2.) CHARACTERISATION OF THE ABDOMINAL MUSCLE PATTERN
The somatic or bodywall muscles of the Drosophila larva have been extensively
studied and this has led to detailed knowledge of their development and
organisation (BATE, 1993; BAYLIES et al., 1998). A particularly high degree of
attention has been dedicated to the abdomen, which is divided into eight or nine
segments (either A1 to A8 and a telson, or more commonly A1 to A9). The
abdominal muscles have been characterised in great detail as to their position,
size and shape, orientation, gene expression patterns and innervation (reviewed
in BAYLIES et al., 1998). All abdominal segments from A2 to A7 show an identical
arrangement of muscles. The extensive knowledge about the formation and the
highly stereotypical organisation of the muscle pattern makes the abdominal
segments of Drosophila an attractive model system to study development, function
and morphology of neuromuscular networks.
Studies of this network from the neuronal angle have allowed the identification
of the motorneurons controlling each muscle in the abdomen from A2 to A7
(LANDGRAF et al., 1997; SINK & WHITINGTON, 1991). Subsequently, these neurons
have been characterised in terms of their origin and development, location, gene
expression patterns and dendritic arborisation (reviewed in LANDGRAF & THOR,
2006). The development of the neuromuscular networks in the abdomen was covered
in Chapter 1; in the following, I shall therefore focus on the organisation of
the muscle pattern as it appears in the larva.
The arrangement of the bodywall muscles in the abdominal segments is highly
stereotypical in segments A2 to A7. There are 30 syncitial muscles per
hemisegment, each of which is a contractile unit and is the equivalent of a
primary or secondary myotube in a vertebrate muscle. Since this muscle pattern
is segmentally repeated, it consists of approximately 600 individual muscles in
the entire larva (BAYLIES et al., 1998).
The muscles are divided into external (many of them are transverse) and internal
(many of them are longitudinal) ones. Based on their position and orientation,
they are classified and named as being dorsal, lateral or ventral (D, L or V);
and oblique, acute, longitudinal or transverse (O, A, L or T). The modern
nomenclature is therefore derived from the features of the muscles, for example,
muscle DO4 is the fourth dorsal oblique muscle starting at the dorsal midline.
The only exception is the segment border muscle (SBM), which runs transversely
from the dorsal to the ventral domain (BATE, 1993).
Segment A1 has a muscle pattern with very few alterations from the described one
of A2 to A7. These alterations include the absence of muscles VT1 and VO6; small
changes in size and orientation of muscles such as DO3; and an additional muscle
VI1 (Ventral Intersegmental). This VI1 appears to be fused with a ventral,
longitudinal muscle that spans the thoracic segments (BATE, 1993; see fig. 2.5
and fig. 2.6).
Since the muscle pattern of A2 to A7 is stereotypical and has served as the
basis for most neuromuscular studies in the embryonic or larval abdomen of
Drosophila, I will later refer to it as the “reference segment”. In these
instances I will neglect the alterations in segment A1, and consider the pattern
from A2 to A7. In these cases, I will also neglect any thoughts about an
ancestral ground plan pattern and whether or not A2-7 is closer to it than A8.
For practical reasons I will treat the A2-7 pattern simply as a
well-characterised functional network that will serve as a reference to study
neuromuscular diversification. The model system for an altered state will be
segments A8 and A9, which are two distinct metameres. However, both are
drastically altered to serve specific functions such as respiration and
defaecation. These segments also terminate the bodywall, which is likely to lead
to additional changes in the neuromuscular morphology. Therefore, I will refer
to them jointly as A8/9 when I discuss features that both segments share.
Since the muscles are the most accessible elements in the neuromuscular network,
I started this project by analysing their pattern in detail in the reference
segment (see figures 2.1 to 2.4) as well as in A8/9 (see figures 2.7 to 2.11).
Previously, such descriptions have led to an extensive map of the complete
muscle pattern (BATE, 1993; CROSSLEY, 1978; CAMPOS-ORTEGA & HARTENSTEIN, 1997).
