of 55
1
Sticks and Stones, a
conserved
cell surface ligand for the Type IIa RPTP Lar
, regulates
neural circuit wiring in
Drosophila
Namrata Bali
1
, Hyung-
Kook (Peter) Lee
1
and Kai Zinn
1
1
Division of Biology and Biological Engineering, California
Institute of Technology
Corresponding author: Kai Zinn (
zinnk@caltech.edu)
, Namrata Bali (nbali@caltech.edu
)
Abstract
Control of tyrosine phosphorylation is an
essential element of many cellular processes, including
proliferation, differentiation neurite outgrowth
, and synaptogenesis
. Receptor
-like protein-
tyrosine phosphatases
(RPTPs) have cytoplasmic phosphatase domains and
cell adhesion
molecule (CAM)
-like extracellular domains that interact with cell
-surface ligands and/or co
-
receptors.
We identified a
new ligand for the
Drosophila
Lar RPTP, the i
mmunoglobulin
superfamily CAM
Sticks and Stones (Sns). Lar is
orthologous
to the three Type IIa mammalian
RPTPs,
PTPRF
(LAR),
PTPRD
(PTPδ), and PTPRS (PTPσ).
Lar and Sns bind to each other in
embryos and
in vitro
. T
he human
Sns ortholog, Nephrin, bind
s to PTPRD and PTPRF.
Genetic
interaction studies show that Sns is essential
to Lar’s functions in several dev
elopmental
contexts in the larval and adult nervous systems.
In the larval neuromuscular system,
Lar
and
sns
transheterozygotes (
Lar/sns
transhets) have
synaptic
defects
like those seen in
Lar
mutants
and Sns knockdown animals.
Lar and Sns reporters are bo
th expressed in motor neurons and
not in muscles, so Lar and Sns likely act in
cis
(in the same neurons).
Lar
mutant
s and
Lar
/
sns
transhets
have identical axon guidance defects in the larval mushroom body in which Kenyon
cell axons fail to stop at the midline and do not branch. Pupal Kenyon cell axon guidance is
similarly affected, resulting in adult mushroom body defects. Lar is expressed in
larval and pupal
Kenyon cells, but Sns is not, so Lar
-Sns interactions in this system must be in
trans
(between
neurons). Lastly, R7 photoreceptor axons in
Lar
mutants and
Lar/sns
transhets fail to innervate
the correct M6 layer of the medulla in the optic
lobe. Lar acts cell
-autonomously in R7s, while
Sns is only in lamina and medulla neurons that arborize near the R7 target layer. Therefore, the
Lar-Sns interactions that control R7 targeting also
occur in
trans
.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
2
Introduction
Neural circuit assembly is a
complex
process that involves axon pathfinding, target selection
,
and synaptogenesis with appropriate synaptic targets
. Many cell
-surface and secreted proteins
are involved in one or
more of these stages. Cell adhesion molecule
s (CAMs) are
one class of
molecules
that play important roles in the process of circuit assembly. Many CAMs are
localized
to nascent
synapses,
where they initiate cell
-cell contact and recruit pre
- and postsynaptic
proteins to direct synapse assembly. CAMs
typically
consist of three domains: an extracellular
domain
(ECD) that interacts with other CAMs
, either homophilically or heterophilically, a
transmembrane domain
, and an intracellular domain that can interact with cytoplasmic or
membrane
-associated
proteins to regulate signaling. Disruption of CAM activities can lead to
neurodevelopmental or neurodegenerative diseases such as autism and schizophrenia
(Sudhof,
2008).
Some well
-known CAMs
involved in circuit assembly include N
-Cadherin
(Bulgakova et
al., 2012; Prakash et al., 2005)
, Dscam (Hattori et al., 2007; Hattori et al., 2008)
, Neurexins
, and
Neuroligins
(Sudhof, 2008)
.
Receptor
-like protein tyrosine phosphatases (RPTPs) are
signaling receptors
with CAM
-
like ECDs that bind heterophilically to ligands
, and cytoplasmic domains with tyrosine
phosphatase enzymatic activity
(reviewed by (Coles et al., 2015)
). Type IIa
(R2A), IIb
(R2B)
and
III (R3) RPTP subtypes have ECDs containing immunoglobulin superfamily (IgSF
) domains
and/or fibronectin type III (FNIII) repeats.
Drosophila
has six RPTPs,
four of
which are
expressed primarily in
the nervous system. Lar is a Type IIa RPTP with both IgSF and FNIII
domains that is orthologous to a 3
-member family of mammalian RPTP
s: PTPRF (Lar), PTPRD
(PTPδ, R
-PTP- δ), and PTPRS (PTPσ, R-PTP-
σ).
C. elegans
has a single Lar ortholog, PTP-
3.
Ptp69D resembles Type IIa RPTPs, having IgSF and FNIII domains, but does not correspond to
any specific mammalian RPTP. Ptp10D and Ptp99A have on
ly FNIII repeats in their ECDs.
There are also two other RPTPs
: Ptp52F
, which is primarily expressed in the gut but also has
functions in neurons, and
Ptp4E
, which is ubiquitously expressed. Ptp10D and Ptp4E are
orthologous to mammalian Type III RPTPs, which are regulators of receptor tyrosine kinases
(reviewed by (Jeon and Zinn, 2015)
). There are no Type IIb RPTPs in
Drosophila
.
