Regulation of the hyperpolarization-activated cationic current Ih in mouse hippocampal pyramidal neurones by vitronectin, a component of extracellular matrix


Corresponding author M. E. Barish: Division of Neurosciences, Beckman Research Institute of the City of Hope, 1450 East Duarte Road, Duarte, CA 91010, USA. Email:


Because the hyperpolarization-activated cation-selective current Ih makes important contributions to neural excitability, we examined its long-term regulation by vitronectin, an extracellular matrix component commonly elevated at injury sites and detected immunochemically in activated microglia. Focusing on mouse hippocampal pyramidal neurones in organotypic slice cultures established at postnatal day 0 or 1 and examined after 3–4 days in vitro, we observed differences in the amplitude and activation rate of Ih between neurones in naive and vitronectin-exposed slices (10 μg ml−1 added to serum-free medium), and between neurones in slices derived from wild-type and vitronectin-deficient mice. The potassium inward rectifier IK(ir), activated at similar voltages to Ih, was not affected by vitronectin. In CA1, differences in Ih amplitude primarily reflected changes in maximum conductance (Gmax): a 23.3% increase to 3.18 ± 0.64 nS from 2.58 ± 0.96 nS (P < 0.05) in vitronectin-exposed neurones, and a 17.9% decrease to 2.24 ± 0.26 nS from 2.73 ± 0.64 nS (P < 0.05) in neurones from vitronectin-deficient slices. The voltage of one-half maximum activation (V½) was not significantly affected by vitronectin exposure (−78.1 ± 2.3 mV versus−80.0 ± 4.9 mV in naive neurones; P > 0.05) or vitronectin deficiency (−83.8 ± 3.1 mV versus−82.0 ± 2.9 mV in wild-type neurones; P > 0.05). In CA3 neurones, changes in Ih reflected differences in both Gmax and V½: in vitronectin-exposed neurones there was a 35.4% increase in Gmax to 1.30 ± 0.49 nS from 0.96 ± 0.26 nS (P < 0.01), and a +3.0 mV shift in V½ to −89.8 mV from −92.8 mV (P < 0.05). The time course of Ih activation could be fitted by the sum of two exponential functions, fast and slow. In both CA1 and CA3 neurones the fast component amplitude was preferentially sensitive to vitronectin, with its relatively larger contribution to total current in vitronectin-exposed cells contributing to the acceleration of Ih activation. Further, HCN1 immunoreactivity appeared elevated in vitronectin-exposed slices, while HCN2 levels appeared unaltered. We suggest that vitronectin-stimulated increases in Ih may potentially affect excitability under pathological conditions.

The hyperpolarization-activated cation-selective current Ih (DiFrancesco, 1993; Pape, 1996; Robinson & Siegelbaum, 2003) is widely distributed in central and peripheral neurones, including hippocampal pyramidal neurones (Halliwell & Adams, 1982; Maccaferri et al. 1993), where it contributes to the resting membrane potential, shapes responses to synaptic input, and influences rhythmic electrical activity. Analysing Ih regulation is thus important for a better understanding of normal and pathological modulation of excitability in the nervous system (Chen et al. 2002; Santoro & Baram, 2003).

How might the properties of Ih be controlled? One aspect of neural regulation involves extracellular matrix (ECM), proteins that can modulate signalling in excitable cells by affecting ion channel activation and intracellular Ca2+ levels (Becchetti et al. 1992; Arcangeli et al. 1993; McPhee et al. 1998; Platts et al. 1998; Wu et al. 1998; Levite et al. 2000; Artym & Petty, 2002; Mistry et al. 2002; Waitkus-Edwards et al. 2002; Wildering et al. 2002).