However, most studies focussed on early embryonic stages; studies investigating
the bodywall muscles in late embryonic or larval stages have relied on
dissections to allow antibodies or curious glances to penetrate the hardened
cuticle. Such invasive practices could only be avoided in anterior segments;
there the gut and fatbody do not interfere with polarised light or Nomarski
optics and the almost crystal-like arrangement of the contractile proteins in
the muscles allows a direct analysis even in living animals.
In the most posterior segments A8/9, dissections are particularly difficult to
perform without disrupting muscles. The presence of the fatbody and the hindgut
have made the analysis in undissected animals with polarised light or Nomarski
optics difficult. To get around such obstacles, I used a fly strain that
expresses GFP, which is inserted into the muscle protein titin (MORIN et al.,
2001; for details, see Material and Methods). If larvae of this strain are
examined under UV light, each individual muscle can be readily observed at very
high resolution and three dimensions even in living animals throughout embryonic
and larval life. This allowed me to map out and directly compare the muscle
patterns of the reference segment and A8/9 in greater detail than before.
2.2 A Map of the Larval Muscle Pattern from A1 to A9
The titin-associated expression of GFP allowed me to map the muscles in the
reference segment and A8/9 and to document their size, position, orientation and
shape. I was able to do this at high resolution and with respect to other
muscles and landmarks such as denticle bands or the posterior spiracles. I could
do this throughout larval development. Comparing the different larval stages
from first to third instar reveals that the muscle pattern as such does not
fundamentally change. Some muscles split and fan out in late second and third
instar larvae (e.g. LT1 in A8), but remain clearly recognisable. The overall
pattern itself remains virtually unchanged throughout larval development. The
figures that I use to show the individual muscles are taken from an early
(feeding) third instar larva. They serve as a basis for a slightly altered
version of the previous muscle map (BATE, 1993). This new map is given in figure
2.13. In some few instances splitting muscles appeared to be two, in which case
I used first instar larvae to verify that the muscle in question was indeed
derived from a single myotube.
At this point I would like to emphasise that the modern nomenclature is solely
based on the position and orientation of the muscles (BATE, 1993). Identical
names describe equivalent muscles only in segments A1 to A7, where there is a
stereotypical pattern, but not between muscles of the reference segment and
A8/9. For example, DT1 in A6 is the equivalent of DT1 in A5. However, the
nomenclature is not meant to imply that it is an equivalent of DT1 in A8, which
is named simply according to its position and orientation. This will be
important to keep in mind not only for the muscle map, but even more so for the
issue of homologies in later chapters.
Unsurprisingly, I did not find any significant differences between the muscle
pattern of the reference segment observed in vivo and the one given in the
literature. In A8/9, however, the direct analysis helped me to put the muscles
into relationship with muscles in A7 and cuticular features such as the
posterior spiracles, the anal pad, a fold in the cuticle ventral to the
posterior spiracles or the anal slit. This allowed me to show the overall
pattern in its three dimensions more accurately. I also mapped two previously
undescribed muscles: a very thin, but long muscle that runs in a slightly
oblique orientation dorsal of DO1, almost aligned with the dorsal midline. I
will refer to this muscle as MO1 for Midline Oblique. The second muscle is
shorter, but thicker and runs longitudinally in the lateral domain, slightly
dorsal of VL1. According to its position and orientation, I will refer to this
muscle as LL1 for Lateral Longitudinal.
I could also show that muscle DO1 has a dominant position among the muscles that
reach out towards the posterior spiracle, which include MO1 and VL1. Muscle TO2
is oriented in a way that it reaches towards the posterior spiracle from the
ventral side. This is difficult to see from the diagram in figure 2.13, but
clearly visible in the image in figure 2.9. Altogether, there are four muscles
per hemisegment that reach out towards the posterior spiracle to an equal
extent: DO1 in the dorsal, VL1 in the lateral and TO2 in the ventral domain.