Drosophila Lar and Ptp69D have well
-characterized functions during neural
development. Motor axon guidance in embryos is altered in mutants for both RPTPs
. They
genetically
interact
with each other and with the o
ther four RPTPs. In
double, triple, and
quadruple mutants there are specific alterations in
motor and
central nervous system (CNS)
axon guidance
(Jeon et al., 2008; Sun et al., 2000).
Lar
mutations affect the structures of
neuromuscular junctions (NMJs) in the larval neuromuscular system (Johnson et al., 2006;
Kaufmann et a
l., 2002). Lar is also required for development of the larval mushroom body
(MB)
lobes (Kurusu and Zinn, 2008)
and for R7 photoreceptor axon targeting in the optic lobe
(OL)
(Clandinin et al., 2001; Maurel
-Zaffran et al., 20
01). Ptp69D works together with Lar to
control
R7 axon guidance and targeting
(Hakeda-
Suzuki et al., 2017; Hofmeyer and Treisman, 2009)
.
Mammalian Type IIa RPTPs function during synaptogenesis, and knockouts of the three
genes have developmental defects causing craniofacial malformations
(Stewart et al., 2013),
severe muscle dysgenesis and loss of motor neurons in the spinal cord
(Uetani et al., 2006).
Ptprs
and
Ptprd
mutant mice exhibit increased paired pulse facilitation, enhanced or reduced
long -term potentiation, respecti
vely, and distinct behavioral alterations
(Horn et al., 2012; Uetani
et al., 2000).
Ptprs
and
Ptprd
mutant mice also exhibit early growth retardation and increased
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
3
neonatal mortality (Uetani et al., 2000)
. The chicken orthologs of PTPRD
and PTPRS
are
expressed in developing motor neurons,
and knockdown of each of these PTPs affected the
dorsal nerve, causing either abnormal fasciculation or a reduced or missing
nerve (Stepanek et
al., 2005).
The CAM
-like structure of the Lar ECD implies that its functions are regulated by
interactions with cell-
surface or extracellular ligands
. We and others identified the heparan
sulfate proteoglycans (HSPGs) Syndecan (Sdc) and Dally
-like (Dlp) as Lar ligands
(Fox and
Zinn, 2005; Johnson et al., 2006)
. Lar interacts
directly
with heparan sulfate, as do mammalian
Type IIa RPTPs
(Aricescu et al., 2002). Sdc and Dlp have genetic interactions with Lar (Fox and
Zinn, 2005; Johnson et al., 2006)
. They are involved in Lar’s regulation of embryonic axon
guidance and larval NMJ development, but
Sdc
and
Dlp
phenotypes are much weaker than
Lar
phenotypes, implying that other ligands are also involved. The HSPGs are not involved in
regulation of
Lar’s functions in
R7 photoreceptor axon guidance
(Hofmeyer and Treisman,
2009).
A number of protein
ligands for mammalian Type IIa RPTPs have been identified. These
include Netrin
-G ligand 3 (NGL
-3) (Kwon et al., 2010; Woo
et al., 2009), Tropom
yosin kinase C
(TrkC) (Takahashi et al., 2011)
, Interleukin-
1 receptor accessory protein
-like 1 (IL1RAPL1)
(Yoshida et al., 2011)
, Interleukin-
1 receptor accessory protein (IL
-1RAcP)
(Yoshida et al.,
2012), Slit
- and Trk-
like family protein (Slitrk) 1-
Slitrk6 (Takahashi et al., 2012; Yim et al., 2013),
synaptic adhesion
-like molecule (SALM) 3
, and SALM5
(Choi et al., 2016; Li et al., 2015;
Wojtowicz et al., 2020)
. Each of these ligands localizes to postsynaptic membrane
s, where they
form heterophilic
trans
complexes with presynaptic RPTPs to regulate cell
-cell adhesion,
presynaptic differentiation and excitatory synapse development
(Takahashi and Craig, 2013)
.
Here we describe the identification of Sticks and Stones (Sns) as a new Lar ligand
, and
show that the Lar
-Sns interaction is conserved between flies and mammals. Sns is a single-
pass transmembrane protein with a large
ECD with
IgSF and FNIII domains.
S ns has orthologs
in
C. elegans
(SYG -2) and mammals
(Nephrin)
. It belongs to a subfamily of IgSF proteins called
Irre cell recognition module (IRM) proteins, which has four
members in
Drosophila
: Sns, Kirre,
Roughest (Rst), and Hibris (Hbs)
(Fischbach et al., 2009). Kirre and Rst are paralogs that
bind
to Sns and Hbs, as well as to each other
(Bour et al., 2000; Galletta et al., 2004; Ozkan et al.,
2013; Shelton et al., 2009). The
C. elegans
orthologs of Sns and Kirre, SYG
-2 and SYG
-1, also
bind to each other (Ozkan et al., 2014). In humans and mice, there is one Sns/Hbs ortholog,
Nephrin, and three Kirre/Rst orthologs (Kirrels or Nephs).
The Sns ECD contains nine Ig domains and a single FNIII
repeat, while the Kirre and
Rst ECDs contain five Ig domains
(Ozkan et al., 2014). All four IRM proteins function together
as ligand-
receptor pairs on the surface of founder cells and fusion competent myoblast
s to
regulate
myoblast fusion
(Bour et al., 2000; Shelton et al., 2009). The four proteins also function
together in
nephrocyte development (Zhuang et al., 2009)
and in ommatidium patterning in the
retina (Bao et al., 2010)
.
