Early capillary flux homogenization in response to neural activation

Abstract

This Brief Communication reports early homogenization of capillary network flow during somatosensory activation in the rat cerebral cortex. We used optical coherence tomography and statistical intensity variation analysis for tracing changes in the red blood cell flux over hundreds of capillaries nearly at the same time with 1-s resolution. We observed that while the mean capillary flux exhibited a typical increase during activation, the standard deviation of the capillary flux exhibited an early decrease that happened before the mean flux increase. This network-level data is consistent with the theoretical hypothesis that capillary flow homogenizes during activation to improve oxygen delivery.

Introduction

Normal functioning of the brain critically depends on its energy supply regulation to meet the spatiotemporally-varying metabolic need, moment by moment, location by location, due to the brain’s small energy reservoir. The traditional paradigm for understanding this energy supply regulation is neurovascular coupling (see Attwell et al. ¹ for review), a range of mechanisms underlying how neural activity results in arterial dilation leading to an increase in local cerebral blood flow. A capillary network connecting arteries and veins works as a direct interface between energy-supplying blood and energy-demanding tissue, and it has been treated as a passive pipeline network in this traditional neuro-vascular coupling paradigm. However, recent studies suggest the possibility that capillaries also actively regulate blood flow in response to neural activation as mediated by pericytes, ² and it has been theoretically proposed that capillary network flow homogenizes during activation to improve the efficiency of oxygen delivery to the tissue. ³ Here, we describe the use of our novel technique of tracing changes in the red blood cell (RBC) flux over hundreds of capillaries during somatosensory activation with sufficient temporal resolution to measure early homogenization of capillary network flux before mean increases in flux.

Material and methods

Animal preparation and somatosensory activation

Sprague Dawley rats (male, 9 weeks old, 250–300 g, n = 3) were initially anesthetized with isoflurane (1.5–2.5%, v/v), and ventilated with a mixture of air and oxygen during surgical procedures. Tracheotomy and cannulation of the femoral artery and vein were done. Following this, the head was fixed in a stereotaxic frame, and the scalp retracted. A craniotomy was performed using a saline-cooled dental drill and a 3 mm × 3 mm area over the somatosensory cortex was exposed. The dura was carefully removed, and the brain surface was covered with agarose gel in artificial cerebrospinal fluid and a glass cover slip and sealed with dental acrylic cement. After surgery, rats were anesthetized with Alpha Chloralose (40 mg/kg intravenous bolus followed by 40–50 mg/kg/h, i.v.) and moved to our OCT system for experiment. Physiological signs such as heart rate, body temperature, and blood pressure were continuously monitored during surgery and during the experiment. All animal experimental procedures were reviewed and approved by the Massachusetts General Hospital Subcommittee on Research Animal Care and conducted in accordance with the Massachusetts General Hospital guidelines and with the ARRIVE guidelines.

We applied electrical stimulation (3 Hz for 4 s) to the forepaw for somatosensory activation. The stimulation amplitude was adjusted for each animal to half of the level where the forepaw showed a clear motion response. Ten runs of the stimulation and resting (30 s for each run) were repeated.

Spectral-domain optical coherence tomography system

We optimized a spectral-domain OCT (SD-OCT) system (Thorlabs, Inc., Newton, New Jersey, USA) for dynamic in vivo imaging of the rodent cerebral cortex as described in a previous publication.^4,5 We employed a large-bandwidth NIR light source (1310-nm center wavelength with 170-nm bandwidth) for a large imaging depth (up to 1 mm in brain tissue) and high axial resolution (3.5 µm). The transverse resolution is 3.5 µm with our 10 × objective (NA = 0.26). Note that the transverse resolution is identical to the axial one providing for isotropic voxels. The Rayleigh range is 29 µm. The imaging speed is 47,000 A-scan/s. This permits us to image 350 × 350 × 700 µm (x,y,z) in 10 s with a pixel pitch of 0.875 × 0.875 µm (double Nyquist sampling).

Identification of the activation center with intrinsic optical signal imaging

A CCD camera attached to our SD-OCT system was used to simultaneously record a 2D pattern of light (570 nm wavelength) reflection from the cortical surface through the cranial window (Figure 1(a)). The intrinsic optical signals (IOSs) ⁶ in response to neural activation were recorded and the maximum signal change for each pixel was displayed for identification of an activated region (Figure 1(b)).