The ECM is quite plastic (Lee & Benveniste, 1999), and synthesis of vitronectin (Hayman et al. 1983) is a common cellular response to injury (Seiffert, 1997). In the nervous system, vitronectin is often found at sites of injury or inflammation (Akiyama et al. 1991; McGeer et al. 1992; Eikelenboom et al. 1994; Yasuhara et al. 1994; Niquet et al. 1996; Seiffert et al. 1996; Walker & McGeer, 1998; Anderson et al. 1999). Microglial activation is a general response to brain injury (see Discussion), and the microglia emerging during brain slice culture display vitronectin immunoreactivity (Aihara & Barish, 2001, 2002).

To further investigate the influence of brain microenvironments on excitability we examined the long-term effects of vitronectin on Ih in CA1 and CA3 pyramidal neurones using organotypic slice cultures of neonatal hippocampus. We observed that over a period of days, manipulation of vitronectin levels affected Ih but not the potassium inward rectifier IK(ir), with added vitronectin increasing Ih amplitude and activation rate, and vitronectin deficiency yielding reciprocal changes. These results raise the possibility that increases in vitronectin levels may contribute to disturbances of excitability characteristic of pathological states.

Preliminary reports of some of these findings have appeared (Barish & Vasilyev, 2000, 2001).



Analyses in which vitronectin was added to cultured naive hippocampal slices utilized outbred Swiss-Webster mice. Effects of the absence of vitronectin were assessed using slices cultured from vitronectin-deficient (vitronectin−/−) mice in which the vitronectin gene was disrupted by homologous recombination (Zheng et al. 1995; gift from Dr David Ginsburg, HHMI, University of Michigan), and their wild-type background strain C57BL/6J.

To prepare organotypic cultures, neonatal hippocampi were isolated from postnatal day 0 (day of birth) or day 1 (P0 or P1) mice (after halothane anaesthesia and decapitation) and 350 μm-thick sections were prepared on a McIlwain tissue chopper. Slices were placed on Millicell CM (PICM03050; Millipore, Bedford, MA, USA) tissue plate inserts as originally described by Stoppini et al. (1991), and grown at 35.5°C in a humidified 5% CO2/air atmosphere using (serum-free) Neurobasal/B27 medium (GibcoBRL, Rockville, MD, USA). When appropriate, slices were exposed to bovine vitronectin (GibcoBRL no. 12172- 011) from 2 h after the time of culture. Reagents were both added to the culture medium and applied in a drop to the surface of the slice, after which the excess fluid slowly passed through the membrane into the underlying reservoir. All procedures involving animals were in accordance with NIH guidelines and approved by the City of Hope Research Animal Care Committee.


I h was recorded from visually identified (using DIC optics and an Olympus BX50WI microscope; Olympus Optical Co., Tokyo, Japan) neurones using standard whole-cell techniques. Electrodes were pulled from borosilicate glass capillaries (TW150F; World Precision Instruments, Sarasota, FL), to resistance of 2−3 MΩ when filled with intracellular solution. The intracellular solution consisted of (mm): 100 K-gluconate; 50 KCl; 5 MgCl2; 1 CaCl2; 5 EGTA; 20 Hepes; pH adjusted to 7.4 with Tris-Cl. Note that no nucleotides are present in this internal solution. The extracellular solution, ACSF, contained (mm): 140 NaCl; 3 KCl; 2 CaCl2; 2 MgCl2; 26.5 NaHCO3; 1.25 NaH2PO4; 20 glucose; bubbled with carbogen (5% CO2–95% O2). Tetrodotoxin (0.5 μm) was added to suppress sodium currents and action potentials and, in many experiments, AP5 (20 μm), CNQX (20 μm) and bicuculline (10 μm) were added to block spontaneous synaptic events. To isolate Ih, BaCl2 (0.5 mm) was added to block inward-rectifier potassium currents (IK(ir)).