The anal pad runs ventrally along the segment border between A7 and A8 along
muscles VT1 and VT2, then bends to the posterior along VL1 and VL2 and finally
along the segment border between A8 and A9, by the posterior insertion sites of
VL2, VL3 and GS2 to the anal slit (see the dotted line in figure 2.13). This
allowed me to designate muscles to A8 or A9, which I later tried to confirm by
showing their innervation.
There are 19 muscles in A8: MO1, DO1 and DT1 in the dorsal domain; LT1, LL1 in
the lateral domain; VL1 to VL4, VA1, VO1 to VO3, GS1 to GS3, VT1 and VT2 in the
ventral domain. There are only four muscles in A9: TO1, TO2, GS4 and TT1. A
classification based on the predicted function of the muscles can combine the
muscles GS1 to GS4 in one group, as they are arranged radially around the anal
slit. The muscle pattern is summarised in the map of figure 2.13. I also used
the vector-based animation program Maya 7.0 (Alias Software) to generate a
3-dimensional model of one hemisegment for A8/9 to show the orientation of the
muscles with respect to each other more comprehensibly (see figure 2.12).
2.3 Innervation and Classification of the Muscles in A8/9
As I have mentioned above, the muscles of the reference segment can be
classified as being longitudinal (including acute and oblique ones) and
transverse (including the SBM) muscles. More commonly, they are divided into
external and internal muscles (see the colour code of the diagram in figure
2.13). This rough division matches with the innervation pattern: External
muscles receive their innervation from the segmental nerve (SN), whereas
internal muscles are innervated through the intersegmental nerve (ISN). The only
two exceptions to the rule are muscle DT1 (external, ISN) and VT1 (external,
TN). The ISN is subdivided into branches ISN (most distally), ISNb and ISNd. The
segmental nerve into SNa and SNc (reviewed in LANDGRAF & THOR, 2006; see figure
2.14; for specimen, see figure 2.16).
Since the innervation therefore appeared to be a mean to determine subclasses of
bodywall muscles, I decided to investigate the division of the nerves reaching
into A8 and A9. To do this, I performed an antibody labelling against the
cell-adhesion protein Fasciclin II (FasII) in “flat preparation” and
“wholemount” embryos to selectively visualise the axons of motorneurons (see
also “Material and Methods”).
The nerves for A8 and A9 exit the ventral nerve cord jointly in embryos at 15
hrs AEL. They are held together by a thick layer of cells that is most likely a
sheath of glia. At this stage, this sheath continues towards the posterior
approximately until the segment border A7/A8. Here the A8 nerve branches off,
whereas the very thick bundle of remaining axons continues to grow to the
posterior. Approaching the hindgut, a small subset of axons branches off again
and grows into the muscle field of A9. Therefore, I will refer to this nerve as
A9. Several fasciclinII expressing axons that leave the CNS together with nerves
A8 and A9 will grow along the hindgut. I will refer to this nerve as HGN
(hindgut nerve; see figure 2.16).
To investigate the development of the sheath that combines these three nerves, I
made flat preparations of embryos from approximately stage 12 until the
described specimens of stage 17. This series revealed that nerve A8
differentiates individually and is originally not combined with nerve A9 and the
HGN. As the ventral nerve cord retracts and increasingly moves to the anterior,
however, nerve A8 gradually gets closer to A9 and the HGN. Eventually some cells
begin to form a sheath that ultimately unites the three nerves at 15 hrs AEL
(see figure 2.15). Beyond that, there is also a transverse nerve present between
A7 and A8; however, it appears to be the most posterior TN, as I could not find
a TN between A8 and A9 (see also Chapter 4).
