Vol. 86, Issue 3, 887-894, March 1999
Efficacy of recombinant human Hb by
31P-NMR during isovolemic
total exchange transfusion
Laurel O.
Sillerud1,
Arvind
Caprihan1,
Nancy
Berton2, and
Gary J.
Rosenthal2
1 The Lovelace Institutes,
Albuquerque, New Mexico 87108; and
2 Somatogen Incorporated, Boulder,
Colorado 80301
 |
ABSTRACT |
The ability of
recombinant human Hb (rHb1.1), which is being developed as an oxygen
therapeutic, to support metabolism was measured by in vivo
31P-NMR surface coil spectroscopy
of the rat abdomen in control animals and in animals subjected to
isovolemic exchange transfusion to hematocrit of <3% with human
serum albumin or 5 g/dl rHb1.1. No significant changes in metabolite
levels were observed in control animals for up to 6 h. The
albumin-exchange experiments, however, resulted in a more than
eightfold increase in Pi and a
50% drop in phosphocreatine and ATP within 40 min. The tissue pH
dropped from 7.4 to 6.8. The decrease in high-energy phosphates obeyed Michaelis-Menten kinetics, with a Michaelis-Menten constant of 3% as
the hematocrit at which a 50% drop in high-energy phosphates was
observed. Exchange transfusion with rHb1.1 resulted in no significant
drop in high-energy phosphates, no rise in
Pi, and no change in tissue pH
from 7.35 ± 0.15 for up to 5 h after exchange. By these criteria,
rHb1.1 at a plasma Hb concentration of ~5 g/dl after total exchange
transfusion was able to sustain energy metabolism of gut tissue at
levels indistinguishable from control rats with a threefold higher
total Hb level in erythrocytes.
oxygen therapeutics; high-energy phosphate; gut metabolism; surface
coil spectroscopy
| |
INTRODUCTION |
CONTAMINATION OF THE HUMAN blood supply with human
immunodeficiency virus, hepatitis, and other pathogens has underscored the need for products that perform as well as blood but do not carry
these attendant risks. Previous attempts to develop oxygen carriers to
avoid transfusion have utilized a number of alternative preparations,
including synthetic perfluorocarbon emulsions (8, 11, 18) or
stroma-free bovine Hb (4, 21) solutions. Current attempts to replace
blood with a recombinant human Hb (5, 10, 17) that would carry neither
antigenic nor infectious risks leave open the following questions: Can
such a molecule deliver oxygen as well as whole blood? Is it
efficacious in maintaining normal tissue metabolism? These questions
have been difficult to address at the cellular level.
31P-NMR spectroscopy is a
sensitive, noninvasive probe of oxygen transport to, and the
bioenergetic status of, cells, tissues, and organs (1, 2, 6, 7, 9,
12-16, 19, 20) that measures the amounts of compounds
[phosphocreatine (PCr), Pi, and nucleotide triphosphates (mainly ATP)] involved in oxidative energy metabolism in tissues. For example,
Pi is the low-energy degradation
product of phosphorus metabolism that accumulates during hypoxia or
ischemia (2, 6, 12, 13, 19).
We applied 31P-NMR spectroscopy,
in real time, to monitor the rat abdomen before, during, and after
complete blood replacement, by means of isovolemic exchange
transfusions, to determine the efficacy of a physiological preparation
of recombinant human Hb (rHb1.1, Somatogen, Boulder, CO). As controls,
we also examined the 31P-NMR
spectra of rats during exchange transfusion with a solution containing
human serum albumin (HSA) and no oxygen carrier and of rats having
undergone only sham cannulation and no exchange. The hypothesis tested
in this study was that rHb1.1 is as efficacious as blood in maintaining
gastrointestinal tissue metabolism in anesthetized rats.
| |
METHODS |
Animals.
A total of 10 Sprague-Dawley rats of either gender (Hilltop Lab
Animals, Scottdale, PA, or animals bred and raised at The Lovelace
Institutes, Albuquerque, NM) were anesthetized with pentobarbital sodium (50 mg/kg; Nembutal, Abbott Laboratories, Chicago, IL), weighed
(283-552 g), and cannulated via the femoral artery and vein with
use of saline-filled polyethylene tubing (PE-10, Clay-Adams, Parsippany, NJ; 0.012 in. ID, 0.025 in. OD). Animals were drawn from
the animal facility and randomly assigned to one of three groups by the
experimenters: control (n = 2), HSA
(n = 3), and rHb1.1
(n = 5). Animals were then placed on a
circulating-water heating pad at 38°C in the bore of the magnet for
31P-NMR spectroscopy measurements.
Additional anesthetic was administered as needed during the course of
the experiments to prevent the animals from feeling pain or moving
during data collection. The need for additional anesthetic was
determined by watching for standard signs of awakening, such as
hindlimb or facial muscle twitching.
Exchange transfusion.
