Reflexes, fictive respiration and cell division in the brain and spinal cord of the newborn opossum, Monodelphis domestica, isolated and maintained in vitro

Studies of the physiological development of the mammalian CNS have been made, until recently, almost exclusively on eutherian mammals and have been hampered by the inaccessibility of embryos at early stages. Many investigations have examined areas of the brain, such as the cerebellum, that arise largely post-embryonically. Others have used avian and amphibian species with more accessible embryos. Slice preparations of immature mammalian CNS have proved especially advantageous for analyzing well-defined circuits and for studies on regeneration in vitro. A start has also been made in developing more complex preparations of immature CNS in which the normal relationships of different brain regions are preserved in vitro. For example, the spinal cord, medulla and midbrain of neonatal rats have been shown to be small enough to survive for several hours in vitro. Moreover, as with invertebrates, the nervous system can be isolated together with effector organs, such as muscles, ribs and lungs, and can continue to display spontaneous rhythmical activity (Onimaru and Homma, 1987; Otsuka and Yanagisawa, 1988).

The present experiments were devised to make use of one specialized type of mammal, a marsupial. Marsupials offer particular advantages since their young are born with a largely undeveloped nervous system that has virtually no cerebellum, an ‘embryonic’ two-layered cerebral cortex and a partially organized spinal cord (Reynolds et al. 1985; Saunders et al. 1989). Marsupials have been ignored, for the most part, because of difficulties in breeding them in the laboratory. However, the recent establishment of breeding populations of the grey, short-tailed opossum Monodelphis domestica in several laboratories has eliminated this difficulty (Saunders et al. 1989; Dziegielewska et al. 1989). This animal is similar to a rat in size. The newborn pups, 1–12 in number, somehow find their way to the nipples. For the next 3 weeks they cling to the mother (which has no pouch), exposed, suckling and breathing.

Our initial aim was to isolate and maintain in culture the entire CNS from the newborn animal. It seemed likely that the CNS at this stage could present the following attractive features: (1) the newborn opossum is less than 1.5 cm long and this suggested that its CNS would be small enough to survive well in vitro; (2) the poorly developed hind-limbs in the newborn opossum suggested the presence of a gradient along the spinal cord, with more mature developed motoneurones in rostral than in lumbosacral segments (Altman and Bayer, 1984; Fulton and Walton, 1986). A long-term aim of these experiments is to develop preparations suitable for studying neuronal circuits involved in generating respiratory rhythms. Although equivalent to a 14-day rat embryo, the newborn opossum breathes and suckles but does little else. The appropriate neuronal circuits must presumably be fully effective.

The techniques we have used include electrical recordings from dorsal and ventral roots and electromyograms from rib muscles attached to the isolated CNS, as well as histological and labelling techniques for examining the progeny of precursor cells dividing over the first few days in vitro. We show that the CNS continues to display reflexes in Krebs’ fluid. In addition the neurones are readily accessible to pharmacological agents such as tetrodotoxin applied to the bathing fluid. By using bromodeoxyuridine (BrdU), we show that cells in the proliferative zone of the spinal cord continue to replicate DNA for 4 days after isolation and maintenance in Krebs’ fluid.

The animals used in this study came from an established colony of M. domestica in Southampton and a new colony established in the Biocenter in Basel. Although these animals will breed throughout the year and often mate within a short time of removal of the young, their reproductive performance is rather variable. In order to produce two pregnant animals per week on a regular basis, at least 20 breeding females and 10 males are required with the young being removed in the first few postnatal days and the females being immediately re-mated. A colony of over 100 animals is required to provide a regular supply of breeding pairs. The animals are kept in plastic boxes supplied with sawdust, woodchips and tissue paper for nesting material. The room temperature is maintained at 26–27°C and the room illuminated for 13 h daily. The diet used consists of cat food (Gourmet) supplemented with cream for lactating females. M. domesticus is poly-oestrus with a cycle of 28 days. Its gestation period is 14 days and litter sizes vary from 1 to 12 pups. A newborn pup detached from its mother is usually eaten by her. On its own the pup dies rapidly, within about 15 min. The young are weaned at 8–9 weeks and reach sexual maturity in 18–20 weeks (Trupin and Fadem, 1982; Fadem et al. 1982; Baggott et al. 1987; Kraus and Fadem, 1987; Saunders et al. 1989). Seventy-five newborn opossums were used for these experiments.

