through the use of picrotoxin. A complex pattern of
nerve evoked excitatory synaptic responses is evoked by
connective stimulation. Included in the pool of axons
that trigger the excitatory responses in A1 cells are the
rapidly conducting lateral (LG) and medial (MG) giant
axons (HoÈrner et al. 1997). As in cray®sh, the LG and
MG axons are believed to be used in escape, and pos-
sibly in ®ghting behavior in lobsters. They are command
interneurons, that in cray®sh act through a segmental
giant interneuron to trigger the readout of escape motor
programs from the ventral nerve cord (Roberts et al.
1982). The LG and MG axons generate long, slow
EPSPs in A1-5HT cells (durations of up to several
hundred milliseconds) that usually trigger action po-
tentials. The latter often arise from a plateau in the
synaptic response.
System properties of serotonergic neurons
Gain setter role in postural regulation
The experiments that began our explorations of the roles
of amines in aggression in lobsters (Livingstone et al.
1980), showed that injections of 5HT and OCT into
living animals triggered the appearance of opposing
static postures that resembled those seen in dominant
and subordinate animals. Serotonin injections triggered
the appearance of tall-standing dominant-looking ani-
mals, in which the postural ¯exor muscles were con-
tracted and the extensors were relaxed, while
octopamine injections resulted in low to the substrate
subordinate-looking animals showing the opposite pro-
®le of muscular responses. Studies with the isolated
nervous system and bath applied amines yielded similar
results: the two amines activated opposing motor pro-
grams. These studies were the origin of our suggestions
that dominance status might be associated with en-
hanced serotonergic neuron function in lobsters, while
subordinate status might result from or might lead to
enhanced octopaminergic neuron function.
To test these suggestions we felt it ®rst necessary to
®nd candidate amine neurons in the lobster CNS, then
de®ne their physiological properties, and ®nally ask
whether the properties of these cells or their targets
changed with changes in social status. The studies ulti-
mately focused on the T5- and A1-5HT neurosecretory
neurons for two reasons: (1) the cells appeared to be the
major source of 5HT circulating in the hemolymph, and
the only route to peripheral tissues responsive to amines
like the exoskeletal muscles and sensory neurons (Beltz
and Kravitz 1987; Glusman and Kravitz 1982; Goy and
Kravitz 1989; Pasztor and Bush 1989); and (2) since the
cells were the source of 5HT acting on muscles, but also
had sets of central endings, we felt that they might be
important in postural regulation as well, which was the
initial observed eect of amine injection. In particular,
we were concerned with whether the ®ring of these ser-
otonergic neurosecretory cells produced the same eect
as injected or bath applied 5HT. In other words, when
activated, did these cells trigger the appearance of a
dominant-looking stance in lobsters? We attempted to
address these questions in isolated tissue preparations by
®ring the A1- and T5-5HT neurosecretory neurons using
intracellular electrodes while recording from the nerve
trunks innervating the postural ¯exor and extensor
muscles in order to monitor the patterns of ®ring of the
excitatory and inhibitory motoneurons innervating the
muscles.
The result we obtained, however, was initially dis-
appointing. We found that increasing or decreasing the
rates of ®ring of single A1- or T5-5HT neurons neither
increased nor decreased the rates of ®ring of motoneu-
rons. With further study, however, the roles of these
neurons in postural control pathways turned out to be
much more interesting than simply serving to turn on or
o particular motoneurons. This was shown in experi-
ments in which ¯exor and extensor command neurons
were activated. Such neurons are identi®ed by teasing
®bers out of connectives of the ventral nerve cord,
stimulating them at high frequency and triggering the
readout of motor programs from the CNS. Flexor
commands increase the rates of ®ring of excitatory
neurons to ¯exors and inhibitory neurons to extensors,
while simultaneously decreasing the ®ring of excitatory
neurons to extensors and inhibitory neurons to ¯exors.
The net result of ®ring a ¯exor command, therefore, is to
cause animals to go into a ¯exed posture, which makes
them stand tall (in intact animals), just as dominant
animals do in approaching a subordinate. Extensor
commands do just the opposite, triggering postures
resembling those seen in subordinate animals.
Fig. 3 The gain-setter role of A1-5HT neurons. Flexor command
neurons excite tonic ¯exor muscles and inhibit tonic extensors through
activation of central motor programs. The same command neurons
increase the rate of ®ring of A1-5HT cells, which enhances the output
of the command through release of 5HT within the central nervous
system (CNS) and increases the strength of contraction of muscle
®bers through release of 5HT into the general circulation (see text).
Thus these spontaneously active neurons act as ``feed forward''
ampli®ers. Further details are presented in the text and see Ma et al.