The cannulas were flushed with heparinized saline and connected to a
dual-channel peristaltic pump (Rabbit, Rainin Instruments, Woburn, MA)
set to a nominal speed of ~1 ml/min. When exchange was desired, the
Hb solution (n = 5) or isotonic HSA
(to which 1 U/ml sodium heparin was added,
n = 3) was pumped into the venous cannula, and blood was removed from the animal by pumping via the
arterial cannula. The outflow was directed into a graduated cylinder,
and the volume was measured every 3-5 min during the exchange.
After ~45 min the hematocrit reached our target value of <3%, and
the pump was stopped.
Hematocrit determination.
Blood samples were removed periodically from the arterial catheter,
placed into 30-mm heparinized microhematocrit tubes sealed with 2.6-mm
Kwik-Seal plugs, and sedimented in a clinical microhematocrit centrifuge for 120 s. The hematocrits were determined from the length
of the column of packed erythrocytes compared with the total blood
column length and are expressed as percentages. The visual precision of
hematocrit determination was estimated to be 0.5 mm or ±1.7%. The
initial hematocrits were 56 ± 4%
(n = 5) for the HSA and control groups
and 54 ± 7% (n = 5) for the rHb1.1 group.
rHb1.1 and other reagents.
rHb1.1 (Somatogen, Boulder, CO) was shipped frozen, stored at
70°F, and thawed in the dark, to avoid light-induced
methemoglobin formation, just before use as a 5 g/dl solution in 5 mM
PBS, pH 7.26 (10). Pentobarbital sodium (50 mg/ml), heparin (1,000 U/ml; SoloPak Laboratories, Elk Grove Village, IL), and HSA (5%
solution, Baxter Healthcare, Glendale, CA) were used as supplied before their expiration dates.
31P-NMR spectroscopy.
31P-NMR spectra were obtained with
the aid of a Nalorac Cryogenics (Martinez, CA) Quest console and
radiofrequency (rf) system and a 31-cm-bore, horizontal, 1.9-T magnet
(Oxford Instruments, Oxford, UK). Rats were placed prone in the magnet
over a 30-mm-diameter custom-built surface coil, which detected signals
from the liver, gastrointestinal tissue, abdominal musculature, and
diaphragm. The static magnetic field was shimmed with the animal in
place by inserting a tuning network that allowed the phosphorus coil to
be retuned for protons. The proton signals from the animal were
observed, and the field gradients were adjusted until the residual
water signal had a full width at half-maximum of <50 Hz. Control
31P-NMR spectra were acquired at
32.5 MHz in 5- to 20-min blocks for 
1 h. Then additional control data
were collected for up to 5 h or the blood was replaced with one of the
two solutions (rHb1.1 or HSA), and the
31P-NMR spectra of the target
organs were followed for 4-6 h. Spectra were accumulated
continuously before, during, and after the exchange. The animals were
weighed again at the end of the NMR procedure and found to have
maintained fluid balance within 2% of their initial weight during the exchange.
The free induction decays after a 35-µs rf pulse (~160° at the
surface of the coil) were collected into 2,000 data points with a 2-kHz
sweep (61.54 ppm) and a recycle time of 2 s. This recycle time slightly
attenuated the PCr signal, because it has a spin-lattice relaxation
time (T1) on the order of 2 s (1). The time-domain data from the
spectrometer's VAX computer (Digital Equipment) were transferred to a
SPARC-2 workstation (Sun Microsystems, Mountain View, CA) and converted
to NMRi (New Methods Research, Syracuse, NY) format, apodized with a
10-Hz filter, Fourier transformed, phased, and baseline corrected. The
peak areas, positions, and widths were determined for all the
resolvable peaks by fitting the signals to Lorentzians with use of NMRi
software. The nonlinear fitting procedure used a convergence limit of
10
6, which required
5-60 iterations before convergence. The reported integrals are
those provided by analytic integration of the Lorentzians from the fits
over a frequency range of five line widths. The widths (Table
1) of the signals (after correction for the
10-Hz line broadening) varied from spectrum to spectrum from 35 to
~60 Hz for the 
-phosphate of ATP, for example. Chemical shifts are reported relative to PCr at a chemical shift of 
= 2.35
ppm.
pH determination.
The pH was determined from the chemical shift difference (
, ppm)
between PCr and Pi according to
the following equation
![pH = p<IT>K</IT> + log [(&Dgr; − &Dgr;<SUB>min</SUB>)/(&Dgr;<SUB>max</SUB> − &Dgr;)]](/content/vol86/issue3/fulltext/887/img001.gif)
|
(1)
|
where
pK = 6.75, minimum (min) = 3.27 ppm, and maximum
(max) = 5.69 ppm. The
normal pH for whole rat blood (1) is 7.38 ± 0.11. This method gave
a resting pH of the normal human forearm muscles of a healthy man of
7.03 in accord with established findings (9). As shown in the control
NMR spectra (see Figs. 4 and 5), the
Pi resonance is small and not
easily assigned in every control spectrum, so we restricted our
reported pH determinations to those control spectra in which there was
an assignable Pi resonance. The
errors reported here for the pH values are purely statistical and arise
from variations in the chemical shift of the resolved Pi signals due to random
electrical noise. We found that the observed signal-to-noise ratio
imposed a statistical accuracy of ±0.10 ppm on the chemical shift
determination for the Pi signals,
and this results in an uncertainty of ±0.10 pH unit. The observed statistical variation in determined pH is similar to this value, as expected.
