 10—source data 2

Similarity z-scores for Ladder neurons.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig10-data2-v2.xlsx

Download elife-79276-fig10-data2-v2.xlsx

We noticed that at least some A08 interneurons that synapses with early-born ELs in A1 resembled early-born ELs in terms of morphology. Therefore, we tested the idea that these A08 interneurons could be NB3-3 progeny in segments other than A1. To do so, we used the multi-color flip out (MCFO) system to highlight the morphology of individual neurons within the expression domain of a GAL4 line (Nern et al., 2015). For GAL4 lines, we used either early-born EL-GAL4 or late-born EL-GAL4 (Wreden et al., 2017). We compared early-born and late-born EL morphologies to the A08 interneurons that formed synapses with early-born EL in A1. We found that a majority were early-born ELs from segments other than A1, including anterior (i.e., thorax) and posterior (i.e., A2) segments (Figure 11A–F, Supplementary file 1). This reveals inter-segmental synaptic connectivity between early-born ELs (Figure 11G). Functionally, ELs are inter-segmentally coordinated (Heckscher et al., 2015). The observed inter-segmental synaptic connectivity could help explain this observation.

Figure 11

Download asset Open asset

Because we found inter-segmental synaptic connectivity between early-born EL temporal cohorts, we wondered if this occurred between other temporal cohorts. We found inter-segmental connections between Basins and Ladders, but not late-born ELs (Figure 11G, Supplementary file 1, Supplementary file 3). In addition, in A1, we found intra-segmental (within-segment) synaptic connections between early-born ELs, Basins, Ladders, but not late-born ELs (Figure 11G, Supplementary file 1, Supplementary file 3). One commonality between early-born, Basins, Ladders, but not late-born ELs, is that they receive synapses from CHOs. This raises the possibility that inter- and intra-segmental synaptic connectivity between these temporal cohort is a feature important for the processing of vibrational stimuli. Further, these data reveal differences in connectivity between temporal cohorts, hinting at complexity in mechanisms driving their wiring.

In summary, we find that Basins neurons are a mid-to-late-born temporal cohort from NB3-5. Ladders are late-born temporal cohort from MNB. And A08 interneurons are early-born ELs from segment other than A1. This supports the hypothesis that in the Drosophila larval nerve cord circuits are assembled by preferential wiring between temporal cohorts. Further, we note that temporal cohorts that form synapses on early-born ELs come from 3 of the 30 neuroblast in the nerve cord. This finding rules out the following two models: (1) that morphologically similar neurons from different lineages synapse onto early-born ELs, and (2) that multiple temporal cohorts from a single neuronal stem cell synapse onto early-born ELs. Instead, it shows that neurons from a limited number of neuronal stem cell lineages selectively synapse onto early-born ELs.

To more fully understand assembly of this feed-forward circuit, we needed to understand the relative birth timings of early-born ELs, Basin, and Ladder interneurons. One model for motor circuit development suggests that early-born neurons wire with each other to generate circuits driving fast movements, whereas later-born neurons wire with each other to generate circuits driving refined movement (i.e., early-to-early and late-to-late model) (Fetcho and McLean, 2010; Ampatzis et al., 2014; Mark et al., 2021). Furthermore, we found that early-born ELs from different segments synapse with each other, raising the possibility that an early-to-early model could also explain the observed pattern of wiring. However, there is suggestive evidence that an exclusive early-to-early model cannot explain the full complexity of wiring patterns seen in motor circuits (Kishore et al., 2014; Menelaou et al., 2014; Song et al., 2018). Thus, here, we test the hypothesis that early-born ELs wire with neurons from other lineages, which are born at the same time as early-born ELs. Note that in the experiments described above we determined the birth order of neurons within a given lineage, but had no information about the relative birth timing of neurons between different lineages.

To determine the relative birth timing of early-born ELs, Basins, and Ladders, we started by asking when during neurogenesis each EL could be generated. We used ts-MARCM with EL-GAL4. We supplied heat shocks at different time points and scored the identity of singly labeled neurons. Heat shock supplied after 7 hr of development labeled a mixture of early-born and late-born ELs, whereas heat shock supplied after 9 hr labeled exclusively late-born ELs (Figure 12A and B). Thus, heat shocks after ~9 hr failed to label early-born ELs. Next, we asked when during neurogenesis the Basins and Ladders are generated. We used ts-MARCM with GAL4 lines that labeled individual Basin or Ladder neurons (Basin-1 or Ladder-D) (Figure 12C and D). We supplied heat shocks at different time points and scored for the proportion of CNS in which clones were induced. In both cases, heat shocks provided at 11 hr of development, or later, failed to label these neurons. Whereas heat shocks provided at 10 hr of development, or earlier, resulted in labeling (Figure 12E). Thus, heat shocks at ~10 hr are sufficient to label Basins and Ladders. Taken together these data strongly suggest that Basins and Ladders are generated after the early-born ELs. Notably, this is inconsistent with the model in the field.

