by Donald Pfaff
What happens when we damage medullary / spinal pathways? The first experiments were very encouraging: NGC lesions reduced lordosis to about 20 percent of preoperative values (Modianos and Pfaff 1976, 1979). Then Frank Zemlan placed electrolytic lesions in NGC, the largest of which damaged virtually all of the NGC on both sides (Zemlan, Kow, and Pfaff 1983). Such lesions brought the occurrence of lordosis down to less than 40 percent of its prelesion values, and eventually, after about three weeks, lordosis recovered. Zemlan interpreted these data as demonstrating that large NGC lesions interrupted descending control mechanisms important both for the initiation and maintenance of lordosis behavior, descending tracts that he had shown run through the lateral columns of the spinal cord (Zemlan et al. 1979).
So as we approach the motor neurons for lordosis, this summary of the evidence applies: Necessary and sufficient for lordosis are an estrogen-dependent signal coming from the VMH to the MCG and the reticular formation just lateral to it. Stimulation of central grey neurons potentiates both the NGC and the lateral vestibular nucleus to activate the motor neurons for the deep back muscles. Indeed, stimulation of each of these two cell groups increases lordosis, and lesions of each of these two cell groups decrease or abolish lordosis.
Motor Mechanisms: How the Behavior Is Finally Produced
Of course, the first step in piecing together the motor response side of the lordosis circuit was to record from the motor neurons whose excitation causes contraction in the deep back muscles that execute lordosis. Rockefeller graduate student Emily Brink did exactly that (Brink and Pfaff 1981). She identified the axons of the motor neurons that control the medial longissimus and lateral longissimus muscles, crucial for lordosis, and first showed that they responded to segmental inputs, namely, to stimulation of the appropriate lumbar level dorsal roots. Then she used medullary reticular formation stimulation and systematically varied the time between its application and the dorsal root stimulation. If the dorsal root (segmental) stimulation preceded reticular (NGC) stimulation, there was no synergy. But in the series of intervals from simultaneity through 1.5 milliseconds, the NGC stimulation increased many-fold (Figure 2.4); the raw data look as though it is between five times and 10 times the facilitation.
For example, if the NGC stimulation preceded the dorsal root stimulation by 1 millisecond, the lateral longissimus motor neurons had a much more amplified output to dorsal root stimulation than if the dorsal root stimulation alone was used. Similar data came from vestibular nucleus stimulation. If the vestibulospinal stimulation preceded the dorsal root stimulation by between 1 and 8 milliseconds, the motor neuronal response to segmental stimulation was greater, with maximal facilitation at the intervals 2 to 5 milliseconds. Then she put it all together—that is, Brink used the conditioning before segmental stimulation in the MCG which, of course, does not have axons projecting all the way to the lumbar deep back motor neurons and must (as we saw earlier) be working through NGC and the vestibular nuclei. For intervals of 10 milliseconds through 100 milliseconds, prestimulation in the MCG increased the motor neuronal response to segmental stimuli a bit more than three times (Brink and Pfaff, 1981). Thus we have the MCG (which receives the estrogen-dependent VMH signals) working through the NGC and vestibulospinal nuclei to facilitate the lordosis-essential deep back motor neuronal responses to the segmental inputs.
Figure 2.4. Conditioning medullary reticular stimulation, activating reticulospinal axons, greatly potentiated the ability of sensory, segmental stimulation to drive activity in axial muscle motoneurons essential for lordosis behavior. (Adapted from Brink and Pfaff 1981.)
These motor neurons are huge. Classically triangular shaped, they are found at or very near the most ventral extent of the ventral horn at lower lumbar and upper sacral levels (Brink, Morrell, and Pfaff 1979). Their locations were discovered using the horseradish peroxidase technique and confirmed by scanning for sites at which microstimulation produced visible twitches of the deep back muscles. Locations of the lowest threshold twitch sites were consistent with conclusions based on the horseradish peroxidase findings. These are the motor neurons whose activity produces lordosis behavior (Figure 2.5).