Figure 1.

Measurement of changes in the capillary flux and its heterogeneity. (a) A CCD imaging of the cortical surface, where blood vessels appear in dark as we used 570-nm illumination (isosbestic point of hemoglobin absorption) (top). Maximum changes in the intrinsic optical signal in response to neural activation (bottom). Scale bar, 500 µm. (b) Maximum intensity projection images of exemplary SIV volumes during baseline (T = 0) over the two ROIs close to and far from the activation center. Scale bar, 100 µm. (c) RBC flux maps of the vectorized capillaries during baseline. (d) Time courses of the RBC flux of individual capillaries in the ROI close to the activation center. (e) Relative changes in the mean and standard deviation of the capillary network flux.

Statistical intensity variation imaging of capillary network flux

We used our previously developed statistical intensity variation (SIV) technique for high-speed imaging of capillary network flux (see Lee et al. ⁷ for technical details and scanning protocols). When properly splitting and merging the region of interest (ROI) as in Lee et al., ⁷ this technique enabled us to measure the red blood cell (RBC) flux [RBC/s] over hundreds of capillary segments in response to brain activation with ∼1-s temporal resolution that was high enough to trace typical hemodynamic responses to neural activation. This means that the flux changes in hundreds of capillaries can be traced individually.

In the first animal, we chose two ROIs, one of which was close to the activation center while the other was far from the center (the black boxes in Figure 1(a)). For each ROI, we obtained a time series of SIV volumes (20 volumes spanning 26 s with a temporal resolution of 1.3 s per volume; Figure 1(b) shows one of them), where each volume was obtained by averaging over 10 individual runs of brain activation, taking care to align the volume time series to the onset time of the stimulation. From a representative angiogram volume that is an average over all time points and runs of brain activation (i.e. 200 volumes), individual capillary segments were identified with the Hessian matrix analysis-based method. ⁷ The RBC flux was estimated at each segment and at each time point from the mean SIV averaged along the segment, finally leading to a series of 20 flux maps (one for each time point) over the vectorized capillaries (Figure 1C for example). These time-series flux maps enabled us to trace a change in the RBC flux for each capillary (Figure 1d).

In the second and third animals, we only chose a single ROI close to the activation center. Compared to the first animal, the imaged volumes were slightly deeper (∼100 µm) and broader (four times in area) in the second and third animals, respectively.

Results

Capillary network exhibited early homogenization

Responses of individual capillaries (Figure 1d) were highly diverse as reported in the literature, ⁸ but measured over hundreds of capillaries nearly at the same time. From these diverse responses from individual capillaries, we estimated the mean and standard deviation of the capillary network flux. The mean flux was obtained at each time point by averaging the flux at that time over all identified capillaries, and a relative, temporal change in this mean was plotted (Figure 1(e), top). A change in the standard deviation was obtained in the same way. To determine the peak and its time, each time series data of the mean change was fit to a Gamma function (equation (1)) as they exhibited relatively typical behaviors of increase followed by relaxation, while each of the other data was fit to a linear combination of two Gamma functions (equation (2)) for more general fitting that covers a biphasic behavior.

${f(t)=a(t-t}_0{)}^2{e}^{-{b(t-t}_0)}{\chi }_{[t_{,\infty )}}(t)$

(1)

${f(t)=a}_1{(t-t}_1{)}^2{e}^{-b_1{(t-t}_1)}{\chi }_{[t_{,\infty )}}(t)\mathrm{\hspace{1em}\hspace{1em}\hspace{1em}}\mathrm{\hspace{1em}\hspace{1em}\hspace{1em}}+a_2{(t-t}_2{)}^2{e}^{-b_2{(t-t}_2)}{\chi }_{[t_{,\infty )}}(t)$

(2)

where χ_A(t) is a step function.

The mean flux exhibited a typical hemodynamic response, increasing to its peak (2.1 ± 1.1%, n = 3; see Figure 2(a) to (d) for data from the second and third animals) at 5.7 ± 0.5 s after the onset of stimulation and then relaxing back to baseline. In contrast, the standard deviation (i.e. capillary flux heterogeneity) exhibited a biphasic response, where it decreased to its minimum (−2.1 ± 1.1%) at 3.7 ± 0.7 s (significantly earlier than the mean’s peak; p < 0.1; Figure 2(e)), increased to its peak (1.5 ± 0.6%) at 12 s, and then relaxed to its baseline. Paired t-tests were used for statistical significance test in this paper.