Signal localization.
To accurately determine the magnetic field profile of the surface coil
and the tissue region probed by the rf excitation, we used rf pulses of
lengths varying from 0 to 400° to perform one-dimensional NMR
imaging, along the surface coil axis, of a phosphoric acid phantom. The
intensity profile (not shown) displayed the expected
1/r3
(where r is the axial distance from the coil center) falloff of
the rf resulting from a 90° (20-µs) pulse and the deeper
penetration and surface suppression of a 160° (35-µs) pulse. The
use of a 160° pulse suppressed the signals from the abdominal
musculature within 8 mm of the coil. The localization provided by the
combination of the surface coil and the surface overpulsing was
sufficient to limit the signal acquisition region to that of the
abdomen, with signals arising from the gastrointestinal tissue, liver, and abdominal musculature.
The identity of the tissue that was probed by the NMR pulse sequence
was further investigated through proton NMR imaging of the abdomen
without repositioning the animal. When the resulting images (not shown)
were combined with the rf profile of the coil (not shown), the NMR
signals arose from ~10% liver, ~80% gastrointestinal tissue, and
~10% abdominal musculature. The liver does not contain PCr (3),
whereas the abdominal musculature and the gastrointestinal tissue
contain significant amounts of PCr. In previous studies of the rat
abdomen (6) we found that the
31P-NMR spectrum taken with a
surface coil placed on the surface of the surgically exposed
gastrointestinal tissue (with a 90° pulse) was indistinguishable
from a spectrum taken using a 200° pulse with a surface coil placed
over the intact abdomen of the rat. Furthermore, the exact origin of
the NMR signals is not a major concern here, because all tissues are
perfused with blood and it is the oxygen transport capacity of this
blood that supports the metabolism, the parameters of which we
ultimately measure by 31P-NMR spectroscopy.
The center of the surface coil was positioned under the midline of the
abdomen of the rat 26 mm caudal to the xyphoid process by using
anatomic landmarks. Our main concern is the time dependence of the
31P-NMR signals, rather than their
absolute magnitudes. Nevertheless, the reproducibility of the coil
placement is illustrated by the fact that the actual signal strengths
determined from the integrals of the phosphorus resonances varied from
animal to animal by <20%, which is only slightly more than the
observed statistical variation from spectrum to spectrum (see above).
In all the studies reported here, each animal served as its own
control, as far as the actual signal strengths were concerned, because
we took several control spectra of each animal before any experimental
interventions and used these control spectra to correct for
animal-to-animal variations in the total intensities of the signals.
| |
RESULTS |
Control 31P-NMR spectra.
A control 31P-NMR spectrum of the
rat abdomen (Fig. 1) displayed well-known
signals, which were fitted to a sum of Lorentzian functions, the
properties of which are given in Table 1, from phosphomonoesters,
Pi, phosphodiesters, PCr, and ATP.
The signal-to-noise ratio for the PCr signal in Fig. 1 is
~14, whereas that for the -ATP signal is ~6. These measurements
determine the statistical variation to be expected from spectrum to
spectrum in the signal integrals of 7 and 16% for PCr and ATP,
respectively. As expected, these signal-to-noise ratios were also found
to dominate the animal-to-animal variation and, therefore, constitute
the main source of error in this study.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 1.
Curve fitting of a control rat
31P-NMR spectrum to Lorentzian
line-shape functions. Acquisition parameters were 300 scans, total time
10 of min, repetition time of 2 s, and exponential filtering at 10 Hz.
Line-fit parameters are given in Table 1, along with peak assignments.
Individual Lorentzians, their sum, experimental spectrum (which is the
only one to contain noise), and difference (which should consist only
of residual noise) are shown.
|
|
The chemical shift of the small Pi
signal ( = 2.93 ppm) implied (Eq. 1) a pH of 7.44 (Fig. 1). The pH from 30 control
determinations in a total of 5 rats was 7.37 ± 0.14 (SD), which is
close to that reported for whole blood (see above) and is in agreement
with that reported (1) for tissue in the abdomen (7.35 ± 0.11).
The control NMR spectra from the rat abdomen were time independent
(Fig. 2) for the entire length of
observation in the magnet (see
METHODS). For example, the control
NMR spectrum in Fig. 2A (at 0 min) is
indistinguishable, within the signal-to-noise ratio, from the spectrum
shown in Fig. 2B (290 min later). The
above-quoted standard deviation (±0.14) for the pH determinations
represents the total variation of the control spectra over 290 min of
observation. In a similar manner, the total temporal variation in the
integrals of the line fits to the control phosphorus spectra was
limited to, e.g., ±11% for PCr and ±10% for -ATP for 30 determinations in a total of 5 animals over 290 min.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 2.