As a side point, I note that the reticulospinal neurons later studied with extracellular single neuron recording by Lee-Ming Kow (Kow and Pfaff 1982) could respond to the type of cutaneous stimulation that produces lordosis. Thus, in the lordosis behavior circuitry, one of the spinal-brainstem-spinal loops runs through NGC. If you take away the motor neuron-facilitating supraspinal descending tracts (Cohen, Schwartz-Giblin, and Pfaff 1987b), the pudendal nerve-evoked excitation of deep back motor neurons is severely reduced. For example, total transaction of the spinal cord at the fourth thoracic level almost abolished the motor neuronal response. After all of a variety of partial transaction surgeries (except those of the lateral columns) the pudendal nerve-evoked response was like that of the control, unoperated animals. However, if the lateral columns were transected, the motor neuron response was virtually abolished, just as with the total transaction. This makes sense because we know that the descending reticulospinal and vestibulospinal tracts run in the lateral column.
Figure 2.5. Locations of cell bodies of motor neurons essential for generating lordosis behavior, in the ventromedial corner of the ventral horn in lumbar spinal cord. (Adapted from Brink, Morrell, and Pfaff 1979.)
Thus, descending facilitation originating in the brainstem’s NGC and vestibular nuclei are both sufficient and necessary to permit and amplify the sensory-motor lordosis-producing connection in the lumbar spinal cord. Further, considering early (short latency) and later (long, 50- to 120-millisecond latency) deep back muscular responses to lordosis-eliciting cutaneous stimulation, estrogen pretreatment greatly (four times) increased the magnitude of the later component. I infer that the hypothalamic estrogen effect has been “read out” through central grey and thence reticulospinal and vestibulospinal systems to empower the sensory-motor lordosis machinery at lumbar levels.
Behavior Quantified
What about lordosis behavior itself, the female behavior essential to permit fertilization? First we used high-speed films of lordosis behavior to time and define the female’s vertebral dorsiflexion—necessary for fertilization and all subsequent steps of reproduction—in response to cutaneous contact from the male’s mounting (Pfaff and Lewis, 1974). But because in conventional movie filming the hair, skin, fat, and other tissues reduce the precision with which the skeletal movements that constitute the physical basis for the behavioral response can be followed, we pursued the analysis of the behavior by collaborating with X-ray cinematographic expert Farish Jenkins at Harvard (Pfaff et al. 1978). Of course, the most striking change at the beginning of the behavior is the rapid elevation of the rump from a convex-up posture toward horizontal and then, at the peak of lordosis, 30 degrees into a concave-up posture. This behavioral onset is triggered initially by the male’s paws on the receptive female’s flanks, but the behavior itself is strengthened in response to the male’s pelvic thrusts.
To appreciate the muscles whose contractions constitute the lordosis response to male mounting, Susan Schwartz-Giblin worked with Michael Chen in the Electronics Laboratory at Rockefeller University to invent a strain gauge and electromyogram amplifier that could be used in unrestrained female rats (Schwartz-Giblin and Pfaff 1980). These recordings revealed that deep back muscles were contracting with high levels of electromyographic activity at the beginning of lordosis, but if the vertebral dorsiflexion of lordosis was sustained (e.g., for 4 seconds) the motor units were actually not showing high activity throughout (Schwartz-Giblin, Halpern, and Pfaff 1984). This method of recording confirmed the effectiveness of light cutaneous stimulation on the flanks followed by pressure on the female’s perineum in triggering long and strong lordoses.
The muscles involved are interesting (Brink and Pfaff 1980). Not to be thought of as a single muscular “motor” that causes the vertebral dorsiflexion of lordosis behavior, they are divid
ed by region (thoracic-lumbar and sacrocaudal) and by type. For example, the longissimus system includes the lateral longissimus and medial longissimus, both of which get their muscular power from their tendons’ insertions onto the vertebrae. Medial longissimus, as the name implies, lies between the lateral longissimus and the most medial system, the transversospinalis muscles. Further, these systems have short fiber components as well as long fibers. Most interesting of all, when stimulated unilaterally they cause a lateral deflection to the side as well as a dorsal movement. When stimulated bilaterally, the two lateral deflections cancel each other out, and we are left with a strong dorsiflexion. Thus, lordosis behavior is a bilaterally balanced system in an obligatory manner.
