INTRODUCTION

Top Abstract Introduction Discussion References

THE ADVENT OF TRANSGENIC AND GENE KNOCKOUT technologies in the past decade has made it possible to create mutant mice with specific genetic modifications as models of various diseases in humans (7, 22). Unfortunately, in some instances, genetic manipulation may provoke unanticipated or undesired side effects that may render the mutants nonviable. As an example, several strains of mutant mice that lack functional N-methyl-D-aspartate (NMDA) receptors have recently been created to expressly investigate the role of this receptor in synaptic transmission, synaptic plasticity, and neural development in the central nervous system (5, 14, 15). One unexpected phenotype of this mutation was that the mutant animals were found to suffer severe respiratory depression shortly after birth (19), and all of the mutant animals eventually died of cyanotic respiratory failure within the first day of life (5, 14, 15). The premature deaths of these mutant animals preclude further studies of the effects of the genetic mutation on postnatal neural development and the corresponding behavioral and physiological consequences throughout infancy and adulthood. Although such difficulties may be circumvented to some extent by the use of emerging technologies such as subregion and cell-type-specific gene knockout (26), this approach is not satisfactory if the brain region of interest directly or indirectly affects respiration and/or other vital functions. Furthermore, mutations of the NMDA-receptor-related genes and other genes (2, 4, 9) that affect respiration are of particular interest in elucidating the mechanisms of respiratory control. Resuscitation of these mutant mice calls for mechanical respiratory support to reverse the symptoms of respiratory failure. Until now, however, there was no mechanical respirator capable of ventilating animals as small as the newborn mouse.

The need for such a ventilator is also prompted by the recent development of several strains of transgenic mice as models of various lung diseases such as pulmonary emphysema (3) and fibrosis (13) and alveolar proteinosis (16). To facilitate the testing of various therapeutic regimens for the management of these lung diseases, it has become necessary to provide mechanical ventilation to these mutant animals during therapeutic treatment. Because the expression of the corresponding phenotypes may have a variable onset, a means of ventilatory support for both neonatal and adult mice is needed.

In this paper, we describe the design and experimental testing of a miniature and versatile mechanical ventilator that is applicable to the newborn and adult mice as well as to other small animals. The miniature ventilator is capable of delivering various ventilatory modes, such as conventional mechanical ventilation (CMV) and high-frequency ventilation (HFV), and circumvents certain difficulties in ventilating such small animals.

	DESIGN RATIONALE

Inadequacies of current ventilators. Current techniques of artificial mechanical ventilation are not suitable for small animals such as newborn mice. Most commercially available human or animal ventilators inflate the lungs by delivering a positive pressure at the airways opening. This approach generally requires endotracheal intubation that may be too invasive and impracticable in the fragile and tiny newborns. Also, endotracheal intubation is problematic in human neonates because some leaks must be allowed in order to minimize the risk of subglottic stenosis (25). A further complication when dealing with newborn mice is the minute tidal volumes involved (5-20 µl), which cannot be accurately measured with ordinary flow or volume transducers or reliably delivered with currently available small-animal ventilators.

Another concern with conventional animal ventilators is that they often require general anesthesia and muscle paralysis to suppress any spontaneous respiratory efforts that may interfere with ventilator operation. Although such difficulties may be circumvented to some extent by the use of spontaneously triggered assisted mechnical ventilation modes (17, 18, 20), the extremely small flow and volume signals encountered in newborn mice make it impossible to match the ventilator support to spontaneous respiratory efforts of the animal.

Proposed ventilator design. In light of the above, we propose a miniature high-frequency negative-pressure mechanical respirator for the ventilation of small newborn animals. This approach combines the classic "iron lung" (body surface negative-pressure ventilation) method of mechanical ventilation with the modern HFV technique to yield a ventilation mode that is minimally invasive and minimally intrusive. Body surface negative-pressure ventilation is ideal for newborns because the only physical encumbrance to the subject is a noninvasive external cuff around the neck region, as opposed to an intratracheal tube. This mode of mechanical ventilation has proved useful in ventilating human neonates (24). It has been suggested that body surface negative-pressure ventilation may be as effective as positive-pressure tracheal methods in allowing pulmonary gas exchange and may even have improved hemodynamic consequences (21).