Figure 12

Download asset Open asset

Because our data were inconsistent with our expectation, we performed an additional set of experiments to confirm our findings. Specifically, we performed dual ts-MARCM using both Basin-GAL4 and EL-GAL4 in the same larva. This approach allows for direct comparison of neurons generated at a given time in a single animal. We provided a series of heat shocks and scored the resulting clones as follows: EL clones were scored for neuron identity (e.g., A08j2), and this was used to assign ELs as early-born or late-born. Basin clones were scored for clone type (e.g., 1:3 or 1 neuron labeled in one color, 3 neurons in another color), and this was used to determine the time window of Basin production. Specifically, if all Basins were labeled in a single color (0:4 clone), the heat shock was provided before the neuroblast began to divide to produce Basins. Alternatively, if Basins were singly labeled (e.g., 1:3, 1:2, etc.), the heat shock was provided to cells as they were dividing to produce Basins. Within our dataset, three animals had labeled early-born ELs. In all of these animals, Basins were labeled in a single color, consistent with the idea that the neuroblast had not yet begun generating Basins (Figure 12F). In 17 animals, we found singly labeled Basin neurons, suggesting that Basins were being generated. In 16 of these 17 animals, late-born ELs were labeled (Figure 12F). Together, these data provide strong evidence that early-born ELs are born earlier than the Basin interneurons.

In summary, we find that between neurons in the feed-forward circuit, early-born EL are born before Basin and Ladder interneurons (Figure 12G). More generally, this demonstrates that circuit outputs from one lineage are born before circuit inputs from other lineages. Further, this rules out an exclusive early-to-early hypothesis that states that neurons born at the same time are circuit partners.

In the Drosophila larval nerve cord, individual neurons make different numbers of synapses onto their various downstream partners. In this section, we focus on neurons that are highly connected to early-born ELs, but not in the three lineages previously discussed. We considered a neuron to be highly connected if it has 10 or more synapses onto early-born ELs (Supplementary file 6). We choose this criterion for highly connected neurons, first, because a majority of synapses onto early-born ELs are made by these neurons, and second, because these neurons are likely to be important drivers of the activity of early-born ELs. In the sections above, we focused on Ladder, Basin, and Early-born EL interneurons, which, together with CHOs, account for the majority of neurons highly connected to early-born EL in A1 (Figure 13A, Figure 13—source data 1, Supplementary file 6). From these data, we learned two ‘rules.’ First, interneurons in the feed-forward circuit come from temporal cohorts in three lineages. Second, output interneurons from one lineage are born before input interneurons from other lineages. Additionally, there are six other highly connected neurons (Figure 13A–C). Here, we characterize the developmental origins of these other neurons to understand the extent to which the rules apply.

Figure 13

Download asset Open asset

Figure 13—source data 1

Summary of amount and timing of important input onto early-born ELs.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig13-data1-v2.xlsx

Download elife-79276-fig13-data1-v2.xlsx

First, we wanted to understand the extent to which the six other highly connected interneurons could be lineage-related. To assign lineage identity, we examined their cell body position and neurite fasciculation pattern. The six other neurons likely come from six different lineages (Figure 13B and C). Of these, one is NB3-5, which generates Basin neurons. This means that NB3-5 neurons contribute more synapses onto early-born ELs than we had previously appreciated. We conclude that together NB3-5, MNB, and NB3-3 generate a majority (75%) of interneurons that are highly connected to early-born ELs. However, the relationship is not exclusive, and interneurons from other lineages do also synapse multiple times onto early-born ELs.

Next, we wanted to understand the extent to which the six other highly connected interneurons could be born after early-born ELs. To do so, we examined the cell body position of each along the medio-lateral axis (Figure 13B’). Of the six, one had a medial cell body close to the neuropil, suggesting that it was early-born within its lineage; four had cell body positions on or close to the lateral edge of the CNS, suggesting that they were late-born with their respective lineages; and one had an intermediate cell body position that is difficult to interpret (Figure 13C). These data, along with our previous birth dating from Basins, Ladders, and ELs, suggest that 70% of highly connected interneurons are born after early-born ELs. We conclude that the large majority of highly connected neurons in this circuit are born after the circuit output neurons (early-born ELs), but that there are exceptions to this ‘rule’.