Figure 2.

Data of the other animals and statistical analysis. (a–d) Data from the second and third animals. Images and data are presented in the same way as Figure 1. (e) Statistical analysis of the peaks and their times in the time courses of the mean and standard deviation flux dynamics. The peak amplitudes and times were determined by fitting data to a linear combination of two Gamma functions (solid lines in Figures 1(e) and 2(d) and (f)). For the analysis of the second peak in the standard deviation, the third animal’s data was not included because it did not exhibit a late increase. (f) Time courses of flux changes in the high-, mid-, low-baseline capillaries. Data from three animals are presented (over the ROI close to the AC for the first animal in Figure 1). The top 15%, middle 20%, and bottom 20% of capillaries were chosen and averaged. (g) Statistical analysis of the peaks and their times in the time courses of the high-, mid-, low-baseline capillaries’ flux changes.

High-baseline capillaries exhibited early decreases

We also analyzed the flux changes separately over three subgroups of capillaries grouped by their baseline flux as shown in Figure 2(f). While the mean flux of mid- and low-baseline capillaries exhibited the typical increase and relaxation, high-baseline capillaries exhibited earlier decreases (−1.0 ± 0.2% at 1.8 ± 0.5 s, 4.0-s earlier than the mid-baseline’s peak; p < 0.01; Figure 2g).

Discussion and conclusion

In this Brief Communication, we used the SIV technique to measure dynamic changes in RBC flux over hundreds of capillaries in the rodent cerebral cortex during somatosensory activation. To our knowledge, no technique thus far has enabled such high-throughput monitoring of capillary RBC flux dynamics as shown in Figure 1(d). The previous techniques generally measured RBC flux over only a few capillaries at a time ⁹ or required a relatively longer acquisition time per capillary.^10–12

Using the SIV technique, we found that the cerebral capillary networks exhibit early flux homogenization in response to neural activation. This behavior is quite different from what we would expect in a simple parallel capillary network. When flows within individual pipes of the parallel network respond passively to an increase in the network’s supply, a time course of the mean flow’s change will be similar to that of the input and the standard deviation also will exhibit a similar time course to the mean. For instance, when the supply to the simple parallel network increases by 10%, pipes of the network will experience approximately similar changes, leading to ∼10% increases in both the mean and standard deviation. On the contrary, the data show the decrease in the standard deviation despite of the increase in the mean. Further investigation will be needed to determine whether this early homogenization originates from active regulation of capillary caliber in response to neural activation or from the impact of capillary connectivity and bifurcations on fluid dynamics.

Considering the above natural response of the standard deviation to the mean, our data suggests that the flux might be homogenized by ∼4.2% than what the simple parallel network would respond. According to a theoretical study ³ on how the capillary transit time heterogeneity affects the maximum oxygen delivery, our ∼4.2% homogenization may correspond to ∼3.9% additional increase in the oxygen delivery, although accurate estimation would need to consider distinction between the transit time and flux.

One of the limitations of the current SIV technique is a relatively small dynamic range. The upper limit in the measured capillary RBC fluxes in this study was ∼40 RBC/s, relatively lower than those in the literature, where 70–80% of capillaries were reported to exhibit lower than 40–45 RBC/s^8,13 or 80% of capillaries exhibited lower than 80 RBC/s. ⁹ As the dynamic range of the measurable RBC flux is a function of the time gap (11 ms in this study), a larger dynamic range will be achieved if one obtains different time gaps in a single volume scan.

When a measurement method has such a relatively low upper limit in its dynamic range, even the simple parallel capillary network exemplified above could exhibit homogenization in response to an arterial input supply increase, by the so-called ceiling effect. For instance, when the supply to the simple parallel network increases by 50% and thus pipes of the network exhibit 50% increases in flow, but if most of these increases lead to higher values than the measurable upper limit and thus measured values are saturated to the upper limit, then the standard deviation would decrease. However, the homogenization observed in this study did not originate from the ceiling effect. First, its peak time was significantly earlier than that of the mean increase (Figure 2(e)). The two peak times would have been similar if the homogenization was due to the ceiling effect. Further, high-baseline capillaries exhibited early decreases in the flux rather than being simply saturated to a certain level (Figure 2(f)). These results support that the observed early homogenization was not due to the low upper limit of the current SIV technique.