Both nerves, A8 and A9, are subdivided into three branches; a distal, an
intermediate and a proximal one. In that, they resemble the division of nerves
in the reference segment. Therefore, I will refer to the individual branches as
ISN (most distal), ISNb (intermediate) and SN (most proximal). This has
interesting implications for the determination of the dorsal musclefield in A9:
ISN in A9 innervates muscle GS4, which appears to be perfectly ventral. To
investigate this contradiction further, I studied the expression of Hb9
(ventrally projecting motorneurons; BROIHIER & SKEATH, 2002) and evenskipped
(dorsally projecting motorneurons, LANDGRAF et al., 1999) in the nerves of A8
and A9. For this, I did flat preparations of embryos from Hb9-gal4 (see figure
3.7b, Chapter 3), RN2-gal4 (evenskipped, specific for two motorneurons per
hemisegment; see figure 3.8, Chapter 3) and CQ-gal4 (evenskipped, U-neuron
specific; see figure 3.7a, Chapter 3) driving mCD8GFP. In A8, the ventral and
lateral muscles including LT1 and DT1 receive innervation through Hb9-expressing
neurons. DO1 is innervated by a neuron of the CQ pattern (see fig. 3.7a, Chapter
3). In A9, the Hb9 neurons project to TO1, TT1 and GS1-3. Both CQ and RN2
neurons project much more distally to muscle GS4. This suggested that at least
muscles GS4 and TO2 are phylogenetically the most dorsal muscles of the segment,
although functionally and morphologically they occupy ventral sites. I later
found evidence through the expression of muscle markers that this is actually
the case (see Chapter 4).
It is worth noting that some muscles that I assigned to A8 (based on the
position of the denticle band in A9, see figure 2.9; the posterior margin of the
anal pad; and the posterior insertion sites of the VL muscles in A8) receive
innervation through nerve A9. These muscles are GS2 and GS3 and possibly GS1,
VO2 and VO3 (see also 2.4 Discussion).
The mapping and classification of the muscle pattern in A8/9 reveals some very
significant differences from the pattern in the reference segment. These
differences presumably reflect altered functions of the most posterior segments,
which are likely to include respiration and the control of the posterior
spiracles; the termination of the bodywall; the presence of the anal pad and
anal slit for defaecation; and the onset of peristalsis that starts with a
possibly simultaneous contraction of muscles in A8/9, which then continues as a
wave of contractions in more anterior segments (Sarah Crisp, personal
The differences in the muscle pattern are particularly obvious in A8, which is a
more “complete” segment than the strongly reduced A9. In A8 one can see that
alterations occur most strikingly in the dorsal and lateral domains: There are
three dorsal muscles (MO1, DO1 and DT1) and only two lateral ones (LT1 and LL1).
These five muscles span an area that contains approximately 15 muscles in the
reference segment. The impressive size of some of these muscles (in particular
of DO1, DT1 and LT1) might compensate for this reduction at a functional level.
This reduction of dorsal muscles also allows speculations about the role of
these areas in locomotion: The longitudinal muscles are likely to act as the
“powerhouses” of peristalsis, since they are all aligned in an
anterior-posterior fashion all along the body and probably contract
simultaneously during peristalsis (Sarah Crisp, personal communication). It
could be that this role of dorsal longitudinal muscles in A8 has been sacrificed
in favour of controlling the motion of the posterior spiracles. In this context,
it is worth noting that the ventral longitudinal muscles in A8 are virtually
identical to those in the reference segment with only one small exception: VL1
in A8 is slightly longer than the other VLs and reaches out towards the
posterior spiracles. VL1 thereby extends to the same posterior level as MO1
(which is tiny compared to any other muscle in this segment and might well be
associated with the dorsal vessel) and more importantly DO1. Together with TO2
in A9, which reaches out to the posterior spiracles from the ventral side up to
the cuticular fold ventrally of the spiracle, these muscles are arranged in a
way that they could retract or orient the spiracles.
