Intrinsic control of muscle attachment sites matching.

Myogenesis is an evolutionarily conserved process. Little known, however, is how the morphology of each muscle is determined, such that movements relying upon contraction of many muscles are both precise and coordinated. Each Drosophila larval muscle is a single multinucleated fibre whose morphology reflects expression of distinctive identity Transcription Factors (iTFs). By deleting transcription cis-regulatory modules of one iTF, Collier, we generated viable muscle identity mutants, allowing live imaging and locomotion assays. We show that both selection of muscle attachment sites and muscle/muscle matching is intrinsic to muscle identity and requires transcriptional reprogramming of syncytial nuclei. Live-imaging shows that the staggered muscle pattern involves attraction to tendon cells and heterotypic muscle-muscle adhesion. Unbalance leads to formation of branched muscles, and this correlates with locomotor behavior deficit. Thus, engineering Drosophila muscle identity mutants allows to investigate, in vivo, physiological and mechanical properties of abnormal muscles.

Introduction

The musculature of each animal is composed of an array of body wall muscles allowing precision and stereotypy of movements. The somatic musculature of the Drosophila larva - about 30 distinct body wall muscles per each hemi-segment which are distributed in three layers, internal, median and external - is a model to study how a muscle pattern is specified and linked to locomotion behaviour. Each muscle is a single multinucleated fibre with a specific identity: size, shape, orientation relative to the dorso-ventral and antero-posterior body axes, motoneuron innervation and attachment sites to the exoskeleton via tendon cells located at specific positions. Intersegmental tendon cells, where a large fraction of muscles is attached, are distributed in three groups, dorsal, lateral and ventral (Bate, 1990; Volk and VijayRaghavan, 1994; Schweitzer et al., 2010; Armand et al., 1994). Internal muscles also attach to muscle(s) in the next segment, forming ‘indirect’ muscle attachment sites (iMAS) (Maartens and Brown, 2015).

Drosophila muscle development proceeds through fusion of a founder myoblast (founder cell, FC) with fusion-competent myoblasts (FCMs) (Deng et al., 2017). FCs originate from asymmetric division of progenitor cells (PCs) themselves selected from equivalence groups of myoblasts, called promuscular clusters (PMCs). Muscle identity ensues from the expression by each PC and FC of a specific combination of identity transcription factors (iTFs) (de Joussineau et al., 2012), established in three steps: First, different iTFs are activated in different PMCs, in response to positional information, and this expression is only maintained in PCs. Second, refinement of the iTF code occurs via cross-regulations between different iTFs in PCs and/or FCs and Notch signalling (Carmena et al., 1998; Carmena et al., 2002; Enriquez et al., 2010). Several PCs can be serially selected from a PMC and give rise to different muscle identities according to birth order, adding a temporal dimension to these regulations. A well-documented example is the distinction between DA2 and DA3 identities (Figure 1A; Dubois et al., 2016; Boukhatmi et al., 2012). Third, transcriptional activation of iTFs in syncytial nuclei after fusion correlates with the activation of identity ‘realisation’ genes acting downstream of some iTFs (Crozatier and Vincent, 1999; Knirr et al., 1999; Bataillé et al., 2010; Bataillé et al., 2017).

Figure 1.

Col CRMs, CRM deletions and col transcription.

(A) Diagrammatic representation of the sequential emergence of 4 different PCs from the Col-expressing PMC (grey), division of the DA3/DO5 PC into 2 FCs and col auto-regulation in the DA3 lineage; the names of each PC and FC are indicated. Accumulation of Col protein is in green. The time windows of mesodermal early (E-CRM) and late (L-CRM) CRM activity are indicated by green lines. Right: muscle pattern of an abdominal segment highlighting DA2 (orange) and DA3 (green). (B) Schematic representation of the col transcribed region (http://flybase.org/cgi-bin/gbrowse2/dmel/?Search=1;name=FBgn0001319). The position of tested GMR and VT fragments are drawn as brown horizontal bars; the numbers are given for those active in DA3. The positions of clusters of in vivo Mef-2, Twi and Tin binding sites are indicated by vertical blue bars, the kn transposon insertion, used for the E-CRM deletion screening, by a vertical arrow. (C) Enlarged view of L-CRM indicating the juxtaposition of PC-specific and autoregulatory DA3-specific CRMs, Col, Mef2 and Twi binding sites, and the col and col deletions generated by CRISPR/cas9 genome editing. (D) Col transcription in wt and mutant embryos, genotypes indicated, visualised by in situ hybridisation to primary transcripts. A detail of the abdominal A2 segment (squared area) is shown in each panel. PMC col transcription is lost in colE embryos; stage 11, col transcription in the DA3/DO5 and DT1/DO4 PCs is detected in all strains; stage 14, DA3 syncytium transcription is lost in col and col embryos. * indicates col transcription in a multidendritic neuron (md).

Genomic sequence spanning the previously mapped E-CRM (start and end indicated in red) and the col deleted fragment (dashed blue line). The nucleotide positions of the col deletion endpoints (brown) are given relative to the col transcription start site. The predicted cleavage sites of the used sgRNAs (sequences underlined) are indicated by yellow arrows. The MiMIC 15480 used for the screening of the deletion is inserted between the red underlined nucleotides. Predicted binding sites for Twist (blue) and Mef2 (purple), conserved between Drosophila species.

Genomic sequence spanning the col (from −2666 to −1339) and col (from −1857 to −1339) deletions. Deleted fragments are represented by dashed lines and their coordinates given relative to col transcription start site. The predicted cleavage sites of the used sgRNAs (sequences underlined) are indicated by yellow arrows. Predicted binding sites for Twist (blue), Mef2 (purple), and Col (green), conserved between Drosophila species.

(A) Schematic representation of col L-CRM (see also Figure 1) indicating the positions of the −2.3 to −1.6 fragment (grey) previously shown to drive expression in the DA3 syncytium and new reporter constructs, −3.3–2.3 and −3.3–2.3 ΔMT with the position of conserved Mef2 and Twi binding sites. (B) Nucleotide sequence of the −3.3–2.3 fragment, positions −3385 to −2273. The >20 bp sequence motifs that are more than 90% conserved at the same relative positions in D. melanogaster, D. pseudobscura and D. virilis are shaded in grey. The 97 bp segment containing both Mef2 (purple) and Twi (blue) binding and deleted in 3.3–2.3ΔMT is underlined. (C) In situ hybridisation to primary reporter transcripts (yellow intron (yi) inserted in the lacZ coding sequence). The 3.3–2.3 fragment drives specific expression in the DA3/DO5 and DT1/DO4 PCs (circled in segment T2 and framed in A2) and the intercalary segment (arrow). Its activity in PCs is lost upon deletion of the region encompassing Mef2 and Twi binding sites (3.3–2.3ΔMT reporter).

The consequence of specific muscle identity defects on locomotion, a question of prime importance for progress on studying human myopathies which affect subsets of muscles, remains largely to be assessed. Genetically controlled muscle identity changes should, in principle, provide suitable models for studying locomotion deficits linked to muscle imbalance. However, mutations for known Drosophila iTFs are embryonic lethal and/or show pleiotropic phenotypes reflecting their multiple expression sites. Here, we took advantage of our previous characterisation of col expression in a single larval muscle, the DA3 muscle and of the two involved cis-regulatory modules (CRMs), early (E-CRM) and late (L-CRM), to generate muscle-specific mutants and circumvent lethality/pleiotropy of null mutants.

CRM deletions show that E-CRM and L-CRM act redundantly at the PC stage, emphasising that PC is a key stage in specification of muscle identity. L-CRM deletion results into loss of col transcription in DA3 FCs and morphological transformation of the DA3 into a DA2-like muscle. Removal of an auto-regulatory cis-module located in L-CRM specifically abolishes col activation in syncytial nuclei fusing with the DA3 FC. This leads to incomplete DA3 transformations and the formation of bifid/branched muscles of mixed DA3/DA2 morphology. In summary, our data show that i) the FC transcriptional program must be propagated to syncytial nuclei for a muscle to adopt a specific morphology; ii) the precise matching of muscle/muscle attachments over the intersegmental border, which leads to a staggered rows pattern, involves a process of selective adhesion controlled by iTFs; iii) branched muscles affect larval locomotion performance.

Branched muscles are typical of late, severe phases of human Duchenne Muscular Dystrophy (Chan and Head, 2011). Drosophila iTF CRM deletion could be an effective setting for creating muscle-specific transformations and branched muscles, as new paradigms to study myopathies specifically affecting subsets of muscles in humans.

Results

Redundant CRMs at the PC step ensure robustness of iTF transcription col transcription in myogenic lineages is first observed in a dorso-lateral PMC from which are sequentially selected several PCs (Dubois et al., 2016). It is subsequently maintained in two PCs, then one FC, the DA3 FC, and is activated in FCM nuclei recruited into the DA3 growing fibre (Crozatier and Vincent, 1999; Figure 1A). Two col CRMs containing embryonic in vivo binding sites for the master myogenic TFs, Mef-2 and Twist (Twi) (Sandmann et al., 2007; Zinzen et al., 2009) were previously identified, which reproduce this sequence of expression in reporter assays: E-CRM, which is active in the PMC and the DA3/DO5 and DT1/DO3 PCs, and L-CRM which is active in these same 2 PCs, the DA3 FC and DA3 syncytial nuclei (Enriquez et al., 2010; Dubois et al., 2007; Figure 1A; Figure 1—figure supplement 1 and Figure 1—figure supplement 2). With the prospect of using CRM deletions to generate muscle-specific mutants and to exclude the possible existence of additional, redundant enhancers (Cannavò et al., 2016), we conducted a systematic analysis of Gal4 reporter lines (Manning et al., 2012; Kvon et al., 2014) covering 36 kb of the col genomic region. A single reporter, GMR13B08, reproduces PMC Col expression and it overlaps E-CRM, and two reporters, GMR12G07 and GMR12H01, display DA3 expression and they both partly overlap L-CRM (Figure 1B and data not shown), attesting to the existence of only two col muscle CRMs.

Figure 1—figure supplement 1.

Genomic sequence of the E-CRM deletion.

Genomic sequence spanning the previously mapped E-CRM (start and end indicated in red) and the col deleted fragment (dashed blue line). The nucleotide positions of the col deletion endpoints (brown) are given relative to the col transcription start site. The predicted cleavage sites of the used sgRNAs (sequences underlined) are indicated by yellow arrows. The MiMIC 15480 used for the screening of the deletion is inserted between the red underlined nucleotides. Predicted binding sites for Twist (blue) and Mef2 (purple), conserved between Drosophila species.

Figure 1—figure supplement 2.

Genomic sequence of the L-CRM deletions.

Genomic sequence spanning the col (from −2666 to −1339) and col (from −1857 to −1339) deletions. Deleted fragments are represented by dashed lines and their coordinates given relative to col transcription start site. The predicted cleavage sites of the used sgRNAs (sequences underlined) are indicated by yellow arrows. Predicted binding sites for Twist (blue), Mef2 (purple), and Col (green), conserved between Drosophila species.

The activity of E-CRM and L-CRM at different phases of muscle development raised the question of their respective roles in defining muscle identity. To address this question, we separately deleted each of them from the genome using the CrispR/Cas9 technology. Deletion of a 2,4 kb fragment removed the E-CRM. Two deletions within the L-CRM were generated: deletion of a 1.3 kb fragment removes the entire core region (Dubois et al., 2007) and deletion of a 0.5 kb fragment removes the Col autoregulation site (de Taffin et al., 2015; Figure 1C; Figure 1—figure supplement 1 and Figure 1—figure supplement 2). The corresponding mutant Drosophila strains are designated col and col, respectively. col strains are homozygous viable and fertile. col strain is homozygous female sterile, a sterility unrelated to col activity, since fertility is restored by placing col over a deficiency (Df(2L)BSC429) (abbreviated Df in the rest of the text and figures) removing the entire col locus (data not shown). As a first step to determine the consequences of deleting either E-CRM or L-CRM, we compared col transcription between wt, col (a protein null mutant), col and col strains, using an intronic probe to detect nascent transcripts (Figure 1D). In col homozygous mutant embryos, col transcription is detected at the PMC and PC stages and lost at the FC stage, showing the key role of autoregulation in the maintenance of col transcription (Crozatier and Vincent, 1999; de Taffin et al., 2015). In col embryos, we found that col is not activated in PMC cells, as could be expected. Yet, col transcription is detected in the DA3/DO5 and DT1/DO3 PCs, showing that inheritance of Col protein synthesised under E-CRM control is not required for activation of col transcription in PCs. A normal pattern is observed in the developing DA3 muscle, at stage 14. Reciprocal to E-CRM deletion, col is transcribed in PMC cells and the DA3/DO5 and DT1/DO3 PCs in col embryos, but no more at stage 14, showing that L-CRM is required for col transcription maintenance.