HFV overcomes the problem of patient-ventilator dysynchrony by producing a small-amplitude, rapidly fluctuating pressure at a frequency much higher than the normal frequency. This ventilation mode, which may be superimposed on the animal's innate respiratory pattern, has been shown to provide adequate pulmonary gas exchange in larger animals (1). Furthermore, high-frequency oscillation of the pressure wave also helps to minimize the effect of leaks in the system, which could be a serious problem for the minute tidal volumes involved.

In a previous study by Harf et al. (8), the combined techniques of HFV and body surface negative-pressure ventilation were shown to be effective in ventilating adult rats. In that method, however, the body pressure was oscillated about atmospheric pressure, resulting in both positive- and negative-pressure swings. Positive surface pressures tend to compress the lungs and airways and may have deleterious effects in the diseased lung. Moreover, such a setup is not satisfactory for the present purpose, considering the size difference between an adult rat (~370 g) and a newborn mouse (~1.3 g).

Design specifications. To meet the ventilatory demands of the newborn mouse under varying pathophysiological states, the miniature ventilator was designed to deliver both CMV and HFV modes with variable tidal volumes ranging from 0 to 0.04 ml over a frequency range from <1 Hz (in CMV mode) to >50 Hz (HFV mode). Special care was given in the design to allow for a continuous resetting of the end-expiratory pressures to a preset level.

To determine an upper limit for the negative body surface pressure, we use the following simplified equation of motion

(1)

where is the instantaneous flow; t₁ and t₂ are the times at the beginning and end of inspiration, respectively; Rrs and Ers are the total respiratory resistance and elastance, respectively; Pbs is the body surface pressure; and Paw is the airway pressure. During negative-pressure ventilation, Paw remains constant at atmospheric pressure throughout the respiratory cycle. For newborn mice, the parameters Rrs and Ers are estimated to be 2.31 cmH₂O · l¹ · s and 554 ml/cmH₂O, respectively, when general allometric scaling formulas derived from various mammalian species are used (23).

Assuming Pbs to be a sinusoidal negative gauge pressure of given amplitude, Eq. 1 may be solved by using direct numeric integration or Laplace transform techniques. For a maximum tidal volume of 0.04 ml, the corresponding maximum pressure swing for Pbs is thus estimated to be ~25 cmH₂O.

	DESIGN DESCRIPTION

Figure 1 shows a full schematic of the miniature mechanical ventilator. The device comprises four major subsystems: the pressure-generating piston assembly, pressure-normalizing piston assembly, motor and controller, and the mouse chamber. The design of each subsystem is detailed below.

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Schematic of miniature ventilator for newborn mice. A special feature of ventilator is the double-piston design that allows for pressure generation during inspiratory phase and pressure normalization during expiratory phase. See text for detailed description.

Pressure-generating piston assembly. The pressure-generating piston assembly consists of a piston fixed translationally at one end via flexible tubing and to a radially adjustable pin (for stroke length adjustments) mounted on an aluminum disk. By rotating the disk at a constant velocity, the piston cycles in a sinusoidal motion.

The assembly dimensions were found by constructing a mathematical model of the piston system. The model assumed that air behaves as an isothermal ideal gas, that there was an imposed sinusoidal piston motion, and that Poiseville and couvette leakage could occur through the piston gap. These considerations led to a system of differential equations in pressure, volume, and chamber air mass. Details of the model description are found in a previous report (12).

The theoretical model was then numerically solved by using a third-order Runga Kutta integration routine available in the computer software package MATLAB (MathWorks, Natick, MA). The variables that could be adjusted were the stroke length, the shaft radius and length, the piston cylinder radius and length, and the volume of the mouse chamber. Each parameter needed to be considered to determine suitable dimensions.

The shaft/cylinder gap size (ideally 0 for no leakage) was limited by achievable tolerances and cost considerations. Another concern was that the stroke length needed to be small (due to size concerns and jamming problems dependent on the piston length) yet large enough for discernible radial piston positions to allow the pressure to be varied accurately between the 0- to 25-cmH₂O range. Finally, by making the shaft and cylinder longer than the stroke length (thereby lengthening the gap), the leakage could be reduced by increasing resistance to Poiseville flow. Again, the lengths could not be too long for practical considerations.

With these limitations, the piston radius and chamber volume were adjusted to ensure that the maximum pressure of 25 cmH₂O could be achieved. After several simulation runs, suitable parameters were obtained (piston shaft radius = 0.00245 m; piston cylinder radius = 0.00255 m; stroke length = 3 cm; nominal gap length = 5 cm; chamber volume = 15 ml).

