NOTE TO REVIEWERS:
The short review by Pasley and Freeman is a very good summary on the topic. I expect that it will serve well the scientific communities interested in neurovascular coupling, brain imaging, and the larger communities of interested people.
Here are a few comments that I expect would be helpful for the authors:
1. First and second paragraphs, ‘Accordingly, during neural activity…suggest that blood flow is controlled directly by energy demands… which trigger vasodilation (or vasoconstriction).’ Whereas the authors mention later the alternative hypothesis of CBF control by neuronal signaling processes involving neurotransmitters, my overall impression is that the reader is lead to believe that CBF is controlled mainly via a feedback mechanism guided by energy demand via a mechanism that is sensitive to the content of metabolism by-products. I believe that the current accepted concept is that both feedforward mechanisms via neuronal signaling and feedback mechanism via ‘sensing’ of the content of metabolism by-products take important roles in control of CBF. I recommend rephrasing these paragraphs for more balance between the two proposed mechanisms.
2. ‘In the cortex, the vascular point-spread function (PSF) has been estimated to be ~2-3 mm (Shmuel et al., 2007).’ Looking at the abstract of Shmuel et al. 2007, these authors in fact: a. Measured an overall PSF of 2.34 mm in the late positive response, with a initial PSF (3 s from stimulus onset) of 1.52 mm. b. Taking in account confounding effects (including the ‘MRI PSF’ mentioned by reviewer b) Shmuel et al. estimated the cortical PSF of the late positive response to be smaller than 2 mm. c. All the papers from Harel et al. (2006) review p. 947 were not done in humans. In addition, the most up to date study mentioned in this review (Park et al., 2004), reported a T2* PSF of 1.7 mm in the cat, similar to the estimate by Shmuel et al., 2007 in humans. The estimate of 0.67 mm in that study refers to perfusion and spin echo measurements. d. The estimated PSF by Shmuel et al. 2007 is in fact consistent with columnar mapping in humans, as discussed by the two most recent papers on human columns (Yacoub et al., Neuroimage 37 (4), 2007; Pei Sun, Keiji Tanaka, Kang Cheng et al., Nature NS 10 (11), 2007).
3. ‘…the function it has to fulfill for continuous brain work is to provide nutrients of metabolism and wash-out waste products (such as CO2 and temperature) on a fine spatio-temporal scale’ – a. I suggest ‘excessive heat’ instead of temperature; b. It is true that the system provides these functions on a fine spatio-temporal scale, but I would argue that it is wrong to state that it ‘has to’ function on this fine scale.
4. ‘For instance, higher level brain areas may be subject to feedback’ – did the authors mean ‘lower level brain areas may be subject to feedback’ or/and ‘higher level brain areas provide feedback’.
5. ‘Because synaptic activity has a much higher… as shown by many PET studies.’ This statement is unclear.
6. ‘...by the fact that a lot agents (such as CO2 and Cafeeine..)’ – note the 2 typos.
Reply to Reviewer A
1. We have attempted to provide a more balanced emphasis by stating the possibility of multiple mechanisms at the beginning of the 2nd paragraph, as well as mentioning several additional proposals, as suggested by other reviews.
2. This section has been re-written to place more emphasis on spatial specificity of the signal and studies of columnar mapping.
3. We have edited this section to clarify this point.
4. This was a misstatement and has been clarified as suggested.
5. This interesting point was provided by another reviewer. It may be the case that CBF is driven by the more metabolically demanding process, but it is also possible that CBF could be driven by a different process altogether, such as in the case that smooth muscle cells are directly innervated by neurons. Given this possibility and the fact that baseline synaptic and spiking activity may be correlated in general, we believe this finding is difficult to interpret in terms of providing a distinction between these two forms of neuronal activity and we have thus omitted it.
6. We have edited this statement.
This is a well written article which will be accessible and useful for a broad audience with a neurobiological background. I only have a few minor comments; these mainly concern the fact that not all aspects of neurovascular coupling have been covered that may be of interest to neuroscientists.
1. In the section on "Physiological consequences of neural activity", it would be useful to cite two review papers that have been quite influential in promoting the idea that increases in local cerebral blood are evoked, in a feedforward fashion, by transmitter release at glutamatergic synapses. These papers are Attwell & Iadecola (Trends in Neurosciences 2002) and Lauritzen (Nature Reviews Neuroscience 2005).
2. In the paragraph on spatial resolution of vascular responses, the authors referred to a vascular point spread function of 2-3 mm. This may create the impression that fMRI cannot be used to measure processes at the columnar level. However, previous studies have indicated that this appears to be possible (e.g. Thompson et al. 2003, Science; Moon et al. 2007, J. Neurosci).
3. So far, the article does not mention that there are quantitative biophysical models of neurovascular coupling which can be used to estimate physiological parameters from measured BOLD data and are important for disambiguating between neuronal and vascular effects in interpreting BOLD responses. A recent review of these models can be found in Stephan et al. 2004, Curr. Opin. Neurobiol.
Reply to Reviewer B
1. These papers have been included as citations.
2. This section has been re-written and now includes several citations regarding columnar-level fMRI measurements.
3. This is an important topic and we have included a brief discussion of these models to conclude the last section.