Both E-CRM and L-CRM activity are detected in PCs. In absence of E-CRM, no Col protein is inherited from the PMC. Therefore, L-CRM activity in PCs cannot be due to col autoregulation. Analysis of new col-lacZ reporter genes (Perry et al., 2010) revealed the existence of a PC-specific CRM located within the −3.3 to −2.3 fragment of L-CRM, i.e., which is separate from the autoregulatory CRM defined by the in vivo Col binding site and contains the in vivo binding sites for Mef2 and Twist (Figure 1C; Figure 1—figure supplement 3; Zinzen et al., 2009). Activity of this PC-specific CRM is transient and lost upon removal of Mef2 and Twist binding sites (Figure 1—figure supplement 3). Overall, we conclude that two CRMs separately drive col transcription in muscle PCs, suggesting that robust iTF expression at the PC stage is critical to confer a muscle its identity.

Figure 1—figure supplement 3.

Mapping of a PC-specific CRM.

(A) Schematic representation of col L-CRM (see also Figure 1) indicating the positions of the −2.3 to −1.6 fragment (grey) previously shown to drive expression in the DA3 syncytium and new reporter constructs, −3.3–2.3 and −3.3–2.3 ΔMT with the position of conserved Mef2 and Twi binding sites. (B) Nucleotide sequence of the −3.3–2.3 fragment, positions −3385 to −2273. The >20 bp sequence motifs that are more than 90% conserved at the same relative positions in D. melanogaster, D. pseudobscura and D. virilis are shaded in grey. The 97 bp segment containing both Mef2 (purple) and Twi (blue) binding and deleted in 3.3–2.3ΔMT is underlined. (C) In situ hybridisation to primary reporter transcripts (yellow intron (yi) inserted in the lacZ coding sequence). The 3.3–2.3 fragment drives specific expression in the DA3/DO5 and DT1/DO4 PCs (circled in segment T2 and framed in A2) and the intercalary segment (arrow). Its activity in PCs is lost upon deletion of the region encompassing Mef2 and Twi binding sites (3.3–2.3ΔMT reporter).

col CRM deletions lead to muscle transformation and branched muscles

Having deleted separately each col CRM allowed to assess the respective roles of iTF transcription before or after the PC step. To compare the different CRM deletions, we placed each of them over the deficiency Df chromosome. We introduced the L-CRM-moeGFP reporter to visualise the DA3 morphology at stage 15 (Enriquez et al., 2012). Control (+/Df) embryos display reporter expression in the DA3/DO5 and DT1/DO3 PCs at stage 11 and the DA3 muscle at stage 15 (Figure 2A). The same pattern is observed in PCs for all col-CRM deletion strains, consistent with transcript analyses (Figure 1D). L-CRM-moeGFP expression at stage 15 shows that the DA3 morphology is normal in about 80% of segments in col/Df embryos (Figure 2A–B), and we did not pursue the analysis of this deletion strain. Low level GFP expression in col/Df embryos, consistent with col transcription data (Figure 1D), shows that the DA3 muscle is most often (85.2% of segments) transformed into a DA2-like muscle (designated below as DA3>DA2; Figure 2A–B), like in col null mutant embryos (Enriquez et al., 2012). In col/Df embryos, i.e, when only the autoregulation module has been deleted, a high number (29%) of branched muscles is observed (Figure 2A–B). Branched muscles correspond to incomplete transformations, with two stable anterior attachment sites, overlapping the DA3 and DA2 sites in wt embryos. The high ratios of either complete (DA3>DA2) or incomplete (branched) transformations in L-CRM deletion mutants demonstrate that an iTF CRM deletion strategy is effective for creating viable muscle-specific identity mutants and explore branched muscle properties.

Figure 2.

DA3 muscle transformations upon col-CRM deletions.

(A) L-CRM-moeGFP expression in stage 11 and 15 hemizygous embryos, as indicated. GFP expression in PCs at stage 11 is similar in all strains. DA3>DA2 transformations (arrow) and branched DA3 muscles (arrowhead) are observed in col embryos. (B) Quantification of the relative proportions of normal DA3, branched DA3, DA3>DA2 transformation and absence of DA3 muscles in wt, col, col and col hemizygous embryos. A minimum of 100 A1-A7 abdominal segments of stage 15–16 embryos were analysed for each genotype. (+/Df: n = 127 segments - 16 embryos; col/Df: n = 170 segments - 23 embryos; col/Df: n = 190 segments - 27 embryos; col/Df: n = 103 segments - 13 embryos).

Re-programming of syncytial nuclei is required for muscle morphological identity

Complete vs incomplete transformations in col versus col deletions suggest that proper level and/or maintenance of iTF expression is crucial for proper muscle development. This led us to compare the pattern of Col protein in growing DA3 syncytium between wt, col and col embryos. In either deletion strain, Col is detected in PCs at stage 11 but not in muscles at stage 15 (Figure 3A). However, at stage 14, some Col protein is still detected in muscle precursors in col, not in col embryos (Figure 3B). To trace the origin of this difference, we examined col transcription in the DA3 PC, FC and stage 14 syncytium using Df/hemizygous embryos which display one hybridisation dot per active nucleus (Figure 3C). In control wt/Df embryos, a dot is systematically detected in the DA3/DO5 PC (20/20 segments; five embryos analysed), the DA3 FC (20/20) and 80% of DA3 nuclei at stage 14 (6 of 7–8 nuclei per fibre on average; 27 segments). A dot is detected as well in the DA3/DO5 PC, in either col (21/21) or col (18/18) embryos, reflecting E-CRM activity (Figure 1D). In col embryos, a col hybridisation dot is detected in the DA3 FC (19/19) and in one nucleus, likely the FC nucleus (11/21 segments), sometimes two nuclei at stage 14. In col embryos, however, col transcription is only detected in a minor fraction of FCs (4/15) and is completely lost at stage 14 (0/6–7 nuclei per fibre on average; 24 segments). Patterns similar to stage 14 are observed at stage 15, while at stage 16, col transcription is detected neither in L-CRM deletion strains nor in control (Figure 3—figure supplement 1). Since col transcription at, and from, the PC stage appears to be nodal to DA3 identity, we measured the level of col transcripts relative to nautilus (nau), the Drosophila MyoD-MRF serving as internal reference (Figure 3D and Figure 3—figure supplement 2). As expected, similar levels of col transcription are found in the DA3 PC, in control (Mean ± sem: 1.22 ± 0.06; n = 18), col (1.15 ± 0.04; n = 26) and col embryos (1.19 ± 0.04; n = 18), which confirms handling by the E-CRM, with only a minor contribution from the L-CRM (Figure 1D). On the contrary, high level of col transcription in the DA3 FC in wt embryos (2.20 ± 0.05; n = 28) is dependent upon L-CRM activity, since a drop is observed in col FCs (1.10 ± 0.04; n = 24), p<0.001. More precisely, it is dependent upon the presence of Mef2 and Twist binding sites since a basal level of transcription is still observed when only the col autoregulation module is deleted (col: 1.41 ± 0.04; n = 25). Quantification of col and nau transcripts shows that the col/nau increase between the PC and FC stages in wt embryo is due to increased col transcription while the level of nau is relatively constant and is unaffected in col L-CRM mutants (Figure 3—figure supplement 2). Together, the data suggest that binding of Mef2 and Twist is required to prime col transcription in the FC nucleus before autoregulation takes off (Figure 3E). Taken with one another, Col immunostaining, FISH data and col transcription quantification indicate that sustained col transcription in the FC nucleus in col embryos (Figure 3C–D), provides enough Col protein for some uptake by other DA3 nuclei at stage 14, and this leads to their partial reprogramming to DA3 identity. Partial reprogramming could, in turn, explain the formation of branched muscles retaining some DA3 morphological characters. Moreover, the col and col expression data and deletion phenotypes show that both iTF transcription in the FC and reprogramming of ‘naïve’ syncytial nuclei contribute to ensure robust muscle morphological identity (Figure 3E).

Figure 3.

Identity reprogramming of syncytial nuclei controls the final muscle morphology.

(A) Col immunostaining of wt, col and col embryos, showing a normal pattern at stage 11 and complete absence of Col protein at stage 15 in both col embryos. (B) Stage14 embryos with a close up view of 4 segments shows low amounts of Col protein (black arrow) in the growing DA3 muscle in col and absence in col embryos (white arrow). Position of the multidendritic neuron (md) is indicated. (C) Col transcription (red dots), Col protein (blue) and L-CRM-moeGFP expression (green) in the DA3/DO5 PC, stage 11, DA3 FC, stage 12, and developing DA3 muscle, stage 14, in wt/Df, col and col; L-CRM-moeGFP embryos. In each panel, col transcripts are shown separately in black and white. col transcription ceases after FC stage in col embryos and does not propagate to other syncytial nuclei in col embryos. (D) Quantification of col primary transcripts level in PC and FC nuclei; orange asterisk: col transcription in FCs is generally lost in col embryos; quantification was done on a small fraction of FCs; n: number of PC or FC analysed, using 5 or six embryos at each stage 11 and 12; 15 col embryos were used for the FC stage (Mean ± sem and ***: p<0.001). (E) Schematic representation of the dynamics of col transcription (red dots) and Col protein (green) in the DA3/DO5 PC, the DA3 FC, muscle precursor, and DA3, DA3>DA2 and branched DA3 muscles in wt, col and col embryos, respectively. Temporal activity of E-CRM and L-CRM is represented by horizontal grey bars.

At stage 15, dots corresponding to col transcription are detected in syncytial DA3 nuclei in wt and col embryos. At stage 16, col transcription is no more detected neither in control nor in L-CRM deletions, except in multidendritic (md) ddaC neurons, indicated by an arrow.

(A) DA3 col (green) and nau (red) transcription dots at the PC and FC stage. Nuclei are labelled by Mef2 immunostaining (blue) and ROIs (regions of interest) used for quantification are circled. One col and two nau dots are observed in heterozygous (wt/Df, col and col) embryos. (B) Calculated dot intensity ratios (mean intensity) between the FC and PC stages. Number of nuclei used to calculate the mean dot intensity: nau: wt PC n = 28, FC n = 19; col PC n = 24, FC n = 26; col PC n = 25, FC n = 18; for col: wt PC n = 19, FC n = 28; col PC n = 26, FC n = 25; col PC n = 18, FC n = 24. nau data indicate a slight decrease between the PC and FC stages in all three conditions. col data show increased transcription between the PC and FC stage in wt embryos.

Figure 3—figure supplement 1.

Col transcription in col and col embryos In situ hybridisation to col primary transcripts in stage 15 and 16 wt/Df, col and col embryos.

At stage 15, dots corresponding to col transcription are detected in syncytial DA3 nuclei in wt and col embryos. At stage 16, col transcription is no more detected neither in control nor in L-CRM deletions, except in multidendritic (md) ddaC neurons, indicated by an arrow.

Figure 3—figure supplement 2.

Nau expression is not affected in col and col embryos.

(A) DA3 col (green) and nau (red) transcription dots at the PC and FC stage. Nuclei are labelled by Mef2 immunostaining (blue) and ROIs (regions of interest) used for quantification are circled. One col and two nau dots are observed in heterozygous (wt/Df, col and col) embryos. (B) Calculated dot intensity ratios (mean intensity) between the FC and PC stages. Number of nuclei used to calculate the mean dot intensity: nau: wt PC n = 28, FC n = 19; col PC n = 24, FC n = 26; col PC n = 25, FC n = 18; for col: wt PC n = 19, FC n = 28; col PC n = 26, FC n = 25; col PC n = 18, FC n = 24. nau data indicate a slight decrease between the PC and FC stages in all three conditions. col data show increased transcription between the PC and FC stage in wt embryos.