Baby opossums were removed from their mothers starting at 1–4 days of age and then anaesthetized with ether, chloroform or by cooling; a photograph of a 2-day-old opossum is shown in Fig. 1A. Anaesthetized young were placed dorsal surface uppermost in a small, Sylgard-filled Petri dish with tiny pins (Minuten Nadelen) placed through the paws. Lateral cuts were made with scissors through the ribs in the body wall. The animal was then covered in Krebs’ fluid bubbled with 95 % O₂/5 % CO₂ and the skin over the brain and spinal cord was removed. The entire CNS was isolated by cutting the muscles and vertebrae from over the dorsal aspect of the spinal cord and the dura mater covering the brain. The brain was isolated further by separating it from the floor of the cranium; cranial nerves and the connections to Rathke’s pouch and to the olfactory bulbs were cut. The dorsal and ventral roots were severed from the cervical to the lumbar region, after which the entire CNS was removed. In some preparations, several ribs and muscles were left attached to the spinal cord so that rhythmical movements of the ribs (breathing) could be observed. Fig. IB shows the relative sizes of the CNS of the adult and the neonate, while Fig. 1C illustrates the incompletely developed structure of the neonatal CNS. The CNS was maintained at room temperature (22–24°C) in Krebs’ fluid containing 11 mmol l⁻¹ glucose for up to 4 days. The level of oxygenation and the pH were maintained by continuously bubbling with 95% O₂, 5% CO₂.

Electrical recordings were made using suction electrodes in Krebs’ fluid or hook silver-wire electrodes in mineral oil at each end of the CNS. Electrical signals were amplified by a low-noise, differential a.c. amplifier (Almost Perfect Electronics, Basel) and displayed on an oscilloscope and on a chart recorder. Signals were digitized using a VR-100 digital recorder (Instrutech Corp., Mineola, USA) and stored on videotape. Suction electrodes for stimulating or recording from dorsal or ventral roots were made with a conventional electrode puller from 25 μl measuring micropipettes (Clay Adams, Accu-Fill 90). Pulled electrodes were scored near their tips with a diamond scribe, the tips broken off, and the ends of the pipettes fire-polished to a diameter of about 20 μm.

The composition of our modified Krebs’ fluid was (in mmol l⁻¹): NaCl 115; KCl 5; CaCl₂ 2; MgSO₄ 1; NaHCO₃ 12 and glucose 11 (Krebs and Henseleit, 1932). Tetrodotoxin (TTX) was added from a stock solution to the Krebs’ fluid; final concentrations were 0.01, 0.1 or 1 μmoll⁻¹. As a test for the potency of the TTX, a frog sciatic nerve was exposed to a TTX concentration of 1 μmol l⁻¹. As expected, blockage of evoked compound action potentials occurred within minutes, but unblocking required washing overnight. MgCl₂ was added to obtain final concentrations of 10 or 20 mmol l⁻¹.

The CNS was placed in Bouin’s fixative immediately after isolation or after maintenance in Krebs’ fluid for periods ranging from hours to days. The fixed preparations were dehydrated, embedded in Epon and sectioned at 1.5–1.7 μm with glass knives on a microtome (Sorval MT2B). The sections were placed on gelatin-coated slides and stained with 1 % Toluidine Blue in 1 % borax at 60 °C for 2 min. Stained sections were differentiated for 10 s in 0.5 % ethanol: 0.5 % acetic acid, rinsed in distilled water, and then air-dried before mounting. In the proliferative zone in all regions of the spinal cord (cervical, thoracic, lumbar and sacral) from newborn up to at least 4-day-old animals, cells were observed that appeared to be dividing or that had just divided, as judged by their rounded shapes and close proximity to the luminal surface.