(1992)
227
By simultaneously isolating ¯exor command neurons
for stimulation from abdominal connectives, and re-
cording from A1- and T5-5HT cells with intracellular
electrodes, and nerve roots with extracellular electrodes,
it was possible to de®ne the role of 5HT neurosecretory
neurons in postural control circuitries (Beltz and Kravitz
1987; Ma et al. 1992). Flexor commands tended to excite
the 5HT neurons, which in turn enhanced the output of
the command. This was demonstrated by either allowing
the 5HT cell to ®re upon command activation, or pre-
venting the cell from ®ring by passing current through
the intracellular electrode. Since the 5HT neurosecretory
neurons have peripheral and central sets of ending, these
neurons not only enhance the command output via
central actions, they also enhance motor eectiveness
through their pre- and post-synaptic actions on neuro-
muscular preparations (Glusman and Kravitz 1982;
Dixon and Atwood 1989; Goy and Kravitz 1989). This is
diagrammed in Fig. 3, which illustrates what we have
termed the ``gain-setter'' role of these neurons. They in
essence act as ``feed-forward'' ampli®ers. If instead of
exciting a ¯exor command neuron, we ®re an extensor
command (these elicit opposite postures to the ¯exor
commands), then the 5HT cells are inhibited. However,
if we force the 5HT cells to ®re by depolarizing the cell
through the intracellular electrode while activating an
extensor command, then the 5HT cells enhanced the
output of the extensor circuitry as well. Thus, the cells
are ``general'' gain-setters, capable of enhancing ¯exor
or extensor motor output, but the circuitry determines
that they are used only to enhance the appropriate
behavioral responses.
Changes in amine neuron function with changes
in social status
Sensory input to LG neurons
The ®rst clear demonstration of an important eect of
social status on synaptic responsiveness to amines comes
from elegant experiments from the Edwards laboratory
(Yeh et al. 1996, 1997) in which they examined the ac-
tions of 5HT on the modulation of synaptic transmis-
sion between mechanosensory aerents of the tailfan
and the LG neuron in cray®sh. In these animals, acti-
vation of mechanosensory aerents leads to a complex
pattern of synaptic activity in LG neurons. The activity
can be broken into a series of components (a, b, c) de-
pending on whether the activation is monosynaptic and
direct, or is through interneurons. In all cases, the input
to the LG neuron is mainly through rectifying electrical
synaptic contacts. Amines were found to modulate this
synaptic input some time ago (Glanzman and Krasne
1983), but the modulation was only recently found to be
dependent on the social status of the animals (Yeh et al.
1997). Cray®sh were divided into three groups for these
studies: isolates, dominants, and subordinates. Isolates
were animals housed alone for 1 month or longer, while
dominant/subordinate pairs were generated by housing
animals together for 12 days or longer. In isolates and in
dominant animals, 5HT had a facilitating eect on
synaptic transmission between the sensory aerents and
the LG, while in subordinate animals, 5HT reduced the
magnitude of the synaptic response. Moreover the fa-
cilitation by 5HT of the response in isolates and domi-
nants was thought to be through dierent kinds of
receptors, as suggested by dierences in the duration of
the response to 5HT. The changes from the kinds
of modulation seen in isolates to that seen in dominants
and subordinates occurred linearly over a 12-day period,
suggesting a gradual change in receptor subtype distri-
bution with the pairing of the animals. Reversals of the
changes caused by the pairings of animals could be
brought about by new pairings of animals (losers with
losers, winners with winners), but the time required for
change depended on the direction of the change. It took
longer for winners to revert to the loser pattern than for
losers to change to the winner pattern.
While it is not yet clear how such changes contribute
to the behavioral dierences seen between dominant and
subordinate animals, recent experiments on synaptic
activation of the A1-5HT cells in lobsters (described
above, HoÈrner et al. 1997), may give some insights into
how these changes ®t into the underlying circuitry (see
Fig. 4). Mechanosensory aerents excite LGs, which in
turn trigger the activation of motor programs for up-
wards and backwards movements used by animals in
escape, but possibly also used in ®ghting behavior. At
least such movements are a prominent part of the high-
level aggression seen in lobster ®ghts. In lobsters, LG
neurons also excite the A1-5HT cells (see above), usually
causing the cells to ®re. This should release 5HT both
within central neuropil regions where it may enhance
motor output, and into the general circulation from
peripheral release sites where it acts on both tonic and
phasic muscles to enhance their eectiveness (Glusman
and Kravitz 1982; Harris-Warrick and Kravitz 1984).
The 5HT also may reach ganglionic sites by release into
the general circulation, but this has not yet been dem-
onstrated. Thus, the circuitry from mechanosensory
neurons to LG to A1-5HT cells to release of 5HT should
be more ecient in dominant and less ecient in sub-
ordinate animals in the presence of 5HT. How this
serves in the behavior remains unknown.
In another series of studies in cray®sh, an altered
excitability is seen in the ®ring of the LG neuron in
animals of dierent social status (Krasne et al. 1997).
The excitability of the LG falls substantially in subor-
dinate animals, but only slightly in dominants. How and
if this interesting eect relates to serotonergic neuron
function, however, remains unknown.
Serotonin in ®ghting behavior in lobsters
In studies using paired socially naõÈve juvenile lobsters,
Huber and Kravitz (1995) devised a quantitative method
228