Recordings were made using an Axopatch 200B amplifier, and digitized using a DigiData 1200 interface and pCLAMP 8 software (all from Axon Instruments, Union City, CA, USA). Current traces were filtered at 1–2 kHz and digitized at 10 kHz. Voltages were corrected for the +10 mV junction potential between the K-gluconate-based internal solution and ACSF. Series resistance was compensated by 50–70%. Recordings were made at room temperature (22–24°C). Data were analysed using pCLAMP 8 and Origin 6 (OriginLab, Northampton, MA, USA).

For construction of I–V relations, leakage current was subtracted by linear extrapolation of the current measured during voltage steps from −40 to −50 mV in CA1 neurones, or −50 to −60 mV in CA3 neurones; within these ranges all currents are linear with voltage. Chord conductances (G) for determination of GV relations were calculated for each voltage (V) from the steady state current (I) and the relation G=I/(VEr), where Er is the reversal potential for Ih (approximately −30 mV). Cell capacitance was calculated by integrating the current transient elicited by 10 mV hyperpolarizing voltage steps from −50 mV beginning at the peak and extending to approximately three times longer than the time required to relax to 98% of the peak, and dividing this measure of total charge (Q) by ΔV. For evaluations of Ih kinetics, because long voltage steps precluded using P/−4 subtraction protocols, capacity transients and steady-state currents recorded during steps from −50 to −60 mV were fitted with exponential functions, and these were used to create an idealized (noise-free) trace. This was then scaled appropriately and used to subtract capacity transients and leakage currents from experimental traces.

All data are presented as mean ±s.d. The statistical significance of the differences seen between cell populations was evaluated by t test (two-tailed, with Welsh correction as appropriate) or ANOVA, using Instat (GraphPad, San Diego, CA, USA).

Kinetic model

For the analysis of CA3 neurones in Fig. 7, two independent conductances to represent fast and slow processes were defined, each having a simple two-state gating scheme incorporating voltage-dependent forward (αV) and reverse (βV) rate constants: αV0 exp(−zV/r) and βV0 exp(zV/r). Here, z is the equivalent charge in the membrane voltage field, r=RT/F (= 25.85 mV) and R, T (= 300 °K) and F have their usual meanings (see Altomare et al. 2003). Values of the constants α0, β0 and z were chosen by using an Microsoft Excel spreadsheet to visually match calculated values of Gv (=αV/(αVV)) and τV (= 1/(αVV)) to GV (Fig. 4) and τ–V (Fig. 5) relations. They were α0= 0.011, β0= 15 and z= 1.4 for the fast process, and α0= 0.001, β0= 12 and z= 1.4 for the slow process. Values of V½ were −66.7 mV for the fast process and −86.8 mV for the slow process. Normalized predicted GV relations (Fig. 7C) were calculated as the sum of the two underlying fast and slow conductance processes weighted by their relative contributions, defined as the ratio Afast: Aslow. Values of τfast and τslow were (ms): 967 and 4278 at −80 mV, 643 and 4493 at −90 mV, 394 and 3593 at −100 mV, 233 and 2394 at −110 mV, 136 and 1465 at −120 mV. The amplitudes of fast and slow processes at each voltage (Afast and Aslow) were determined by linear fits to plots of experimentally measured Afast and AslowversusVm (Fig. 5). Values of Afast and Aslow for the naive case were (pA): 4.4 and 23.5 at −80 mV, 19.7 and 29.9 at −90 mV, 34.9 and 36.4 at −100 mV, 50.2 and 42.8 at −110 mV, 65.4 and 49.2 at −120 mV. Values for the +vitronectin case were (pA): 21.7 and 36.5 at −80 mV, 42.0 and 38.0 at −90 mV, 62.4 and 39.5 at −100 mV, 82.7 and 41.0 at −110 mV, 103.1 and 42.4 at −120 mV. Idealized currents were calculated as the sum of two exponential functions: I=Afast exp(−tfast) +Aslow exp(−tslow).

Figure 7.