The isolated CNS of 1-to 4-day-old M. domestica was maintained for a short period (2h) or for longer (up to 4 days) before incubation in Krebs’ fluid containing 30 μmol l⁻¹ bromodeoxyridine (BrdU), an analogue of thymidine. The full procedure for BrdU labelling was as follows: after soaking in BrdU for 2 h, the CNS was rinsed with fresh Krebs’ fluid, fixed immediately in 70 % ethanol for 1 h and rinsed again for 1 h in phosphate-buffered saline containing 1 % Triton X-100 (PBS–TX). The CNS was cut into lumbar, thoracic, cervical and brain regions. DNA strands were separated with 2moll⁻¹ HC1 for 30min to 1h, the acid was neutralized with sodium borate (Na₂B₄O7) for 1 min and then the preparation was rinsed for 30 min with PBS–TX. Anti-BrdU (1:20; Becton-Dickinson) in PBS–TX containing 10 % horse serum (PBS–TX–HS) was added to brain, cervical and lumbar regions and left overnight at 4°C. Controls were carried out omitting (1) the BrdU, (2) the primary antibody or (3) the DNA hydrolysis step and using a rhodamine-conjugated secondary antibody instead of the peroxidase anti-peroxidase (PAP) described below. After rinsing with PBS-TX-HS for 30 min, secondary antibody (goat anti-mouse IgG, 1:100 in PBS–TX–HS) was added for 2h to both the control and experimental preparations for 2h after which the preparations were rinsed for 30min with PBS–TX–HS. Mouse PAP diluted 1:500 in PBS–TX–HS was added to the preparations for 2h. The preparations were rinsed in PBS–TX–HS for 30 min and then for 3 min in Tris-buffered saline (TBS). The preparations were incubated in diaminobenzidine (DAB) made up in TBS (15 mg of DAB to 20 ml of TBS) for 20 min before 1 drop of 3 % hydrogen peroxide in TBS was added. The peroxidase reaction was allowed to proceed until cells adjacent to the spinal canal became dark brown. The reaction was stopped by rinsing with TBS. The preparations were dehydrated in an ethanol series (70–100 %), cleared in methyl salicylate, and mounted in ‘Permount’ (Fisher). In a few preparations, the spinal cords were embedded in ‘Tissue-Tek’ (Miles, Elkhart, IN, USA), frozen on dry ice and sectioned at 20μm on a cryostat before labelling with anti-BrdU.

Three techniques were used to judge whether the CNS remained viable after isolation: the ability of the spinal cord to conduct action potentials for extended periods (days) ; the presence of mitotic cells in the proliferative zone of the spinal cord; and the continued ability of blast cells to incorporate BrdU.

Compound action potentials were recorded from the isolated CNS of 1-to 4-day-old opossums to assess ‘through-conduction’ and the viability of neurones. The conducted volleys were usually biphasic with an initial step (see Figs 5 and 6). The shape depended on the types of electrodes used and their locations, and varied considerably from preparation to preparation. The recording arrangement is shown in Fig. 2A. The compound action potential recorded from the ventral brain region of a CNS that had been isolated from a 2-day-old opossum and maintained in Krebs’ fluid for 4 days at room temperature is shown in Fig. 2B. Compound action potentials were also obtained in the opposite configuration by stimulating the brain and recording from lower thoracic regions. Crushing the cord between the recording and stimulating electrodes eliminated all evoked compound action potentials. Not surprisingly, after 4 days in vitro the conducted volley was smaller than that observed on the day of isolation (see below, Fig. 5), but our results clearly indicated survival of electrical excitability (see also Discussion).

Anatomical sections were made of isolated newborn opossum CNS that had been maintained in Krebs’ fluid for periods ranging from hours to days, to determine whether dividing cells were still present in the spinal cord (Sidman et al. 1959). Fig. 3 shows an example of a CNS from a 3-day-old opossum that had been maintained in Krebs’ fluid for 6.5 h before fixation in Bouin’s solution. In each region of the spinal cord - cervical (Fig. 3A), upper thoracic (Fig. 3B) and lumbar (Fig. 3C) - rounded cells were present in the proliferative zone along the spinal canal. The proliferative zone itself was thicker and denser in more caudal regions (Fig. 3C). A section through the lumbar cord is shown at high magnification in Fig. 3D. Four cells that by their relative position to one another and by their shapes appear to have divided are marked by arrows. Further evidence for active division is presented below.

The isolated CNS from neonates was exposed to 30 μmol l⁻¹ BrdU for 2 h either immediately after removal or after 1–4 days in Krebs’ fluid. BrdU is an analogue of thymidine that is incorporated into DNA during replication. An example of BrdU labelling in an isolated CNS from a 1-day-old opossum that had been kept for 1 day in oxygenated Krebs’ fluid at room temperature (22–24 °C) is shown in Fig. 4A–C. Cells that had incorporated BrdU were present in the proliferative zone. In rostral regions of the spinal cord - cervical and upper thoracic - incorporation of BrdU occurred mainly in the dorsal half of the proliferative zone (Fig. 4A,B). In lower thoracic (Fig. 4C) and lumbar regions (not shown), cells that had incorporated BrdU were more uniformly distributed in the dorsal and ventral regions of the spinal cord. Higher regions of the CNS, such as medulla, midbrain and cerebral vesicles, also contained BrdU-labelled cells in abundance (not shown).

