Worm Breeder's Gazette 10(1): 91

These abstracts should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.

Using the Laser to Assign Functions to Neurons in the Head Ganglia

J. Thomas, G. Garriga, L. Avery and B. Horvitz

We have recently been attempting to assign functions to various 
neurons, many of which are located in the head ganglia near the nerve 
ring.  We use a Laser Sciences Inc.  nitrogen-pumped laser (described 
in WBG 9(2), 1986, p. 110) to kill specific neurons identified on the 
basis of position in L1 worms, then let the animals grow to adulthood 
and test behavior.  In combination with the complete wiring diagram (
White et al., 1986), this method provides a powerful approach to 
understanding the function of the nervous system.
We can envision at least four potential pitfalls with this approach. 
First, we may fail to eliminate functionally the neuron that is 
killed.  Second, we may cause damage to neighboring cells or processes,
especially those located near the site of the laser strike.  Third, 
the cell that we identify on the basis of position might not 
invariably be the same cell; for example, it might occasionally be 
switched with some neighboring cell.  Fourth, the cell in a particular 
position might not be correctly correlated with the wiring diagram.  
Of these pitfalls, we have collected some information about the first 
three.  The fourth problem is more difficult to address, and largely 
depends on the assignments by Sulston et al.  (1983).
The evidence so far suggests that killing a neuron with the LSI 
laser results in functional elimination of the target cell.  First, 
when we kill a neuron (by this we mean all members of a class of 
neurons) with a previously described function, we reliably obtain the 
expected phenotype for loss of the neuron's function (based on genetic 
studies and/or embryonic kills; Chalfie et al., 1986; M.  Chalfie, C.  
Desai, personal communications).  Such cells include AVA (backward Unc;
6), RIP (loss of Mec inhibition of pumping; 2), HSN (egg-laying 
defective; 3), PLM (tail Mec; 5), ALM (reduced head touch sensitivity; 
3), and PVC (tail Mec; 3).  (Parenthetical remarks after each neuron 
indicate the behavioral phenotype and the number of animals analyzed.) 
In other cases, we have observed a behavioral defect associated with 
killing a particular neuron.  When we kill this neuron in more than 
one animal we consistently observe the same behavioral defect.  These 
cells include ASH (12), ASJ (2), PVQ (3), and M4 (150).  (For the 
functions of ASH, ASJ, and PVQ see the accompanying Thomas and Horvitz 
newsletter entry.) A related concern is how fast a killed neuron loses 
function.  The best evidence for this time course is for the 
pharyngeal motorneuron M4.  This neuron is required for peristalsis of 
the pharyngeal isthmus (Avery and Horvitz, WBG 9(2), 1986, p.  57), a 
function that can be assayed at any stage of development.  There is a 
pronounced, but not always complete, deficit in M4 function 5 hours 
after laser killing in young L1 larvae, and M4 is invariably 
nonfunctional after 24 hours.  Similarly, when PVC (3), AVA (6), or 
PLM (5) is killed during the early L1 stage, animals acquire the loss-
of-neuron-function phenotype within 24 hours (the earliest time tested)
.  These results suggest that laser killing in the early L1 typically 
eliminates neuron function well before adulthood.
We also have evidence that laser damage to cells or processes near 
the target cell is not generally a problem.  Even when we kill neurons 
in the head ganglia, where neuron cell bodies and processes are 
closely packed, we do not observe effects on other behaviors.  For 
example, no kill (other than AVA) has caused a backward Unc phenotype, 
characteristic of AVA animals.  The most extensive analysis has been 
done for the cell ASH, which is required for normal osmotic avoidance (
see accompanying Thomas and Horvitz newsletter entry).  Neurons with 
cell bodies that surround ASH on all sides (ADF (2), AWC (3), AUA (2), 
AIB (1), and ASE (3)) have been killed with no effect on osmotic 
avoidance.  In addition, many neurons with a process that runs 
alongside the ASH process in the amphid bundle (ADL (2), AWC (3), and 
ASK (4)) or in the nerve ring (AIB (l) and ADF (2)) or both (ASE (3)) 
have been killed with no discernible effect on ASH function.  These 
results should be taken with a grain of salt, since most C.  elegans 
neurons (including AVA and ASH) are bilateral.  Since probably only 
one of the pair need function for normal behavior, we might require 
damage of both cells to see a phenotype.  However, this redundancy 
also works to our advantage, since only damage to both sides will 
produce spurious results.
The same set of laser experiments show that, for several neurons in 
the head, position in the early L1 is sufficient to identify the same 
cell in different animals.  These cells are RIP (2), ASH (12), ASJ (2),
and AVA (6).  For each of these cases, when the cells in these 
positions are killed in different animals they cause the same 
behavioral defects in the adult.  We have also tested cells in the 
tail ganglia (PLM (5), PVC (3), and PVQ (3)) with the same result.  
This evidence indicates that the same functional cell type occupies 
the same position during the L1 stage in different animals.  This rule 
may not apply to all neurons, but as more cells are assigned invariant 
positions, the potential variability of the remaining neurons becomes 
more restricted.
Although we cannot directly address the final possible pitfall (
correlation of the killed neuron with the wiring diagram), we can 
point out that the behavioral phenotype observed for each kill fits 
well with the target cell's assigned connectivity.
Finally, we offer some general comments on the usefulness of this 
method.  The position of most neurons in the head ganglia and 
elsewhere can be learned fairly easily by careful inspection of a 
number of animals and frequent comparison with the excellent diagrams 
found in Sulston et al.  (1983).  Some neurons are easier to identify 
than others, and a few (around the posterior bulb of the pharynx) may 
not be possible to identify because of their variable positions (as 
noted in Sulston et al., 1983).  However, the general impression one 
gets, after sufficient observation time to learn the patterns, is the 
remarkable reproducibility of the relative positions of most neurons.  
It is quite reasonable, for a cell one knows well, to kill a 
particular type of neuron and confirm the kill in more than 30 animals 
in one day.