Lateral longissimus, the largest component of the dorsal vertebral musculature, was studied further using histochemical techniques to figure out the sources of metabolic energy used for a strong lordosis response that even supports the weight of the male (Schwartz-Giblin, Rosello, and Pfaff 1983). Among rat mating encounters, frame-by-frame film analyses could show a 225-gram female rat performing lordosis behavior and supporting the entire weight of a 500-gram stud male that was mounting with all four feet were raised off the substrate. Thus, we determined for lateral longissimus its fiber composition, fiber size, muscle spindle distribution, and, most importantly, the distribution of fast-twitch-glycolytic (FG) fibers, fast-twitch oxidative-glycolytic (FOG) fibers, and slow-twitch oxidative (SO) fibers. Lateral longissimus contains predominantly FG fibers; SO fibers were concentrated superficially in the L2–L6 region. We inferred that for the forceful, ballistic movement of the vertebral column during lordosis behavior, fast motor units with large tension strengths would be required, and precise control (as might be expected from slow muscle units with regulation by muscle spindles) would not be required.
As an example of these deep back muscles, strong contraction of lateral longissimus is necessary and sufficient for lordosis. Dorsal root stimulation evoked a electromyographic response with latencies between 1 and 2.5 milliseconds, consistent with the latency of a monosynaptic reflex (Schwartz-Giblin, Femano, and Pfaff 1984), but polysynaptic responses predominated and had both early and late discharges (the latter enhanced by estrogen injections to the female). Conversely, as expected, surgical ablations of deep back muscles by clipping away the muscles from their tendinous attachments abolished lordosis (Brink, Modianos, and Pfaff 1980).
Summary: The Circuitry Is Hierarchical and Modular
Summarizing the discussion thus far, Figure 2.6 shows the lordosis behavior circuit, a social behavior essential for reproduction, the first neural circuit determined for any vertebrate behavior (Pfaff and Schwartz-Giblin 1988). The most obvious property of the neural circuit is its modular structure.
Spinal Module
The degree of sophistication of the circuitry within and between spinal segments (Pfaff and Schwartz-Giblin 1988) is not to be underestimated, as you would guess from the prodigious accomplishments of the previous generation, such as neurophysiologists Elzebieta Jankowska and Anders Lundberg. Sensory inputs from the behaviorally and electrophysiologically defined cutaneous receptive fields are processed through complex dorsal horn circuitry. As you move from the most superficial part of the dorsal horn, the substantia gelatinosa toward Rexed layer V, as determined by the British physiologist Patrick Wall, the receptive fields become larger, more multimodal, and more obviously related to behavioral regulation. While we have studied segmental regulation of lordosis-relevant motor neurons, we have also emphasized powerful descending influences from reticulospinal and vestibulospinal axons that facilitate motorneuronal activity both directly and through interneurons.
Our data show that deep back muscle motor neurons for lordosis can be excited either directly by descending reticulospinal and vestibulospinal axons or indirectly by heightening throughput from the pudendal nerve. But spinal circuitry is not sufficient for lordosis. Transections of the lateral columns, especially the anterolateral columns, abolish lordosis. A brainstem module is required.
Lower Brainstem Module
Although individual spinal segments primarily do local business, input-output relations among segments would tend to produce an amplification of response to a limited excitatory stimulus. Nevertheless, a lower brainstem module is needed to coordinate the molar behavioral response, lordosis, across many spinal segments. The data cited above prove that descending signals in the reticulospinal and vestibulospinal tracts do the job.
