I(f), a diastolic depolarizing current activated by hyperpolarization, is a key player in cardiac pacing. Despite the fact that I(f) has been known for over 20 years, the encoding genes, namely HCN1 to 4, have only recently been identified. Functional data imply that different HCN isoforms may coassemble to form heteromeric channel complexes, but little direct evidence is available. Subunit stoichiometry is also unknown. Although the pore region of HCN channels contains the glycine-tyrosine-glycine (GYG) signature motif found in K+-selective channels, they permeate both Na+ and K+. In the present study, we probed the functional importance of the GYG selectivity motif in pacemaker channels by replacing this triplet in HCN1 with alanines (GYG(349-351)AAA or HCN1-AAA). HCN1-AAA did not yield functional currents; coexpression of HCN1-AAA with wild-type (WT) HCN1 suppressed normal channel activity in a dominant-negative manner (55.2+/-3.2%, 68.3+/-4.3%, 78.7+/-1.6%, 91.7+/-0.8%, and 97.9+/-0.2% current reduction at -140 mV for WT:AAA cRNA ratios of 4:1, 3:1, 2:1, 1:1, and 1:2, respectively) without affecting gating (steady-state activation, activation and deactivation kinetics) or permeation (reversal potential) properties. HCN1-AAA coexpression, however, did not alter the expressed current amplitudes of Kv1.4 and Kv2.1 channels, indicating that its suppressive effect was channel-specific. Statistical analysis reveals that a single HCN channel is composed of 4 monomeric subunits. Interestingly, HCN1-AAA also inhibited HCN2 in a dominant-negative manner with the same efficacy. We conclude that the GYG motif is a critical determinant of ion permeation for HCN channels, and that HCN1 and HCN2 readily coassemble to form heterotetrameric complexes.