Muscle attachment: tendon attraction and muscle-muscle matching

We next investigated in more detail how ectopic muscle attachment sites at the origin of transformed and branched muscles are selected and stabilised. Double staining of ready-to-hatch (st17) wt embryos for F-actin and Ilk-GFP, a component of muscle attachment sites (Zervas et al., 2011; Sarov et al., 2016) shows that the DA3 posterior edge anchors to dorsal, and anterior edge to lateral intersegmental tendon cells (Figure 4A), giving its final acute shape (Bate, 1990). Moreover, the DA3 and DA2, as well as the DA2 and DA1 muscles precisely align with each other over each intersegmental border, forming heterotypic muscle-muscle attachment (iMAS; [Maartens and Brown, 2015]) at the origin of the staggered rows disposition of DA muscles. No DA3/DA3 (homotypic) iMAS surface is observed (Figure 4A). On the contrary, in col embryos, the anterior edge of DA3>DA2 transformed muscles anchors to dorsal tendon cells instead of lateral intersegmental tendon cells, leading to a ‘dual DA3>DA2 morphology’, DA2-like at the anterior, and DA3 at the posterior edge. This dual identity leads to iMASs between adjacent DA3>DA2 muscles (Figure 4A). As a consequence, DA2 iMASs are shifted dorsally and the general pattern of DA muscles is affected.

Figure 4.

Multistep muscle attachment to selected tendon cells.

(A) F-actin (green) and Ilk-GFP (red) staining of stage 17 control and col embryos, showing heterotypic and homotypic muscle attachment sites, respectively. A drawing illustrates muscle matching in control embryos and mismatching in col embryos, with DA1, DA2 and DA3 muscles coloured in blue, yellow and green, respectively and attachment sites as red lines. (B) Snapshots of live imaging DA3 muscle development, using L-CRM-moeGFP expression (green). Tendon cell precursors express stripe-Gal4; UAS-RFP (red). Embryos were filmed during 4 hr (Sup. Videos 1 and 2) and Z sections collected every ~2–2.5 min. The outlines of the developing muscles are schematised for each stage. In both control (left, one segment shown) and col embryos (right, two segments shown), the posterior muscle end reaches the intersegmental border first. At stage 14, wt DA3 contacts both dorsal and lateral tendon cells at its anterior end. In col embryos, the anterior dorsal projections fail to retract, leading to ectopic DA3>DA2 attachment and branched muscles.

Video 1.

Live-imaging of wt embryos.

The DA3 muscle is visualised by L-CRM-moeGFP (green) and tendon cell along the entire intersegmental border by stripe Gal4;UASmCD8RFP (red). Embryos were filmed during 4 hr. See also the legend of Figure 4.

Video 2.

Live-imaging of col embryos.

The DA3 muscle is visualised by L-CRM-moeGFP (green) and tendon cell along the entire intersegmental border by stripe Gal4;UASmCD8RFP (red). Embryos were filmed during 4 hr. See also the legend of Figure 4.

To explore the dynamics of DA3 muscle attachment, we live-imaged wt and col embryos, starting at stage 12 (defined as t0 in Videos 1 and 2). The DA3 muscle is visualised by L-CRM-moeGFP and tendon cell along the entire intersegmental border by stripe Gal4;UASmCD8RFP (Volohonsky et al., 2007). In wt embryos between stages 12 and 13 (Video 1; Figure 4B), the DA3 muscle precursor extends protrusions towards dorsal tendon cells, posteriorly, and both dorsal and lateral tendon cells, anteriorly, diverging from a bipolar extension scheme (Schnorrer and Dickson, 2004). At stage 14, the posterior DA3 edge makes contacts with dorsal tendon cells before the anterior edge(s) reaches the intersegmental border, a time gap previously observed during live-imaging of a ventro-lateral muscle (Gilsohn and Volk, 2010). In addition, numerous filopodia emanate from the DA3 dorsal surface and contact the DA2 ventral surface, contributing to give the DA3 muscle precursor its transient angled-shape. At late stage 14, the posterior DA3 attachment widens parallel to the intersegmental border. Its anterior attachment to lateral tendon cells in turn becomes stable, whereas dorsal anterior filipodia appear to be repelled from the intersegmental border where the DA3 muscle in the preceding segment is attached. Remaining filopodia are limited to the rims of DA3 anchoring, possibly suggesting a mechanism of homotypic repulsion. In col embryos, the main axis of the DA3>DA2 muscle precursor elongation is more longitudinal than wt, already at stage 12 (Video 2; Figure 4B). At stage 13 and like in wt, many protrusions emanate from the DA3 dorsal surface, but unlike in wt, protrusions contacting dorsal tendon cells do not retract at later stages when they contact the DA3>DA2 muscle from the preceding segment. Rather, stabilisation of contacts made by these protrusions prefigures abnormal DA3>DA2/DA3>DA2 iMAS formation (Figure 4A). In some segments, contacts with lateral tendon cells are also stabilised, resulting into branched muscles. In summary, imaging of live wt embryos illustrates different steps involved in establishment of the acute DA3 orientation: posterior attachment to dorsal tendon cells at the same time as anterior exploration of dorsal and lateral tendon cell. This is followed by DA3 attachment to lateral, and retraction from dorsal tendon cells (Figure 4B, Video 1), cumulating into heterotypic DA3/DA2 iMAS stabilisation (Figure 4A and B). This retraction does not take place in col embryos (Figure 4B and Video 2).

Muscle attachment: heterotypic adhesion

Further determining how DA3>DA2/DA3>DA2 homotypic attachment could interfere with the staggered rows pattern of DA muscles, required to visualise at the same time the DA3 and DA2 muscles and tendon cells contours. Vestigial (Vg) is expressed in DA muscles (Deng et al., 2010; Tixier et al., 2010). We screened the vg regulatory landscape and characterised one vg CRM active in the DA3 and DA2 (and VL1) muscles, VgM1 (Figure 5—figure supplement 1). Expressing together VgM1-mCD8GFP-H2bRFP, L-CRM-moeGFP and stripeGal4:UAS-mCD8RFP, and co-staining of stage 16 control embryos for α-Spectrin, a protein enriched at muscle attachment sites, shows that the posterior edge of DA3 precisely aligns with the anterior edge of DA2 (Figure 5A, Figure 5—figure supplement 1), suggesting heterotypic attractive cues. This attraction is already observed at stage 14, when the dorsal DA3 and ventral DA2 surfaces contact each other via numerous protrusions (Figure 5A and B; Figure 5—figure supplement 2). Strikingly, at stage 15, the contact zone gets away from intersegmental borders and the DA3/DA3 and DA2/DA2 homotypic connections disappear, to give way to heterotypic DA3/DA2 stable iMAS formation. In col embryos, the initial steps, stages 14 and 15, are similar to wt, except for the absence of stable connection of DA3>DA2 to lateral tendon cells, as seen by live imaging (Video 2). At stage 16, DA3>DA2/DA3>DA2 and DA3>DA2/DA2 iMASs are privileged, while similar to control, no homotypic DA2/DA2 iMAS forms, suggesting homotypic repulsion (Figure 4; Figure 5A). Together, the data indicate that the precise pattern of iMASs and muscles is contributed by a combination of attractive and, possibly, repulsive cues downstream of muscle iTFs (Figure 5B).

Figure 5—figure supplement 1.

Identification of a vg CRM active in the DA2 and DA3 muscles.

(A) Schematic representation of the vestigial (vg) transcribed region (https://flybase.org/cgi-bin/gbrowse2/dmel/?Search=1;name=FBgn0003975). GMR fragments tested for their enhancer activity are drawn as brown horizontal bars; GMR69G04 and GMR69G05 are active in DA3 and DA2 muscles. The fragment of overlap between GMR69G04 and GMR69G05, positions 12889557–12891520 on Chr2R, is called vgM1 (B) vgM1-moeGFP expression in the DA2, DA3 and VL1 muscles in wild type embryos at stages 14 and 16 (schematised on the right); higher magnification views show the close apposition of the DA3 ventral and DA2 dorsal sufaces at stage 14 and their tandem arrangement at stage 16. Of note, vgM1 expression seems to be more robust in DA2 than DA3.

Figure 5.

DA3 heterotypic and homotypic muscle attachment sites (iMAS) in wt and L-CRM mutants.

(A) Immunostaining of wt and col VgM1-moeGFP-H2bRFP; srGal4 >mcd8 RFP; L-CRM-moeGFP embryos for GFP (green), RFP (red) and Spectrin (blue). DA3 (DA3>DA2) and DA2 muscles are indicated in each panel. At stage 14, the wt DA3 and col DA3>DA2 attachment sites partly align over the intersegmental border (circled). Their dorsal surface closely contacts DA2 via numerous protrusions; at stage 15, likely because homotypic DA3/DA3 and DA2/DA2 repulsion, these protrusions are restricted away from intersegmental borders in wt embryos while maintained in col. At stage 16, stable heterotypic DA3/DA2 contacts in wt, and both homotypic DA3>DA2/DA3>DA2 and DA3>DA2/DA2 contacts are stabilised. Anti-Spectrin staining is shown in black and white on the right of each panel, to indicate the position of the dorsal iMASs. (B) Schematic drawings of DA muscle development (green) in wt and colembryos; interpreted from data in Figure 4 and supplementary videos. Arrows indicate attractive cues, broken lines repulsive cues, thin circles attachment initiation, thick circles, attachment stabilisation. DA3 MAS initiation starts earlier at its posterior than anterior end. Homotypic repulsion between muscles of same identity leads to DA3 posterior attachment to lateral tendon cells. Repulsion does not operate upon DA3>DA2 transformation, leading to stable homotypic DA3>DA2 and DA3>DA2/DA2 iMASs.

(A) Schematic representation of the vestigial (vg) transcribed region (https://flybase.org/cgi-bin/gbrowse2/dmel/?Search=1;name=FBgn0003975). GMR fragments tested for their enhancer activity are drawn as brown horizontal bars; GMR69G04 and GMR69G05 are active in DA3 and DA2 muscles. The fragment of overlap between GMR69G04 and GMR69G05, positions 12889557–12891520 on Chr2R, is called vgM1 (B) vgM1-moeGFP expression in the DA2, DA3 and VL1 muscles in wild type embryos at stages 14 and 16 (schematised on the right); higher magnification views show the close apposition of the DA3 ventral and DA2 dorsal sufaces at stage 14 and their tandem arrangement at stage 16. Of note, vgM1 expression seems to be more robust in DA2 than DA3.

Only the GFP channel is shown, in black and white, to highlight protrusions emanating from the DA3 and DA2 muscles which contact each other (white arrows). Similar protrusions are seen between the DA2 and DA1 muscle (open arrows) which expresses VgM1-moeGFP in the thoracic T2 muscle (blue open arrows). The DA3/DA2 and DA2/DA1 muscle-muscle attachments are circled in white at one segment boundary. The observed protrusions are very dynamic as observed in videos and schematised in Figure 5B.

Figure 5—figure supplement 2.

Interactions between the developing DA2 and DA3 muscles via dynamic protrusions Immunostaining of a stage 15 VgM1-moeGFP-H2bRFP; srGal4 >mcd8 RFP; L-CRM-moeGFP embryo.

Only the GFP channel is shown, in black and white, to highlight protrusions emanating from the DA3 and DA2 muscles which contact each other (white arrows). Similar protrusions are seen between the DA2 and DA1 muscle (open arrows) which expresses VgM1-moeGFP in the thoracic T2 muscle (blue open arrows). The DA3/DA2 and DA2/DA1 muscle-muscle attachments are circled in white at one segment boundary. The observed protrusions are very dynamic as observed in videos and schematised in Figure 5B.