Finally, for one of the other highly connected other interneurons, we wanted to experimentally validate our lineage assignment and more directly determine its birth time relative to early-born ELs. We chose to focus on neuron A03 upstream of A08m because it is inferred to be progeny of NB7-1, and the lineage NB7-1 is extremely well-characterized. NB7-1 delaminates early during neurogenesis (Broadus et al., 1995). The first six divisions of NB7-1 generate motor neurons and undifferentiated motor neuron siblings (Seroka and Doe, 2019; Lacin et al., 2009). Then, NB7-1 divides approximately 14 more times, generating interneurons (Meng et al., 2019), one of which is thought to be neuron A083 upstream of A08m (Mark et al., 2021). We labeled neuron A03 upstream of A08m using a NB7-1-GAL4 driver line to create MCFO clones, which validates the lineage assignment (Figure 14A and B). Then, in the connectome, we measured the cortex neurite length of neuron A03 upstream of A08m and the other interneurons within the same hemilineage as a proxy for birth order (Figure 14C and D, Figure 14—source data 1). At the earliest, neuron A03 upstream of A08m is the sixth-born interneuron, and because there are six divisions of NB7-1 before interneurons are made, neuron A03 upstream of A08m must be born on or after the 12th division (Figure 14E). To compare the birth timing of neuron A03 upstream of A08m to the birth timing of early-born ELs, we present the following logic. NB3-3 and NB7-1 are generated at ~400 min and ~230 min after egg laying, respectively (Doe, 1992). Neuroblasts divide every 45 min. And so, NB3-3 is four divisions behind of NB7-1 ([400 min –230 min]/45 divisions/minute = 3.8 divisions). Early-born ELs are made during the second to sixth divisions of NB3-3, which corresponds to the fifth to ninth divisions of NB7-1 (Figure 14E). Because NB7-1 produces neuron A03 upstream of A08m no earlier than the 12 divisions, we conclude that it must be born after the early-born ELs. In addition, we noticed that three other interneurons in the same hemilineage as ‘A03 upstream of A08m’ synapse onto early-born ELs, albeit less strongly. The cortex neurite length of these neurons shows that they largely have birth times adjacent to neuron A03 upstream of A08m (Figure 14D), which is consistent with the idea that these neurons are a temporal cohort. Using the same logic as described earlier, we find that all the NB7-1 interneurons that synapse onto early-born ELs are born after the early-born ELs themselves (Figure 14E). Thus, analysis of NB7-1 shows that neurons from a third lineage follows the patterns identified for Ladder and Basin temporal cohorts. Furthermore, these data show that additional temporal cohorts have neurons that synapse onto early-born ELs, but not as extensively as neurons in the Ladder, Basin, or early-born EL temporal cohorts.

Figure 14

Download asset Open asset

Figure 14—source data 1

NB7-1 naming and cortex neurite length.

https://cdn.elifesciences.org/articles/79276/elife-79276-fig14-data1-v2.xlsx

Download elife-79276-fig14-data1-v2.xlsx

The Drosophila larval connectome is a resource that can be used to understand circuit assembly (Meng and Heckscher, 2021). However, because the connectome is an anatomical dataset, a major challenge is to develop approaches that connect anatomy to development. In this article, we develop several approaches. For example, we optimized ts-MARCM for use in Drosophila embryos (Figure 1). ts-MARCM is the gold standard for determining birth order in Drosophila (Yu et al., 2009b). However, for technical reasons, ts-MARCM has been used only in adults. We discovered ts-MARCM clones can be robustly generated in early stage larvae with the addition of an amplifying and immortalizing gene cassette (Figure 1). In addition, we independently validated several recently developed methods for inferring developmental origin based on anatomical features in the Drosophila larval connectome (Mark et al., 2021). Specifically, our validations include use of neurite bundles as a proxy for lineage-relatedness (Figures 3B, C, 9A, E, 10B, 14C) and use of cortex neurite length as a rough proxy for birth order within a lineage (Figures 3E, 9H and 14B). Finally, we adapted network science methods (distance analysis) to characterize the patterns of connectivity in connectome data (Figures 5, 9 and 10). These approaches should be useful for Drosophila neurobiologists and beyond.

We also developed NB3-3A1L/R as the first entire lineage for which birth order, morphology, and input connectivity is known at single neuron and synapse precision (Figures 2—4, Supplementary file 1). One reason this is important is because NB3-3 has been extensively studied in embryos and much is known about molecular marker expression of NB3-3 progeny at the single-neuron level (Tsuji et al., 2008; Wreden et al., 2017; Baumgardt et al., 2014). Because our dataset achieves cellular resolution, good guesses about the embryonic molecular–larval morphological pairings can be made using single-neuron birth order as a point of cross-reference. Such integrated data generates detailed and testable predictions. For example, our data predict that Castor expression in the late-born ELs promotes projection neuron morphology. Additionally, here, we generated a new NB3-3-GAL4 line that can be used to manipulate gene expression in NB3-3 (Figure S1). Thus, NB3-3A1L/R is a model lineage in which transcription factor expression in neuronal stem cells can be linked to circuit assembly and tested for function, and we have generated tools that will enable hypothesis testing in the future.