31P-NMR spectra of a control rat
abdomen. A: spectrum (10 min) taken
just after animal was placed into magnet bore at completion of proton
shimming. B: spectrum taken 290 min
after animal was placed into magnet bore. These spectra are
indistinguishable within limits of signal-to-noise ratio.
|
|
These measurements represent the background against which our data for
the HSA and rHb1.1 experiments are to be compared and establish the
limits on variability of the metabolite amounts detectable with the NMR
methods. In this manner the
31P-NMR data from control animals
serve as negative controls, in that only sham intervention was
performed. The possibility remains that
31P-NMR is incapable of detecting
metabolic stress in these animals. To establish the ability of the
chosen 31P-NMR methods to report
metabolic stress, we embarked on a series of positive control
experiments in which all the blood of three rats was replaced by
isovolemic exchange transfusion with HSA, which contains no added
oxygen carriers.
Exchange transfusion with HSA.
To positively determine that detectable changes in the
31P-NMR spectra of the rat abdomen
were associated with a reduction of oxygen delivery, we exchanged all
the blood of three rats at a nominal rate of 1.2 ml/min with a
physiological solution of HSA. The exchanged volume was a linear
function of time: an example linear fit to the data from a typical HSA
run gave a slope of 1.17 ± 0.02 ml/min. The residual hematocrit of
the blood, monitored in the outflow catheter, decreased exponentially
(half time ~ 8 min) with time so that when ~40 min had elapsed, the
residual hematocrit had dropped to <3% and the pump was switched off
(see METHODS for details).
The animals experienced respiratory arrest 43 ± 11 min (or ~5
exchange half times) after exchange initiation. The abdominal 31P-NMR spectrum was monitored
before the start of exchange, during the ~40 min of exchange
transfusion, and for an additional 40 min after the exchange was
terminated, because the spectra continue to change postmortem (see Fig.
4). The 31P-NMR spectrum (Fig.
3) of the animal before the exchange was identical to the control spectrum (Fig. 2), whereas that taken 40 min
after the start of the exchange was dominated by an increased Pi (>800%) signal (Fig.
4) along with a twofold drop in the PCr signal (Fig. 4)
and essentially identical behavior of the ATP signal (not shown).
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 3.
Effect of exchange transfusion with human serum albumin (HSA) on
31P-NMR spectrum of a rat abdomen.
Control spectrum taken 10 min before exchange and spectra taken at
succeeding 10-min intervals after start of HSA exchange are shown. Note
large increase in Pi ( ~ 2.5 ppm) and decrease in phosphocreatine (PCr, = 2.35
ppm) and ATP ( = 5, 10, and 18 ppm) as
erythrocyte population in blood was reduced. Exchange was completed at
40 min at a residual hematocrit of 1.6%. Initial pH determined from
chemical shift of Pi at 10
min was 7.6 ± 0.1; final pH was 6.8 ± 0.1. Rat died 41 min
after initiation of exchange process.
|
|
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 4.
Time course of integrals of
31P-NMR signals from
Pi and high-energy phosphates
(average of integrals of -ATP and PCr resonances) from rats that
were isovolemically exchange transfused with HSA. Exchange was begun at
0 min and continued until hematocrit reached <3% at ~40 min.
Values are means ± SD from 2-4 measurements at each time
point.
|
|
The PCr data shown in Fig. 4 represent the response of the
gastrointestinal tissue to hypoxia and are similar to the results found
for ischemia induced by ligation of the superior mesenteric artery reported earlier (14). The rise in
Pi and the drop in PCr result from
a shift to the right in the creatine kinase and ATPase
equilibria
|
(2)
|
where Cr is creatine. We will concentrate our results on the
behavior of PCr, because PCr serves to buffer, in a sense, the ATP
concentrations in the gastrointestinal tissue through the creatine
kinase reaction (Eq. 2) in the
smooth muscle. Changes in the amounts of PCr in the tissues of interest
can be expected to occur before or simultaneously with changes in ATP levels.
The animal's hematocrit, as well as the
31P-NMR spectra of the
gastrointestinal tissue, was monitored continuously during exchange with HSA in the magnet. This experiment, therefore, provides a unique
opportunity to correlate the observed changes in the
31P-NMR data with changes in the
hematocrit. Our results for Pi (Fig. 5) and for PCr (Fig.
6) show that as the hematocrit was lowered
from its normal value to ~25%, there was only a small, <10%, drop
in the average PCr 31P-NMR signal
(Fig. 6) and a modest, ~50%, rise in the
Pi resonance (Fig. 5). At
hematocrits less than ~25%, however, major changes occurred for both
metabolites. A quasi-exponential increase in Pi was observed (Fig. 5) as the
hematocrit approached zero, and the PCr began to plummet in a nonlinear
manner (Fig. 6).
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 5.
Effect of hematocrit on Pi
31P-NMR signals for rats
isovolemically exchange transfused with HSA ( ,
n = 2) or recombinant Hb (rHb1.1,  ,
n = 4). Data are presented as
percentages of control values at high hematocrit (~57%). For rats
transfused with HSA, Pi increase
with lowered hematocrit was fitted to an exponential of form:
Pi(h) = [Pi(0) Pi(57)]eh/5 + Pi(57), where h is hematocrit;
for rats transfused with rHb1.1, linear fit is discussed in
RESULTS.
|
|
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 6.