Muscle strength lines in larvae

To investigate the impact of embryonic muscle patterning defects on Drosophila larval crawling, we first recorded the fraction of DA3, DA3>DA2 and branched muscles in 3rd instar wt and col larvae expressing GFP under control of the Myosin heavy chain (Mhc) promotor region (Figure 6A). It conforms to the statistics in late embryos (Figure 2B), except for an increased proportion of branched muscles, which could reflect under-evaluation of their number in embryos, due to threshold detection limit of L-CRM-moeGFP expression in thin fibres (Figure 6B). We then examined the pattern of MASs, using scanning electron microscopy (SEM) of dissected larval filets (Figure 6C; Figure 6—figure supplement 1). In wt larvae, as seen in stage 17 embryos (Figure 4A), the anterior edge of the DA3 muscle is anchored to a nodal lateral attachment site shared with the LL1 muscle, while its posterior edge aligns with the anterior edge of the DA2 muscle in the next posterior segment. Precise heterotypic DA2/DA1 iMASs are also visible over each intersegmental border. The regular alignment of the DA1, DA2 and DA3 muscles both draws strength lines spanning three adjacent segments, and regular tension surfaces between segments (Figure 6C; Figure 6—figure supplement 1). In col larvae, the anterior DA3 muscle attachment has shifted from lateral to dorsal to form a DA3>DA2/DA3>DA2 homotypic iMAS (Figure 6C). This leads in turn to the formation of narrow, ectopic DA2/DA2 contacts and the DA1-DA2-DA3 strength line is distorted (Figure 6—figure supplement 1). In case of branched DA3 muscles, two strength lines co-exist (Figure 6C and Figure 6—figure supplement 1). In conclusion, SEM analyses of larval muscles show that loss of DA3 identity leads to ectopic iMASs between several dorsal internal muscles, distortion of the staggered ends architecture of DA muscles and of their alignment between consecutive segments.

Figure 6.

Muscle mismatching in L-CRM mutant larvae.

(A) Mhc-GFP expression in control, col and col live 3rd instar larvae; lateral views, anterior to the left. Examples of DA3, DA3>DA2 and DA3 branched muscles are circled in white. (B) Quantification of the relative proportions of DA3, branched DA3, DA3>DA2 and absence of DA3 muscles in A1 to A7 segments of wt/Df, col/Df and col/Df; Mhc-GFP larvae; (wt/Df, n = 375 segments/27 larvae; col/Df, n = 320/23; col/Df, n = 264/19). (C) Scanning electron microscopy of filleted larvae showing the dorsal and dorso-lateral muscles, schematically colour-coded below. In wt larvae, the DA muscles are parallel to each other within a segment with precise matching of the DA3/DA2 and DA2/DA1 MASs at each posterior segmental border. In col larvae, DA3>DA2 muscles show homotypic, dorsal MASs and the DA3/LL1 connection is lost. The posterior MASs of branched DA3 is composite.

The dorsal midline is represented by a yellow dashed line. Upper panels: four abdominal segments of wt and col L3 larvae. Lower panels: muscle orientations are schematised by coloured lines, DA1 blue, DA2 yellow, DA3 green, LL1 red and SBM grey. In a wild type larva, the DA1, DA2 and DA3 muscles are oriented parallel and draw a force line over three consecutive segments and straight tension lines between two adjacent segments. In col larvae, DA3>DA2 and branched DA3 transformations these lines are broken due to the formation of homotypic DA3>DA2 MAS and either loss or reduction of the DA3/LL1 connection.

Figure 6—figure supplement 1.

Scanning electron microscopy of wt and col internal muscles Larval fillets cut longitudinally along the ventral midline to expose the dorsal and dorso-lateral internal muscles.

The dorsal midline is represented by a yellow dashed line. Upper panels: four abdominal segments of wt and col L3 larvae. Lower panels: muscle orientations are schematised by coloured lines, DA1 blue, DA2 yellow, DA3 green, LL1 red and SBM grey. In a wild type larva, the DA1, DA2 and DA3 muscles are oriented parallel and draw a force line over three consecutive segments and straight tension lines between two adjacent segments. In col larvae, DA3>DA2 and branched DA3 transformations these lines are broken due to the formation of homotypic DA3>DA2 MAS and either loss or reduction of the DA3/LL1 connection.

Branched muscles result in subtle locomotion defects

Drosophila larval crawling relies upon ordered abdominal body wall muscle contractions (Heckscher et al., 2012). The distorted muscle patterns observed in col larvae raised the question of whether it impacted on locomotion. To address this question, we compared the locomotion of +/Df, col and col larvae using FIM (FTIR-based Imaging Method) and the FIMTrack software, which allows tracking simultaneously many larvae and quantitatively describing a variety of stereotypic movements (Risse et al., 2013; Risse et al., 2014; Risse et al., 2017). Here, we focused on three parameters: crawling speed, stride length, stride duration. We first recorded the ‘walking rate’, also called crawling speed, during the first 20 s, after larvae have been dropped on the agarose gel. At that time, larvae engage an ‘escape response’ corresponding to an active crawling phase (Figure 7A). Box-plot graphs (left) show intra-variability for each of the three genotypes. Beyond this variability, we observe, however, that col larvae display a significantly reduced crawling speed on average 1.05 ± 0.034 mm/sec (n = 108), compared to 1.15 ± 0.031 mm/sec for +/Df controls (n = 118), (p=0.03) (Figure 7A). To further investigate the origin of this speed reduction, we measured two crawling speed parameters: stride length and stride duration (Figure 7B–C). A significantly shorter stride was measured for col larvae (1.08 ± 0.025 mm) compared to control +/Df larvae (1.17 ± 0.024 mm), (p=0.008). Furthermore, stride duration was extended, from (1.09 ± 0.014 s) for +/Df larvae to 1.15 ± 0.017 s for col larvae. For both crawling speed, stride length and stride duration, col larvae (n = 112) display intermediate values. Yet, differences with either control or col/Df larvae fall below the significance threshold level, suggesting that mis-orientation of the DA3>DA2 muscle does not, by itself, significantly impair the efficiency of segment contraction. Since larval crawling integrates information provided by neuronal networks, relayed by synaptic connections between motoneurons (MNs) and muscles, the mobility phenotype of col/Df larvae could indicate defects motor innervation of DA3>DA2 muscles. The DA3 and DA2 muscles are innervated by the intersegmental nerve (ISN) which fasciculates motor axons reaching dorsal muscles (Hoang and Chiba, 2001; Landgraf and Thor, 2006). To examine the DA3>DA2 and branched DA3 innervation, we used anti-HRP and phalloidin staining to view ISN motoneuron projections and muscles, respectively (Figure 7—figure supplement 1). In wt embryos, the MN projection which innervates DA3 leaves the ISN ventral to the DA3 position and orients left such that the neuromuscular junction locates to the first anterior third of the muscle (100%; n = 28 segments). In case of a complete DA3>DA2 transformation, a MN projection is still observed (100%; n = 15 segments), which leaves the ISN at a more dorsal position than wt, reflecting the dorsal shift of the DA3>DA2, relative to DA3 muscle. In case of branched muscles, the lower branch is always innervated (100%, n = 30 segments). Only in 20% of the cases, the second, upper branch is also innervated.

Figure 7.

Branched muscles result in specific locomotion defects.

(A) Left, Tukey’s diagrams (box-plot graph) showing the walking rate in mm/s (crawling speed) of wt/Df (n = 118), col (n = 108) and col/Df (n = 112) larvae. Each point represents the average measurement for one larva, recorded during 20 s. 50% of points are located within the Tukey’s diagram. The red line gives the median, the narrowed area, the confidence interval of the median (95%). Right, same data as left, showing the mean speed ± standard error of mean (SEM) for each genotypes. (B) Stride length in mm. (C) Stride duration in s. The number of larvae (n) tested for each genotype is indicated in (A). Only significant differences are indicated (*p<0.05 and **p<0.01).

Phalloidin and anti-HRP staining of wt and col 3rd instar larval fillets, showing internal DA muscles (green) and intersegmental nerve (ISN) projections (white), schematically colour-coded below. ISN projection to the DA3 muscle is circled by a red dashed line. Absence of projection is indicated by a red cross.

Figure 7—figure supplement 1.

DA muscles innervation in L-CRM mutant larvae.

Phalloidin and anti-HRP staining of wt and col 3rd instar larval fillets, showing internal DA muscles (green) and intersegmental nerve (ISN) projections (white), schematically colour-coded below. ISN projection to the DA3 muscle is circled by a red dashed line. Absence of projection is indicated by a red cross.

These innervation data indicate that fully transformed DA3 muscles are innervated, but only one branch of branched muscles is, in most cases. It remains to be established whether this asymmetric innervation contributes to the reduced crawling speed of col larvae.

Discussion

The stereotyped set of 30 somatic muscles in each abdominal segment which underlies Drosophila larval crawling has been thoroughly described many years ago (Bate, 1990; Bate and Rushton, 1993). One essential aspect laid out at that time was the concept of founder cell (FC), i.e., the assignment to form a distinctive muscle to a single founder myoblast able to recruit other myoblasts by fusion. The morphological identity of each muscle is foreseen by a specific pattern of iTF expression in its FC (Tixier et al., 2010; Frasch, 1999). Here, we engineered a CRM deletion strategy connecting iTF expression to the larval musculature architecture and locomotion.

iTF transcription in muscle PCs; redundant CRMs

Some iTFs are transiently transcribed during Drosophila muscle development, for example, Kr and nau, others such as col, ladybird and slouch/S59, at every step of the process (Dubois et al., 2016; Knirr et al., 1999; Bataillé et al., 2017; Dubois et al., 2007; Michelson et al., 1990; Jagla et al., 2002). Reporter analyses indicated that col transcription is controlled by two sequentially acting CRMs, of overlapping activity at the PC step, suggesting a handover mechanism between CRMs at this stage (Enriquez et al., 2012). However, deletion analyses show that L-CRM activity in PCs is not dependent, but redundant with E-CRM activity, and can be separated from col positive autoregulation that is specific to the DA3 lineage. This leads to a new model where iTF code refinement at each step of muscle identity specification, PMC >PC, PC >FC and FC >syncytial nuclei, is driven by a separate CRM. Distribution of the PC-identity information into two CRMs further supports the idea that iTF regulation at the PC stage is nodal to muscle identity specification (Carmena et al., 1998; Dubois et al., 2016; Enriquez et al., 2012; Jagla et al., 2002; Nose et al., 1998; Kumar et al., 2015). Our former analysis started to decrypt the combinatorial control of each dorso-lateral muscle identity, which involves at least eight different iTFs, in addition to Nau/MRF [Dubois et al., 2007]. Persistent, low level expression of Col protein in the DT1 and LL1 muscle precursors upon removal of the col autoregulatory module suggests the existence of both positively and negatively acting iTF-responsive elements in this module. Our present analysis concentrated on the DA3 lineage where col expression is maintained, whereas col is expressed in several PCs. Morphological transformations of other dorso-lateral muscles are observed at variable frequency in col protein null mutants (Enriquez et al., 2010), suggesting that E-CRM activity provides robustness to the combinatorial control of identity of these muscles and that robustness is likely also contributed by other iTFs (Dubois et al., 2016; Schnorrer and Dickson, 2004). PC/FC specific CRMs have only been functionally identified for a handful of muscle iTFs (Rivera et al., 2019). Computational predictions identified, however, several thousand putative muscle enhancers and uncovered extensive heterogeneity among the combinations of transcription factor binding sites in validated enhancers, beside sites for the core intrinsic muscle regulators Tin, Mef2 and Twi (Sandmann et al., 2007; Gisselbrecht et al., 2013; Cusanovich et al., 2018). Dissecting whether step-specific and redundant/distributed CRM configurations (Cannavò et al., 2016; Hong et al., 2008; Frankel et al., 2010) apply to many muscle iTFs and underlie the progressively refined control of final muscle patterns is a future step.

Distinctive muscle morphology requires identity reprogramming of fused myoblasts

The process by which iTFs determine the final morphological features of each muscle, is not fully understood. Fusing FCM nuclei generally adopt the iTF protein code of the FC nucleus, while propagation of iTF transcription and activation of realisation genes is muscle lineage-specific (Boukhatmi et al., 2012; Crozatier and Vincent, 1999; Knirr et al., 1999; Bataillé et al., 2010; Bataillé et al., 2017; Bourgouin et al., 1992). We have previously shown that Col protein import precedes activation of col transcription in fused FCM nuclei, and correlates with the activation of realisation genes, a sequence of events termed syncytial identity reprogramming (Bataillé et al., 2017; Dubois et al., 2007). Upon loss of col transcription in the DA3 FC (col embryos), there is a complete DA3>DA2 transformation. When col transcription is maintained (col embryos), in the DA3 FC but not propagated to other syncytial nuclei, there is an incomplete DA3 transformation into branched muscles. From this, two conclusions can be drawn: 1) Final selection of myotendinous connection sites is intrinsic to FC identity. 2) Identity reprogramming of syncytial nuclei is required for robustness of this selection and precise muscle patterns.