A circuit motif is a pattern of synaptic connections between a set of specific neuron types that can be found across brain areas and across species (Braganza and Beck, 2018). Circuit motifs have been suggested to represent the physical substrates of ‘computational primitives’ (Marcus et al., 2014). There a are small number of fundamental, recurring circuit motifs (e.g., feed-forward, feedback, lateral inhibition, etc.; Luo, 2021). And so, a new conceptual approach in this article is to ask how are specific circuit motifs assembled during development. In this study, we identified a new feed-forward circuit motif (Figure 7). Generally, feed-forward motifs are characterized as a pattern of connectivity in which one neuron (or neuron type) provides both direct and indirect input onto a second neuron (or neuron type). Feed-forward circuit motifs are common, found in many animals (e.g., nematodes, insects, mouse) and in many brain regions (e.g., somatosensory systems, olfactory systems, neocortex) (Schafer, 2016; Anton and Homberg, 1999; Harris and Shepherd, 2015). Thus, feed-forward motifs are fundamental to neural signal processing. Feed-forward motifs can be further subdivided into feed-forward inhibitory or feed-forward excitatory circuit motifs, depending on the transmitter types of the neurons involved(Luo, 2021). Notably, anatomically the pattern of synaptic connectivity between neurons is the same in either motif subtype, and so in this study we do not distinguish between the two.

Here, we find that, in general, early-born ELs get direct excitatory synapses from CHOs and indirect (excitatory or inhibitory) input from chordotonals via Basins and Ladder interneurons (Figure 7). Notably, there are differences in connectivity patterns between early-born ELs. For example, although all early-born ELs get direct input from chordotonals and Ladders, A08j1-3s get left–right symmetrical inputs, whereas A08x and A08m get asymmetrical inputs (Figure 7). This could correspond to functional differences. For example, A08x and A08m may be involved in left–right asymmetrical signal detection. Teasing apart the diversity of computations performed by each individual EL interneurons is the domain of future studies.

Because we characterized the development of a feed-forward circuit in Drosophila, we have information that allows us to search for similar circuits in other insect species. For example, in Drosophila, ELs express the transcription factor, Even-skipped (Eve, Figures 1–2). In Locust and other insects, lateral interneurons also express Eve (Bevan and Burrows, 2003). In Drosophila, MNB generates H-shaped, inhibitory interneurons (Ladders), which get input from sound/vibration sensitive, CHOs (Figures 7 and 11, Supplementary file 3). In Locust, MNB generates H-shaped, inhibitory interneurons, which encode sound stimuli (Thompson and Siegler, 1991). Further, in both Drosophila and Locusts, MNB interneuron progeny express the transcription factor Engrailed (Schmid et al., 1999; Kearney et al., 2004; Patel et al., 1989). Thus, the neuronal components of the Ladder to early-born EL circuit are conserved, which raises the possibility of circuit-level conservation.

A major unanswered question in developmental neuroscience is how the mechanisms that generate neuronal diversity contribute to the formation of functional neuronal circuits (Sagner and Briscoe, 2019). Part of the answer lies in the observation that temporal cohorts are subunits of lineages, linked both to larval circuit anatomy/function as well as to the embryonic gene expression programs that generate neuronal diversity (Meng et al., 2019; Meng et al., 2020; Seroka and Doe, 2019). Thus far, temporal cohorts had been looked for and found in 8 out of 30 lineages in the nerve cord (Wreden et al., 2017; Meng et al., 2019; Meng et al., 2020; Mark et al., 2021). Here, we identify three additional temporal cohorts—Basins, Ladders, and Notch OFF NB7-1 interneurons—in three lineages—NB3-5, MNB, and NB7-1 (Figures 8—10 and Figure 14), bringing the number to 11. This underscores the idea that temporal cohorts are common.

One open question about temporal cohorts was to what extent are temporal cohort borders associated with sharp changes in connectivity. Previous studies had identified temporal cohorts and linked them with function and connectivity (Wreden et al., 2017; Meng et al., 2019; Meng et al., 2020; Mark et al., 2021). However, these studies lacked the resolution to distinguish between a ‘graded’ or ‘sharp’ wiring transition models. We identified four temporal cohorts (early-born ELs, late-born ELs, Basins, Ladders) in three lineages (NB3-3, NB3-5, MNB), all of which have sharp changes in connectivity correlated with temporal cohort borders (Figures 5 and 9–10). For the Basin temporal cohort, we note that Basins have similar connectivity patterns with two additional NB3-5 progeny, ‘Down and Back’ and ‘Crescent’ neurons. Crescent is adjacent to Basins in birth order, whereas Down and Back is born much earlier. From this, we learn two things: (1) there can be significantly similar connectivity between neurons in one hemilineage that have nonadjacent birth times. (2) There can be significantly similar connectivity between neurons with adjacent birth times, but of different morphological classes. Another example of morphological variants with similar connectivity are late-born ELs, which can be either local or projection neurons. This underscores the idea that temporal cohorts are subunits of hemilineages defined by birth within a tight time window, rather than defined by similar neuronal morphology per se. Further, we note that within a temporal cohort, sequentially born neurons are often, but not always, the most similar in terms of connectivity. For example, within the early-born EL temporal cohort, the fifth-born neuron (A08m) is more similar to the first-born neuron (A08x) than its temporal neighbor (A08j2), the fourth-born EL. Thus, our data, combined with previous studies, show how temporal cohorts are developmental units related to circuits both at the functional and anatomical levels.