C. elegans
SYG-2 and SYG
-1 regulate the formation of synapses by
HSNL neuron onto vulval muscles
(Ozkan et al., 2014; Shen, 2004). SYG
-1 acts
presynaptically, while SYG
-2 acts in the guidepost epithelial cells to direct presynaptic
component assembly at the site of their interaction
(Shen, 2004). The mammalian orthologs of
Sns an
d Kirre, Nephrin and Neph1/Kirrel1, are required for the formation and functioning of the
kidney slit diaphragm
(Liu et al., 2003; Ruotsalainen et al., 1999)
.
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
4
In this paper, we show that interactions with Sns control Lar’s functions in MB lobe
development
and R7 photore
ceptor axon targeting.
Transheterozygous a
nimals lacking one
copy of the wild
-type
Lar
and
sns
genes have MB and R7 phenotypes identical to those of
Lar
homozygotes. The relevant Lar-
Sns interactions appear to be in
trans
, because Lar and Sns are
expressed in different neurons. Sns also regulates Lar’s functions in NMJ development, but in
that system the two proteins are expressed in the same neurons and likely interact in
cis
.
Results
Identification of Sns as a Lar binding partner
Cell -surface protein (CSP) interactions mediated by
ECDs are often of low affinity, having K
d
s in
the micromolar range and fast dissociation rates. This usually precludes identification of such
interactions through methods such as affinity purification/mas
s spectrometry (AP/MS), since
they do not form stable complexes.
Low- affinity
CSP- CSP interactions create stable adhesive
interactions between cells through avidity effects, since there are many copies of each protein at
cell interfaces. Successful
in vitro
detection of low
-affinity CSP interactions often requires taking
advantage of avidity through clustering. Clustering is achieved by incorporating multimerization
domains into soluble ECD fusion proteins
, which can be used for ELISA
-based assays and cell
staining. The RPTPs were included in a
n ELISA-
based
global extracellular interactome (ECIA)
screen of all
Drosophila
cell -surface proteins containing IgSF and/or
FNIII domains
(Ozkan et
al., 2013). However, interactions above background were not detected for any RPTP in this
screen, which was conducted using unpurified cell supernatants.
We developed the
live -dissected
embryo staining screen as a way to identify low
-affinity
binding partners
among neural CSPs in an assay in whi
ch they are expressed in a normal
cellular context. This screen may be more sensitive than the ECIA because it takes maximum
advantage of avidity effects. Multimeric ECD fusion proteins are added to live
-dissected
embryos and incubated for ~2 hours at room temperature
, during which time complexes of
fusion proteins with overexpressed CSPs can coalesce (“cap”) into dense patches. The
embryos are then washed directly with paraformaldehyde, which crosslinks these patches and
freezes complexes into place. The complexes are
then visualized with fluorescent secondary
antibody
(Fox and Zinn, 2005; Lee et al., 2009)
.
We first used this
method
to identify the heparan sulfate proteoglycan (HSPG)
Syndecan
(Sdc) as a ligand for Lar, using a deficiency (
Df
) screen
to find a region
of the
genome whose deletion eliminated Lar ECD staining and then narrowing down the region to a
single gene,
Sdc
(Fox and Zinn, 2005)
. We later developed a gain
-of-function (GOF) version of
the screen in which we crossed 300 lines bearing “EP
-like” (UAS
-containing)
P
elements
upstream of CSP genes to a strong pancellular driver, tubulin (tub)-
GAL4.
This collection of
lines had first been used for an
in vivo
screen in which we crossed each line to a muscle GAL4
driver and then searched for motor axon phenotypes conferred by high
-level muscle expression
(Kurusu et al., 2008). Using the GOF screen, we identified Stranded at second (Sas), a large
CSP expressed in epidermal cells, as a ligand for Ptp10D
(Lee et al., 2013).
To conduct the GOF screen with Lar ECD fusion proteins, we first needed to eliminate
Lar binding to Sdc, which is mediated by short basic sequences in the first Lar Ig domain that
bind to heparan sulfate
(HS). Accordingly, for the screen described here, we used a mutant,
HS2 , that eliminates HS binding
(Fox and Zinn, 2005). We expressed a dimeric LAR
HS2
-alkaline
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
5
phosphatase (henceforth called Lar
-AP) fusion protein at high levels using the baculovirus
system and
stained live
-dissected stage 16 embryos with unpurified supernatant, which reduces
background.
We screened crosses for all 300 EP-
like lines to tub
-GAL4 in this manner.
As shown in Figs. 1A
-B, we observed very faint CNS axon staining in wild
-type (WT)
embryos, but saw bright staining in both the CNS
axon ladder and the periphery when tub-
GAL4
was crossed to a
line that has an insertion of an EP
-like element
~200 bp 5’ to the transcription
start of
sns (s ns
EY08142
)
. This indicates that
ectopic expression of Sns
in neural, ectodermal, and
muscle cells
confers binding
to the
Lar ECD. However, it does not prove that Lar and Sns bind
directly to each other, since the results could
also be explained if Sns ectopic expression
induced expression or stabilization of another protein that actually binds to Lar.
To address this issue, we
performed
“ reverse-
binding”
experiments
to determine
whether Lar binds directly
to S ns. To do this, we
used tub-G
AL4 to drive pancellular expression
of Lar in embryos
, using a transgenic UAS-L
ar construct. To confirm that Lar was
overexpressed, we stained WT and tub>Lar stage 16 embryos with anti-
Lar antibody. In
Tub>Lar embryos, Lar was stro
ngly expressed on CNS axons as well as in the periphery, while
in WT embryos Lar is localized to longitudinal axon tracts and there is no expression in the
periphery
(Figs. 1C
-F).