Identity shifts, a source of branched muscles

Live imaging of lateral-oblique and ventral transverse muscles development distinguished 3 phases of muscle elongation (Schnorrer and Dickson, 2004): 1) FC migration, stage 12, ending with the first FC/FCM fusion event and stretching of the muscle precursor along a given axis; 2) bipolar myotube elongation, characterised by the presence of extensive filopodia at both axis ends, in search for attachment sites, stages 13 to 15; 3) maturation of myotendinous attachment, stage 16. The formation of DA3>DA2 and branched muscles in CRM mutants recalls a transient exploration by the wt DA3 muscle precursor of both dorsal and lateral tendon cells, a process deviating from the bipolar migration/attachment scheme (Schweitzer et al., 2010; Enriquez et al., 2012; Schnorrer and Dickson, 2004; Bahri et al., 2009). Interestingly, ectopic Col expression in the DA2 muscle leads to reciprocal DA2>DA3 transformation as well as branched muscles (Boukhatmi et al., 2012). This indicates that selection of dorsal versus lateral tendon cells is a highly controlled process. A few molecules involved in targeted attachment of subsets of muscles to specific tendon cells have been identified: Kon tiki/Perdido, a single pass transmembrane protein and the PDZ protein DGrip for proper elongation of ventral longitudinal muscles (Schnorrer et al., 2007); the ArfGAp protein Git for sensing integrin signaling and halting elongation of Lateral Transverse (LT) muscles once their attachment site has been reached (Bahri et al., 2009; Richier et al., 2018). Robo/Slit signaling attracts muscles at segmental borders, the Slit ligand being expressed by tendons, and Robo and Robo2 receptors by elongating muscles. slit also acts as a short-range repellent contributing to the collapse of leading-edge filopodia when a muscle reaches the tendon extracellular matrix (Ordan and Volk, 2015; Ordan et al., 2015). In vivo imaging showed that the DA3 muscle fails to stop at the segment border in slit mutants and sometimes branches (Ordan et al., 2015). However, this is not observed in col; the DA3 slit branching pattern is rather similar to that frequently observed in nau mutants, with two posterior attachment sites in place of one (Dubois et al., 2016; Boukhatmi et al., 2012). Which realisation genes downstream of iTFs are responsible for the precision of DA attachment sites, that is, proper balancing attraction/repulsion cues, should be the focus of future studies.

Muscle staggered ends; heterotypic versus homotypic interactions?

In addition to attaching to tendon cells, it was previously shown that internal muscles attach to each other (Maartens and Brown, 2015; Bate and Rushton, 1993). Detailed imaging in both embryos and larvae shows that the DA3/DA2 and DA2/DA1 attachment sites precisely match over each segmental border, such that larval DA1, DA2, and DA3 align over three consecutive segments. Recording DA3 and DA3>DA2 muscle development by a combination of live imaging and immunostainings shows that DA3/DA2 matching in wt embryos results from both heterotypic DA3/DA2 attraction and, possibly, homotypic repulsion. Prior attachment of the posterior DA3 edge to dorsal tendon cells leads to retraction of filopodia issued from the homologous muscle in the next adjacent segment while stabilisation of heterotypic contacts results into DA3/DA2 iMAS formation over the intersegmental border. A similar process results in alignment of DA2 with DA1. Interestingly, DA3>DA2 establish both homotypic iMASs, and heterotypic iMASs with DA2, suggesting preserved attraction to tendon cells and partial loss of preferential heterotypic adhesion. Prior posterior attachment was previously observed during development of abdominal adult muscles (Currie and Bate, 1991). Whether this temporal sequence is instrumental in the precise matching of muscles over each segmental border and whether competition between muscles for the same tendon cells is also involved remain to be assessed. Cell matching is a widely used process during embryogenesis to construct complex tissue architecture. Selective filopodia adhesion has recently been shown to ensure precise matching between identical cardioblasts and boundaries between different cell identities in the Drosophila heart. In this case, homotypic matching is linked to differential expression by each cell type of the adhesion molecules, Fasciclin III and Ten-m (Zhang et al., 2018). Transcriptome analyses of specific muscles at different developmental times should allow to identify attractive and, possibly, repulsive molecules, acting in muscle precise matching.

Branched muscles impact on crawling speed

Drosophila larval crawling is a well-suited paradigm to link muscle contraction patterns and locomotor behaviour. Longitudinal, acute and oblique muscles within a larval segment contract together and, as they begin to relax, the contraction is propagated to the next segment, creating a peristaltic wave from tail to head (forward locomotion), or head to tail (backward locomotion) (Heckscher et al., 2012). The rhythmic movements of locomotion are part of behavorial routines that facilitate the exploration of an environment. Exploratory routines alternate straight line movement also called ‘active crawling phase’, with change of direction, the ‘reorientation phase’ (Günther et al., 2016; Berni et al., 2012; Lahiri et al., 2011). The active larval crawling phase requires an intense, prolonged muscular effort. In this study, we focused on crawling parameters during this phase. The crawling speed during the escape response is not significantly reduced in col/Df, compared to control +/Df larvae, indicating that muscle contraction is properly controlled and that a mechanical compensation mechanism for the DA3 mis-orientation could occur. However, it is significantly reduced in col/Df larvae. This seems paradoxical because the number of DA3>DA2 transformed muscles is higher in col/Df larvae. col/Df larvae present many branched muscles, however. While DA3>DA2 are always innervated, only one branch of branched muscles is, most of the time, raising the possibility that branched muscles do not contract properly, or with a gap in time (see Zarin et al., 2019). From these different observations, we can conclude: i) single muscle transformations only moderately impact crawling speed, raising the possibility of biomechanical compensation by other muscles; ii) branched muscles could be less efficient than fully transformed muscles. At this point, the reason why, – either mechanic weakness, improper innervation or impaired Ca2+ wave propagation, antagonistic force lines upon muscle contraction - may only be object of speculation.

The generation of branched muscles in Drosophila identity mutants is one important finding as branched muscle fibres accumulate in humans, following muscle regeneration after damage or in Duchenne muscular dystrophy patients (Chan and Head, 2011). It opens the possibility to investigate, in vivo, how physiological properties of branched fibres differ from morphologically normal fibres and associated mechanical instability in an otherwise normal muscle pattern.

Materials and methods

Fly strains

All Drosophila melanogaster stocks and genetic crosses were grown using standard medium at 25°C. The strains used were whiteLCRM 4–0.9 (Enriquez et al., 2010), col (Crozatier and Vincent, 1999), sr-Gal4 (obtained from G. Morata, Madrid, Spain). The 12 kn and 10 vg Janelia-Gal4 lines (GMR) (Pfeiffer et al., 2008), UAS-mcd8RFP, Mhc-GFP, Df(2L)BSC429, kn (BDSC_67516) (Nagarkar-Jaiswal et al., 2015), vasa-cas9 (BDSC_51324), Ilk-GFP (w; P{PTT-GB}Ilk) (BDSC_6831), lines were provided by the Bloomington Drosophila Stock Center. The col and Df(2L)BSC429 strains were balanced using CyO,{wg} or CyO,{dfd-YFP} and homozygous mutant embryos or larvae identified by absence of lacZ or YFP expression, respectively.

CRM deletions generated by Crispr/Cas9

Genomic col target sites were identified using http://tools.flycrispr.molbio.wisc.edu/targetFinder/ (Gratz et al., 2014). Prior to final selection of RNA guides (gRNA) for deletions of col CRMs, genomic PCR and sequencing of DNA from kn and vasa-cas9 flies was performed to check for polymorphisms in the targeted regions. Guides targeting E-CRM and L-CRM were inserted in the pCFD4: U6:3-gRNA vector (Addgene no: 49411) as described (Port et al., 2014); (see http://www.crisprflydesign.org/wp-content/uploads/2014/06/Cloning-with-pCFD4.pdf). All guides were verified by sequencing. The sequences of the oligonucleotides used to construct each gRNA expression plasmid are given in Figure 1—figure supplements 1 and 2. To delete the core region of L-CRM, vasa-cas9 embryos were microinjected with gRNAs in pCFD4 (200 ng/μl). To delete the E-CRM, kn embryos were injected with gRNA in pCFD4 (150 ng/μl) and pAct-Cas9U6 (400 ng/μl). Each adult hatched from an injected embryo was crossed to the balancer stock snaCyO, {wg} and 100–200 F1 fly were individually tested for either col CRM deletion by PCR on genomic DNA. A pre-screening for E-CRM deletion was based on the loss of yellow carried by Mimic kn.

Reporter constructs

The yellow intron (yi), FlyBase ID #FBgn0004034 (position: 356918–359616) was inserted in the lacZ coding region between aa (Tyr 952) and aa (Ser 953) by standard PCR-based cloning position. The resulting fragment was cloned downstream of L-CRM inserted in a pAttB vector, and micro-injected in embryos for chromosomal insertion at position 68A4. VgM1-moeGFP was constructed by PCR amplification of the GMR69G04 and GMR69G05 overlap. The 1.4 kb amplicon (named VgM1) was inserted upstream of moeGFP to generate the pAttB VgM1-moeGFP construct. It was inserted at position 68A4 on the third chromosome.

Immunohistochemistry

Antibody staining, in situ hybridisation with intronic probes and phalloidin staining were as described previously (Dubois et al., 2007). Primary antibodies were: mouse anti-Col (1/50; Boukhatmi et al., 2012; Dubois et al., 2007), anti-LacZ (1/1000; Promega), mouse anti α-Spectrin (1/200; Hybridoma Bank), rabbit anti-GFP (1/1000; Torrey Pines Biolabs), chicken anti-GFP (1/500; Abcam), Phalloidin-Texas RedX (1/500; Thermofisher Scientific). Secondary antibodies were: Alexa Fluor 488-, 555- and 647- conjugated antibodies (1/300; Molecular Probes) and biotinylated goat anti-mouse (1/2000; Vector Laboratories).

Motoneurons and muscles visualisation

To both visualise motoneuron axonal pathways and muscles, fillets of control and col third instar larvae were incubated overnight with Alexa 594-conjugated anti-HRP (1/300; Jackson Immunological Research) and Alexa Fluor 488 Phalloidin (1/500; Thermofisher Scientific), at 4°C. To prepare fillets, third instar larvae placed in myorelaxant buffer (Yalgin et al., 2011), were cut longitudinally on the ventral side to expose the dorsal and dorso-lateral musculature. Fillets were then fixed 1 hr in 4% formaldehyde and washed in PBT.

In situ hybridisation

In situ hybridisation with Stellaris RNA FISH probes were done as described by the manufacturer for Drosophila embryos (https://www.biosearchtech.com). The FISH probe sets for col and nau were designed using the Stellaris probe designer (https://www.biosearchtech.com/stellarisdesigner) and labelled with Quasar 670 Dye (col) and Quasar 570 Dye (nau) (Stellaris Biosearch Technologies). One set of 48 oligonucleotides was designed against the first col intron to detect primary nuclear transcripts. Another set of 48 oligonucleotides was also designed against the first and third nau introns. When antibody staining and FISH were combined, the standard immuno-histochemistry protocol was performed first, with 1 U/μl of RNase inhibitor from Promega included in all solutions, followed by the FISH protocol. Confocal sections were acquired on Leica SP8 or SPE microscopes at 40x or 63x magnification, 1024/1024 pixel resolution. Images were assembled using ImageJ and Photoshop softwares.

Quantification of col transcription

To quantify the level of col nuclear transcripts in FCs and PCs, we calculated the ratio between col and nau hybridisation signals using intronic probes. Before using nau as internal reference we verified that nau transcription level is not modified in col L-CRM mutants. The same laser parameters were set for all intronic probes and at least five different embryos at each stage 11 and 12 were recorded. Optimal Z stacks were acquired at × 40. ImageJ was used to analyse the data. For each stack, a Sum slices projection was generated. Each region of interest (ROI), corresponding to a DA3 nucleus, was manually drawn, based on Mef-2 immunostaining. The same ROI served to determine the intensity of nau and col signals on the green and red channels, respectively (Figure 3—figure supplement 2A). A threshold was applied to each channel to remove background. Data plots and statistical analyses were performed with Prism 5.0 using unpaired t-test.