A second open question about temporal cohorts was the extent to which they are copies of each other. One reason this is an interesting question relates to circuit evolution. For the evolution of gene function, a popular model is a ‘duplicate and diverge’ model. Similarly, temporal cohorts could be duplications within a lineage whose function could then diverge, thereby driving circuit evolution. Such an idea motivated us to ask to what extent are temporal cohort copies. For example, do early-born ELs process chordotonal stimuli in a manner identical to how late-born ELs process proprioceptive stimuli? Our data suggest a more complex picture. Early-born and late-born ELs differ in their connectivity patterns—including left–right and following pair similarities (Figure 5B) and inter-and intra-segmental connectivity patterns (Figure 11G–H), suggesting that early-born and late-born ELs are likely to process information differently. Independent of the evolutionary implications, our data reveal previously unknown diversity in the structure of connectivity between neurons within adjacent temporal cohorts, which may indicate a diversity in underlying circuit assembly mechanisms.

A third open question is what sets up the borders of temporal cohorts. Previous work used motor neuron temporal cohorts of NB7-1 and NB3-1 as a model to address this question (Meng et al., 2019; Meng et al., 2020). In these stem cells, mis-expression of temporal transcription factors, Hunchback, Pdm, and Castor, modulates the number of motor neurons in a temporal cohort, without changing the size of the lineage. Thus, temporal transcription factors are able to regulate motor neuron temporal cohort borders. But it remains unclear how temporal transcription factors do so. Two, not mutually exclusive possibilities are that they act as transcriptional co-factors to induce differential gene expression programs and/or that they act as pioneer factors to alter the chromatin landscape. For NB3-3, we note that during the 7th to 11th divisions, which generate late-born ELs, NB3-3 expresses Grainyhead (Tsuji et al., 2008), which raises the possibility that Grainyhead may define the late-born EL cohort. Testing this idea is an important future direction.

It has been hypothesized that nerve cord circuits are assembled by preferential connectivity between distinct temporal cohorts (Meng and Heckscher, 2021). Our data provide experimental support for this hypothesis. Specifically, we find that interneurons from three temporal cohorts wire together to form a feed-forward circuit—early-born ELs from NB3-3, mid-to-late-born Basins from NB3-5, and late-born Ladders from MNB (Figures 8—10). Although a vast majority of both the total synaptic input and the strongly connected individual neurons come from these lineages, there are also neurons from other lineages that synapse on early-born ELs (Figures 13–14).

Notably, one other study provided limited supported for the hypothesis that Drosophila nerve cord circuits are assembled by preferential connectivity between distinct temporal cohorts (Mark et al., 2021). This study focused on the Jaam-to-late-born EL-to-Saaghi circuit (Heckscher et al., 2015). Jaams are later-born interneurons in the NB5-2, Notch OFF hemilineage, and Saaghis are later-born interneurons in the NB5-2 Notch ON hemilineage (Mark et al., 2021). These data raised several possibilities: (1) there could be global alignment between lineages (e.g., all NB3-3 neurons get synapses from NB5-2 neurons). (2) Notch ON/OFF pairs of neurons might be pre-/postsynaptic partners of neurons within a temporal cohort. (3) Birth order-matched temporal cohorts from different lineages might selectively wire together (e.g., early-to-early and late-to-late connectivity). However, our data demonstrate that none of these possibilities are generally true. Instead, we find a diversity in the manner in which temporal cohorts associate. The one consistent theme is that a limited number of temporal cohorts highly interconnect.

For most of this discussion, we labeled neurons as ‘early-born’ and ‘late-born.’ These labels refer to the birth order of neurons within a lineage. However, these labels do not refer to the absolute time at which a neuron is born. This is because in the Drosophila nerve cord neuroblasts are generated over a large span of embryogenesis (Broadus et al., 1995). Thus, early-born neurons from one lineage can be generated at the same time as later-born neurons from a different lineage. Our study is unique in that it determined both neuronal birth order and birth time.

One of our unexpected findings is that circuit outputs in one lineage are most often born before circuit inputs from other lineages. Broadly speaking, nerve cord and spinal cord contain many local circuits, which output to the brain (e.g., circuits processing somatosensory stimuli) or which output to muscles (e.g., circuits that generate motor patterns). Here, early-born ELs are the outputs of a somatosensory processing circuit and transmit information to the brain. Early-born ELs are born before their local, nerve cord inputs—Ladders, Basins, NB7-1 Notch OFF interneurons (Figures 12—14). However, from a first principles perspective, we would expect the opposite. This is because 22 of 30 neuroblasts in the nerve cord are born before NB3-3, and many of them divide multiple times to produce neurons before NB3-3 begins to generate early-born neurons (Doe, 1992). Therefore, there should be more neurons born before early-born ELs compared to those born after early-born ELs. And so, by chance alone we would expect early-born ELs to get more input from neurons born earlier or at the same time.

