Haemodynamic and neural responses to hypercapnia in the
awake rat
Chris Martin, Myles Jones, John Martindale and John Mayhew
Centre for Signal Processing in Neuroimaging and Systems Neuroscience, Department of Psychology, University of Sheffield,
Western Bank, Sheffield S10 2TN, UK
Keywords: field potentials, functional magnetic resonance imaging, metabolism, regional cerebral blood flow, whisker ⁄ barrel cortex
Abstract
The relationship between localized changes in brain activity and metabolism, and the blood oxygenation level-dependent (BOLD)
signal used in functional magnetic resonance imaging studies is not fully understood. One source of complexity is that stimulus-
elicited changes in the BOLD signal arise both from changes in oxygen consumption due to increases in activity and purely
‘haemodynamic’ changes such as increases in cerebral blood flow. It is well established that robust cortical haemodynamic changes
can be elicited by increasing the concentration of inspired CO2 (inducing hypercapnia) and it is widely believed that these
haemodynamic changes occur without significant effects upon neural activity or cortical metabolism. Hypercapnia is therefore
commonly used as a calibration condition in functional magnetic resonance imaging studies to enable estimation of oxidative
metabolism from subsequent stimulus-evoked functional magnetic resonance imaging BOLD signal changes. However, there is
little research that has investigated in detail the effects of hypercapnia upon all components of the haemodynamic response (changes
in cerebral blood flow, volume and oxygenation) in addition to recording neural activity. In awake animals, we used optical and
electrophysiological techniques to measure cortical haemodynamic and field potential responses to hypercapnia (60 s, 5% CO2). The
main findings are that firstly, in the awake rat, the temporal structure of the haemodynamic response to hypercapnia differs from that
reported previously in anaesthetized preparations in that the response is more rapid. Secondly, there is evidence that hypercapnia
alters ongoing neural activity in awake rats by inducing periods of cortical desynchronization and this may be associated with
changes in oxidative metabolism.
Introduction
The neuroimaging technique of blood oxygen level-dependent
(BOLD) functional magnetic resonance imaging (fMRI) is based on
the haemodynamic response to changes in neural activity (Logothetis
et al., 2001). Much work has been carried out in recent years to
identify the spatiotemporal structure of the haemodynamic response
[which comprises changes in cerebral blood flow (CBF), cerebral
blood volume (CBV) and oxygenation] and its relationship to
underlying changes in metabolism and neural activity. It is well
established that increasing the level of inspired CO2 (inducing
hypercapnia) has marked affects upon CBF and cortical haemody-
namics (Kety & Schmidt, 1948; Reivich, 1964; Grubb et al., 1974)
and the findings of numerous studies have led to the widely-held
assumption that hypercapnia is ‘metabolically neutral’ (e.g. Kety &
Schmidt, 1948; Novack et al., 1953; Sundt et al., 1976; Nioka et al.,
1987). Hypercapnia protocols have thus been employed to investigate
the effects of baseline CBF changes on the haemodynamic responses
to stimuli (Hoge et al., 1999a; Posse et al., 2001; Stefanovic et al.,
2006); as a tool to investigate cerebrovascular reactivity in normal and
diseased states (Kastrup et al., 2001; van der Zande et al., 2005;
Silvestrini et al., 2006); and in fMRI studies to enable the stimulus-
evoked cerebral metabolic rate of oxygen (CMRO2) to be estimated by
calibrating responses to the haemodynamic changes elicited under
hypercapnic challenge (Davis et al., 1998; Mandeville et al., 1998,
1999; Hoge et al., 1999b; Kim et al., 1999; Schwarzbauer & Heinke,
1999; Liu et al., 2004; Stefanovic et al., 2005).
In animal models, techniques such as optical imaging spectroscopy
(OIS) and laser Doppler flowmetry (LDF) have proved invaluable in
this field as they permit, at higher temporal and spatial resolution than
fMRI, independent measurement of each component of the haemo-
dynamic response which together comprise the BOLD fMRI signal
(e.g. Jones et al., 2001). However, together with most direct measures
of neural activity, these techniques are invasive and necessitate the use
of anaesthetic agents. This has a number of drawbacks (Martin et al.,
2006) and, in response, we and others have developed awake animal
models for research in this field (Lahti et al., 1998; Shtoyerman et al.,
2000; Berwick et al., 2002; Martin et al., 2002, 2006; Brevard et al.,
2003). Research using these models has highlighted a number of
important differences in response characteristics (e.g. magnitude,
spatial extent and temporal structure) that are probably attributable to
the effect of anaesthesia on neural, metabolic and haemodynamic
processes. The specific effects of hypercapnia upon CBF, CBV, the
concentrations of deoxyhaemoglobin (Hbr) and oxyhaemoglobin, and
neural activity in the awake state are, however, unknown and yet are
critical to the interpretation of both human and animal BOLD fMRI
studies that employ hypercapnia. The present study uses OIS, LDF
and electrophysiological recording to investigate the effect of hyper-
capnia upon individual components of the cortical haemodynamic
Correspondence: Dr Chris Martin, as above.