Typical simulation runs are given in Fig. 2. By comparing the high-gap and no-gap scenarios, it can be seen that leakage reduces the pressure swing attainable for a given piston. Because all systemic leakage was not modeled, it is necessary that the piston be overdesigned with a relatively large safety margin to ensure the full 25-cmH₂O pressure specification.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Model predictions of pressure waveform in mouse chamber corresponding to a gap size of 0.0 cm (no gap; A) and 0.06 cm (high gap; B). Simulation parameters are as follows: stroke frequency, 2 Hz; stroke length, 3 cm; nominal gap length, 5 cm.

Additionally, with a negative pressure being generated during the inspiratory phase, there is always mass transfer into the chamber due to leakage. Therefore, on recoil of the piston during the expiratory phase, a positive pressure would be generated, causing the subsequent swing to start from an elevated level. Eventually, this baseline shift settles as the gauge pressure oscillates around 0 cmH₂O. The time constant of this shift is determined by the size of the leakage. Such a pressure shift is detrimental to ventilation in two ways. First, the upward shift means the desired negative pressure cannot be reached without changing the piston parameters. More importantly, the positive pressure generated (which is equal in magnitude to the negative pressure on equilibration) serves to compress the body cavity, thereby closing the animal's airways. In the present design, this baseline shift is corrected by a pressure-normalizing valve, as described below.

Pressure-normalizing piston assembly. In larger ventilators, the leakage is not a major concern, since the tidal volumes are much larger than the mass transfer due to leakage. However, the small volume in neonatal mouse respiration makes it imperative that the leakage problem be considered.

To eliminate the effects of mass transfer, the chamber pressure must be normalized with each cycle. This was accomplished via a cam valve assembly (Fig. 1), which includes a cam and a hollow shaft closed at both ends. The shaft is concentric with a cylindrical port extending 0.7 cm into the mouse chamber. The camshaft has an inlet slit 0.7 cm from the end proximal to the mouse chamber and another outlet on the length of shaft outside the chamber. The cam, which is coaxial to the pressure-generating piston disk, drives the shaft providing a 0.7-cm stroke. It is designed to provide the maximal stroke throughout the entire positive-pressure phase of the pressure-generating piston while retracting during the negative phase. This action allows the chamber pressure to equilibrate with atmospheric pressure throughout the positive phase.

The cam is designed to have a maximum outer radius of 2 cm, equal to the maximum radial offset of the pressure piston (a 4-cm stroke was allowed for the prototype design). This guarantees size consistency regardless of later design changes. Another requirement is that the cam stroke needs to be as large as possible to minimize leakage through the normalizing valve slit during the negative-pressure swing (the larger the stroke, the further the slit could be placed from the chamber end, thereby increasing resistance to mass transfer). A limiting factor on the stroke length, other than the overall size limit of 2 cm on the cam radius, is the need to keep the cam pressure angle <30° (10). With this in mind, a cam with a base circle of 1.3 cm was chosen (0.7-cm stroke), which offered a pressure angle of 28.5°.

A cam follower (essentially a flat surface perpendicular to the shaft) provides the contact surface between the shaft and cam. The follower is guided by tracks to keep undesirable force components from bending the shaft.

A spring with a sufficiently large spring constant was used to ensure that the cam follower would remain in contact with the cam surface. This arrangement would reduce the chattering and wear of the cam assembly and, moreover, would keep the valve timing in synchronization with the pressure-generating piston.

Motor/controller. A servobrush motor (Kollmorgan U9M4T) is used to drive the ventilator. The motor is velocity controlled with a tachometer feedback by an amplifier (KXA-48). The motor shaft is connected to a steel shaft, which drives the cam and pressure piston assemblies via an aluminum joiner. Both shafts are press fit in the joiner and held snugly with screws. The 3-mm shaft rotates in two greased ball bearings that are separated by a 2-cm gap.

