Reply to Heck et al.: Signal amplification at the rhodopsin-to-transducin·phosphodiesterase step in rod phototransduction

In PNAS (1), we estimate ∼12–14 active transducin·phosphodiesterase complexes (G_T*·PDE*s) produced per active rhodopsin (Rho*) in mouse rods. This is the effective gain—more informative by not including the empty gain from G_T*s failing to activate PDE. Nearly all previous estimates were on the total number of G_T* produced per Rho* merely because no one before us could measure the individual G_T*·PDE*-triggered electrical event for direct comparison with the single-Rho* response.

We do not state in our paper that previous measurements of the rate of G_T* production per Rho* were all incorrect, but rather they spanned a wide range of values with no consensus regarding which value is valid, owing to the different experimental preparations, methodologies, temperature, interpretations, etc. [see SI appendix in our paper (1)]. Most importantly, unlike previous work, we obtained our data exclusively from intact, live rods, and thus our results are more straightforward to interpret. Heck et al.’s Letter (2) selects without justification particular values from a wide range to claim agreement with ours and fails to recognize the significance of our findings in the broader history of the problem.

Regarding specific point i in Heck et al.’s Letter, the authors again selectively discuss just one model [namely, their own (3, 4)] out of several about functional symmetry/asymmetry of PDE. This model is neither recent (see ref. 5) nor widely accepted in the field currently because of limited experimental evidence. Our paper provides an unbiased discussion of this and other models.

In point iia, Heck et al. question our analysis of the data in figure 3A, right (1), pointing out the disparity in waveform between the ensemble variance and the ensemble mean square of the flash responses. We do not completely disagree about possibly some small disparity between the two waveforms, but these measurements were difficult, given the small variance (10 times smaller than that in WT experiments). Moreover, instead of stemming necessarily from variability in the underlying elementary events as Heck et al. suggest, the discrepancy may come from variability in the timing of activation of individual G_T molecules by REY-Rho* due to the much reduced affinity between REY-Rho* and G_T.

Point iib questions a constancy of the G_T*·PDE*-triggered events produced by apo-opsin (Opn*), proposing instead a stochastic variability in event size/waveform, but admits that it is not straightforward to take this into account in the analysis. Given no a priori evidence for stochastic variations, we extracted the average event size and shape by adopting the simplest model of fairly constant events. This possibility is not unfounded but guided by our previous findings on olfactory transduction (6), where the response to G_olf*⋅AC*(adenylyl cyclase)—functionally equivalent to G_T*⋅PDE* in vision—appears quite constant in amplitude and waveform. Furthermore, the time-integrated single-G_T*⋅PDE*–triggered responses from our two independent methods match each other well (figure 7B in ref. 1).

In sum, relative to the “historical,” textbook-dogma value of 500 (7), we do consider our value very low and would still consider it low even if our estimate were, say, higher by another factor of 2.