We then stained WT and t
ub>Lar
embryos with a pentameric
Sns- AP fusion pro
tein,
containing the ECD of S
ns fused to
a COMP pentamerization domain and
AP (Ozkan et al.,
2013). S
ns- AP
5
stained the
CNS in WT embryos
(Fig . 1C’) . Interestingly, midline glia (arrow)
were more
strongly stained
than axons and cell bodies. This pattern does not rese
mble Lar
antibody staining (Fig. 1C),
indicating
that Sns
has another binding partner in embryos
, perhaps
Kirre or Rst
. There is also weak Sns
-AP
5
staining in the peripher
y (Fig . 1E’), where Lar is not
expressed.
Embryos with ectopic expression of Lar
driven by tub-
GAL4 show
ed a three
-fold
increase in
Sns- AP
5
staining in the CNS compared to WT control embryos
(Fig s. 1D’ , H). S ns-
AP
5
staining was increased at sites where motor
axons exit the ventral nerve cord (
VNC;
arrowhead, Figure 1D’) and in midline glia.
Staining was also increased in the periphery (Figs.
1E’ , F’). Some of this ectopic staining colocalizes with ectopic Lar, such as at the VNC exit
points. However, there is only weak
staining on CNS axons, where Lar is most prominently
expressed. This indicates that either Lar on axons is inaccessible to Sns-
AP
5
, or that there is
another protein expressed in midline glia and exit junctions that facilitates Sns-
AP
5
binding to
Lar. In any case, the reverse binding experiment provides strong evidence that Lar and Sns
directly interact with each other.
To confirm direct binding, we conducted ECIA experiments with
dimeric Lar
-Fc and
pentameric Sns-
AP
5
proteins made in
Schneider
2 (S2) cells. Sns-
AP
5
“prey” exhibited specific
binding to Lar
-Fc “bait” coupled to the surface of an ELISA plate
, showing that the two proteins
do interact directly
in vitro
(Fig. 1
P). This binding signal is weak compared to what is typically
observed for strongly interacting partners such as Sns and Kirre
, being
only 3-
fold over
background, but is nevertheless statistically significant. This may indicate that Lar
-Sns
interactions are o
f low affinity (tens or hundreds of micromolar).
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
6
Mammalian orthologs of L
ar and Sns bind to each other in embryos
and
in vitro
Lar is orthologous to the Type IIa
RPTP
subfamily, which has three members in mouse and
human: PTPRF
, PTPRD, and PTPRS
. Sns is orthologous to Nephrin. To
determine whether
binding between Lar and
Sns is evolutionarily conserved
, we tested whether Nephrin binds to
PTPRD, PTPRS or PTPRF
in live -dissected embryos. We made a transgenic line
with a UAS-
linked
full- length
human
Nephrin cDNA (NPHS1) construct
. We made
AP
5
fusion proteins
containing the ECDs of PTPRD, PTPRS and
PTPRF
in S 2 cells.
We tested the binding patterns of each of the three AP fusion
proteins in WT,
tub>NPHS1, and tub>Sns
embryos. PTPRD
-AP
5
gave the clearest signal, with strong staining in
midline glia and weak staining in the rest of the VNC (Fig. 1I).
When Nephrin was ectopically
expressed using tub-
GAL4 (tub>NPHS1), midline glial st
aining was not changed much, but
there was a 3
-fold increase in staining intensity relative to WT in the VNC as a whole (Figs. 1J,
K). These data suggest that a PTPRD binding partner is expressed in WT midline glia, or
that
midline glial membrane
s bind nonspecifically to this probe.
PTPRF-
AP
5
showed almost no
staining of WT embryos, but stained the VNC
and
midline glia
in tub>NPHS1 and tub>Sns embryos
(Figs.
1L-N ). Staining
intensity was increased
by 18-
20- fold relative to WT when Nephrin
or Sns were expressed (Fig. 1O). PTPRS-AP
5
showed little staining in
WT or tub>NPHS1 embryos.
We then tested
PTPRF and PTPRD for binding to Sns
and Nephrin
in vitro
. We
observed that Fc dimers for both human proteins bound to fly Sns
-AP
5
, and the signal
-to-noise
ratio was about the same as for Lar.
PTPRF-
AP
5
also bound to Nephrin-
Fc, and there was a
small increase over background for
Nephrin and PTPRD-
AP
5
(Fig s. 1P, Q)
. In summary, these
data indicate
that the Lar
-Sns interaction
is evolutionarily conserved
for at least two
of the three
mammalian Lar orthologs
.
Lar and Sns are co-
expressed
in larval motor
neurons
To characterize
the cells that express Lar and Sns, we created T2A
-GAL4 lines derived from
intronic
MiMIC
insertions in the two genes. We used the method described in
(Diao et al., 2015)
to insert a T2A-
GAL4 “Trojan” exon in
to
MiMIC
elements
in coding intr
ons of
Lar
and
sns
. The
expression of Lar and Sns was visualized by crossing the resulting T2A-
GAL4 lines
to a UAS-
EGFP reporter. Note that, in coding intron T2A
-GAL4 lines, expression of GAL4 requires in
-
frame readthrough from the coding region, and therefore reports on the rate of initiation of
translation from the correct ATG. Thus, these GAL4s are translational, not just transcri
ptional,
reporters. In the third instar larval VNC,
Lar-T2A -GAL4>
UAS-E
GFP (Lar>GFP)
expression was
seen in motor neurons (large paired cells)
and in a large number of interneurons
(Figs. 2A, B).