What are the hypotheses about the importance of birth order of output versus input neurons? In Drosophila, birth order is linked to two things: (1) lineage-intrinsic factors such as dynamically changing programs of gene expression in the stem cell, and (2) lineage-extrinsic factors or the dynamic environmental context into which neurons are born. Intrinsic factors, or extrinsic factors, or both may be playing a role in the assembly of this feed-forward circuit. A potential intrinsic mechanism is that of a temporal transcription factor ‘matching code,’ in which early-born ELs, Ladders, and Basins would all be derived from the same temporal transcription factor window. For NB3-5 and MNB, temporal transcription factor expression is only partially characterized. But tantalizingly both early-born ELs and a subset of Ladders are born during a period in which their respective neuroblasts express the temporal transcription factor, Castor (Tsuji et al., 2008; Kearney et al., 2004). A potential extrinsic mechanism is that early-born EL dendrites may provide some type of signal that promotes later born neurons to synapse. There is evidence for such communication among Drosophila nerve cord neurons (Valdes-Aleman et al., 2021). Finally, it will be interesting to understand if sequential assembly is an absolute requirement, or if instead it facilitates rapid, efficient, or robust circuit assembly.

Studies of circuit assembly are still in their infancy. It is known that lineage–circuit relationships differ depending on circuit anatomy. But the converse is not known. That is, do all circuits of the same anatomy have a common lineage–circuit relationship? In the case of this study, we focused on a single feed-forward circuit in the Drosophila larval nerve cord. And the specific question it raises is: To what extent is sequential addition of temporal cohorts from different lineages the only mechanism used to assembly all feed-forward motifs? Currently, our answers are only partial and speculative. We note that there are so many connections made among neurons, it is unlikely that just one simple phenomenon that can explain the full complexity. The goals of the current research must be to identify rules and the circumstances where those rules apply. For example, our data rule out a ‘strict’ early-to-early, late-to-late model, meaning that this model alone cannot explain the wiring we observe in this circuit. And yet, an early-to-early model does apply to inter-segmental wiring among temporal cohorts of the same lineage. Further, this early-to-early wiring phenomenon occurs alongside a sequential addition phenomenon. We speculate there could be additional phenomena underlying assembly of this simple motif. For example, beyond local interneurons, we did not determine the birth date of sensory neurons, which provide the initial input to the circuit, nor did we investigate the developmental origins of central brain neurons, which receive the output from the circuit. We do note there is some evidence to support the generality that for assembly of feed-forward circuit motifs, presynaptic interneurons are born after their postsynaptic partners. Specifically, motor neurons are always circuit outputs (to muscle), and, in general, they are among the first neurons to be born during neurogenesis (Meng and Heckscher, 2021). Moreover, this pattern holds true in both Drosophila nerve cord and spinal cord (Schmid et al., 1999; Fetcho and McLean, 2010). Therefore, our data raise the possibility that one of many fundamental rules for circuit assembly is that feed-forward circuits are assembled sequentially from circuit output to circuit input.

Fly stocks were maintained at 25°C on standard cornmeal molasses medium. Experimental crosses were set up at the indicated temperatures. See below for details. See Supplementary file 4 for details of fly lines.

Larvae at different stages were dissected in Baines’ solution. The dissected larval brains were adhered to poly-lysine-coated coverslip and fixed for 7 min in 4% paraformaldehyde solution (Electron Microscopy Services, Hatfield, PA) as previously described (Heckscher et al., 2014). Larval brains were then washed three times with phosphate-buffered saline containing 0.1% Triton X-100 (PBT), blocked 1 hr in PBT containing 2% normal goat serum (NBT), and stained with primary antibody in PBT at 4°C overnight. After washing, samples were stained with secondary antibody at room temperature for 1 hr, washed again, and applied with increasing percentages of ethanol in water (30, 50, 70, 95, 100) series to replace PBT. Samples were then immersed into xylene for clearance, and then mounted in DPX (Sigma, MI). Images were acquired using either a Zeiss LSM 800 confocal microscope, and were processed and analyzed using ImageJ. For a list of primary antisera, see Supplementary file 4. Secondary antibodies were obtained from ImmunoResearch (Bar Harbor, ME) and were used at a 1:400 dilution.

