Matrix metalloproteinase 1 modulates invasive behavior of tracheal branches during ingression into Drosophila flight muscles

Tubular networks like the vasculature extend branches throughout the bodies of animals, but how developing vessels interact with and invade tissues is not well understood. We investigated the underlying mechanisms using the developing tracheal tube network of Drosophila indirect flight muscles (IFMs) as a model. Live imaging revealed that tracheal sprouts invade IFMs directionally with growth-cone-like structures at branch tips. Ramification inside IFMs proceeds until tracheal branches fill the myotube. However, individual tracheal cells occupy largely separate territories, possibly mediated by cell-cell repulsion. Matrix metalloproteinase 1 (MMP1) is required in tracheal cells for normal invasion speed and for the dynamic organization of growth-cone-like branch tips. MMP1 remodels the Collagen IV-containing matrix around branch tips and promotes degradation of Branchless FGF in cultured cells. Thus, tracheal-derived MMP1 may play dual roles in sustaining branch invasion by modulating ECM properties as well as by shaping the distribution of the FGF chemoattractant.


42
Indirect flight muscles (IFMs) of flying insects display the highest known metabolic rates in 43 the animal kingdom (Weis-Fogh, 1964). In Drosophila, two sets of IFMs, the dorsal-44 longitudinal muscles (DLMs) and the perpendicularly oriented dorso-ventral muscles (DVMs) 45 are anchored to the thoracic cuticle and move the wings indirectly by deforming the thoracic 46 exoskeleton rather than by acting directly on the wings. Each adult IFM is approximately 1 47 mm long and 100 µm wide (Spletter et al., 2018) and contains about 1000 nuclei (Rai & 48 Nongthomba, 2013). To supply these large muscles with sufficient oxygen, an extensive 49 network of gas-filled tracheal tubes not only superficially enwraps the IFMs, but also invades 50 the myotube interior. This remarkable physiological adaptation minimizes the distance for 51 oxygen diffusion from tracheoles to muscle mitochondria (Weis-Fogh, 1964; Wigglesworth & 52 Lee, 1982) and provides efficient gas exchange for aerobic respiration to sustain flight over 53 long time periods (Götz, 1987).

62
Tracheal invasion into IFMs depends on the attraction of tracheal branches by Bnl FGF 63 secreted on the muscle surface, followed by a switch to release of FGF from the interior 64 transverse (T)-tubule system (Peterson & Krasnow, 2015). The T-tubule system is a network during larval growth (Glasheen et al., 2009) (Levi et al., 1996).

86
MMP2 was shown to restrict FGF signaling through a lateral inhibition mechanism that 87 maintains highest levels of FGF signaling in tracheal tip cells (Wang et al., 2010). Moreover,

101
Tracheae invade flight muscles in a non-stereotyped, but coordinated manner

102
To understand the mode of IFM tracheation, we first analyzed tracheal branch pathways on 103 and within IFMs. We focused our analysis on DLMs, which receive their tracheal supply from 104 thoracic air sacs (Fig. 1A). Stochastic multicolor labeling of tracheal cells (Nern et al., 2015) 105 revealed that multicellular air sacs converge into unicellular tubes (

117
1B). Interestingly, however, the number of tracheal branches was relatively uniform along 118 DLM myotubes ( Supplementary Fig. 1A, n=6). These findings suggest that branches 119 originating from tracheal terminal cells uniformly fill the available myotube volume in a 120 manner that is non-stereotyped, but tightly coordinated with myotube morphology.  Table 1). However, 127 certain features were more uniform among cells. At least 95% of the branches of a given cell 128 were aligned with the myotube axis ( Fig. 1F; n=31) and the direction of branches was often 129 biased towards one end of the myotube (Fig. 1G, Supplementary Fig. 1B). Stochastic 130 multicolor labeling revealed that individual tracheal terminal cells occupy largely non-131 overlapping territories (Fig. 1B'). Interestingly, at the borders of such territories, branches 132 from different cells were occasionally in close proximity or in direct contact (Fig. 1H,H').

133
These findings suggest that invading tracheal cells fill the available space within the 134 myotube, but minimize overlaps, possibly mediated by contact-dependent repulsion between 135 tracheal cells.

144
prior to entry of tracheal branches into myotubes, motor neurons have already innervated 145 IFMs ( Fig. 2A,A''). Furthermore, the distribution of tracheal branches was largely distinct from 146 that of motor neurons (Fig. 2B

198
We confirmed the specificity of the RNAi effect using two independent dsRNAs targeting

353
The ability of tracheal cells to enter the IFM myotubes is likely to depend on permissive and

363
Classical electron microscopy studies suggested that tracheoles enter the IFMs through 364 plasma membrane invaginations that are continuous with T-tubules, and then spread through 365 the T-tubule network (Smith, 1961a(Smith, , 1961bWigglesworth & Lee, 1982). Other muscle types 366 that lack these membrane invaginations are not invaded by tracheal branches (Peterson & 367 Krasnow, 2015). Surprisingly, however, we did not find evidence that a normally organized T-  (Woolley, 1970). Thus, mitochondrial wrapping may 385 represent a common mechanism to sustain the extensive energy demands of specialized 386 motile cell types such as flight muscle or sperm.

MMP1 modulates invasive behavior of IFM tracheal cells 388
In addition to the muscle-derived factors discussed above, we show that the matrix 389 metalloprotease MMP1 is required in IFM tracheal cells for their normal invasive behavior.

506
Protein extracts were separated on 12.5% SDS polyacrylamide gels (35 μg protein per lane) 507 and electro-transferred to PVDF membranes. Bound secondary antibodies were visualized 508 using the ECL-system (Amersham). Three independent samples were analyzed.

509
Quantification of Western blot bands was performed using GelAnalyzer2010a with 510 background subtraction.             were segmented using the Surface tool of Imaris with "split touching objects" enabled.

599
Analysis of segmented mitochondria was performed using the Vantage tool of Imaris.

601
For phenotypic analyses, sample size (n) was not predetermined using statistical methods,