Sns- T2A -GAL4>UAS
-E GFP (Sns>GFP) expression was also seen in
motor neurons (Figure
2C), as well as in a pattern of interneurons that appeared different from those expressing the
Lar reporter (Figure 2D). We stained VNCs for Even
-skipped (Eve), which labels the aCC and
RP2 motor neurons, and found that the Sns repor
ter is expressed in RP2 (known as MNISN
-1s
in larvae
; Supp.
Fig. 1).
We characterized
Lar and Sns expression in motor neurons by examining larval NMJs
labeled with GFP reporter. The 30 body wall muscles in third instar larvae are innervated by 32
motor n
eurons in each hemisegment. There are two types of glutamatergic motor neurons, 1b
and 1s. Each muscle is usually innervated by a single 1b motor neuron. There are two 1s motor
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
7
neurons, MN
ISN -1s and MNISNb/d-
1s, in each hemisegment
. Each 1s motor neuron innervates
an entire field of muscles, with MN
ISN -1s innervating more dorsal muscles including muscles 4,
3, 2 and 1, whereas MNISNb/d
-1s innervates more ventral muscles including muscles 7, 6, 13
and 12. In addition to
1 b and 1s, two other types of motor neurons innervate a subset of
muscles
. Each
Type II motor neuron innervate
s several muscles
, whereas the Type III motor
neuron only innervates muscle 12. We compared the expression of Lar and Sns using their
respective T2A
-GAL4s in
the different types of motor neurons.
Both Lar and Sns were broadly expressed in several different types of motor neurons.
Lar>GFP was expressed in most 1b motor neurons, and in both 1s motor neurons (Figs.
2I , K).
However, Lar
>GFP was not expressed in
either Type II or Type III motor neurons (Fig
s. 2 E, E’).
Lar>GFP expression was stronger in 1s motor neurons compared to 1b motor neurons.
Specifically, at the muscle 7/6 NMJ stronger Lar
>GFP expression
can be seen in 1s boutons
compared to 1b boutons (Fig
. 2 G, G’). No expression was seen in
muscles. These data are
consistent with earlier findings on Lar expression
(Johnson et al., 2006; Kaufmann et al., 2002)
.
Sns>GFP
was also expressed in both 1b and 1s motor neurons.
Expression was
stronger in the 1b motor neurons that target more ventral muscles, including muscles 7/6, 13
and 12 (Fig.
2J). Weak Sns>GFP
expression could be seen in 1b motor neurons that target
more dorsal muscles, including muscles 4 and 3
(Figure 2L). Strong Sns>GFP
expression was
seen in both 1s motor neurons
, MNISN
-1s and MNISNb/d-
1s ( Figure 2J
). Similar to Lar>GFP,
Sns>GFP
was not exp
ressed in the Type III motor neuron. However, unlike Lar
>GFP , Sns>GFP
expression was seen in Type II motor neurons (Figs. 2
F, F’). Similar to Lar
>GFP expression
being stronger in 1s motor neurons, we observed stronger Sns
>GFP expression in 1s boutons
compared to the 1b boutons on the same NMJ (Fig
s. 2H, H ’). No Sns>GFP
expression was
seen in muscles. Thus, Sns is expressed in 1b, 1s and Type II types of motor neurons and is
co- expressed with Lar in most 1b and 1s neurons.
Lar and Sns genetically interact to
shape
morphogenesis of NMJs
Previous studies have shown that NMJs require appropriate levels of Lar for proper
development
(Johnson et al., 2006; Kaufmann et al., 2002). Reduction of Lar levels reduce
s the
number of synaptic boutons at the muscle 7/6 NMJ. We asked whether the interaction between
Lar and Sns is required for proper development of the NMJ
using genetics
. Sns is required for
myoblast fusion in developing embryos, and
sns
null mutant animals lack body wall muscles
and do not survive to the first instar larval stage.
To examine genetic interactions, we combined an
sns
null mutation with two different
Lar
null mutations to analyze NMJ phenotypes in transheterozygote (transhet) animals.
s
ns
xb3
is a
point mutation resulting in a stop codon in the ECD of the SNS protein.
Sns
xb3
mutant embryos
lack S
ns protein (Bour et al., 2000). We tested two different alleles of
Lar
with
sns
xb3
:
Lar
13.2
and
Lar
451
. Both alleles have been described as being presumed null mutations (Clandinin et al.,
2001; Krueger et al., 1996).
Lar
13.2
mutants
were shown to
display abnormalities at the muscle
7/6 NMJ and
in the larval
MB (Johnson et al., 2006;
Kurusu and Zinn, 2008)
.
Lar
451
mutants
were characterized for R7 photoreceptor axon guidance defects
(Clandinin et al., 2001). We
analyzed the NMJs at muscle 7/6 in
Lar
13.2
/
sns
xb3
and
Lar
451
/
sns
xb3
transhets
and used a semi
-
automated macro in FIJI to quantify several different parameters at the 7/6 NMJ, including total
NMJ area, total NMJ length, longest branch length, number of boutons and number of branches
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
8
(Nijhof et al., 2016). To dissect out the roles of Lar and Sns in 1b vs. 1s NMJ formation, we
performed separate analyses for the 1b and the 1s NMJ arbor at each NMJ.
Lar/+
and
sns/+
heterozygote control animals did not display any overt abnormalities at
the 7/6 NMJ (Fig.
