In our view, the title suggested by the Otero and Rodriguez groups could lead the reader to think that the Ih current is, by itself, responsible for BOTH of the two main dynamic aspects of the rod bipolar cell's response: i. attenuation at low frequencies and ii. attenuation at high frequencies.
Band-pass behavior arises from the interaction of the 'inductive' element Ih, with the 'Resisto-Capacitive' cellular membrane. The former confers high-pass filtering characteristics, while the latter low-pass [ref. 31]. In other words, in the hypothetical case of a cell without capacitance (i.e. C=0), Ih would transform it into a high-pass filter. Throughout the paper we refer, although not always explicitly, to the band-pass behavior of rod bipolar cells as a consequence of the above-mentioned role of Ih.
Figure 4 (right column) displays the predicted effect that a blockade of Ih has on the impedance profile of a rod bipolar cell at potentials negative to -75mV: attenuation at low frequencies is abolished, while that at high frequencies is essentially unchanged. What Otero et al. may refer to in this figure is the slight enhancement with Ih just to the right of the resonance peaks. This enhancement, which may or may not arise depending on the specific electrical parameters of the cell, is not masked but rather relies entirely on the presence of the slow membrane properties!
In summary, the current title of the paper emphasizes the main contribution of Ih to rod bipolar cell behavior: that of attenuating low-frequency input, thus making responses more transitory in nature.

As Otero, Rodriguez and colleagues point out, Ih is one of several known ionic currents conducive to resonance in neurons [ref. 41]. On the other hand, the actual expression of resonance in a specific cell endowed with Ih, requires its electrical parameters to be within a certain domain [eq. 23 in ref. 31]. Moreover, other types of ion channels could also contribute at the same, or at different membrane potentials (as exemplified in our paper).
Our approach of modeling individually recorded cells allowed us to evaluate to what extent Ih contributed to their band-pass profiles, at any given potential. As reported in the paper, model and experimental profiles were found to match very well at potentials more negative than about -70 mV (Figure 3C), thus demonstrating that Ih is largely--perhaps entirely--responsible for that domain of resonant behavior.

Yes, this is one important point made by our paper. Band-pass filtering of light-evoked signals all along the rod pathway could, in conjunction with amplifying mechanisms, preserve or enhance behaviorally relevant temporal components of the original stimulus [ref. 25]. There is evidence to show that a similar mechanism operates already at the level of the rod inner segments, where the role of the voltage-dependent currents (Ih and Ikx) is to accelerate the voltage response with respect to the slow photocurrent [ref. 24].

The large spread in the iBP values shown in Figure 3D is simply due to the intrinsic variability found in the rod bipolar cell population. The red curve, which displays the iBP predicted by the average cellular model, shows that the voltage range where Ih may underly band-pass filtering overlaps with the negative range of experimentally observed band-pass behavior (<-75mV), but not with the positive range (>-70mV). The remarkably good predictive power of the model on a cell-by-cell basis, is instead shown in the examples of Figures 3C and 4, as well as in Figure 5A.

Judging channel isoform identity by comparing time constants across dramatically varying experimental conditions such as patch pipette solutions, bath temperature, tissue vs. heterologous expression system, appears as a tricky operation at best. The kinetics of Ih in rod bipolars lies somewhat in between the two published ranges (see Discussion), and could be fully compatible with being mediated by either HCN1 or HCN2.
Based on these considerations we thus turned to immunohistochemistry. In full agreement with previously published data, we found that the HCN1 isoform is expressed in rods, with NO SIGN of an expression by rod bipolar cells! Our results, which point instead to an expression of HCN2 in rod bipolars, match those of references [19] and [21] in the rat using a different antibody, with the difference being the exact localization of these channels within the neuron. Essentially, the best currently available evidence has it that rod bipolars express HCN2, but not HCN1.
In principle our analysis of the functional role of Ih in rod bipolars of the mouse could have been published alone without the study on channel identity. On the other hand we believe that the results obtained with the immunohistochemistry represent an important and intriguing addition to the electrophysiology. Of particular interest is the fact that HCN2 appear to have a very localized expression at the level of sites of synaptic input, in striking analogy to what is found in other systems [ref. 73]. As discussed in the paper, the apparent conflict between the isopotentiality of rod bipolars and the spatially restricted location of HCN channels, could be the tantalizing sign of a modulatory interaction (structural or by means of a short range diffusing molecule) with some element of the post-synaptic complex.