Phenotype quantification at embryonic and larval stages

To quantify embryonic phenotypes, L-CRM-moeGFP embryos were immunostained with a primary mouse anti-GFP (1/500) (Roche) and secondary biotinylated goat anti-mouse (1/2000) (VECTASTAIN ABC Kit). Stained embryos were imaged using a Nikon eclipse 80i microscope and a Nikon digital camera DXM 1200C. A minimum 100 A1-A7 abdominal segments of stage 15–16 embryos were analysed for each genotype. (+/Df: n = 127 segments - 16 embryos; col/Df: n = 170 segments - 23 embryos; col/Df: n = 190 segments - 27 embryos; col/Df: n = 103 segments - 13 embryos). To quantify larval phenotypes, wandering L3 larva displaying Mhc-GFP reporter line were immobilised between slide and coverslip, and left and right larval sides imaged using Nikon AZ100 Macroscope at 5x magnification. Minimum 260 abdominal segments were analysed for each genotype. (+/Df: n = 375 segments - 27 larvae; col/Df: n = 320 segments - 23 larvae; col/Df: n = 264 segments - 19 larvae).

Live imaging embryonic muscle development

Embryos were bleach dechorionated and stage 12 embryos manually picked, laterally orientated and mounted on a coverslip coated with heptane glue to prevent drift during imaging. A drop of water was placed on the embryos to maintain their survival. Images were collected on a Leica TCS-SP8 confocal using a 25X water immersion lens. Sections were recorded every 130 to 150 s for the wt embryos and every 120 to 160 s for the col embryos, and z-stacks collected with optical sections at maximum 1 µm interval. Image processing was performed with Fiji (http://fiji.sc/wiki/index.php/Fiji) and custom programming scripts in Fiji. The z-stacks projections were corrected in x and y dimensions by manual registration using a reference point tracking.

Scanning electron microscopy (SEM)

To prepare fillets, third instar wild type and homozygous col larvae raised at 25°C were dissected in myorelaxant buffer, according to Gratz et al., 2014. Larvae were cut longitudinally on the ventral side to preserve and expose the dorsal and dorso-lateral musculature. Fillets were then fixed 1 hr in a 4% formaldehyde/2.5% glutaraldehyde mixture in 1X PBS, washed in water and dehydrated gradually in ethanol. Fillets were dried at the critical point (Leica EM CPD 300 critical point apparatus), covered with a platinum layer (Leica EM MED 020 metalliser) and imaged with a Quanta 250 FEG FEI scanning microscope.

Behavioral analysis

We conducted locomotion assays by tracking the trajectory of larvae using the FIM method (Risse et al., 2013). Wandering third instar larvae were gently picked up with a paintbrush and transferred to an agar plate. The larvae were then videotaped using a digital camera (Baumer VCXG53M); lentille (Kowa LM16HC); infrared filter (IF093SH35.5). Each video containing 5 to 10 larvae per run, on a 1% agarose gel, was recorded at five frames/sec for 20 s. Individual larva were tracked using the FIMTrack software (Risse et al., 2014), which provided the position across time of five points regularly spaced along the spine of each animal, from head to tail. Analysis was done by using MATLAB software. Peristalsis cycles were obtained using the derivative of the spine length (i.e., sum of the distance across successive point along the spine) through time, which provide a time series smoothly oscillating around zero. For each peristalsis cycle, we measured Stride length (centroid displacement across each cycle) and Stride duration (cycle duration). Walking rate was obtained by measuring the distance of the centroid (3rd spine point) across successive frame. These values were averaged for each individual across the 20 s of recording. Statistical comparisons between genotypes were computed using a linear model with GenoT as fixed effect, and individual larva as a statistical unit.

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The authors study the gene encoding a muscle identity transcription factor, Collier, through examination of its cis-regulatory modules. They demonstrate how selection of attachment sites in muscle and locomotion are related to muscle identity and through transcriptional reprogramming of syncytial nuclei.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Intrinsic control of muscle attachment sites matching" for consideration by eLife. Your article has been reviewed by a Senior Editor, and three reviewers. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

We do appreciate that some of the results are interesting and novel. However, we are not persuaded that they are a significant advance and of sufficient interest for eLife readers at this stage. Also, the manuscript should be re-written so that the conclusions are not over-interpreted. In addition, quantification of the FISH analysis must be provided and compared to the protein distribution and intensity. These are amongst the important experiments suggested below that would make the study more widely relevant. We feel that these could take well over two months to do well. We will be pleased to consider a fresh submission, in due course, that addresses all reviewers' concerns, should the authors wish to consider eLife.

Reviewer #1:

The manuscript by Carayon et al. addresses an intriguing question of whether the acquisition of muscle identity could impact muscle function and mobility of an organism. To tackle this they generate a series of regulatory region deficiencies that affect transcription of Drosophila muscle iTF col required for the formation and properties of DA3 muscle. In particular, they generate two viable col regulatory regions mutant lines: one in which a complete DA3 to DA2 transformation takes place and another line with deletion of col autoregulatory region in which DA3 is not fully transformed to DA2 and has branched morphology.

Interestingly, mutant larvae with branched DA3 muscles display several mobility defects demonstrating that inappropriate acquisition of identity of one muscle could have a deleterious systemic effect. From this perspective, it is a valuable and novel work that merits to be published in eLife.

There are a few aspects that require some clarifications:

1) Is the problem of identity propagation observed at stage 14 in col context also seen at later stages in DA3 muscle? In other words, is it only a delay in switching on col or permanent incapacity to make fused nuclei transcriptionally active for col? FISH experiment with intronic col probe on stage 16 embryos would help to clarify this point.

2) Some fraction of DA3 muscles in col embryos does not display branching phenotype. Could this normal DA3 morphology be associated with a partial or total identity propagation? Could authors document such cases? This would further strengthen the link between identity propagation and branching.

3) Why one nucleus is still able to transcribe col and produce Col protein in the context autoregulation is missing. Also, is Col protein expression level in the col-positive nucleus of col DA3 similar to that in wt DA3?

4) Authors show that larvae with branched DA3 move slower than larvae with DA3 fully transformed to DA2 indicating that muscle branching has more impact on muscle function than muscle transformation. This could be the case, however other muscle properties and in particular muscle innervation could contribute to mobility phenotype. Some views showing how branched versus fully transformed DA3 muscles are innervated would be of help.

Reviewer #2:

The manuscript of Carayon et al. investigates how the identity of a muscle cell shapes the choice and structure of muscle attachment to its tendon. Previous work from the group has identified specific muscle enhancers in the identity gene collier that controls its expression during the process of founder specification and muscle morphogenesis. This new manuscript adds to this body of work by making specific genomic deletions of these enhancers via Crispr technology. The group subsequently analyzes the phenotypes that result from the selective removal of these enhancers, through confocal imaging of both fixed and live Drosophila samples, by SEM and by locomotion assays. These new reagents allow the investigators to map the changes in the patterns of collier expression, assess the contribution of each enhancer to collier expression, and assess one important aspect of muscle identity – the attachment to the tendon cells. Of the key conclusions, they find that (1) as expected from previously published work in the field, FC identity controls selection and development of muscle attachment to the tendon; (2) the robustness of the muscle pattern depends on the reprogramming of myonuclei after fusion (as suggested by earlier published work); (3) the morphogenesis of the tendon-muscle interaction requires several steps in the process, that when altered, leads to the development of branched muscles; and (4) these alterations in muscle attachment (branched muscles) can lead to defects in muscle activity as measured by locomotion of the organism.

Overall, this work is solid. However, the issue for me becomes whether this work provides significant enough advancement for the muscle biology field to merit publication in eLife. Unfortunately, I believe that this manuscript would be better suited for a more specialized journal. To merit consideration in eLife, one would want to see, for example, how collier regulates the dynamic attachment process that is described.

Reviewer #3:

In this manuscript, Alexander Carayon et al. analyze the consequences of deletion of transcription cis-regulatory modules of a single muscle identity Transcription Factor (iTF), Collier (Col), which in Drosophila embryo regulates the identity of muscle DA3. Previously the authors characterized 3 cis-regulatory regions of the Col gene, responsible for early, late, as well as autoregulatory transcriptional control of Col gene in DA3 muscle. They now have deleted each of these sites and ask how it affects Col transcription in the DA3 muscle, and what would be the outcome in terms of muscle morphology, attachment, and larval movement. They show that the CRM for late expression of Col (and to some extent the autoregulatory CRM) are essential for providing the DA3 identity since in its absence, this muscle is either abnormal or partially transformed into DA2 muscle. Although the contribution of Col to the identity of the DA3 had been reported previously, the present study adds important information regarding the functional contribution of the different Col regulatory regions to the muscle identity, attachment sites, and overall larval movement. These results are clean and convincing.

Essential revisions:

1) Overinterpretation of the data: in the abstract, the authors write that " We show that both selection of muscle attachment sites and muscle/muscle matching is intrinsic to muscle identity and requires transcriptional reprogramming of syncytial nuclei". Whereas they indeed show that selection of muscle attachment sites is intrinsic to muscle identity, there is no direct evidence that it requires transcriptional reprogramming of syncytial nuclei.

2) The Col protein, as well as FISH analysis in the col or col are both intriguing because they indicate that Col transcription is retained in a single nucleus (presumably the founder nucleus) within the entire syncytium of DA3. Do the authors have an explanation as to why the Col protein, being produced in the syncytial cytoplasm, does not translocate to the neighboring nuclei?

3) It will be also informative to demonstrate where this nucleus resides relative to the other nuclei, and whether this impacts on the outcome in terms of the observed attachment sites. In Figure 3, the authors should show the other nuclei of the DA3 muscle in each of the mutants. This can be easily achieved by co-staining the muscles with GFP-NLS driven by col-Gal4 driver.

4) Why the transformation of DA3 to DA2 is relatively rare? Could it be that additional iTFs exist in the DA3 muscle?

5) Do the authors have any idea as to why the col is homozygous female sterile. Is it due to an off-target effect?

6) Quantification of the FISH: it would be important to add quantification data as to the reduction in the col transcripts in the mutants shown in Figure 3.

7) The authors should explain better what they mean by homotypic repulsion.

8) Did the authors check that muscle genes, such as MHC, TpnC, actin etc, are expressed normally and in the mutant DA3 muscle, and that its innervation occurs ordinarily? Can they rule out that the muscle does not function due to physiological reasons, rather than the aberrant attachments?

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for re-submitting your article "Intrinsic control of muscle attachment sites matching" for consideration by eLife. Your article has been reviewed K VijayRaghavan as the Senior Editor, and two reviewers. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

Carayon et al. describe an interesting functional analysis of distinct cis-regulatory modules (CRMs) controlling the expression of the collier (col) gene in a single DA3 muscle. Previous analysis from the Vincent lab identified a critical role for col in providing the identity of the DA3 muscle. In the present study, the authors analyzed the functional contribution of specific CRMs for the temporal col expression and function, in the DA3 muscle, following their specific deletion. The analysis revealed that deleting the Late CRM (L-CRM, col), which contains both Mef2 and Twi binding sites, as well as an autoregulatory Col binding sequences, led to the transformation of the DA3 into the DA2 muscle. Deleting a smaller region within this sequence responsible only for the autoregulatory Col activity led to abnormal split muscle DA3. The authors show that whereas in col DA3 muscle the entire transcription of col is undetectable at stage 14, in fused multinucleated DA3 muscle, the deletion of only the auto-regulatory Col sequences led to partial col transcription. Further image analysis of live embryos indicated the sequence of events leading to the split DA3 phenotype. The authors speculate that Col late expression in all the nuclei of DA3 is required for proper detachment of DA3 from the DA2 muscle at its dorsal site and further attraction to the LL1 attachment site. Additional analysis was performed to extract the crawling behavior of the mutant larvae. This analysis implied that the transformation of DA3 to DA2 did not affect significantly the larval step length, however, larvae with split DA3 did show a change in stride length, which was also correlated with aberrant innervation of the split muscle.

Essential revisions:

Overall, the results regarding the functional contribution of the distinct CRMs are convincing and nicely presented. However, the conclusions regarding muscle-muscle repulsion and the model explaining the sequence of events leading to DA3 transformation are less convincing. The link to larvae crawling behaviour is also puzzling due to the stronger phenotype obtained by partial deletion of Col, relative to its complete deletion, and also due to the effect on muscle innervation. These could be discussed in the 'Discussion' section.

1) Figure 2B – the numbers of embryos analyzed is not indicated.

2) Figure 3B – in col st 14 there are extra nuclei labeled below the DA3 muscle. These are not stained in the control. What are these?

3) Figure 3C – why does Col protein not label all nuclei in the control DA3? Why is the green label stronger in col relative to col.