Relationship between hematocrit and high-energy phosphates (PCr
resonance) for rats that underwent isovolemic exchange transfusion with
HSA (, n = 2) or rHb1.1 (5 g/dl,
 , n = 4). Curve through
HSA data is a fit to Michaelis-Menten equation (Eq. 3), with
Vm = 104 ± 4% and Km = 2.8 ± 0.3%; line through rHb1.1 data is a linear least squares fit
with a slope of 0.119 ± 0.119 and an intercept of 107 ± 9%.
|
|
The PCr data (Fig. 6) show an apparent change of slope at a hematocrit
of ~25%: a linear fit for hematocrits ranging from normal to 25%
gave a slope of 0.017 ± 0.094 and an intercept of 101.3 ± 3.1%. The slope determined for a hematocrit of >25% was not
significantly different from zero, indicating that rats have considerable reserve capacity to withstand hemodilution up to a nearly
twofold reduction in hematocrit. The fact that the intercept was not
statistically distinguishable from 100% also indicates that there is
no effect of a twofold reduction in the hematocrit in anesthetized animals.
The observed nonlinear relationship between PCr concentration and
hematocrit (Fig. 6) is well described by kinetics of the Michaelis-Menten form
|
(3)
|
where
v(x)
is the PCr amount as a function of hematocrit,
Vm is the maximum
PCr (nominally 100% in this case), and
Km is the
apparent Michaelis-Menten constant (hematocrit) at which the amount of
PCr drops to 50% of its maximum value. The PCr data for the HSA
experiments in Fig. 6 were fitted to a Lineweaver-Burke plot, which
yielded a Vm of
104 ± 4% and a
Km of 2.8 ± 0.3%. This implies that anesthetized rats require a ~15-fold drop in
hematocrit to reduce oxidative phosphorylation to one-half of its
normal value.
As erythrocytes were removed from the circulation by exchange of HSA
for blood, the hematocrit dropped, oxygenation of the tissues declined,
and the tissues became acidotic (Fig. 7),
presumably from the glycolytic accumulation of lactic acid. The average
pH before the exchange at a hematocrit of 57% was 7.44 ± 0.12, and this fell below 6.8 as the hematocrit approached zero. In a pattern resembling that observed for PCr (Fig. 6; see above), there was little
change in pH for a hematocrit greater than ~25% and a marked drop
below this value. The HSA exchange transfusion produced a useful status
of tissue hypoxia easily detectable via
31P-NMR, which we can now compare
with exchange transfusion with a buffered solution of 5% rHb1.1.
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 7.
Relationship between tissue pH of rats during exchange transfusion with
HSA () and rHb1.1 (). Line through rHb1.1 data is a linear least
squares fit to data, with a slope of (2.1 ± 1.4) × 103 and an intercept of
7.35 ± 0.12 (see RESULTS). Curve
through HSA data is drawn to guide eye.
|
|
Exchange transfusion with rHb1.1.
The 31P-NMR spectra of the four
rats that underwent the exchange transfusion procedure with rHb1.1
(Figs. 8 and 9)
remained stable at control values for >5 h after exchange. There was
no time-dependent drop in the high-energy phosphates (e.g., PCr; Fig.
8) and no concomitant rise in Pi
(Fig. 8), in contrast to the behavior observed with HSA exchange
transfusion (Fig. 4). When linear least squares fits to the
time-dependent rHb1.1 data were performed (Fig. 8), only the slope of
the PCr curve was significantly different from zero:
Pi slope = (3.9 ± 10.8) × 103 integral
units/min; PCr slope = (38.1 ± 13.3) × 103 integral units/min. The
PCr data (Fig. 8) show a small increase with time after rHb1.1
exchange, in contrast to the decline seen with HSA (Fig. 4).
View larger version (33K):
[in this window]
[in a new window]
|
Fig. 8.
Time course of integrals of
31P-NMR signals from
Pi () and PCr () from 4 rats
that were isovolemically exchange tranfused with rHb1.1. Exchange was
begun at 0 min and continued until hematocrit reached <3% at ~40
min. Heavy black lines through data are least squares fits, with
Pi slope = 3.9 ± 10.8 × 103 integral
units/min and PCr slope = 38.1 ± 13.3 × 103 integral units/min.
Dashed lines without data points are HSA data replotted from Fig. 4 for
comparison.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 9.
31P-NMR spectra for a rat
isovolemically exchange transfused with rHb1.1 before exchange
(A) and 290 min after exchange
(B).
|
|
Just as there was no time-dependent effect of exchange transfusion with
rHb1.1 on the 31P-NMR spectra of
the gastrointestinal tissue, there was no significant deterioration of
the bioenergetic status of the animals when the 31P-NMR data were examined as a
function of the lowered hematocrit. Exchange transfusion with rHb1.1
had no significant effect on the PCr or
Pi concentrations in these animals
(Figs. 5, 6, 8, and 9). A linear least squares fit to the
Pi data as a function of the
hematocrit in Fig. 5 gave a slope of 0.248 ± 0.269 and an
intercept of 104 ± 6%, whereas fitting the PCr data (Fig. 6) gave
a slope of 0.119 ± 0.119 and an intercept of 107 ± 9%.
