Figure 1 shows the hemodynamic responses measured using NIOI and/or fMRI in human occipital cortex during visual stimulation. Using flash-photo stimulation, the Hb species concentrations measured with NIOI increased from their basal levels, but the increases were minimum at a temporal frequency of 8 Hz (Fig. 1A). The same response tendency was observed in all subjects. This result was the opposite to results obtained using PET [16] or fMRI [17, 18], which showed the maximum increase in rCBF in the visual cortex occurred at around 8 Hz. To investigate this discrepancy between changes in Hb species concentrations and rCBF, we performed simultaneous NIOI and fMRI measurements on two subjects using a black/white annular checkerboard (because we were unable to use our photo stimulator in the MRI system) (Fig. 1B). Changes in the blood oxygenation level-dependent (BOLD) fMRI signals of individual subjects showed a maximum at stimulus frequencies around 1.4–4.7 Hz, while Δ[oxy-Hb], Δ[deoxy-Hb], and Δ[total-Hb] measured with NIOI showed a minimum around these frequencies. It should be noted that the frequency of stimulation using flickering checkerboards is considered to be twice the white-to-white or black-to-black frequency of the checkerboard, if the visual stimulus basically comes from pattern reversal. Thus, the frequency of 1.4–4.7 Hz corresponds to 2.8–9.4 pattern reversals/sec. Therefore, the present results obtained using a checkerboard stimulus are consistent with our NIOI results and the results of other studies obtained using flash-photo stimulation [16–18]. Moreover, the use of this stimulus highlighted the following two findings: (1) changes in [deoxy-Hb] for the checkerboard stimulation decreased from the basal level (Δ[deoxy] < 0), whereas those for the flash-photo stimulation increased from the basal level (Δ[deoxy] > 0), and (2) the responses of the Hb parameters and BOLD signals dissociated at around 8 Hz. One possible explanation for the former finding is that the difference in the [deoxy-Hb] reflects differences in the metabolic and circulatory conditions in the visual cortex during the resting state. The subjects were asked to close their eyes during the flash-photo stimulation, whereas during the checkerboard stimulation they were asked to keep them open. It has been reported that the metabolic rate of glucose in the visual cortex (CMRGlc) decreased from the basal level when subjects closed their eyes, whereas during the checkerboard stimulation (with the eyes open) it increased above the basal level [19]. As for the second finding, the result indicates that the dissociation between rCBF and rCBV occurs at a temporal frequency around 8 Hz, since the temporal frequency dependence of the BOLD signal responses corresponded well with the rCBF responses [16–18], whereas the responses of the Hb parameters, especially [total-Hb], reflected changes in the rCBV. Moreover, it has been reported that electrical [20] and CMRO2 responses [21] showed a maximum at a frequency around 4 Hz (i.e., 8 pattern reversals/sec) when a checkerboard was used for visual stimulation. These results suggest that there are physiological requirements for the dissociation of stimulus-induced responses of rCBF and rCBV above a certain level of neuronal activity.
If the above dissociation phenomenon applies generally to any type of regional brain activation, it should occur regardless of stimulus type, cortical area, or animal species. To test the above hypothesis, we examined stimulus-induced hemodynamic responses in the somatosensory cortex of rats during electrical stimulation of the peripheral nerve (see Materials and Methods). Figure 2 shows the spatiotemporal profile of stimulus-induced activation in the somatosensory cortex of the rat. Figure 2A shows a CCD image over the left parietal cortex viewed through the cranial window. The vessel labeled A is a second-order (parietal) branch of the middle cerebral artery (MCA), and vessels B and C are its tributaries (third-order branches). Vessel B predominantly supplies the hindlimb somatosensory area and vessel C predominantly supplies the trunk area (left branch). One of the tributaries of vessel C formed an interarterial anastomosis with the anterior cerebral artery (ACA), denoted by a broken line. The stimulus-induced maximal neuronal activation obtained with changes in the membrane potential was localized in the hindpaw area at 28 msec after stimulus onset (Fig. 2B, traces 1 and 2). It then propagated over the hindlimb area (Fig. 2B, trace 3). In contrast, Figure 2C shows an absorbance change at a wavelength of 577 nm (ΔA577), which mainly reflects [oxy-Hb], indicating that the change in rCBF spread beyond the hindlimb area at 6 sec after stimulus onset. This wide distribution of CBF correlated well with the widespread increase in intravascular pO2 measured using albumin-bound oxygen-sensitive phosphorescence dye, although the maximal increases in rCBF and pO2 were observed over the hindlimb area (Fig. 2D).