Motor selection depended on the ability to provide adequate torque over the 0- to 100-Hz operating range. From the inertial, damping, and restoring forces for the pressure-generating and pressure-normalizing piston systems, the maximum and minimum torques in a given cycle were determined to be between 0.1 and 0.06 Nm. To ensure a constant rotational frequency with this torque variation, the Kollmorgan U9M4T DC servobrush motor was chosen to provide a continuous torque rating of more than three times the torque variance with a small mechanical time constant. The motor can operate continuously over a range from 0 to 4,000 revolutions/min (67 Hz), providing a torque of ~40 oz · in. (0.28 Nm) with no ventilation. With proper cooling and power amplification, stroke frequencies of up to 100 Hz could be reached.

Mouse chamber. The mouse chamber consists of two compartments. The first is a 3-ml plastic cylinder that is detachable from the apparatus. It is connected via a three-way valve to the pressure-generating piston assembly. The unconnected end allows a neonatal mouse to be positioned inside the cylinder. To stop leakage, a rubber cuff is first placed around the animal's neck and then sealed with a nontoxic dental adhesive. Then, the mouse's body is placed within the chamber, with the rubber fitting snugly in place.

A second compartment is connected to the third stem of the three-way valve and houses the chamber end of the normalizing valve. This compartment has an adjustable volume of 20 ml, which acts as a dead space. By varying the dead-space volume, the mouse chamber pressure can be adjusted to some extent while the ventilator is in operation, although adjustment of the stroke length is necessary if larger pressure calibrations are desired. A solid-state sensor (Motorola MPX 2010) is connected to the chamber and allows for real-time monitoring of the pressure signal.

The required size of the mouse chamber was determined in conjunction with the pressure-generating piston design. The latter required a chamber volume of 15 ml, considering the parameters chosen (piston dimensions and stroke length). The actual chamber as designed is adjustable from ~5 to 25 ml. This wide range was selected, since the theoretical model assumes that leakage occurs only through the piston gap. In reality, other leakage sources (such as via the neck cuff) may be present. Therefore, a variable volume allows for pressure calibration. Additionally, the adjustable volume allows for the ventilator pressures to be varied while in operation, since stroke length adjustment would require the motor to be stopped. Finally, a large dead space in the chamber may help to suppress the pressure fluctuations created by the valve operation.

	PERFORMANCE

Mechanical testing. To test the pressure range of the ventilator, the mouse chamber orifice was sealed with a stopper, and the pressure profile during operation was sampled with a data-acquisition program (CODAS; Dataq Instruments, Akron, OH) on an IBM-486 computer. Figure 3A shows the ventilator operation at 4-Hz frequency and a pressure swing of 22.5 cmH₂O. The function of the normalizing valve can be seen from the flat regions in the pressure profile, where a positive pressure would otherwise be generated. Figure 3B shows the pressure waveform at a frequency of 40 Hz and a pressure swing of 3.5 cmH₂O (corresponding to HFV). Again, the normalizing valve effectively keeps the baseline pressure from drifting.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Pressure profiles of mouse chamber during operation of ventilator at low frequency and high pressure (4 Hz; 22.5 cmH2O; A) and high frequency and low pressure (40 Hz; 3.5 cmH2O; B). Note that pressure was effectively normalized at 0 cmH2O in expiratory phase.

The need for the normalizing valve is further demonstrated in Fig. 4, which reveals the baseline drift as predicted in the theoretical simulations. After the valve was abruptly plugged during normal operation, the baseline pressure gradually shifted upward, reaching a new equilibrium after 0.3 s. The relatively small time constant indicated that there were other sources of leakage in the system apart from the modeled gap. Indeed, after the valve was plugged, the amplitude of the pressure wave increased, suggesting that some of the leak was from the valve itself.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Sample pressure plot indicating baseline drift when normalizing valve was abruptly plugged at 0.4 s.

The function of the 20-ml dead space can be seen in Fig. 5. Initially, the connection to the 20-ml dead space was clamped. On a sudden reconnection of the dead space the resulting drop in pressure can be seen. Under the given conditions, the pressure was nearly halved. As discussed above, this mechanism can be used for calibrating and adjusting the pressure wave independently while the ventilator is in operation.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Effect of dead-space volume on pressure wave. Dead space was initially disconnected from mouse chamber and was abruptly added at t = 0.45 s.