Neither of these slopes is significantly different from zero, and the intercepts are indistinguishable from 100% as well. There was no
significant bioenergetic deterioration, even when essentially all the
endogenous erythrocytes were removed from the circulation and the sole
oxygen carrier became the added rHb1.1.
A similar lack of effect of lowered hematocrit was observed on the pH
of the animals exchanged with rHb1.1 (Fig. 7). The average pH of 7.35 ± 0.15 derived by pooling the
Pi chemical shift data (n = 68) from all the rHb1.1-exchanged
rats is consistent with that previously reported (7.38 ± 0.11) (10)
and is identical to that found in the control animals (7.37 ± 0.11, see above). The slope of the line fitted to the hematocrit-dependent pH
data was (2.1 ± 1.4) × 103
min1, with an intercept of
7.33 ± 0.03. The fact that the pH remained constant at a baseline
value as the hematocrit approached zero indicates that rHb1.1 supports
normal tissue pH regulation, even when it completely replaces blood.
In addition, there was no significant change in the pH with time during
or after the exchange of rHb1.1 for the blood for up to 6 h. The slope
of the least squares line fitted to the time-dependent pH data (not
shown) and the total pH change were 6.71 ± 17.29 × 105
min1 and 0.08 ± 0.07, respectively. Neither the slope nor the total pH change over the entire
range of hematocrits from 0 to 57% was significantly different from zero.
| |
DISCUSSION |
The present experiments report the results of a systematic assessment
of the efficacy of a 5 g/dl physiological solution of rHb1.1 as an
oxygen carrier in support of whole animal bioenergetics in an
isovolemic total exchange transfusion model. The basis for these
studies is the ability of 31P-NMR
to directly monitor tissue metabolites, including PCr,
Pi, and ATP, in vivo as blood is
exchanged for other fluids. Support of high-energy phosphate levels
depends on oxygen delivery to the tissues (18), and monitoring the
31P-NMR spectrum is an excellent
method for assessing oxygen delivery and the ability of oxygen
therapeutics to sustain life in a whole animal approach (1, 2, 6, 7, 9,
12-16, 19, 20).
The studies reported here utilized control animals subjected to only
sham manipulation, animals in which the entire erythrocyte population
was replaced with HSA, and animals in which the entire erythrocyte
population was replaced with rHb1.1. The control animals served to
establish the background of statistical variation against which the
other two experimental groups could be compared and judged for
significance. The HSA group served to demonstrate that NMR methods
could indeed measure hypoxic stress in this setting. The group of
animals exchanged with rHb1.1 provided a direct test of the ability of
this preparation to supply oxygen to hypoxia-sensitive gastrointestinal tissues.
Prior non-NMR studies of oxygen therapeutics have utilized exchange
transfusion with, among other substances, perfluorocarbon emulsions and
have reported that rat brain pH is 7.36-7.58, depending on the
fraction of oxygen in the inspired gas (11). The survival time for rats
subjected to isovolemic exchange transfusion with perfluorocarbon
emulsions was 13 ± 2 h when the residual hematocrit was reduced to
<2% (11).
Our HSA experiments were designed to act as positive controls in the
sense that they revealed the behavior of the
31P-NMR spectra of the animal in
response to exchange hemodilution with a solution that was not expected
to be able to support life. They showed that the
31P-NMR spectrum of the abdomen is
capable of detecting tissue hypoxia and served to establish the
magnitude of the metabolic changes expected when tissue is perfused
with material containing no added oxygen carrier. These data
established a worst-case scenario, in that if rHb1.1 were completely
incapable of oxygen transport, then the
31P-NMR data from animals
exchanged with rHb1.1 would resemble data from the animals exchanged
with HSA. The control experiments in which only cannulation and no
exchange was performed established the other limit on the possible
performance of an oxygen therapeutic. Here, no change was made in the
ability of the circulatory system to transport oxygen, and the control
data establish just what "no change" means in a statistical
sense. If rHb1.1 were to perform as well as whole blood, then we would
expect to see no statistically significant difference between the
control spectra and the spectra from the animals exchanged with rHb1.1
as a function of time.
The observed response of the rat gastrointestinal tissue to hypoxia
produced via exchange transfusion with HSA is similar to that observed
in the rat gastrointestinal tissue (2, 6) during ischemia
produced through occlusion of the superior mesenteric artery, where it
was found that the PCr level dropped to zero and the ATP level dropped
by 70% after 30 min of ischemia. The pH fell from 7.3 to 6.7 over the course of 60 min, with an attendant rise in
Pi of 130% (2, 6).
Isovolemic exchange transfusion with rHb1.1 prevented deterioration in
all measured and calculated bioenergetic parameters. The
31P-NMR spectra (Fig. 9) before,
during, and after the exchange display no significant changes. There
was neither a rise in Pi nor a
drop in high-energy phosphates (Fig. 8) during or after the exchange
procedure. The pH was stable for >5 h at 7.35 ± 0.15, which is
identical to that found by Bittl et al. (1) in normal rats (7.38 ± 0.11). The amounts of high-energy phosphates were also independent of
hematocrit (Fig. 6). The most important facet of the PCr data derived
from exchange transfusion with rHb1.1 was the fact that the levels of
PCr did not decrease with time.