Model demonstrating that differences in V½ and activation rates can arise from changes in the relative proportions of fast and slow process without changes in underlying voltage dependence
A, fast and slow τ–V relations (lines) calculated as described in Methods from parameters chosen to approximate data (symbols) acquired from CA3 neurones (shown are averages of values for the naive and vitronectin-exposed neurones of Fig. 5B3). B, calculation of values for Afast and Aslow (lines) by linear regression of the measurements in Fig. 5B3 (symbols). C, normalized GV relations predicted from the forward and backward rate constants (αv and βv) used to calculate the fast and slow τ–V relations in A, in the proportions Afast: Aslow from values calculated in B. Fits of the Boltzmann relation to these curves yield V½=−82.8 mV and k= 7.3 for the predicted naive case, and V½=−80.4 mV and k= 7.0 for the predicted vitronectin-exposed case V½=+2.4 mV). D, values of t½ (symbols) emerging from currents calculated using values of τfast and τslow from A and values of Afast and Aslow for naive and vitronectin-exposed neurones from B. These predicted values are similar to but slower than those shown in Fig. 2D3, and they imply a more positive V½ since the maximum activation time (in a two-state system) will occur at V½. Quantitative differences in V½ may arise because the values of τfast and τslow used in the model (drawn from Fig. 5B3) were determined for currents activated by 15-s-long voltage steps, while the values of t½ in Fig. 2D3 were measured from currents during 3-s-long voltage steps. It has been observed, consistent with this pattern, that V½ measurements are sensitive to test pulse duration, with shorter pulses favouring more negative values of V½ (Chen et al. 2001b; Ulens & Tytgat, 2001).

Figure 4.

Voltage dependence of Ih activation
AC, normalized conductance-voltage (GV) relations for pyramidal neurones: naive or wild-type (open symbols) and vitronectin-exposed or vitronectin-deficient (filled symbols). Chord conductance values (G) for each cell (see Methods) were fitted with Boltzmann relations of the form G=Gmax/(1 + exp[(VV½)/k]) (where Gmax is the extrapolated maximum conductance, V is the test voltage, V½ is the half-activation voltage, and k is the slope factor), and then normalized to that cell's Gmax. Mean values of Gmax are indicated (see also Table 1), and the positions of mean V½ are indicated by the vertical dotted lines. The Boltzman relations shown were calculated for the values of mean Gmax, V½, and k given in Table 1. Differences in Gmax were significant for all variations in vitronectin exposure. Only in CA3 neurones did V½ show a statistically significant difference with vitronectin exposure – the differences in CA1 neurones being smaller but consistent in direction with vitronectin exposure or deficiency. This analysis utilized the same cell population as Fig. 2 and Table 1.

Figure 5.

Kinetic properties of Ih activation
A, pairs of capacity- and leak-subtracted currents from representative naive and vitronectin-exposed (+vn) Swiss-Webster CA3 pyramidal neurones, scaled to match the steady state amplitudes of the naive traces. For each trace, the activation time course was fitted by the sum of two exponential functions (shown superimposed in red): Ih=Afast exp(−tfast) +Aslow exp(−tslow), where Afast, τfast, Aslow and τslow are the amplitudes and time constants of fast and slow activation components, respectively. For these calculations, values for amplitudes and time constants were unconstrained. The rapid downward deflections in the current traces are synaptic currents. B and C, characteristics of fast and slow kinetic components in naive and vitronectin-exposed CA1 (left) and CA3 (right) neurones, and in wild-type and vitronectin-deficient CA1 neurones (centre). Data from control neurones are shown in black and data from experimental neurones are shown in red or green. Fast and slow component data are shown using circles and squares. B13, the activation time constants τfast and τslow did not differ significantly between naive and vitronectin-exposed neurones or wild-type and vitronectin-deficient neurones (P > 0.05 at all voltages). C13, vitronectin sensitivity of the fast and slow component amplitudes Afast and Aslow. Afast was significantly larger at most voltages in vitronectin-exposed CA1 and CA3 neurones, and Aslow showed a similar trend. Reciprocally, Afast was significantly smaller in vitronectin-deficient CA1 neurones, with Aslow showing a parallel trend. Notice that Afast increased with hyperpolarization (as expected from the increase in driving force), while Aslow was relatively constant at more hyperpolarized voltages. nCA1:Swiss-Webster= 6 for naive, 5 for + vitronectin; nCA1:C57BL/6J= 15 for wild-type, 17 for vitronectin−/−; nCA3:Swiss-Webster= 3 for naive, 6 for +vitronectin. Data are mean ±s.d.; *P < 0.05, **P < 0.01, and ***P < 0.001.