Four controls were done to ensure that the BrdU labelling was specific: (1) the primary antibody was omitted (Fig. 4D); (2) the CNS did not receive BrdU; (3) the DNA strands were not separated; and (4) a rhodamine-conjugated secondary antibody was used instead of peroxidase anti-peroxidase. The first three controls showed no labelling of precursor cells (see Fig. 4D) and the fourth control with the rhodamine-conjugated secondary antibody showed essentially the same labelling as the sections processed with peroxidase. BrdU incorporation by blast cells also occurred after 4 days in Krebs’ fluid; the pattern of labelling was similar after 1, 2 or 4 days in Krebs’ fluid.

Together these results indicate that the isolated CNS of the newborn opossum continues to conduct action potentials for days in Krebs’ fluid and that precursor cells continue to replicate DNA and to divide.

Tetrodotoxin (TTX) and Mg²⁺ were applied to the bathing fluid to determine the efficacy of blocking agents on conduction through the nervous system. Compound action potentials were blocked rapidly (within 2 min) and completely by 0.1–1μmoll⁻¹ 11X (Fig. 5B). Increasing the amplitude or duration of the stimulus to 10 times maximal did not produce conduction in the presence of TTX. Within a few minutes of removal of 11X from the bathing fluid, through-conduction was completely restored (Fig. 5C). The compound action potentials conducted through the CNS were partially blocked by high Mg²⁺ concentrations (10 or 20 mmol l⁻¹; Fig. 6) applied to the bathing fluid. These records show that most of the action potential components conducted to the recording electrode were generated through synaptic interactions. The conduction velocity of the fastest unblocked peak was approximately 0.3 m s⁻¹, as shown in Fig. 6B. As with TTX, the effects of Mg²⁺ reversed rapidly after washing the preparations. The half-time for recovery was about 4min.

Recordings were made from ventral roots of the spinal cord to monitor spontaneous and evoked activity in young opossums 1–4 days of age. Fig. 7 illustrates impulses recorded from the ventral root following dorsal root stimulation. This animal was 3 days old and the preparation had been maintained in vitro for 8 h. Stimulation of the ipsilateral dorsal root of the same cervical segment gave rise to complex volleys with variable latencies in the ventral root. These responses were completely eliminated by 10–20 mmol l⁻¹ Mg²⁺ added to the bathing fluid. Increasing the strength of the stimulus to 10 times maximal failed to evoke responses in this solution. At the same time all spontaneous asynchronous activity in the roots was abolished by high Mg²⁺ concentrations. A few minutes after replacing the bathing fluid with normal Krebs’ solution spontaneous and evoked activity returned. These results demonstrate the persistence of reflex responses to stimulation in the isolated spinal cord. Similar recordings have been made from preparations maintained for more than 48 h in vitro.

The body temperature of a newborn opossum is probably close to that of its mother from which it suckles, approximately 32°C. All the previous experiments were made at room temperature (23°C). Raising the temperature to 28°C invariably approximately halved the amplitude of reflex responses recorded in ventral roots. At the same time synchronous spontaneous action potentials ceased or became lower in frequency. These effects were promptly reversed by cooling to 23°C once again.

Spontaneous, rhythmical bursts of activity were recorded from ventral roots in cervical and thoracic segments of isolated opossum nervous systems in Krebs’ fluid. Fig. 8A shows such rhythmical neural activity on a slow time base. Rhythmical bursts of this type continued under our present experimental conditions with Krebs’ fluid at 23°C for at least 8h after the CNS had been isolated. Systematic searches for rhythmical activity were not made at later times. In ‘semi-intact’ preparations in which several ribs were left attached to the spinal cord with their innervation intact, rhythmical movements were recorded by suction electrodes applied to intercostal muscles. Bursts of activity in contralateral ventral roots were correlated with contractions of the rib muscles. Spontaneous neural activity in the interburst interval resulted in only small movements; strong, synchronized muscle contractions were evoked during the burst. Fig. 8 shows recordings from a ventral root corresponding to thoracic segment T2 contralateral to the ribs and from intercostal muscles between segments T5 and T6. Rhythmical activity from the ventral root preceded activity in the rib muscles. Both interburst and burst activity were reduced by raising the temperature from 23 to 28°C.

An index of the viability of the isolated CNS of the opossum is provided by our finding that compound action potentials can be evoked by stimulation of the CNS maintained in vitro for up to 4 days in Krebs’ fluid. Many evoked action potentials were conducted across synapses, as shown by the blocking effects of high Mg²⁺ concentrations. The fastest action potentials that were recorded in normal Krebs’ fluid persisted in high-Mg²⁺ solutions and had a conduction velocity of 0.3 ms⁻¹, presumably reflecting lack of myelination and small fibre diameters. Mg²⁺, as well as TTX, diffused rapidly through the tissues from the bathing fluid, as did BrdU. The half-time of diffusion of approximately 5 min for T1X and Mg²⁻¹⁻ seems short, especially since no attempts were made to remove the pia mater. An unexpected result was the quick reversibility of TTX action. Other mammalian neuronal sodium channels in the peripheral or the central nervous system do not usually show such rapid recovery after washing out TTX. Recently, R. R. Stewart, D.-J. Zou and J. G. Nicholls have been able to prolong survival with excellent through conduction for more than 7 days by bathing the CNS in enriched media under sterile conditions (unpublished observations).