E-mail: c.j.martin@shef.ac.uk
Received 23 May 2006, revised 17 August 2006, accepted 23 August 2006
European Journal of Neuroscience, Vol. 24, pp. 2601–2610, 2006 doi:10.1111/j.1460-9568.2006.05135.x
ª The Authors (2006). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd

response, and upon cortical neural activity, in an animal model free
from the complexities of anaesthesia.
Materials and methods
Experimental overview
Experiments were conducted to record neural and haemodynamic
responses to hypercapnia in awake animals. All recordings were made
from the somatosensory ‘barrel’ cortex. This region was chosen to
permit better comparison with previous, similar research studies and
also because of its popularity as a model system for studying
neurovascular coupling. Neural responses were recorded from five
animals over a total of 15 experimental sessions. Changes in CBVand
oxygenation were thus recorded using OIS in eight animals over a
total of 10 experimental sessions, and changes in CBF were recorded
using LDF in six animals (all from the spectroscopy group) over a
total of eight sessions. Each session consisted of three hypercapnia
trials (see below). Animals were handled extensively and trained to
accept comfortable restraint prior to surgical preparation. Details of the
training procedures are described in Martin et al. (2002). In all cases,
animals were female hooded Lister rats weighing between 200 and
300 g, kept in a 12-h dark ⁄ light cycle environment at a temperature of
22 C with food and water ad libitum. Some of these animals were
also used in experiments to investigate cortical responses to somato-
sensory stimulation (see Martin et al., 2006). All procedures were
carried out with the approval of both The University of Sheffield local
ethics committee and the UK Home Office under the Animals
(Scientific Procedures) Act 1986.
Surgical preparation for imaging experiments
Many of the details of the awake animal training, preparation and data
collection procedures are reported in Martin et al. (2002). After
training, animals were anaesthetized with an intraperitoneal injection
of ketamine and xylazine (1.1 mL ⁄ kg). Under surgical anaesthesia,
the skull was exposed and a section overlying somatosensory barrel
cortex was thinned to translucency. Location of barrel cortex was
guided by both stereotaxic co-ordinates and the highly regular pattern
of vasculature overlying the cortex in this region (Cox et al., 1993;
Moskalenko et al., 1996). A coating of clear cyanoacrylate was
applied to the thin cranial window to provide increased strength. At
least two skull screws were then secured into burr holes drilled in the
contralateral skull, one anterior and another posterior to the thinned
skull region. A chronically implanted chamber was then placed over
the thinned region of skull and secured in dental cement which also
engulfed the skull screws. All wounds were closed and the animals
were treated with an analgesic (Rimadyl, 0.05 mL). Animals were
then left to recover for 3–5 days.
Due to technical limitations, it was not possible to perform OIS and
LDF concurrently in awake preparations and so separate OIS and LDF
experimental sessions were conducted alternately. For each OIS
experiment, animals were placed into a harness and an endoscope
‘docking unit’ was screwed into the implanted chamber. The harness
was then suspended on a frame and, to reduce head movements, a
pneumatically operated clamp secured the implanted chamber. The
experimental apparatus is illustrated in Fig. 1. The imaging chamber
was filled with saline (warmed to 37 C) and a 3-mm medical
endoscope (Endoscan Ltd) was screwed into position in the chamber.
The endoscope, which provided both illumination of the cortex and
transmission of the remitted images to a digital camera, permitted fast
configuration of the imaging apparatus. This is necessary to keep
animal restraint time to a minimum and avoids the need to finely
adjust the position of either the animal or the camera for optimal image
transmission. A spectrograph was mounted on the camera with the slit
(250 lm wide) positioned over barrel cortex. For LDF experiments,
the animal was placed in the harness and the imaging chamber was
clamped as described for OIS experiments. An LDF probe (Probe 403,
Fig. 1. Diagram illustrating the experimental apparatus. Further details can be found in Martin et al. (2002). LDF, laser Doppler flowmetry.
2602 C. Martin et al.
ª The Authors (2006). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd
European Journal of Neuroscience, 24, 2601–2610