For immunostaining of cultured hippocampal slices, the tissue and adjacent culture membrane were excised from the Millicell insert and fixed in 4% paraformaldehyde in PBS (phosphate-buffered saline; pH 7.4) for 2 h at 4°C, and then rinsed in PBS (3 times, 15 min each). Slices were then permeabilized with 0.1% Triton-X (Sigma, St Louis, MO, USA) in PBS containing 3% BSA (bovine serum albumen) and 5% normal goat serum for 1 h at room temperature, and then incubated for 12–14 h at 4°C in primary antibody in PBS with 3% BSA: anti-HCN1 (AB5884; 1.5 μg ml−1) or anti-HCN2 (AB5378; 2.0 μg ml−1) (both from Chemicon, Temecula, CA, USA). After rinsing in PBS (3 times, 20 min each), sections were incubated in fluorescein-conjugated goat anti-rabbit IgG (Zymed, South San Francisco, CA, USA) diluted 1: 100 in PBS containing 5% normal goat serum for 1 h at room temperature. Finally, sections were rinsed in PBS (3 times, 20 min each) and mounted in Vectashield (Vector Laboratories, Burlingame, CA, USA). Images were collected on a laser scanning confocal microscope (Zeiss 310; Jena, Germany) using 10× air and 40× oil-immersion objectives. Montage acquisition software (written locally) was used to scan large areas at high resolution.

There are several indications that the immunoreactivity observed using antibodies directed against HCN1 and HCN2 showed adequate specificity. First, in our previous study using these same antibodies, we observed that all HCN1 or HCN2 immunoreactivity on mouse brain sections was sensitive to pre-incubation with the appropriate peptides (see Figs 9 and 10 of Vasilyev & Barish, 2002). In addition, as indicated by the manufacturer (product data sheets for Chemicon AB5584 and AB5378;, these affinity-purified antibodies were raised against unique synthetic peptides, reacted only with bands of appropriate molecular weight on Western blots of rat brain membranes, and were blocked by pre-incubation with the immunizing peptide.

For comparisons of HCN subunit expression, groups of cultured slices (naive and vitronectin-exposed (+vitronectin)) were processed in parallel, and images were acquired using identical confocal microscope parameters. A semiquantitative analysis of immunofluorescence was performed by establishing a common threshold luminance for each group that was positioned just above the background level (determined to include the entire slice while eliminating blank areas), and then extracting all pixel luminance values above this threshold (Optimas 6.2; Media Cybernetics, Silver Spring, MD, USA). The threshold luminance was then subtracted from each pixel luminance value, and the luminance of each pixel in each image was then normalized relative to the mean (threshold-subtracted) luminance of the naive slices. Statistical comparisons of mean relative pixel luminance (t test) and of cumulative probability of relative pixel luminance (log rank test) were made using Prism 4 (GraphPad). These computations were greatly simplified by the fact that each of the millions of (8-bit) pixels in these images could only have one of 256 values.