3A, B, C). Quantification of 1b NMJ parameters showed no differences
between
WT and
Lar
13.2
/+,
Lar
451
/+, or
s
ns
xb3
/+ animals in 1b NMJ area (Fig
. 3G), number of 1b
boutons (Figure 3H), 1b NMJ length (Figure 3I), longest 1b branch length (Figure 3J) or the
number of 1b branches (Figure 3K). On the other hand, the two different
Lar
and
sns
transhets
displayed severe NMJ abnormalities, similar to
Lar
null animals (
Lar
13.2
/
Lar
451
). Quantification of
NMJ parameters showed that
Lar
13.2
/
sns
xb3
transhet NMJs had a 34% reduction in 1b NMJ area,
39% reduction in number of 1b boutons, 42% reduced total 1b NMJ
length, 33% reduced
longest 1b branch length and 45% reduced number of 1b branches (Fig
. 3).
Lar
451
/
sns
xb3
transhets showed slightly stronger phenotypes at the 7/6 NMJ with 48% reduced 1b NMJ area,
64% reduced number of 1b boutons, 57% reduced total 1b NM
J length, 48% reduced longest
1b branch length and 64% reduced number of 1b branches (Fig
. 3). Similar reductions in NMJ
parameters were observed in
Lar
mutants,
which showed 54% reduced NMJ area, 70%
reduced number of 1b boutons, 62% reduced total 1b NMJ
length, 52% reduced 1b longest
branch length,
and 72% reduced number of 1b branches. There was no
significant
difference
between the stronger
Lar
451
/
sns
xb3
transhet and
Lar
mutants
for any of the NMJ parameters
measured, indicating that Lar and Sns probably function in the same genetic pathway. We also
observed similar NMJ abnormalities on other muscles as well, including the muscle 13 and 12
NMJs (data not shown). There was no differenc
e in the size of muscles in the transhets or the
Lar
mutants. This
suggests that the Lar
-Sns interaction is not required for the role of Sns in
myoblast fusion during embryonic development.
We next asked whether the number of synapses was altered in
Lar/s ns
transhets and
Lar
mutants at the 7/6 1b NMJ. We used
antibodies against
the active
-zone protein
Bruchpilot
(Brp) to label active zones in boutons and performed quantitative analyses of Brp
-positive
punctae using the NMJ FIJI Macro. We quantified the total number of Brp punctae in 1b boutons
at the muscle 7/6 NMJ and found that
Lar
451
/
sns
xb3
transhets and
Lar
13.2
/
Lar
451
mutants had 64%
and 71% fewer Brp punctae per NMJ
, respectively,
compared to WT NMJs (Fig
. 3L). This
indicates that there is no compensatory increase in the number of synaptic active zones in
response to a reduced NMJ size and number of boutons.
During the course of NMJ
development, synaptic maturation occurs later than NMJ expansion and arborization. Our
results
suggest that Lar and Sns are required for both processes of NMJ morphogenesis: initial
NMJ expansion and arborization as well as synapti
c active zone formation.
Lar and Sns are also required for the morphogenesis of 1s motor neuron arbors at the
muscle 7/6 NMJ. 1s NMJs were severely affected in
Lar/s ns
transhets and
Lar
mutants.
Lar
13.2
/
sns
xb3
transhets
displayed 50% reduction in 1s NMJ area, 52% reduction in the number
of 1s boutons and 46% reduction in the 1s NMJ length (Supp
. Fig. 2).
Lar
451
/
sns
xb3
transhets
displayed even stronger phenotypes, with 75% reduced 1s NMJ area, 78% reduced 1s bouton
number,
and 65% reduced 1s NMJ length (Supp.
Fig. 2).
Lar
mutants
had similar reductions in
1s NMJ size (65% reduced 1s NMJ area, 82% reduced 1s number of boutons and 82% reduced
1s NMJ length)
.
We confirmed the NMJ abnormalities seen in
Lar/ sns
xb3
transhets by analyzing a
sns
deficiency
(
Df
) allele which lacks the entire
Sns
gene. We observed similar NMJ abnormalities
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
9
in
Lar
13.2
/sns
Df
animals
to those seen in
Lar
13.2
/
Sns
xb3
and
Lar
451
/
Sns
xb3
transhets (data not
shown). Overall, our data shows that in
teraction between Lar and Sns is required for 1b and 1s
NMJ morphogenesis.
Lar and Sns act in
cis
at the NMJ
While
sns
is expressed in body wall muscles during
the period of muscle fusion, its RNA levels
decrease in late embryos (Bour et al., 2000
). We did not observe any expression of either Lar or
Sns reporters
in muscles at the third instar larval stage. To
confirm that both Lar and Sns act in
motor neurons to regulate
NMJ development, we performed neuron
-specific RNAi knockdown
for both Lar and
Sns and measured the same NMJ parameters as
in the transhets and mutant
analyses. We used a pan-
neuronal driver
, elav
C155
-GAL4 (C155-
GAL4)
to drive UAS
-RNAi lines
for either
Lar
or
s ns
. We tested two different RNAi lines for both
Lar
and
sns
. Neuronal
knockdown of Sns caused NMJ abnormalities similar to those seen in
Lar/ sns
transhets and
Lar
mutants (Fig
. 3O). Quantification of NMJ parameters revealed that 1b NMJ area was reduced
by 25% (
s
ns
KK RNAi line) and 23% (
sns
GD RNAi line) (Fig.
3Q), the
number of 1b boutons
was reduced
by 22% and 25% in the two
s
ns
RNAi lines (KK and GD respectively) (Fig
. 3R),
and 1b total NMJ length was reduced by 28% and 27% in the two RNAi lines
, respectively (Fi
g.