Spinal reflexes and spontaneous rhythmical activity also persisted in the isolated CNS for several hours. In this respect the opossum brain resembles other preparations, such as freshly dissected, isolated brainstem-spinal cord preparations of the neonatal rat and neonatal rat brain slices in which reflexes and respiratory activity are apparent (Suzue, 1984; Otsuka and Yanagisawa, 1988; Feldman and Smith, 1989; Greer et al. 1989). The rib movements in the opossum preparations in culture presumably represented existing respiratory commands issuing from the CNS. In a suitable medium it seems likely that the activity would be maintained for longer periods. This end-point has not yet been determined; nor have we investigated the viability of isolated preparations removed from animals older than 4 days, by which time the CNS appears larger and better formed, with an obvious cerebellum.

Spontaneous and evoked activity became brisker and stronger at lowered ambient temperatures. At the behavioural level it is not unreasonable to suppose that falling off the mother must represent a disastrous threat to the pup. If, as a result of the drop in body temperature, neuronal activity increased, this could help survival by producing stronger movements towards the home base, at least temporarily. In sensory systems, such as muscle spindles, lowered temperature also increases spontaneous activity (Lippold et al. 1960).

A remarkable result was that the structure of the CNS observed by light microscopy appeared normal for several days after isolation. Moreover, blast cells in the proliferative zone continued to divide for up to 4 days in a medium as deficient as Krebs’ fluid. Dividing cells were abundant, but our results do not show whether the progeny of the blast cells continued their normal differentiation and migration or whether they died without reaching their normal targets. Recent results by Temple (1989) indicate that, under appropriate conditions in microculture, single blast cells can continue to divide and their progeny to differentiate. Similarly, although the BrdU technique clearly labelled blast cells produced in the CNS days or hours after isolation, we do not know whether BrdU itself interfered with migration and differentiation. There is evidence that DNA is altered by BrdU (Stubblefield, 1975) and that cell differentiation can be blocked (Tapscott et al. 1989); nevertheless, preliminary experiments made with opossum preparations indicate that after BrdU incorporation at least some blast cell progeny continue to be produced and that they can migrate to more lateral locations in the spinal cord (D.-J. Zou and R. R. Stewart, unpublished results). Tests are now being made to compare birth-dating of cells using BrdU and [³H]thymidine.

We conclude that the isolated opossum CNS offers distinct advantages for physiological, pharmacological and molecular studies of circuitry and development. Unlike other CNS preparations, the entire nervous system is removed with the various parts in their normal relationship. Moreover, it shows unusual hardiness by surviving for so long in Krebs’ fluid, which represents a minimal medium, containing as it does only salts and glucose. With preparations bathed in media better suited for continued electrical activity, cell division, growth and metabolism, it is already becoming possible to aim for prolonged survival times and for development of CNS structures in vitro. The delayed maturation of caudal regions of the cord presents opportunities for comparing development of motor neurones as the limbs form. In addition to studies on development, the preparation holds promise for analysis of respiratory mechanisms. This simplified CNS preparation is not only accessible to drugs, with its major structures intact, but has no cerebellum overlying its fourth ventricle and is undisturbed by pulsations from blood vessels. Our results suggest that the opossum brain may provide an opportunity for studying the cellular basis for rhythmical activity in a manner similar to that exploited in simple invertebrate preparations (Stent and Kristan, 1981).

We thank Colin Bunce and Chris Snart at the Boldrewood Animal House in Southampton and, in Basel, Eric Hardman, Sonia Chitvanni and Beat Christen for their outstanding efforts in starting and maintaining our colony of Monodelphis. Special thanks to our colleague Dr W. B. Adams for his essential help with the electronics, the design of the project and for reading the manuscript. We are grateful to Ms J. Wittker for her secretarial assistance and to Mr P. Baettig for photographing the figures. We also thank C. Baptista and T. Gershon in the Macagno laboratory at Columbia University for giving us their protocol on BrdU labelling. SDE was supported by a Fogarty Senior International Fellowship and a US Public Health Service Grant (NS 12211).