Ih in CA1 and CA3 pyramidal neurones

In hippocampal pyramidal neurones, Ih is activated during voltage excursions negative to the resting potential. We examined Ih using whole-cell voltage clamp techniques, and, as illustrated in Fig. 1A, separated time-dependent Ih from time-independent (at this time scale) IK(ir) based on the differential sensitivity of the latter current to Ba2+. With IK(ir) blocked by bath application of 0.5 mm Ba2+, all time-dependent current recorded at voltages between −60 and −140 mV (centre) was sensitive to the antagonist ZD7288 (Gasparini & DiFrancesco, 1997) (not shown), and therefore identified as Ih.

Figure 1.

Separation of Ih and IK(ir)
A, two hyperpolarization-activated currents, Ih and IK(ir), were distinguished on the basis of differential sensitivity to low concentrations of Ba2+. On the left is total current recorded during 2.5-s-long steps from −40 mV to voltages between −50 and −110 mV (in 10 mV increments); in the centre is Ih, recorded in the presence of 0.5 mm Ba2+; on the right is Ba2+-sensitive IK(ir) determined by subtracting the previous two sets of traces. The Ba2+-resistant current was identified as Ih by its slow activation and deactivation kinetics, and its sensitivity to the blocker ZD7288 (100 μm) (not shown). These traces were not leak- or capacity-subtracted; the fast downward transients are synaptic currents not blocked in these recordings. B, current-voltage (I–V) relations for Ih (•) and IK(ir) (▾) in CA1 and CA3 pyramidal neurones; measurements were taken at the end of 2.5-s-long hyperpolarizations from −40 mV. Ih amplitude was larger in CA1 pyramidal neurones (statistically significant at all voltages negative to −70 mV and almost so at −60 mV), its activation threshold was more positive (arrows), and its amplitude relative to IK(ir) was significantly greater (see Results). (n= 9 for CA1, 17 for CA3). All data in this figure were taken from slices cultured for 3–4 days from P0–P1, and are presented as mean ±s.d.; *P < 0.05, **P < 0.01, and ***P < 0.001.

I h in CA1 pyramidal neurones was significantly larger than in CA3 neurones at all voltages negative to and including −70 mV (Fig. 1B), and had a more positive activation threshold (near −50 mV versus near −60 mV in CA3 neurones; arrows). Further, Ih was proportionally larger than IK(ir) in CA1 neurones; the ratio of Ih to IK(ir) was 2.2 ± 1.0 in CA1 versus 1.0 ± 0.6 in CA3 at −110 mV (P < 0.001; all data are presented as mean ±s.d., n values indicated in legends where appropriate). Differences in this balance between Ih and IK(ir) may help differentiate excitability in CA1 and CA3 pyramidal neurones (see also Takigawa & Alzheimer, 2003).

Ih sensitivity to vitronectin

To investigate regulation of Ih, we either cultured P0–P1 hippocampal slices from Swiss-Webster mice in the presence or absence of added vitronectin (10 μg ml−1 to serum-free medium), or cultured slices from vitronectin-deficient mice (Zheng et al. 1995) of the same age and compared currents to those of the C57BL/6J background strain. In either case, we examined Ih and other currents after 3–4 days in culture. Figure 2A1 and 2 show representative traces from naive and vitronectin-exposed neurones CA1 and CA3 pyramidal neurones; Ih was recorded during hyperpolarizing steps to a test voltage of −100 mV. These records are superimposed to illustrate the larger amplitude and more rapid activation of Ih in vitronectin-exposed neurones.

Figure 2.

