To the editor

— Stroh et al. recently proposed that diffusion measurements may be used to direct rational drug design1. They focused on brain-derived neurotrophic factor (BDNF), a neurotrophin nearly identical to nerve growth factor (NGF) in size (27 kDa), shape and charge. Despite the similarities, BDNF distribution in brain is often more limited than that of NGF after central administration. Stroh et al. concluded that chemical modification of BDNF with polyethylene glycol (PEG) produces a conjugate with enhanced diffusion properties in rat brain slices, as compared with native BDNF1. Two problems are evident from these authors' data.

First, the free diffusion coefficient, Df, measured for tetramethylrhodamine-labelled BDNF (R-BDNF) was far lower than expected. Tetramethylrhodamine conjugates are prone to aggregation2,3 so this finding should have aroused concern. Correlations2 predict a size of 360–630 kDa for R-BDNF based on the value of Df (4.57 × 10−7 cm2 s−1) reported by Stroh et al.1, strongly suggesting aggregation. Had R-BDNF been stable in solution, Df should have been similar to that earlier measured by Stroh et al.4 for NGF (12.6 × 10−7 cm2 s−1; predicted size2 17–30 kDa).

Second, the effective diffusion coefficient, Db, determined for 2 kDa PEG in neostriatum by Stroh et al.1 (19.4 ± 15.1 × 10−7 cm2 s−1; mean ± s.d.) yielded a surprisingly low tortuosity (= (Df/Db)1/2) of 1.02. A tortuosity of 1.88 had been reported previously for a slightly larger PEG in a hippocampal slice preparation5. Because tortuosity describes tissue hindrance to diffusion, a value of 1 suggests PEG can diffuse in brain, an environment containing many obstacles, with no more difficulty than it does in water. If true, this would be a very important finding, so we attempted to verify it using integrative optical imaging2. Our preliminary data indicate, however, that PEG diffusion is significantly hindered, both in neostriatum and neocortex (Table 1). Tortuosity measured with PEG appears similar to that obtained with 74 Da tetramethylammonium (1.54 and 1.62 in neostriatum5 and neocortex6, respectively). We suspect the accuracy of the Stroh et al. data1 was compromised by the substantial variability in some of their measurements.

Table 1 PEG diffusion parameters measured in free medium (0.3% agarose) and in rat brain slices.

We agree that diffusion analysis can be used to screen drugs and chemically modified drug conjugates. However, we must stress that high signal-to-noise measurements and careful interpretation are keys to unlocking the power of diffusion analysis for rational drug design.