4) Figure 3D – The quantification of transcription at stage 14 is missing. We are puzzled by the question of why late CRM is not active when the autoregulatory is missing. Mef2 should be still active in all nuclei, should it not?

5) Figure 3D – there is no information of how the fluorescence quantification was done. Did the authors perform background subtraction? Did the authors check whether nau transcription is affected or not by col depletion? They should provide images and quantification of nau FISH as well or give clear reasons of why these could be future work if they are unable to do this now.

6) Figure 5 – The authors describe numerous membrane protrusions, which we do not see in the image provided, and no quantification of these is described. No information regarding homotypic DA3/DA3 and DA2/DA2 is provided. Are these real homotypic adhesions? Again, please give clear reasons for why these could be future work if they are unable to do further experiments now.

7) Figure 5 – The images are very nice, however, we do not know about repulsion or attraction. It is speculation. There could be other explanations. Please tone down and discuss various possibilities

The suggested experiments can be done speedily, assuming that the FISH data are already available. In addition, they should tone down the sentences regarding attraction/repulsion because they do not bring strong evidence for that, other than the live imaging, which is nice but descriptive.

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

[…]

There are a few aspects that require some clarifications:

1) Is the problem of identity propagation observed at stage 14 in colΔL0.5 context also seen at later stages in DA3 muscle? In other words, is it only a delay in switching on col or permanent incapacity to make fused nuclei transcriptionally active for col? FISH experiment with intronic col probe on stage 16 embryos would help to clarify this point.

The suggested experiment has now been done and the results are presented in a supplementary figure (new Figure 3—figure supplement 1) and recorded in Results section subsection “(Re)programming of syncytial nuclei is required for muscle morphological identity”.

Briefly, in wt embryos, col transcription in all nuclei of the DA3 muscle is switched off at stage 16. The same complete switch off is observed in col embryos, showing that there is no delay in activation of col transcription in low Col level conditions, but rather a low number of fused nuclei which is transcriptionally reprogrammed to DA3 identity.

2) Some fraction of DA3 muscles in colΔL0.5 embryos does not display branching phenotype. Could this normal DA3 morphology be associated with a partial or total identity propagation? Could authors document such cases? This would further strengthen the link between identity propagation and branching.

We agree with the reviewer that normal DA3 morphology observed in col is likely associated with at least partial identity propagation. A proxy is L-CRM-moeGFP expression (Figure 2A), which depends upon Col levels through its autoregulation site. To more directly link DA3 nuclei reprogramming to normal versus branched DA3 formation would require visualization of col transcription when selected muscle attachment sites have been stabilized (see Figure 4B). As discussed above, col transcription has already ceased at that stage, precluding this experiment.

3) Why one nucleus is still able to transcribe col and produce Col protein in the context autoregulation is missing. Also, is Col protein expression level in the col-positive nucleus of colΔL0.5 DA3 similar to that in wt DA3?

The question of the transcriptional outcome of the FC relative to final muscle identity is indeed a key question. To address this question, we have chosen to quantify the col transcription rather than Col protein levels, using stellaris single RNA probes, because of higher accuracy. Results, now presented in new Figure 3D and in subsection “(Re)programming of syncytial nuclei is required for muscle morphological identity” show that the level of col transcription in the DA3 FC is nodal for transcriptional reprogramming of syncytial nuclei and proper muscle identity. This is an important addition to the previous version; thanks to the reviewer.

4) Authors show that larvae with branched DA3 move slower than larvae with DA3 fully transformed to DA2 indicating that muscle branching has more impact on muscle function than muscle transformation. This could be the case, however other muscle properties and in particular muscle innervation could contribute to mobility phenotype. Some views showing how branched versus fully transformed DA3 muscles are innervated would be of help.

We thank the reviewer again for their suggestion. Labelling motoneuron axons using anti-HRP staining, shows that fully transformed DA3>DA2 muscles are innervated. However, in most cases, only one branch of branched muscles is (new Figure 7—figure supplement 1). This innervation defect could, in part, explain the impaired locomotion specifically observed in larvae with numerous branched muscles, as now stated in subsection “Branched muscles result in subtle locomotion defects”.

Reviewer #2:

[…]

Overall, this work is solid. However, the issue for me becomes whether this work provides significant enough advancement for the muscle biology field to merit publication in eLife. Unfortunately, I believe that this manuscript would be better suited for a more specialized journal. To merit consideration in eLife, one would want to see, for example, how collier regulates the dynamic attachment process that is described.

How Collier regulates DA3 muscle attachment and identifying Col targets in this process is certainly a priority next step. We already tested several candidate genes, either via phenotypic analyses (e.g., slit), or expression analyses (e.g., members of the Dip and Dpr families; Orban et al., 2013; Tan et al., 2015), focusing on Col direct targets (de Taffin et al., 2015) and their proposed partners, but without success. There are hundreds of extracellular protein variants and this is a long-standing investment. The syncytial nature of muscles when the attachment process takes place and dynamic reprogramming of syncytial nuclei has been, so far, an obstacle to transcriptome and ChIP-seq analyses of single muscle, in any organism. New toolsare being developed in our lab to profile in parallel several dorsal muscles in wt and mutant embryos, but we feel that this goal is beyond the scope of our present analysis which connects iTF CRM regulation and locomotion.

Our detailed analysis of DA3 muscle morphological development, both by live-imaging with a dedicated reporter and on fixed embryos and larvae (modified Figure 4, Figure 5 and Figure 6), provides a novel, unprecedented view of the muscle attachment site selection and attachment process (modified scheme Figure 5B). Data in the previous version suggested, for the first time, that homotypic repulsion and heterotypic attraction are involved in precise muscle-muscle attachments leading to the exquisite staggered muscles pattern. In the present manuscript, we visualized both the DA2 and DA3 muscles in wt and mutant embryos using a newly identified driver (new Figure 5A) and the data fully support our previous conclusions.

Reviewer #3:

[…]

Essential revisions:

1) Overinterpretation of the data: in the abstract, the authors write that " We show that both selection of muscle attachment sites and muscle/muscle matching is intrinsic to muscle identity and requires transcriptional reprogramming of syncytial nuclei". Whereas they indeed show that selection of muscle attachment sites is intrinsic to muscle identity, there is no direct evidence that it requires transcriptional reprogramming of syncytial nuclei.

We agree with the reviewer that there is only indirect evidence that muscle identity requires transcriptional reprogramming of syncytial nuclei. We toned down the sentence in the Abstract and modified highlight 2, accordingly.

2) The Col protein, as well as FISH analysis in the colΔL1.3 or colΔL0.5 are both intriguing because they indicate that Col transcription is retained in a single nucleus (presumably the founder nucleus) within the entire syncytium of DA3. Do the authors have an explanation as to why the Col protein, being produced in the syncytial cytoplasm, does not translocate to the neighboring nuclei?

Immunostainings show thepresence of more Col protein in syncytial nuclei in col than col mutants (Figure 3B), and this correlates with the severity of the phenotype, branched muscle versus DA3>DA2. We also found this result puzzling. We have now measured the level of col transcription in the DA3 FC in wt and each CRM mutant (new Figure 3D). It shows that this level is significantly lower than wt in both col and col mutants, at stage 12, when there is a single nucleus. Beyond the lower level in col3 than col at that stage, what differs between these two mutant strains is the maintenance of col transcription in the syncytial muscle precursor, which is only observed in col mutants (Figure 3C). We infer from these different results that the uptake of Col protein by newly fused nuclei reflects the level of newly synthesized Col protein from the FC. One hypothesis is that, in conditions of low transcription/low translation, uptake is mainly limited to the FC nucleus, because of diffusion rate of the protein. In both col and col3 mutants, and unlike in wt, there is no other source of Col protein than the FC because of the deletion of the Col autoregulation site. We have now revised the text (subsection “(Re)programming of syncytial nuclei is required for muscle morphological identity”) to make this point clearer.

3) It will be also informative to demonstrate where this nucleus resides relative to the other nuclei, and whether this impacts on the outcome in terms of the observed attachment sites. In Figure 3, the authors should show the other nuclei of the DA3 muscle in each of the mutants. This can be easily achieved by co-staining the muscles with GFP-NLS driven by col-Gal4 driver.

This is an interesting question, which we are presently unable to answer. Previous studies of col and other genes transcription in the DA3 muscle have led to the conclusion that the FC nucleus occupies a specific, central position at stages 14 and 15, when the muscle is angled shape (Bataillé et al., 2017). The cease of transcription did, however, not allow determining its position at stage 16. The absence of col transcription, after stage 12 makes impossible to determine whether the FC nucleus adopts a different position in col mutants and no reporter we have tested so far allows to specifically track the DA3 FC nucleus in the DA3 syncytium.

4) Why the transformation of DA3 to DA2 is relatively rare? Could it be that additional iTFs exist in the DA3 muscle?

We do not feel that the transformation of DA3 in DA2 muscle in col mutants is rare (more than 85% in the embryos and 60% + plus 27% branched muscles) in larvae. Previous experiments indicated a similar penetrance for col null mutant embryos (Enriquez et al., 2012), suggesting that other iTFs may contribute DA3 identity. Other iTFs are indeed known to be expressed or/and required in the DA3 muscle (Tixier et al., 2010; Dubois et al., 2016) including Vestigial see Figure 5—figure supplement 1). We added a sentence in subsection “iTF transcription in muscle PCs; redundant CRMs”.

5) Do the authors have any idea as to why the colΔE is homozygous female sterile. Is it due to an off-target effect?

We verified that col female sterility (abortive oocyte development, data not shown) is not due to an additional mutation at the col locus, suggesting indeed that it is due to an off-target effect.

6) Quantification of the FISH: it would be important to add quantification data as to the reduction in the col transcripts in the mutants shown in Figure 3.

We thank the reviewer for their important remark. We have now quantified the col transcription level in nuclei of PC and FC cells. Results are presented in new Figure 3D and described (subsection “(Re)programming of syncytial nuclei is required for muscle morphological identity”), and show that the level of col transcription in the DA3 FC is nodal for transcriptional reprogramming of syncytial nuclei and proper muscle identity.

7) The authors should explain better what do they mean by homotypic repulsion.

We name homotypic repulsion the fact that two muscles of the same identity cannot adhere to each other and form indirect musclemuscle attachments. Internal muscle-muscle attachments are only heterotypic, i.e., form between muscle of distinct identities, e.g., DA3/DA2 and DA2/DA1. Repulsion between two DA3 muscles in wt embryos is both supported by movies (Video 1) and stainings of fixed embryos, which show that exploring filopodia issued from one DA3 muscle are repelled by the DA3 muscle of the next anterior segment. Repulsion does not operate upon (partial) loss of DA3 identity (Video 2). We extensively rephrased the text to better explain this point in the revised version (subsection “Muscle attachment: tendon attraction and homotypic repulsion”).

8) Did the authors check that muscle genes, such as MHC, TpnC, actin etc, are expressed normally and in the mutant DA3 muscle, and that its innervation occurs ordinarily? Can they rule out that the muscle does not function due to physiological reasons, rather than the aberrant attachments?

As far as we can tell, the sarcomeric structure of transformed or branched DA3 muscles is normal (see the new supplementary Figure 7—figure supplement 1). We have previously established that activation of “generic” muscle differentiation and attachment genes, such as Mhc and Kon are synchronously activated in all muscle nuclei, making unlikely that is controlled by iTFs (Bataille et al., 2017). Following the suggestion of reviewers 3 and 1, we looked at “DA3” innervation in col and col mutants. Labelling motoneuron axons using anti-HRP staining shows that fully transformed DA3>DA2 muscles are innervated. However, in most cases, only one branch of branched muscles is (new Figure 7—figure supplement 1). This innervation defect could, in part, explain the impaired locomotion specifically observed in larvae with numerous branched muscles, as now stated in subsection “Branched muscles result in subtle locomotion defects”.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:

Overall, the results regarding the functional contribution of the distinct CRMs are convincing and nicely presented. However, the conclusions regarding muscle-muscle repulsion and the model explaining the sequence of events leading to DA3 transformation are less convincing. The link to larvae crawling behaviour is also puzzling due to the stronger phenotype obtained by partial deletion of Col, relative to its complete deletion, and also due to the effect on muscle innervation. These could be discussed in the 'Discussion' section.