Reciprocal sensitivity of Ih to vitronectin exposure and deficiency
A1 and 2, representative traces recorded at −100 mV from naive and vitronectin-exposed Swiss-Webster CA1 and CA3 pyramidal neurones. Note that currents from vitronectin-exposed neurones are larger and activate more rapidly. As in Fig. 1, traces were not leak- or capacity-subtracted, and the downward transients are not-blocked synaptic currents. B13, I–V relations illustrating the larger Ih observed in vitronectin-exposed CA1 and CA3 neurones, and the reciprocally smaller Ih of vitronectin-deficient CA1 neurones. C13, ○ plot percentage differences in Ih amplitude at each voltage for the I–V relations in panels B13, computed as ([(Ih(experimental)/Ih(naive/wild-type)) − 1]× 100). These are reasonably independent of voltage for vitronectin-exposed and vitronectin-deficient CA1 neurones, but monotonically increase with voltage in vitronectin-exposed CA3 neurones. The continuous line shows the differences predicted from the values of Gmax, V½ and k derived from Boltzmann analyses of GV relations (Fig. 4AC), the grey diamonds show the differences predicted from variations in Gmax only, and the grey lines show the differences predicted from variations in V½ and k only. See the text for further details. D13, times for currents to reach one-half steady-state amplitude (t½), and their reciprocal sensitivity to vitronectin exposure or deficiency. The inflection point for CA3 neurones (D3) roughly corresponds to the position of V½ (see Fig. 4C). nCA1:Swiss-Webster= 27 for naive, 25 for + vitronectin; nCA1:C57BL/6J= 14 for wild-type, 21 for vitronectin−/−; nCA3:Swiss-Webster= 21 for naive, 33 for + vitronectin. Data from these same cells were used for construction of Fig. 4 and Table 1. Data are mean ±s.d.; *P < 0.05, **P < 0.01, and ***P < 0.001.

I h amplitude in pyramidal neurones was sensitive to vitronectin: significantly larger in vitronectin-exposed neurones, and reciprocally, smaller in neurones from vitronectin-deficient slices (current-voltage, I–V, relations of Fig. 2B13). In CA1 neurones, percentage differences in mean Ih amplitude (open circles in Figs 2C1 and 2) were reasonably independent of voltage between −80 and −120 mV, showing increases of 20–38% with vitronectin exposure, and decreases of 20–25% for vitronectin deficiency. In CA3 neurones, increases in Ih amplitude showed a pronounced curvature with voltage, from 28% at to 120 mV to 104% at −80 mV (Figs 2 and 3).

Figure 3.

Preferential vitronectin sensitivity of Ih
Current amplitudes and densities in a population of CA1 and CA3 pyramidal neurones for which measures of Ih, IK(ir) and whole-cell capacitance (Cm) were all available. Currents were recorded during steps to −100 mV from −40 mV, and separated as described in Fig. 1. A, current amplitude; B, whole-cell capacitance Cm (as an index of somatic and proximal dendritic membrane area); C, current density expressed as pA/pF. Only Ih amplitude and density differed significantly between naive and vitronectin-exposed neurones. nCA1= 9 for naive, 9 for + vitronectin; nCA3= 17 for naive, 33 for + vitronectin. Parenthetically, vitronectin deficiency did not significantly affect Cm of CA1 neurones (61.2 ± 14.2 pF versus 60.7 ± 5.7 pF for wild-type; P > 0.05). Also, Cm of naive Swiss-Webster CA1 pyramidal neurones (47.8 ± 8.0 pF, n= 9) was significantly smaller than that of naive CA3 neurones (62.0 ± 15.8 pF, n= 17; P < 0.05) and that of CA1 neurones in slices from wild-type C57BL/6J mice (60.7 ± 5.7 pF, n= 15; P < 0.001). Data are mean ±s.d.; *P < 0.05, **P < 0.01, and ***P < 0.001.

I h activation was also sensitive to vitronectin: faster and slower with vitronectin exposure and deficiency, respectively (Fig. 2D13). For CA1 neurones, mean times to reach one-half the amplitude at the time of repolarization were 15–18% shorter between −80 and −120 mV in vitronectin-exposed neurones, with the maximum difference at −90 mV (P < at least 0.05 at all voltages), and 15–28% longer in neurones from vitronectin-deficient slices, with the maximum difference between −90 and −100 mV (P < at least 0.05 at all voltages). For CA3 neurones, one-half activation times were 9–30% shorter between −90 and −120 mV, with the maximum difference at −100 mV (P < at least 0.05 at all voltages).

