Screen Shot 2018-02-01 at 12.17.10 PM.png

TRACT: Tips, Troubleshooting & Limitations


Please, notice that in its current implementation, TRACT may work to perform transneuronal tracing (or genetic manipulation) in some circuits, but not in others.  In addition, for your specific research question it may be necessary to do some fine tuning to achieve an appropriate signal-background ratio.  We would greatly appreciate if you could share your experiences with us, so that we can modify the TRACT components to make it more reliable. If you have any problems or questions, please contact us at:, and we will try to help you.


(i) Methods to detect transneuronal tracing:

(a)    Detection by immunostaining: In some cases the transneuronal induction is strong enough to be directly detectable simply by observing the samples under a fluorescent microscope (for example, with reporters that are 20xUAS-6xGFP, or other multiples of UAS).  However, we routinely do immunostaining against the reporter to increase the intensity of the signal. This is the protocol that we use for immunostaining:

After incubation at 29C for one day, the adult Drosophila were dissected, and the brains were removed in 1x PBS under a dissection microscope. For PER immunostaining, the flies were dissected one hour before the lights turned on (around 7 AM). Brains were fixed by immersing them in a 4% paraformaldehyde solution in PBS for 15 minutes at room temperature. Brains were washed in PBS three times for 10 mins each, followed by permeabilization with PBS +0.5% triton X-100 (PBST) for 30 mins and blocking with 5% horse serum in PBST for 30 mins. The brain samples were stained with antibodies against GFP (rabbit polyclonals (AB3080 1:1000, AB3080P 1:1,500 (Millipore)), or chicken polyclonal (Abcam), diluted at 1:1000), mcherry (rat monoclonal 5F8 (Chromotek) diluted at 1:1,000), Brp (mouse monoclonal nc82 (DSHB) diluted at 1:50), ChAT (mouse monoclonal 4B1 (DSHB) diluted at 1:200), GABA (rabbit polyclonal A2052 (Sigma) 1:200), PER (guinea pig polyclonal PA1140, gift from Dr. Sehgal, University of Pennsylvania), V5 (mouse monoclonal R960-25 (invitrogen) diluted at 1:300), OLLAS (rat monoclonal L2 NBP106713 (Novus) diluted at 1:300) diluted in 5% horse serum/PBST. Brains were Incubated with primary antibodies overnight at 4C, washed 3 times in PBST, incubated with secondary (goat secondary antibodies (Life Technologies) 1:500, except for rabbit anti-GFP AB3080P (Millipore), where the secondary was used at 1:750) for 1.5 hours at room temperature, washed in PBST and mounted on glass slides with a clearing solution (Slowfade Gold antifade reagent (Invitrogen)).
Stained brains were imaged with Confocal microscopes (Olympus Fluoview 300 or Zeiss 710) under 20x or 40X objectives. In a typical experiment, we imaged 150 sections with an optical thickness of 0.3-0.5 µm from dorsal or ventral sides. Confocal stacks were processed with Fiji to obtain maximal projections.

Notice: there are big differences in the intensity of staining produced by different anti-GFP antibodies. 

For brains with nSyb-nlgSNTG4 and 5xUAS-CD4::tdGFP, we recommend using Rabbit anti-GFP A3080p (1:1500) or chick anti-GFP (1:1000). Chick Ab will generate higher background than Rb Ab in general.

For brains with elav-nlgSNTG4 and 5xUAS-CD8::GFP, we recommend using Rabbit anti-GFP A3080 (1:1000).


(ii) Tips:

(a) Which ligand should I use?

 We have tested a number of domains to target the ligand to presynaptic sites, and our experiments indicate that syb and sdc appear to be the best candidates.

Antennal lobe:

Sdc:  works reliably in het or homozygote animals (single or double copy of the ligand)

Syb: we have observed that the multiglomerular PNs can be detected in the homozygote animals (2 copies), but not with the hets.

Circadian system:

            Both sdc and syb label more postsynaptic neurons as homozygote animals.

Syb: appears to label more neurons than sdc

Sdc: when expressed in PDF neurons, it triggers GFP expression in neighboring PDF neurons.  This could mean that: (i) PDF neurons are actually connected by synapses to other PDF neurons, or (ii) that sdc may produce false negatives because PDF neurons are close to each other (but they may not have synapses between them).

Summary: We would recommend using the syb:CD19 ligand because: (i) it is more precisely localized in the presynaptic terminal than sdc, and (ii) the transneuronal labeling that it produces appears to be consistent with the existing data.  If possible, always compare het and hom for the ligand, because it is likely that the results will be more consistent between animals.