The KDR-based permanent labeling plasmid, pPL[KD] (nSyb promoter and 5′UTR-KDRT-stop-KDRT-IVS-Syn21-GAL4-VP16-p10 3′UTR), was constructed in the backbone of nSybKOG (gift from Dr. Tzumin Lee) by restriction enzyme-mediated molecular cloning (Awasaki et al., 2014). The GAL4 coding region in nSybKOG was replaced by IVS-Syn21-GAL4-VP16-p10 3′UTR sequence. The IVS and p10 fragments were derived from the vector pJFRC81 (Addgene #36432), Syn21 sequence (Pfeiffer et al., 2010) was added by PCR amplification, and GAL4-VP16 was derived from the vector pBPGAL4.2::VP16Uw (Addgene #26228). The resulting pPL[KD] plasmid was incorporated into the docking site VK27 by integrase-mediated site-specific integration, performed by GenetiVision Corporation (Houston, TX). Details of molecular cloning and construct sequence are available upon request.

The pEntr-EMS plasmid was generated from a HiFi reaction (NEB) using the pEntr3c Gateway cloning (Invitrogen) plasmid digested with KpnI and NotI restriction endonucleases and four PCR reactions spanning the EMS promoter sequence found in Estacio-Gómez et al., 2013. The primer pairs used were (1) CGACTGGATCCGGTACCccagacagaactccatactccac and gtcgttaaacAAATGAATTGCCATAAGCG; (2) caattcatttGTTTAACGACCAACGCTC and tccggatggtCGAGCGGGATTTATGAGC; (3) atcccgctcgACCATCCGGATCTGGGCAAAAC and aatgaaaaccGTAAAAAATGCAGCCAACAAAGGG; (4) cattttttacGGTTTTCATTCCTTTTTGCG and GTCTAGATATCTCGAGTGCGGCCGCgtgtagtatggccgtcttctttgc.

The pEntr-EMS was combined with the Gateway cloning destination vector pBPzpGal4AD using an LR reaction to generate pBP-EMSzpAD.

In this study, we used ts-MARCM to determine both neuronal birth order and estimate neuronal birth time. Birth order is determined in reference to other neurons within a given lineage, for example, first-born, second-born. Birth order is also often referred to as a neuron’s temporal identity. Neuronal temporal identity is assessed by expression of temporal transcription factors. Consequently, determining birth order can be used to infer temporal transcription factor expression. Birth time refers to when during neurogenesis a neuron is born. Birth timing is usually reported in units of minutes or embryonic stage. Determining neuronal birth timing provides context into which a neuron is born (e.g., availability of other early-born neurons, transient signaling cues). In Drosophila embryos, neuron birth timing and birth order are not identical because during embryogenesis neuroblasts start to divide at different times. For example, NB3-5 is generated relatively early in embryogenesis and generates a large number of neurons, whereas MNB is generated much later and generates fewer neurons. And so, MNB can be generating first-born neurons at the same time that NB3-5 is generating much later-born neurons. We tried to birth date neurons using photo-activation and EdU/BrdU labeling, but were unsuccessful. So, instead, we estimated Basin birth timing using a series of carefully timed heat shocks to generate ts-MARCM clones. See below for details.

We used the pan-EL driver, EL-GAL4, which is expressed in all ELs (Heckscher et al., 2015). Embryos of the genotype hsFLP; UAS-rCD2.RFP, UAS-GFPi, FRT40A/UAS-mCD8-GFP, UAS-rCD2i, FRT40A; EL-GAL4/pPL[KD] were used. Addition of the pPL[KD] amplifies expression from EL-GAL4 (Figure 1E and Figure 2B, C). This optimization step was required to drive high levels of RNAi constructs that are at the core of the ts-MARCM strategy (e.g., UAS-rCDi). High RNAi levels are needed such that ‘leaky’ expression of membrane reporters (e.g., UAS-rCD3.RFP) is suppressed. Also, high levels of membrane-tethered reporter protein are needed to observe the detailed morphology of the entire neuron arbor.

To determine the birth order and birth timing of Basins, we used the pan-Basin driver, 72F11-GAL4. Embryos of genotype hsFLP; UAS-rCD2.RFP, UAS-GFPi, FRT40A/UAS-mCD8-GFP, UAS-rCD2i, FRT40A; 72F11-GAL4/+ were collected for 1 hr on apple juice plates. Collections were aged for 6–13 hr after egg collections. The collected samples were exposed to 37°C heat shock for 20, 30, or 35 min. Heat-shocked samples were then incubated at 29°C to boost GAL4 activity. Larvae were dissected at late L1 stage. We scored a total of 113 clones.

We used the neuron-specific drivers, 78F07-GAL4 and 20B01-GAL4. These drivers expressed solely in Ladder D and Basin 1, respectively, but not in other neurons within the same lineage. Embryos of genotype hsFLP; UAS-rCD2.RFP, UAS-GFPi, FRT40A/UAS-mCD8-GFP, UAS-rCD2i, FRT40A; 78F07-GAL4 and hsFLP; UAS-rCD2.RFP, UAS-GFPi, FRT40A/UAS-mCD8-GFP, UAS-rCD2i, FRT40A; 20B01-GAL4 were used.