Figure 2.6. Three experimental launching points were used to discover the complete working circuit for producing lordosis behavior. 1) The cutaneous stimuli that are adequate to start the behavior (lower left). 2) The motor response, dependent on deep back muscles (lower right). 3) The ventromedial hypothalamic neurons that express the estrogen receptor-α gene (top). The circuit is bilaterally symmetric, plotted here on one side for clarity. (Adapted from Pfaff and Schwartz-Giblin 1988.)
Of course, signals ascending from the spinal to the lower brainstem module play an important role. Neuroanatomical studies show cord neurons backfilled from the NGC, for instance. And Lee-Ming Kow recorded NGC neuronal responses to lordosis-relevant cutaneous stimulation both in anesthetized and free-moving unanesthetized female rats.
The sensory summation is as follows. As part of our discussion of these ascending signals we can state a principle within a principle. Cutaneous pathways for lordosis behavior converge. As the signals ascend, the cutaneous receptive fields become larger, and the sensory submodality specificity becomes sharper.
Then we find selective distribution. That is, as sensory information travels from the primary sensory surface into the lordosis circuitry, it is clear that not all sensory information is transferred. First, as part of the growth of receptive field size, spatial information is lost. Second, as the information ascends the neuraxis in the lordosis circuit, the magnitude of the sensory response compared with the background activity caused by other sources declines. Convergence from other parts of the circuit is required for high levels of activity.
Pfaff and Schwartz-Giblin listed (1988) several properties that demonstrated a congruence between lower brainstem module neuronal properties and the requirements for lordosis behavior. Put briefly, they included the facilitation of deep back muscle activity by electrical stimulation of the NGC or lateral vestibular nucleus, the decrease in lordosis after lesions of either of these cell groups, the facilitation of responses to pudendal nerve inputs, bilateral symmetry, and the termination of reticulospinal and vestibulospinal axons across several spinal levels, as required for lordosis. Put briefly, the lower brainstem module integrates behavioral and postural changes across spinal segments.
As mentioned previously, the hierarchical nature of lordosis motor control is clear. The hypothalamus energizes the MCG which, in turn, empowers reticulospinal and vestibulospinal descending signals to turn on lordosis-relevant motor neurons (Pfaff et al. 1990).
Midbrain Module
The proof of existence of a midbrain module is neuroanatomical, in part. On the descending side of the lordosis circuit, crucial estrogen signal–carrying axons do not descend below the midbrain, so midbrain neurons as such must participate in the circuit. VMH neurons increase the excitability of MCG neurons that project in turn to the medulla. On the ascending side we have seen evidence of fibers from the anterolateral columns of the spinal cord ascending to reach the MCG; from Yasuo Sakuma’s work in the laboratory, central grey neurons respond to lordosis-relevant somatosensory input. From the lower brainstem module, axons from NGC to midbrain are dense and intense.
My usual way of thinking about the midbrain module is that it acts as a neurophysiological version of an automobile transmission. That is, hypothalamic dynamics are slow. Endocrine signals to the hypothalamus change over hours or days. VMH neurons fire slowly—in fact, stimulation over thirty pulses per second blocks them. In contrast, many other neurons in the lordosis behavior circuit—and in fact in most of the brain—are fast. Thu
s, analogous to an automobile transmission, MCG neurons take in slowly changing hypothalamic outputs and put out signals that work well with rapidly firing, rapidly reacting brainstem reticular and vestibular neurons.
In summary, neurons in the central grey of the midbrain facilitate lordosis, and central grey neurons reduce lordosis. Peptides synthesized in the hypothalamus and preoptic area such as gonadotropin-releasing hormone (Chapter 5) and oxytocin (Chapter 6) play a major role.
Forebrain Module
When forebrain systems have been studied for their roles in the regulation of lordosis behavior, clear-cut effects have been inhibitory. The most efficient way to sever forebrain outputs descending toward the diencephalon and midbrain is to sever those axons by what is called the roof deafferentation technique. This procedure greatly enhances lordosis. Decorticate females perform lordosis. Lesions of the septum and surgical removal of the olfactory bulb greatly increase lordosis.