In vivo testing. We have tested the performance of the miniature respirator in ventilating newborn mice in postnatal day 0. A strain of mutant mice with targeted deletion of the R1 subunits of the NMDA receptor (NMDAR1 mutant mice) (15) was used as the experimental model. Typically, the homozygous mutants exhibited marked respiratory depression and died of respiratory failure within 20 h after birth. During the experiment, the mutant animals were differentiated from the normal (wild-type and heterozygous) animals by their characteristically low respiratory frequency and the absence of milk in the stomach. At the end of the experiment, the genotypes of the mutant and normal animals were verified by cutting a piece of the animal tail under metafane anesthesia. Tail DNA was then extracted and genotyped by polymerase chain reaction analysis with suitable primers, as described previously (15).

Testing of the miniature ventilator was performed on both normal (control) and NMDAR1 mutant mice. The primary purpose of these tests was to ascertain whether the ventilator could prolong the lives of the mutants. The normal controls in the preliminary test ensured that the ventilator per se was not harmful to the mice. As expected, all the normal mice tested (n = 3) survived equally well on the ventilator as without ventilator.

Two litters of newborn mice of mixed (Mendelian) genotypes were used. In each trial, one mutant mouse was placed in the device, and a mutant littermate was used as pairing control. Ventilatory frequencies between 10 and 20 Hz, with pressure swings of 5-6 cmH₂O, were used for these trials. Figure 6A shows sample recordings on a newborn mutant mouse with ventilation parameters of 10 Hz and 5 cmH₂O. The ventilator pressure was negative during inspiration and was maintained at zero by the normalizing valve during expiration, in agreement with the simulation runs.

View larger version (28K): [in this window] [in a new window]   Fig. 6.   In vivo testing of mouse ventilator. A: pressure profile in an R1 subunit of N-methyl-D-aspartate receptor mutant mouse. Ventilator was set at 10 Hz, 5 cmH2O. B: life spans of 2 mutant mice on ventilator were significantly prolonged compared with their mutant littermates. Dashed line indicates maximum life span of all mutants previously reported (see Ref. 15).

As shown in Fig. 6B, the life span of the mutants was prolonged with ventilatory assistance. In both instances, the ventilated mutant mice outlived their unventilated mutant littermates, surviving beyond the absolute life expectancy of 20 h in these mutant animals. Indeed, the mutants died only after being weaned from the ventilator.

	DISCUSSION AND CONCLUSIONS

Top Abstract Introduction Discussion References

We have presented the design and testing of a miniature ventilator suitable for the mechanical ventilation of animals as small as the newborn mouse. The ventilator requires minimally invasive and minimally intrusive experimental procedures without the need for any surgery or intubation. Furthermore, it uses commonly available pressure transducers for measurement and calibration of its operation, without the need for highly sensitive flow or volume sensors.

The ventilator is unique in several ways. Because the device can achieve large pressure fluctuations over a wide frequency range, it is effective in both normal and HFV modes. Such adaptability is needed, since newborns of this nature have never been tested before, and the proper ventilation techniques are not certain. Additionally, the simple double-piston design (pressure piston and valve piston) effectively normalizes the chamber pressure, allowing negative pressures to be generated efficiently. The valve mechanism also keeps positive pressures from developing around the body surface, which would be detrimental to respiratory efforts.

Our intitial test results have indicated that the life span of the ventilated NMDAR1 knockout mice could be prolonged for >10 h. Still, more studies must be performed to determine the optimal ventilatory parameters for such newborn mice and to verify the long-term consequences of this form of mechanical ventilation.

The present method of negative-pressure ventilation requires that a cuff be placed around the neck of the newborn animal. Although this technique is much less invasive than a tracheotomy, improper placement of the cuff may run the risk of compressing the animal's airways, thereby increasing the airway resistance. A possible alternative is to place the cuff around the torso and simply oscillate the animal's abdomen. Previous studies have shown this to be a highly effective method of negative-pressure ventilation (1).

Another important consideration is temperature control in the neonates. In the present study, unventilated mice were kept in an incubator, since the mechanisms for body temperature regulation are not well developed at such a young age. Inclusion of automatic temperature-control capability in the mouse chamber is straightforward and requires only minor modifications in its design.

Finally, in the present setting of body surface mechanical ventilation, positive pressures tend to close the airways, whereas negative pressures serve to open them. Therefore, a negative-pressure source connected to the outlet of the normalizing valve would create a negative end-expiratory pressure, helping to open the airways. This strategy is analogous to the use of positive end-expiratory pressures in modern positive-pressure ventilators and may help to improve arterial oxygenation in animals with diseased lungs (11).