3S).
Neuronal knockdown of Lar caused similar NMJ a
bnormalities as seen in
Lar/s ns
transhets and
Lar
mutants (Fig
. 3P). One of the two
Lar
RNAi lines
, TRiP HMS00822, showed
similar reductions in 1b NMJ area, number of boutons and total NMJ length
to those
seen with
both
s ns
RNAi lines
, with 1b NMJ area reduced 30%, number of 1b boutons reduced 29% and
total NMJ length reduced 30% (Fig
s. 3Q-S). The second
Lar
RNAi line,
TRiP HMS02186
, had
stronger RNAi effects
, as adult flies with neuronal knockdown of Lar mediated by that line were
not viable and died during mid-
pupal stages. We found stronger effects of Lar knock
-down by
this RNAi line on NMJ development
, with 1b NMJ area reduced 58% (Fig
. 3Q), number of 1b
boutons reduced 50% (Fig.
3R), and total NMJ length reduced 54% (Fig
. 3S). Taken together,
our data suggest that Lar and Sns interact in
cis
in motor neurons to regulate
NMJ
development.
Lar and Sns reporters do not colocalize in the larval mushroom body
Next, we analyzed Lar and Sns expression in the larval
brain, specifically focusing on the MB,
as Lar has been shown to be required for proper development of the larval MB
(Kurusu and
Zinn, 2008). Lar was shown to be expressed in Kenyon cells
(KCs), the principal cells of the
MB, using antibody staining against LAR protein. Here, we confirm that Lar is expressed in
larval KCs using Lar>GFP (Fig
. 4A). Lar
>GFP is also seen in MB axons that compose the two
lobes of the larval MB, the dorsal lobe and the medial lobe (Fig
s. 4B, B ’). A confocal
z
-projection
through the entire larval MB is shown in Figure
4B, B ’. The MB lobes are visualized using
an
antibody against
Fasciclin II (FasII
), which specifically labels the MB neuropil (Figure
4B). A
single optical slice shows that Lar>GFP labels both
the dorsal (arrow) and the medial
lobes of
the larval MB
(Figure 4C’)
.
Sns>GFP was not detected in KCs
(Fig . 4D), but
was seen in
many
other neurons in the
larval central brain
. No Sns reporter expression could be seen in
either the dorsal or the medial
lobes of the MB (Figs.
4E, E ’). A single optical slice of the MB lobes shows no Sns>GFP
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
10
expression in the MB neuropil (Fig
s. 4F, F’). In order to determine if Sns is expressed in
neurons postsynaptic to MB axons, we used a
dendrite-
specific marker
(UAS -Drep2)
(Andlauer
et al., 2014)
to l abel dendrites of Sns-
expressing neurons. We observed dendrites of Sns-
expressing neurons enveloping the dorsal lobes of the mushroom body (Supp
. Fig. 1), indicating
that Sns-
expressing neurons are postsynaptic to Lar-
expressing
KCs. Next, we asked whether
Sns>GFP was expressed in glial cells in addition to neurons. We performed immunostaining
using Sns>GFP and anti
-Repo to label glial cells. We did not see any co
-localization between
GFP and Repo (Supp. Fig.
1), showing that Sns is expressed only in neuron
s.
Lar and Sns genetically interact to shape the formation of the larval mushroom body
Our group has previously shown that Lar is required for the formation of the larval
MB, with Lar
mutants displaying two distinct defects in the third instar MB
. In WT
animals,
KC axons bifurcate
and form two separate neuropil lobes, the medial lobe and the dorsal lobe. Axons in the medial
lobe stop at the midline and do not cross over to th
e contralateral side. In
Lar
mutants, these
medial lobe axons fail to stop at the midline, instead crossing over to the contralateral side and
forming a fused medial lobe across the midline. A second defect is seen in
Lar
mutants
in which
KC axons do not b
ranch
properly,
resulting in reduced or absent dorsal lobes
(Kurusu and Zinn,
2008). L
igand (s) involved in
Lar’s actions in the larval MB
have not been identified. We
investigated whether Sns is required for Lar’s role in the development of the larval MB
by
phenotypic analysis of
Lar/sns
transhets.
We analyzed the medial and dorsal lobes in
Lar/s ns
transhets and
Lar
mutants, along
with their respective heterozygote controls. FasII antibody staining specifically labels both
medial and dorsal lobes. We analyzed 3D reconstructions of FasII
-stained larval MBs to
visualize the lobes in their entirety. Each optical section
of confocal
z
-stacks through the
MBs
was analyzed for the medial lobe fusion phenotype. Supp
. F ig. 3 shows single optical slices with
medial lobe axons either intact or crossing the midline.
For clarity, data on medial lobe and dorsal lobe phenotypes is
displayed in two types of
bar graphs, showing % of MBs (or animals)
without the phenotype (
e.g.,
Fig. 4M
), and % of MBs
with the phenotype (
e.g.,
Fig. 4M
’). Het erozygous
control animals did not show any abnormal
phenotypes in the larval
MB, with largely normal medial and dorsal lobes (Figs.
4G-H , Supp.
Fig. 3). However,
Lar
13.2
/
sns
xb3
transhets displayed fused medial lobes in 37%
of animals, and
Lar
451
/
sns
xb3
animals had fused medial lobes in 41% of animals.