With respect to hemodynamics, it is unknown whether rats in the HSA or
the rHb1.1 group maintained a normal distribution of extracellular
volume during or after the exchange transfusions. It is known that none
of the animals gained or lost significant amounts of fluid, because the
weights of the animals did not change significantly (<2%) over the
course of the experiments (see
METHODS). Preservation of euvolemia
is not a major concern for the HSA experiments, since these studies
were designed to reveal the response of the animals to a serious, fatal
hypoxic stress. If euvolemia was not maintained during exchange with
HSA, then that constitutes a portion of the overall response to
hemodilution with HSA. In the case of the rHb1.1 rats, it is clear from
the observation of no detectable tissue hypoxia that if euvolemia was
not maintained, this had no metabolic consequences in terms of
phosphorus compounds measured with NMR.
One possible mechanism for the animals to compensate for hemodilution
is deposition of additional erythrocytes into the circulation by
contraction of the spleen. It is obvious that this mechanism cannot
provide significant amounts of erythrocytes during hemodilution with
HSA, since the animals are not observed to compensate for the
hemodilution. However, the 31P-NMR
data cannot be used to rule out a potential autotransfusion mechanism,
which might compensate during hemodilution with rHb1.1. To test the
ability of the rats to autotransfuse, we hemodiluted a rat to a
residual hematocrit of 3% with HSA and allowed the rat to rest at that
hematocrit for 30 min. When the pump was restarted, we again measured
the hematocrit and found that it was 3.7%. The estimated precision of
the hematocrit measurements was ±1.7% (see METHODS). This increase of the
hematocrit from 3.0 to 3.7% from putative autotransfusion is therefore
not statistically significant and is unlikely to provide a significant
source of compensatory oxygen-carrying capacity during hemodilution.
Finally, free Hb is osmotically active and could have had an
uncontrolled effect on the relationship between hematocrit and Hb
concentration in this study. Had this been the case, however, we would
have found a systematic difference between the hemodilution curves from
the HSA group and those from the rHb1.1 group. We specifically sought
evidence for such an effect in weight-matched groups of rats from the
HSA and rHb1.1 pools; however, the data (not shown) revealed no
significant difference in the hemodilution curves.
The experiments reported here demonstrate the efficacy of rHb1.1 in
vivo in the extreme situation of total blood replacement and make it
clear that rHb1.1 sustains vital energy-producing functions of tissues
in anesthetized animals at levels that are indistinguishable by
31P-NMR from those found when
whole blood is present, even though the rHb1.1 concentration (5 g/dl)
was only one-third that of normal blood (~15 g/dl).
| |
ACKNOWLEDGEMENTS |
We acknowledge the capable technical assistance of Vicki White (The
Lovelace Scientific Institutes) in cannulation and handling of the
animals and Kent Springer (Somatogen). We thank Dr. Eiichi Fukushima
(The Lovelace Scientific Institutes) for support and advice.
| |
FOOTNOTES |
This research was supported by Somatogen.
Portions of this work were previously presented at the meeting
"Current Issues in Blood Substitute Research," 30-31 March 1995, San Diego, CA.
The costs of publication of this
article were defrayed in part by the
payment of page charges. The article
must therefore be hereby marked
"advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Address for reprint requests: L. O. Sillerud, 304 Las Colinas Lane NE,
Albuquerque, NM 87113 (E-mail: laurel{at}unm.edu).
Received 5 January 1998; accepted in final form 22 October 1998.
| |
REFERENCES |
1.
Bittl, J. A.,
J. DeLayre,
and
J. S. Ingwall.
Rate equation for creatine kinase predicts the in vivo reaction velocity: 31P-NMR surface coil studies in brain, heart and skeletal muscle of the living rat.
Biochemistry
26:
6083-6090,
1987[Medline].
2.
Blum, H.,
J. J. Summers,
M. D. Schnall,
C. Barlow,
J. S. Leigh,
B. Chance,
and
G. P. Buzby.
Acute intestinal ischemia studies by phosphorus nuclear magnetic resonance spectroscopy.
Ann. Surg.
204:
83-88,
1986[Medline].
3.
Canioni, P.,
and
B. Quistorff.
Liver physiology and metabolism.
In: NMR in Physiology and Medicine, edited by R. J. Gillies. San Diego, CA: Academic, 1994, p. 374-388.
4.
Hess, J. R.,
C. E. Wade,
and
R. M. Winslow.
Filtration-assisted exchange transfusion using 
Hb, an erythrocyte substitute.
J. Appl. Physiol.
70:
1639-1644,
1991[Abstract/Free Full Text].
5.
Hoffman, S. J.,
D. L. Looker,
J. M. Roerich,
P. E. Cozart,
S. L. Durfee,
J. L. Tedesco,
and
G. L. Stetler.
Expression of fully functional tetrameric human hemoglobin in Escherichia coli.
Proc. Natl. Acad. Sci. USA
87:
8521-8525,
1990[Abstract/Free Full Text].