We used both the pan-Basin driver, 72F11-GAL4, and the pan-EL driver, EL-GAL4. Embryos of genotype hsFLP; UAS-rCD2.RFP, UAS-GFPi, FRT40A/UAS-mCD8-GFP, UAS-rCD2i, FRT40A; 72F11-GAL4/EL-GAL4 were used.

We labeled single neurons using MCFO (Nern et al., 2015). MCFO stochastically labels with epitopes the membranes of cells within a GAL4 pattern. To obtain single-cell clones, adult flies laid for 24 hr on apple juice caps. Caps were heat-shocked in a water bath at 37–39°C for 15–30 min and incubated at 25°C for 4–5 hr. First-instar larvae were dissected. Their brains were stained for HA, Flag, and V5 epitopes to visualize single-cell clones. Larvae were also stained for Eve protein to confirm the identity of each single-cell clone as an EL interneuron or to assign segmental identity to each clone. We generated >100 single-cell clones and saw each neuronal morphology in a minimum of two separate larvae. Each clone was analyzed in dorsal and posterior views.

To assign neurons to a specific neuroblast lineage, larvae of genotype Basin-1-LexA/UAS-mCherry.NLS; NB3-5-GAL4/LexAop2-CsChrimson-mVenus were used, where NB3-5-GAL4 was one of three different GAL4 lines (49C03-GAL4, ham-GAL4, 59E09-GAL4). We used three different GAL4s to label NB3-5 because any one GAL4 line might have unexpected off-target labeling. We reasoned that each line should have different off-target labeling, and so if we saw co-localization of markers in all three cases, this could be taken as strong evidence that Basin 1 was in the NB3-5 lineage. Lines also included FLP-based permeant labeling (UAS-FLP, actin promoter and FRT-stop-FRT-GAL4), which labels all progeny from a given neuroblast. Larvae were raised at 29°C on the apple juice plate. The larvae were then dissected in the L3 stage for further staining and imaging processing.

We used a similar logic to determine neuroblast origin for Ladders. Larvae of genotype Ladder-D-LexA/UAS-mCherry.NLS; MNB-GAL4/LexAop2-CsChrimson-mVenus were used, where MNB-GAL4 was one of three different GAL4 lines (13G03-GAL4, 70D06-GAL4, and 65H11-GAL4).

For calcium imaging experiments, all larvae were within 6 hr of age on the day of recording and collected 48–54 hr after hatching. Larvae expressing GCaMP6m were rinsed with water and placed ventral side up on agarose pads with a 22 mm × 22 mm coverslip placed on top. Pads were made by pouring 3% agarose into a well. Recordings began with a 30 s period of no stimulus followed by a 30 s period of sound stimulus and ending with a final 30 s period of no stimulus. A Visaton FR12, 4 Ohm speaker (5 inches diameter) and a PYLE PCA2 stereo power amplifier was used to project sound. For further details, refer to Marshall and Heckscher, 2022. Images were acquired on a Zeiss LSM 800 confocal microscope using 0.1–0.2% 488 nm laser power with the pinhole entirely open. Images were acquired at 3 frames per second using a 10× (0.3 NA) or 20× (0.8 NA) objective. The calcium signal was continuously collected before, during, and after the stimulus. Extracting changes in GCaMP6m fluorescence amplitude was done using Fiji as in Marshall and Heckscher, 2022. A region of interest (ROI) that included the larval nerve cord was manually drawn, and the mean fluorescence within the ROI was calculated for each time point.

The connectome dataset used in this study is a CNS reconstruction from a 6-hr-old first-instar larva, described in Ohyama et al., 2015.

Since we observed a broad variation in synaptic strengths and input onto ELs, to disambiguate the identity of inputs from their respective strength, we repeated distance analyses using input identity only. To do so, we treated inputs to a neuron as a binary vector, with 1 for the presence and 0 for the absence of input, respectively. Otherwise, analysis was identical to nonbinary. Broadly speaking, results from binary and nonbinary analyses were similar. One notable exception is that binary analysis did not identify statistically significant differences.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank the Janelia Fly EM Project Team for the gift of the EM volume, and the Janelia visitor program for their support. We would like to thank the following for contributing to EM annotation: Akria Fushiki, Albert Cardona, Andreas Schoofs, Antia Burgos, Aref Arzan Zarin, Avinash Khandelwal, Brittany Kemp, Casey M Schneider-Mizell, Chris Q Doe, Elizabeth Barsotti, Ingrid Andrade, Jamie Macleod, Larisa Maier, Laura Herren, Maarten F Zwart, and Xinyu Tang. This work was supported by NIH grants NS105748 (to ESH), EY022338 (to JNM), and T32 HD044164 (to ZDM), and NSF grant DGE-1746045 (to JLM).

© 2022, Wang, Wreden et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.