Lar
mutants displayed fused
medial lobes in 6
9% of animals (Fig
s. 4I- L, M, M’). We also observed
dorsal lobe
phenotypes
in
Lar/s ns
transhets, with 34%
of dorsal lobes
either reduced or missing in
Lar
13.2
/
sns
xb3
transhets
and 85% of dorsal lobes reduced or missing in
Lar
451
/s ns
xb3
animals.
84% of
Lar
mutants
displayed
this branching defect, lacking dorsal lobes
(Figs. 4I, K, L, N, N’). We did not observe a
strong correlation between the two phenotypes occurring in the same animal. An animal with a
fused medial lobe did not necessarily also display reduced or absent dorsal lobes.
We next performed pan
-neuronal RNAi
knockdown
for Lar and Sns using the two RNAi
lines for each gene to investigate whether knocking down each gene individually also results in
the MB abnormalities seen in
Lar/sns
transhets and
Lar
mutants. Knocking down Sns resulted
in medial lobe fusion in 58%
of animals using the KK line and 42% of animals using the GD line
(Supp.
Fig. 3).
Lar
RNAi knock
-down resulted in 100% of animals showing fused medial lobes
with the stronger RNAi line (HMS02186) and 54%
of animals with fused medial lobes using the
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint
11
second RNAi
line (HMS00822). We observed weaker penetrances of the branching defect with
RNAi, with only 11%
reduced or
missing dorsal lobes
upon Sns knockdown with the KK line and
21% reduced or
missing dorsal lobes
with the GD line (Supp.
Fig. 3). Similarly, only 27%
of
dorsal lobes were reduced or missing
upon Lar knock-
down with the second RNAi line
(HMS00822), while the stronger Lar RNAi line displayed 61%
penetrance
(Supp. Fig. 3). Thus,
knocking down Lar and Sns causes similar developmental
abnormalities as those seen in
transhets and
Lar
mutants.
These
data show that Lar interac
ts with Sns to regulate the formation of the larval
MB.
The Lar-
Sns interaction in this context is likely to be in
trans
,
as we do not observe any
Sns>GFP expression in larval
KCs or in the MB lobes. Lar>GFP
, on the other hand, is strongly
expressed in larval
KCs and the reporter is visible in both
the medial and the dorsal lobes. We
observe dendrites of Sns-
expressing neurons encircling the dorsal lobes of the mushroom body,
indicating that Sns might be in
neurons that
are postsynaptic to Lar-
expressing
KC axons.
Expression pattern
s of Lar and Sns in the pupal and adult mushroom body
To further
clarify the relationship
s between
the Lar and Sns expression patterns
, we examined
Lar>GFP and Sns>GFP in the adult
MB. Adult KCs
can be broadly classified into three neuronal
types based on the lobes they innervate.
g
neurons are born before the third instar larval stage
and form the adult
g
lobe.
a
’/
b
’ neurons are born during the late third instar larval stage and form
the
a
’ and the
b
’ lobes, which project dorsally and medially respectively.
a
/
b
neurons are born
during early pupal stages and form the
a
/
b
lobes, which project dorsally and medially, par
allel to
the
a
’ and
b
’ lobes.
a
/
b
lobes can be visualized using FasII antibody staining, while the
γ and
a
’/
b
’ lobes can be visualized using Trio antibody staining
.
We performed immunostaining for either FasII or Trio combined with
anti -GFP staining
to la bel either Lar>GFP and Sns>GFP
adult brains. We found strong Lar
expression in the
a
/
b
lobes of the adult
MB, with strong co
-localization of FasII with Lar>GFP (Figs.
5A, A’). Single
optical slice
s through the adult
MB show strong L
ar>GFP
expression in both
the
a
and
b
lobes
(Fig s. 5B, B’). However, we did not observe any expression in the
a
’/
b
’ lobes
, which were
labeled by Trio antibody staining (Fig
s. 5C, C’).
Single optical slice
s showed clear Lar>GFP
expression in
a
/
b
lobes, but no detectable expression in
a
’/
b
’ lobes (Fig
s. 5D, D’). We observed
weak S
ns>GFP
expression in
the
a
/
b
lobes. Figs. 5E, E’ show a
single optical slice. There was
no co-
localization of
Sns>GFP and Trio antibody staining (Fig
s. 5F, F’).
Thus, ne
ither Lar or Sns
is detectably expressed
in the
adult
a
’/
b
’ lobes.
We next investigated
the developmental profile of L
ar and Sns expression in the MB
during pupal stages. We analyzed Lar
>GFP and Sns>GFP
expression in the MB
in 24APF,
40APF and 72APF pupal brains. We observed very little Lar
>GFP in the 24APF MB (Supp.
Figs. 4A, A’). We did see Lar expression on growth cones of
a
/
b
KC axons
near the midline in
the
b
lobes (Supp
. Figs. 4B, B’, arrows). At 40APF, moderate Lar>GFP
expression can be seen
in both
a
/
b
lobes as well as
a
’/
b
’ lobes
. A s ingle optical slice
shows clear Lar>GFP
expression
in
a
/
b
and
a
’/
b
’ lobes
(Supp.
Figs. 4F, F’)
Lar expression peaks at 72APF in the
MB, with strong
Lar>GFP expression seen in all lobes, the
g
lobe,
a
/
b
lo
bes and
a
’/
b
’ lobes
. Single optical slices
show that Lar
>GFP is expressed in both
a
/
b
lobes and
a
’/
b
’ lobes at 72APF (Supp
. Figs. 4J- K’).
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint
this version posted November 5, 2020.
;
https://doi.org/10.1101/2020.11.03.367540
doi:
bioRxiv preprint