Proper crawling of the larva requires integration of information from neuronal networks, relayed by stereotypic synaptic connections between motorneurons and muscles (Landgraf and Thor, 2006). One possible interpretation of the colmutant locomotion phenotype is that muscle innervation of branched muscles – one connection for two branches- may interfere with its timing of contraction. In case of complete DA3 >DA2 muscle transformation in colmutant larvae, the transformed muscle is innervated by the ISN and could contract normally. It will be interesting in the next future to profile the transformed and branched muscle activity in crawling (following the study of Zarin et al., eLife, 2019), or other movements such as larval rolling.

Possible interpretations are discussed in subsection “Branched muscles impact on crawling speed”.

1) Figure 2B – the numbers of embryos analyzed is not indicated.

The number of embryos analyzed in Figure 2B was indicated in subsection “Phenotype quantification at embryonic and larval stages”. It is now also given in the legend of the Figure 2.

2) Figure 3B – in colΔL0.5 st 14 there are extra nuclei labeled below the DA3 muscle. These are not stained in the control. What are these?

We thank the reviewer for pointing out this “ectopic” expression. It corresponds to nuclei of other muscles which, in addition to DA3/DO5, originate from progenitors transcribing col (DO4/LL1 and DT1/DO3). At stage 14, under normal conditions, Col expression is only maintained in the DA3 muscle and barely visible in these other muscles. In colmutant embryos where the col binding site is removed, low level of Col expression are still detected in the DT1 and LL1, and possibly D05 nuclei. We previously noticed a similar “ectopic” expression of a 2.6-0.9 late col CRM reporter where the Col binding site was mutated (unpublished data; Author response image 1), while col expression in the DA3 is strongly reduced. Our interpretation is that either mutation or removal of the Col binding site removes a binding site for a repressing factor (perhaps competing with Col for binding) contributing to Col repression in lineages other than DA3. We have previously reported that several TFs contribute to the lineage-specific transcription/repression of col transcription (Figure 7 in Dubois et al., 2016), and this putative repressor remains to be identified (this has been added in subsection “iTF transcription in muscle PCs; redundant CRMs”). Of note Col expression in stage 14 nuclei is dependent in all muscles upon the presence of Twi and Mef2 binding sites in the late CRM (compare coland col). This is mentioned in subsection “(Re)programming of syncytial nuclei is required for muscle morphological identity”.

Author response image 1.

LacZ immunostaining of stage 14 embryos, carrying a late col CRM reporter (p2.

6-0.9LacZ). Left: The Col binding site is intact. LacZ expression is predominantly detected in the DA3 muscle. Right: The Col binding site is mutated (nucleotides in red). Low level of LacZ expression is detected in the DA3 as well as DT1, DO5 and LL1 muscles.

3) Figure 3C – why does Col protein not label all nuclei in the control DA3? Why is the green label stronger in colΔL0.5 relative to colΔL1.3.

We agree with the reviewer that the green channel may obscure the blue channel in some panels of Figure 3C, for example wt/df stage 14. Author response image 2 is the original image showing only the blue and red channels, which shows that Col protein (blue) is present in all nuclei transcribing col (red channel). We chose not to show the blue channel alone in submitted Figure 3C not to overload it.

Author response image 2.

Of note, we have previously shown that Col protein is detected in all DA3 nuclei in stage 14 wt embryos. What we observed, however, was a mixture of ON/OFF nuclei for transcription of identity transcription factors and realization genes, suggesting that col auto-regulation and activation of downstream genes is Col threshold-dependent (Dubois et al., 2007; Bataillé et al., 2017).In the Figure 3C, we use L-CRM-moeGFP expression to visualize the Col-expressing progenitors and DA3 muscle contours. L-CRM-moeGFP activity is depending on the Twi and Mef2 binding sites, from the progenitor stage (see Figure 1—figure supplement 3 and Dubois et al., 2007), explaining why more GFP accumulates in colthan col mutants.

4) Figure 3D – The quantification of transcription at stage 14 is missing. We are puzzled by the question of why late CRM is not active when the autoregulatory is missing. Mef2 should be still active in all nuclei, should it not?

As shown in Figure 3C and summary (Figure 3E), col transcription is no more detected in colmutants at stage 14, preventing comparative measurements.

Previous analysis of col mutants has shown that Col is required for maintenance of its own transcription in a few tissues where it is expressed, among which the DA3 muscle lineage beyond the FC stage (Crozatier et al., 1999; Dubois et al., 2007). Mef2 and Twist are also required for L-CRM activity (this manuscript, see reply to point 2). Our working model is that Mef2 and Twist early binding to the L-CRM (Sandmann et al., 2007) primes the late col CRM for Col autoregulation (subsection “(Re)programming of syncytial nuclei is required for muscle morphological identity”).

5) Figure 3D – there is no information of how the fluorescence quantification was done. Did the authors perform background subtraction? Did the authors check whether nau transcription is affected or not by col depletion? They should provide images and quantification of nau FISH as well or give clear reasons of why these could be future work if they are unable to do this now.

We have now modified subsection “Quantification of col transcription “ and provide a new supplementary figure (Figure 3—figure supplement 2) showing that nau transcription is not particularly modified in ΔCRM col mutants.

6) Figure 5 – The authors describe numerous membrane protrusions, which we do not see in the image provided, and no quantification of these is described. No information regarding homotypic DA3/DA3 and DA2/DA2 is provided. Are these real homotypic adhesions? Again, please give clear reasons for why these could be future work if they are unable to do further experiments now.

7) Figure 5 – The images are very nice, however, we do not know about repulsion or attraction. It is speculation. There could be other explanations. Please tone down and discuss various possibilities

Following the reviewer’s suggestions, we now added a supplementary figure (Figure 5—figure supplement 2), stage 15 embryo, to show the membrane protrusions issued from DA muscles and contacting each other.

We think that the accumulation of MoeGFP at the matching surface between the DA2 and DA3 muscles supports our conclusion of heterotypic adhesion. Triple staining for the tendon cell and muscle contours and α-spectrin supports the model of “indirect” muscle-muscle attachment underneath tendon cells schematized in Maartens and Brown, 2014 (Author response image 3). We postulate that stabilization of the posterior, before anterior attachment, a developmental sequence also observed during development of abdominal adult muscles (Currie and Bate, 1991) may be instrumental in the precise matching of muscle-muscle attachments.

Author response image 3.

Conversion of the heterotypic DA3/DA2 matching into homotypic DA3>DA2/DA3>DA2 adhesion in col mutants suggests that DA3>DA3 homotypic repulsion is alleviated. The alternative possibility that heterotypic adhesion is not privileged over homotypic adhesion is also plausible. Future work aims at identifying surface molecules which could mediate heterotypic adhesion and/or homotypic repulsion, using DA3, DA2 or DA3>DA2-specific RNA-seq experiments. From our previous studies one the dynamics or reprograming of naïve myoblast nuclei during syncytial elongation (Bataille et al., 2017) the time window may be relatively short, making the experiments challenging.

The text has been modified to tone down conclusions on these aspects and discuss various possibilities; subsection “Muscle staggered ends; heterotypic versus homotypic interactions?”.

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Strain, strain background (Drosophila melanogaster)	white[1118]	Bloomington Drosophila Stock Center	BDSC Cat# 3605, RRID:BDSC_3605
Strain, strain background (Drosophila melanogaster)	Df(2L)BSC429	Bloomington Drosophila Stock Center	BDSC Cat# 24933, RRID:BDSC_24933
Strain, strain background (Drosophila melanogaster)	vasa-cas9VK00027	Bloomington Drosophila Stock Center	BDSC Cat# 51324, RRID:BDSC_51324
Strain, strain background (Drosophila melanogaster)	sr-Gal4			obtained from G. Morata, Madrid, Spain
Strain, strain background (Drosophila melanogaster)	Col1	Our lab		Crozatier and Vincent, 1999
Strain, strain background (Drosophila melanogaster)	Mi{MIC}knMI15480/SM6a	Bloomington Drosophila Stock Center	BDSC Cat# 67516, RRID:BDSC_67516	MiMic Line
Strain, strain background (Drosophila melanogaster)	GMR69G03	Bloomington Drosophila Stock Center	BDSC Cat# 39500, RRID:BDSC_39500	GMR line, GMR located in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G04	Bloomington Drosophila Stock Center	#46616	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G05	Bloomington Drosophila Stock Center	#39501	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G06	Bloomington Drosophila Stock Center	#39502	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G07	Bloomington Drosophila Stock Center	BDSC Cat# 47956, RRID:BDSC_47956	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G08	Bloomington Drosophila Stock Center	BDSC Cat# 46617, RRID:BDSC_46617	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G09	Bloomington Drosophila Stock Center	BDSC Cat# 46618	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G10	Bloomington Drosophila Stock Center	BDSC Cat# 39503	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G12	Bloomington Drosophila Stock Center	BDSC Cat# 46619	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR69G03	Bloomington Drosophila Stock Center	BDSC Cat# 46620, RRID:BDSC_46620	GMR line, GMR in the Vg gene
Strain, strain background (Drosophila melanogaster)	GMR12A09	Bloomington Drosophila Stock Center	BDSC Cat# 47319	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR12G07	Bloomington Drosophila Stock Center	BDSC Cat# 47854, RRID:BDSC_47854	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR12H01	Bloomington Drosophila Stock Center	BDSC Cat# 48528, RRID:BDSC_48528	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13A11	Bloomington Drosophila Stock Center	BDSC Cat# 49248	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13B06	Bloomington Drosophila Stock Center	BDSC Cat# 48544, RRID:BDSC_48544	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13B08	Bloomington Drosophila Stock Center	BDSC Cat# 48546, RRID:BDSC_48546	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13C09	Bloomington Drosophila Stock Center	BDSC Cat# 48555, RRID:BDSC_48555	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13C11	Bloomington Drosophila Stock Center	BDSC Cat# 48556, RRID:BDSC_48556	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13F08	Bloomington Drosophila Stock Center	BDSC Cat# 48576, RRID:BDSC_48576	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR13F10	Bloomington Drosophila Stock Center	BDSC Cat# 48578, RRID:BDSC_48578	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR47D05	Bloomington Drosophila Stock Center	BDSC Cat# 47605, RRID:BDSC_47605	GMR line, GMR in the Kn gene
Strain, strain background (Drosophila melanogaster)	GMR46H09	Bloomington Drosophila Stock Center	BDSC Cat# 54712, RRID:BDSC_54712	GMR line, GMR in the Kn gene
Antibody	anti-col (Mouse monoclonal)	Our lab		1:50 Krzemień et al., 2007
Antibody	anti-LacZ (Mouse monoclonal)	Promega	Promega Cat# Z3781, RRID:AB_430877	1:1000
Antibody	anti-spectrin (Mouse monoclonal)	Hybridoma Bank	DSHB Cat# 3A9 (323 or M10-2), RRID:AB_528473	1:200
Antibody	anti-GFP (Rabbit polyclonal)	Biolabs	Torrey Pines Biolabs Cat# TP401 071519, RRID:AB_10013661	1:1000
Antibody	anti-GFP (Chicken polyclonal)	Abcam	Abcam Cat# ab13970, RRID:AB_300798	1:500
Antibody	Phalloidin-texas red	Thermofisher Scientific	Cat# Cat#T7471	1:500
Antibody	Alexa fluor antibodies 488, 555 and 647	Molecular probes		1:300
Antibody	Alexa fluor 594 anti HRP	Jackson Immunological research	Jackson ImmunoResearch Labs Cat# 123-585-021, RRID:AB_2338966	1:300
Antibody	Alexa fluor phalloidin	Thermofisher Scientific	Thermo Fisher Scientific Cat# A12381, RRID:AB_2315633	1:500
Antibody	Biotinylated goat anti-mouse	Vector Laboratories	Vector Laboratories Cat# BA-9200, RRID:AB_2336171	1:2000
Software, algorithm	ImageJ	Schneider et al., 2012	ImageJ, RRID:SCR_003070	https://imagej.nih.gov/ij/
Software, algorithm	FIMTrack	Risse et al., 2014		https://www.uni-muenster.de/PRIA/en/FIM/
Software, algorithm	MATLAB	MathWorks	MATLAB, RRID:SCR_001622	https://www.mathworks.com/products/matlab-online.html
Other	FISH probes labelled with Quasar dye 670 (col)	Biosearch Technologies and this study		Kn first intron
Other	FISH probes labelled with Quasar dye 570 (nau)	Biosearch Technologies and this study		Nau first and third introns