Previous experiments in which I had fed second instar larvae coloured yeast had
shown that defaecation does not occur continuously, but once every few minutes
in one quick motion. Since the four GS muscles run radially around the anal
slit, they are likely to be involved in defaecation. I tested this by directly
investigating the anal region of a larva from the GS203 line under UV light. The
observed activity suggests that the anal slit is passively closed and that
defaecation probably involves the contraction of all GS muscles, VO2 and VO3. I
could film defaecation in a freely moving larva, but the narrow focal plane did
not allow conclusions about the role of individual muscles in the control of
Another interesting issue is the conflict between the position of muscles and
their innervation that I have mentioned before: Muscles in A8 are innervated
through nerve A9 (for example, GS2 and GS3), ventral muscles are innervated by
dorsally projecting motorneurons (for example, TO2 and GS4). If these muscles
form a functional unit controlling behaviours such as the opening of the anal
slit during defaecation, they might originate from A9 and could be anchored in
the domain of A8 only to accommodate a very specific purpose.
Considering the anal slit as marking the phylogenetically posterior end of the
body, one can think that it has moved ventrally and then anteriorly in the
course of evolution. Thereby, it would have forced the ventral muscles of A9 to
the side and to bend around the anal slit in a similar fashion as the A9 nerve
actually does. This speculation would be supported if muscles like GS2 and GS3
were found to originate from A9 and later migrate to A8. Further experiments
could trace the development of individual muscles through the expression of
specific markers, similar to the “dorsal” expression profiles I could find in
TO2 and GS4 (see Chapter 4). This model could also explain the apparent lack of
ventral muscles in A9 (since I assigned muscles like GS2 and GS3 to A8, where
their insertion sites are).
The detailed map of the muscle pattern in A8/9, its innervation and the direct
comparison to the muscle pattern in the reference segment provides me with two
different, easily accessible variations of neuromuscular networks that share a
common basis. Moving on from here, I then performed experiments that allowed me
to identify the motorneuron partners that control individual muscles and come
back to my original questions: What are the mechanisms that drive
diversification and what are the cues that tell individual cells in the network
how to behave?
The muscle map is the first step towards a thorough identification of segmental
differences. In theory, one could expect three kinds of alterations to occur in
a neuromuscular network compared to one in a different metamere: The loss of
elements; the conservation of elements; or the differentiation of novel
Here in this map one can already find beautifully clear examples for all of
these three possible alterations (summarised in fig. 2.17): Muscle DA1 is in a
prominent position in the reference segment, but it has no obvious homologue in
A8. Therefore, it appears to be missing in this segment. Muscle LT1 in A8 is
bigger than any lateral, transverse muscle in the reference segment; its shape
and orientation is distinct. Even if it had an equivalent muscle in A7 (I will
show in Chapter 4 that it does, namely muscle VT1), it has gained a set of new
features that make this element a novel one with respect to the reference
segment. The VL muscles in A8, in particular VL2 to 4, do not show features that
make them particularly distinct from their equivalents in the reference segment.
They are aligned with the other VLs, most likely a driving force in forward
locomotion. Thus, the VL muscles are an example where elements of the
neuromuscular network have been conserved.
In the following chapters I will demonstrate equivalent examples for these
principles in the central nervous system and explore some of the mechanisms that
underlie this segment-specific diversification.
Figure 2.1 to 2.4: The muscle pattern of the reference segment
The pattern of the somatic muscles is stereotypically repeated from segments A2
to A7 with small variations in A1. Using early (feeding) third instar larvae of
a line that expresses GFP associated with a muscle protein, one can visualise
individual muscles under UV light and analyse them in great detail (n >> 10).
In all figures, a red label refers to external muscles, yellow to more internal
ones and blue to muscles even more internal. Note that some muscles feather out
in 3rd instar larvae – in such cases, the unity of a muscle was confirmed in
Figure 2.1: Dorsal view of the muscle pattern. The black arrow indicates
the dorsal midline.
Figure 2.2: Dorso-lateral view of the muscle pattern.
Figure 2.3: Ventro-lateral view of the muscle pattern. Note the
Figure 2.4: Ventral view of the muscle pattern.
Figure 2.5 and 2.6: Features specific for A1
The abdominal segment A1 is slightly different from A2-A7. The differences are
shown in these two figures.