Figure 3A shows representative temporal profiles of changes in the hemodynamic parameters measured using LDF in the maximally activated hindpaw area (marked with a white circle in Figs. 2C and 2D), in which maximal changes in RBC flow (rRBCFlow), RBC velocity (rRBCVeloc), and RBC number (rRBCMass) were observed 5~7 sec after stimulus onset. To investigate the relationship between stimulus intensity and degree of change in these hemodynamic parameters, the LDF measurements were performed in the activated hindpaw area at various current intensities and stimulus periods (Fig. 3B). Since it has been reported that electrical stimulation of the peripheral nerve of rats showed a maximal response (without tetanus) of the hemodynamic parameters at around 5 Hz [22], the stimulus frequency was kept at 5 Hz during this experiment. The index of stimulus intensity (horizontal axis in Fig. 3B) was assumed as a function of a product of the current intensity and stimulus period. At a lower stimulus intensity (SI ≤ 2), rRBCFlow, rRBCVeloc, and rRBCMass all increased (P < 0.001) due to functional hyperemia, while at a higher stimulus intensity (SI > 2), rRBCFlow and rRBCVeloc increased, while rRBCMass decreased (P < 0.001). These results demonstrated that the dissociative response between rCBF and rCBV in the human visual cortex (Fig. 1) also occurred in the somatosensory cortex of rats in response to different stimuli. This finding strongly suggests that Grubb's relationship between CBF and CBV (CBV = 0.8*CBF0.38) [10] does not always apply, especially to the relationship between regional changes in CBF and CBV (probably above a certain level of neuronal activation). In addition, the saturation of the increase in rRBCFlow indicates that the maximal level of increase in O2 inflow into the activated area remains at a certain level (about 30%, see Fig. 3B) because the increased rRBCFlow reflects the inflowing oxygenated RBC.
Figure 4A shows changes in the blood flow rate and diameter (inset) of the pial arterioles supplying the blood to the hindpaw area (see Fig. 2A). The blood flow and diameter of the afferent vessel A and its tributaries (B, which supplies the hindlimb area, and C, which supplies the trunk area) increased just after the onset of stimulation (Fig. 4A), but the pial arterioles lying in other areas remained unchanged. The tissue blood flow in the hindpaw area measured using the laser-Doppler flowmeter increased in accordance with the increase in blood flow in vessel B measured using the FLDAM. The blood flow and diameter of vessel B showed long-lasting increases, probably due to the metabolic effect of the neuronal activation in the hindpaw area. It should be noted that vessel C does not supply blood to the activated area, but the change in blood flow in vessel C was almost the same as that in vessel B. These results suggest that the increases in the diameter and blood flow in vessel C are regulated so that vessel C plays an active role as an "escape route" to prevent excess inflow of O2 into vessel B. The increase in pO2 more than 10 mmHg may be undesirable in the capillary bed and venules in the activated area (cf., Fig. 2D). These results may account for the finding that the maximal level of increase in O2 inflow into the activated area remained at a certain level (i.e., the asymptotic increase in rRBCFlow in Fig. 3). The stimulus conditions used in this study (a rectangular pulse of 3.8 mA intensity and 0.5 msec duration at 5 Hz) considerably exceeded motor and sensory thresholds, which may have been sufficient to excite Aα, Aβ, and Aδ axons and C-fibers [23]. In turn, this may have been enough to elicit the maximal increase in rRBCFlow and a maximal level of O2 inflow in the hindpaw area [24]. These results are summarized in Fig. 4B. The diameter and blood flow of the afferent vessel A and its tributaries, B and C, increased just after onset of the stimulation due to a fast and transient neurogenic regulation. The blood supply to the activated area was maintained by a delayed and lasting metabolic factor. Vessel C does not supply blood to the activated area, but the change in blood flow in vessel C (27% increase) was almost the same as that in vessel B (30% increase), suggesting that vessel C plays an active role as an "escape route" to prevent excess O2inflow into vessel B and the activated area. This regulation could be achieved by a dissociation between rCBF and rCBV responses, called a flow-mass regulatory mechanism. A third mechanism, possibly a high-O2 sensing mechanism and/or fluid shear stress, may act as a regulator for the mechanism in addition to the neurogenic and metabolic regulation.
In conclusion, our results indicate the existence of a flow(rCBF)-mass(rCBV) regulatory mechanism, and that changes in blood flow during brain activation are not tightly regulated to supply only to an the activated area. This loose regulation serves to prevent intense functional hyperemia and excess O2 inflow into the focally activated area. A high-O2 sensing mechanism and/or fluid shear stress are proposed as a regulator of the flow-mass regulatory mechanism, which may elicit a neurogenic modulation of vascular tonus in the activated cortex. Collapse of the flow-mass regulatory mechanism (e.g., arteriosclerosis) would result in excess inflow of oxygen into the focally activated area, triggering the production of harmful reactive oxygen species and leading to the accumulation of irreversible neuronal damage (Fig. 5).