Author's response

The aim of the brain-derived neurotrophic factor (BDNF) study in question1 was not to provide a validated set of diffusion coefficients. Rather, in this study we sought to exploit our method featuring multiphoton microscopy to guide the design of an improved therapeutic agent. Therefore, the experimental design and presentation emphasize this broader goal.

Previous to this, we had published a related study of the nerve growth factor (NGF) with the aim of validating the new method using multiphoton microscopy to obtain tissue diffusion coefficients2. In that work we designed the experiments and performed the analysis to emphasize the values of diffusion constants, and to test that these constants were similar to historical data or within predicted ranges.

In contrast, our specific goal in investigating the free diffusion coefficient for BDNF was to compare relative values of diffusion behaviour for our preparations in the brain. In accordance with this goal, we neither designed the experiments nor conducted the appropriate analysis to test against a predicted free diffusion coefficient. Although we agree with the points made by Thorne et al., and we appreciate the value of the new measurements of PEG diffusion that they provide, we believe that the data in Table 1 of ref. 2 demonstrated the relative differences in diffusion coefficient sufficiently to support the approach to protein modification that we pursued, even with the difficulties that Thorne et al. identify.

With regard to the specific comments of Thorne et al., we considered the possibility that aggregation influenced the diffusion measurements. (The influence of aggregation on protein diffusion is a familiar issue for us: see Radomsky et al.3) Our NGF diffusion measurements, which Thorne et al. cite as a benchmark for comparison against the BDNF results, were likewise obtained using a tetramethylrhodamine (TMR) conjugate1; aggregation was clearly not an issue in that data. In addition, gel electrophoresis of our BDNF preparations did not reveal an unexpected, high-molecular-weight product (the conjugate migrated as expected), the TMR–BDNF conjugate was bioactive, and the material was extensively centrifuged before measurement. All of these observations led us to conclude that the labelled proteins were suitable for study.

Regarding the comments on polyethylene glycol (PEG), our goal was not to validate a brain diffusion constant for PEG. It was rather to contrast to the behaviour of the PEG–BDNF conjugate, and to provide evidence to the reader that we were monitoring the latter rather than free PEG diffusion in the brain. We feel that Table 1 and Figs 3d and e of the manuscript1 convincingly demonstrate that we were not measuring the diffusion of free PEG in tissue, but the diffusion of a BDNF–PEG conjugate. We agree that the tortuosity reported for PEG in the correspondence by Thorne et al. (∼1.6) is different from ours (∼1.0), and more in line with past measurements using their technique. However, we note that Thorne et al. contrast their results from a single-photon methodology with our multiphoton results, which could perhaps account for some of the discrepancy, as suggested by some recent reports4. More measurements, and side-by-side comparisons, are probably needed to fully address this issue.

Although we agree that it is important to be careful in the interpretation of diffusion data, we do not find that variability in our measurements is an issue that compromises our ability to design an improved drug.