Figure 2.5: Certain muscles, such as DO3, are longer than in A2-A7.
Figure 2.6: A1 has an additional muscle, VI1. There is no muscle VT1
in A1, nor a muscle VO6. VT1 in A2 is labelled.
Figure 2.7 to 2.11: The muscle pattern in A8 and A9
The figures to show the muscle pattern in A8 and A9 are oriented from dorsal to
Figure 2.7: Dorsal view of the muscle pattern. The black arrow indicates
the dorsal midline.
Figure 2.8: Lateral view of the muscle pattern. Note the denticle
bands as landmarks (white arrows, A8 and A9).
The red arrow in the labelled figure points to what appears to be a branch of a
fanned muscle, most likely VL3 or VL4. The muscle of origin is not entirely
clear in this specimen.
Figure 2.9: Ventro-lateral view. Note the denticle bands as landmarks
(green arrow: A8; purple arrow: A9).
Figure 2.10: Ventral, anterior view of the muscle pattern. Note the
anal slit that is highlighted through GFP expression in the visceral mesoderm
Figure 2.11: Ventral, posterior view of the muscle pattern. Note how far
TO2 reaches out to the posterior; this is not clear from the diagram or from
dissected specimens, but a striking feature in vivo. Z-projection of several
exposures with different focal planes.
Figure 2.12: 3D models of the muscle pattern to represent the arrangement of
individual muscles more clearly.
(A) Inside out (with posterior pointing to the left) view; (B) dorsal, (C)
lateral and (D) ventral view.
Figure 2.12: 3D models of the muscle pattern to represent the arrangement
of individual muscles more clearly.
(A) Inside out (with posterior pointing to the left) view; (B) dorsal, (C)
lateral and (D) ventral view.
Figure 2.13: Map of the muscle pattern in A7, A8 and A9
This map gives the muscle pattern of one hemisegment of A8/9 and the reference
segment (A7) with the name for each individual muscle. Nomenclature with minor
changes after (BATE, 1993).
It refers to the position (dorsal, lateral or ventral) and orientation (oblique,
transverse, acute or longitudinal) of individual muscles. There are four gut
suspension muscles (GS) and three muscles with a terminal position (TT1, TO1 and
External muscles are red, more internal muscles are yellow and blue. Dorsal is
up, ventral midline is down. The dashed line highlights the approximate position
of the anal pad. Posterior spiracles are marked with PS.
Figure 2.14: Innervation of the nerves in A7, A8 and A9
Diagrammatic outline of the nerves present in A8/9 and the reference segment. TN
is the transverse nerve, SN the segmental nerve, ISN the intersegmental nerve,
HGN the hindgut nerve. Nomenclature as reviewed in (LANDGRAF & THOR, 2006).
Figure 2.15: The “merging” of A8, A9 and the HGN
Top: At approximately 14 hrs AEL in an RN2 background (as a marker), the
three nerves of A8, A9 and the HGN are still distinct. The colour of the arrows
indicating individual nerves matches with the label of the peripheral muscle
field. Segments in the CNS are labelled separately, based on the position of the
Bottom: Approximately one hour later, A8, A9 and the HGN exit the VNC
jointly, connected through what is most likely a glial sheath. The individual
nerves branch off as they approach their target segment (n for these
observations > 8).
Figure 2.16: FasII staining reveals the subdivision of nerves
Staining for the cell-adhesion molecule Fasciclin II reveals the subdivision of
the nerves A7, A8, A9 and the hindgut nerve (HGN; n > 8).
Figure 2.17: Diagram illustrating principles of diversification
The three types of segmental variation with examples in the muscle field:
(1) Red: Loss (example: DA1 in A7, absent in A8)
(2) Blue: Drastic alteration and appearance of novel features (example: LT1 in
A8, equivalent to VT1 in A7)
(3) Yellow: Hardly any alteration (VL1-4 in A7 and A8)
return to "thesis