(b) Activity dependence of labeling:

 We have successfully used two different ligands that confirmed synaptic transneuronal tracing: syndecan (sdc) and synaptobrevin (syb).  In theory, syb should show some activity-dependence for labeling, and we have observed this to be the case for some circuits, but not others (olfactory system, circadian system).

(c) Fine tuning:

Transgene copy number: We have observed that in some cases, having the appropriate number of copies of ligand and/or receptor could have an important effect on the intensity of the transneuronal labeling.

For example, we can detect multiglomerular PNs (mPNs) with the syb:CD19 ligand with two copies of the ligand (homozygotes), but not in heterozygotes. In the same case, 2 copies of Syb::CD19 driven by pdf driver provided stronger GFP induction than 1 copy of it.

Screen Shot 2018-02-01 at 4.29.32 PM.png

Ligand-independent, receptor-independent background: In other cases we have observed that the reporter used can produce different results.  In particular, we have noticed that some reporters have a high level of background, even when there is NO TRACT receptor in the fly.

Screen Shot 2018-01-31 at 5.04.27 PM.png

Ligand-independent, receptor-dependent background: We have observed that when the receptor is expressed under the elav enhancer, we see very strong background (ligand-independent) in several brain areas. For example, in the image below (elav-nlgSNTG4), with the lexA driver and LexAop ligand there is robust induction in the LNs of the antennal lobe (right image).  However, without driver there is a strong GFP background induction in the mushroom body (left image). 

Screen Shot 2018-01-31 at 5.05.54 PM.png

Driver-independent, ligand-dependent, receptor-dependent background: We have observed situations, in which the ligand (mCD19 or mCD19:: mch expressed from a LexAop promoter), is expressed and induced GFP at the AMMC even if there is NO lexA driver.  Presumably this is due to a position effect of the LexAop promoter.

Screen Shot 2018-02-02 at 12.09.23 PM.png


(iii) Current limitations:

 We have successfully performed anterograde transneuronal tracing between olfactory receptor neurons and antennal lobe principal neurons, and also between PDF+ neurons and some of their postsynaptic partners in the central brain.

However, we also know that in its current implementation TRACT does NOT reliably work for other circuits, in particular the optic lobe  and the mushroom body. 

We have observed that with the nSyb enhancer, the TRACT receptor is very weakly expressed in the mushroom body, and in the optic lobe.

Optic lobe: For example, notice in the image below that the TRACT receptor is almost undetectable in the optic lobe. In this image we did immunostaining against the V5 tag located in the intracellular domain of the TRACT receptor.  The antennal lobe expresses the TRACT receptor very strongly, but the optic lobe barely expressed the receptor.   

Screen Shot 2018-01-31 at 4.39.28 PM.png

As a result of this limitation, for example, when the starter cells are R7 neurons (with an Rh3-lexA ligand) we see no cells labeled in the medulla, although we would expect hundreds of postsynaptic targets there.

Screen Shot 2018-01-31 at 4.47.31 PM.png

Mushroom body: In this single optical plane of the brain, you can detect that the mushroom body expresses the TRACT receptor at very low levels.  As a result of this lack of receptor expression, we have NOT been able to detect any signal in Kenyon cells when the starter cell is a PN.

Screen Shot 2018-01-31 at 4.49.38 PM.png

Central brain: It is possible that some neurons that have very extensive axonal branches may only have few synaptic contacts with each of its postsynaptic targets. In those cases, it is possible that TRACT may fail to detect those postsynaptic partners.


(iv) Work in progress:

we are currently working on the TRACT system to optimize it, to make it more reliable, and to enable it to be used for different applications.

(a) retrograde tracing: the initial version of TRACT could only be used for anterograde tracing (from presynaptic to postsynaptic neuron).  We are currently generating new tools so that TRACT can also be used in the retrograde direction.

(b) comprehensive tracing: as mentioned in the “current limitations” section, we have observed that the syb promoter is not pan-neuronal.  We are generating new strains with different promoters in an attempt to achieve true pan-neuronal labeling.

(c) Alternative drivers-reporters:  the initial version of TRACT used LexA to express the ligand and GAL4/UAS to report receptor activation.  To allow more flexibility to the system we are currently generating new strains in which the expression of the ligand will be under Gal4 drivers, and the output of the receptor will be LexA/LexAop or QF2/QUAS. 

(d) as noted in the “current limitations” section, we have observed some background siganlling in some brain regions.  We are testing new strategies so that we can increase the transneuronal signalling while reducing the backgournd.

We are actively working in these optimizations of TRACT, and we will make these flies available as soon as we confirm that they work.