6.
Icenogle, M. V.,
E. Fukushima,
J. A. Jackson,
and
L. O. Sillerud.
Detection and measurement of ischemic bowel in rats by means of 31P surface coil spectroscopy.
Proc. Soc. Magn. Reson. Med.
5:
899-900,
1986.
7.
Katz, L. A.,
J. A. Swain,
M. A. Portman,
and
R. S. Balaban.
Intracellular pH and inorganic phosphate content of heart in vivo: a 31P-NMR study.
Am. J. Physiol.
255 (Heart Circ. Physiol. 24):
H189-H196,
1988[Abstract/Free Full Text].
8.
Kline, R. A.,
W. Negandank,
L. McCoy,
and
R. Berguer.
Beneficial effects of isovolemic hemodilution using a perfluorocarbon emulsion in a stroke model.
Am. J. Surg.
162:
103-106,
1991[Medline].
9.
Kushmerick, M. J.,
R. A. Meyer,
and
T. R. Brown.
Phosphorus NMR spectroscopy of cat biceps and soleus muscles.
Adv. Exp. Med. Biol.
159:
303-325,
1982.
10.
Looker, D. L.,
D. Abbott-Brown,
P. Cozart,
S. Durfee,
S. Hoffman,
A. J. Mathews,
J. Miller-Roerich,
S. Shoemaker,
S. Trimble,
G. Fermi,
N. H. Komiyama,
K. Nagai,
and
G. L. Stetler.
A human recombinant haemoglobin designed for use as a blood substitute.
Nature
356:
258-260,
1992[Medline].
11.
Lowe, K. C.,
D. C. McNaughton,
and
R. N. Hardy.
Changes in intravascular fluid composition following blood replacement with perfluorocarbon emulsion in the rat.
Klin. Wochenschr.
63:
1028-1034,
1985[Medline].
12.
Marcovitz, P. A.,
A. M. Aisen,
L. E. Fencil,
S. D. Swanson,
and
A. J. Buda.
Metabolic effects of adenosine on regional myocardial ischemia by phosphorus-31 nuclear magnetic resonance spectroscopy.
Am. Heart J.
124:
1205-1212,
1992[Medline].
13.
Martin, L. F.,
A. O. Peter,
D. M. Fehr,
J. R. Landis,
J. Cotter,
and
R. W. Briggs.
31P-NMR evaluation of postischemic renal ATP and pH levels after ATP-MgCl2 treatment in rabbits.
Am. J. Surg.
164:
132-139,
1992[Medline].
14.
Masson, S.,
and
B. Quistorff.
The 31P-NMR visibility of ATP in perfused rat liver remains about 90%, unaffected by changes of metabolic state.
Biochemistry
31:
7488-7493,
1992[Medline].
15.
Murray, P. A.,
T. J. J. Blanck,
M. C. Rogers,
and
W. E. Jacobus.
Effects of halothane on myocardial high-energy phosphate metabolism and intracellular pH utilizing 31P-NMR spectroscopy.
Anesthesiology
67:
649-653,
1987[Medline].
16.
Rosenberg, J. M.,
G. B. Martin,
N. A. Paradis,
R. M. Nowak,
D. Walton,
T. J. Appelton,
and
K. M. A. Welch.
The effect of CO2- and non-CO2-generating buffers on cerebral acidosis after cardiac arrest: a 31P-NMR study.
Ann. Emerg. Med.
18:
341-347,
1989[Medline].
17.
Shen, T.,
N. T. Ho,
V. Simplaceanu,
M. Zou,
B. N. Green,
M. F. Tam,
and
C. Ho.
Production of unmodified human adult hemoglobin in Escherichia coli.
Proc. Natl. Acad. Sci. USA
90:
8108-8112,
1993[Abstract/Free Full Text].
18.
Sylvia, A. L.,
and
C. A. Piantadosi.
O2 dependence of in vivo brain cytochrome redox responses and energy metabolism in bloodless rats.
J. Cereb. Blood Flow Metab.
8:
163-172,
1988[Medline].
19.
Taylor, M. D.,
J. A. Jackson,
and
L. O. Sillerud.
31P-NMR analysis of metabolic alterations in the gerbil brain following hypoxia/ischemia (Abstract).
Proc. Soc. Magn. Reson. Med.
4:
292,
1985.
20.
Thompson, C. H.,
G. J. Kemp,
D. J. Taylor,
J. G. G. Ledingham,
G. K. Radda,
and
B. Rojagopalan.
No effect of blood transfusion on muscle metabolism
Q. J. Med.
85:
897-899,
1992[Abstract/Free Full Text].
21.
Vlahakes, G. J.,
R. Lee,
E. E. Jacobs,
and
W. G. Austen.
Efficacy of a new blood substitute based on ultra-pure polymerized bovine hemoglobin.
Eur. J. Cardiothorac. Surg.
3:
353-354,
1989[Medline].
J APPL PHYSIOL 86(3):887-894
8570-7587/99 $5.00
Copyright © 1999 the American Physiological Society
TOP
ABSTRACT
INTRODUCTION