Note that neurones derived from vitronectin-deficient mice were themselves responsive to vitronectin. For example, Ih amplitude at −100 mV in vitronectin-exposed vitronectin−/− CA1 neurones was 269.0 ± 40.2 pA versus 209.5 ± 31.9 pA for neurones in control vitronectin-deficient slices (P < 0.05; n= 5 for vitronectin−/−, 4 for +vitronectin), and in CA3 neurones was 73.6 ± 10.2 pA versus 54.8 ± 14.3 pA in control vitronectin−/− neurones (P < 0.05; n= 6 for vitronectin−/−, 7 for +vitronectin).

Exposure to vitronectin did not appear to affect somatic and proximal dendritic membrane area, as assessed by measurements of whole-cell capacitance (Cm). There were no statistically significant differences in Cm between naive and vitronectin-exposed CA1 neurones or CA3 neurones (Fig. 3B), nor between CA1 neurones in wild-type and vitronectin-deficient slices (Fig. 3 legend).

Further, the effects of vitronectin were confined to Ih; neither the amplitude or density of IK(ir) were altered by vitronectin exposure, which therefore shifted the balance between Ih and IK(ir) in favour of Ih. This is illustrated in Figs 3A and C, where Ih and IK(ir) amplitudes at −100 mV are compared. At this voltage, Ih amplitude will be sensitive to variation in both Gmax (maximum available conductance) and V½ (voltage of one half-activation as determined by fits to Boltzmann relations). In CA1 and CA3 neurones for which measures of IK(ir) along with Ih were available, mean Ih amplitude in vitronectin-exposed neurones was significantly larger, by 31.5% in CA1 and by 73.2% in CA3. Ih densities followed: 31.6% larger in CA1 and 70.8% larger in CA3. In these same neurones, there were no significant differences in IK(ir) amplitude or density.

Voltage dependence and kinetics of Ih activation

We examined the voltage-dependence of Ih activation by constructing conductance-voltage (GV) relations using chord conductances calculated from steady-state current amplitudes (see Methods). Shown in Fig. 4AC are GV relations for CA1 and CA3 pyramidal neurones: naive or wild-type (open symbol), and vitronectin-exposed or -deficient (closed symbol). For these comparisons, conductance measurements for an individual cell were normalized relative to its Gmax. The continuous lines are fits of these values to Boltzmann relations; the positions of mean V½ (Table 1) are indicated by the vertical dotted lines. A and C show data for CA1 and CA3 neurones exposed to vitronectin (Swiss-Webster mice); B shows data for CA1 neurones in vitronectin-deficient slices (C57BL/6J mice).

Table 1.  Boltzmann analyses of GV relations with variation in vitronectinThumbnail image of

In CA1 neurones (Figs 4A and B; Table 1), only Gmax was significantly sensitive to vitronectin, being 23.3% larger in vitronectin-exposed neurones and 17.9% smaller in vitronectin-deficient neurones. In these cells the positions of GV relations (as indicated by V½) were not significantly altered by vitronectin exposure (displacement of +1.9 mV from −80.0 ± 4.9 mV to −78.1 ± 2.3 mV; P > 0.05) or vitronectin deficiency (displacement of −1.8 mV from −82.0 ± 2.9 mV to −83.8 ± 3.1 mV; P > 0.05). In CA3 neurones (Fig. 4C; Table 1), Gmax was 35.4% larger in vitronectin-exposed cells, and GV relations showed a further statistically significant +3.0 mV displacement (from −92.8 ± 3.7 mV to −89.8 ± 5.8 mV; P < 0.05).

The behaviours of the GV relations in Fig. 4AC are reflected in the percentage changes in Ih amplitude shown in Fig. 2C13, where the continuous lines are calculated from differences in the I–V relations predicted from the three Boltzmann parameters Gmax, V½ and k. Clearly, in CA1 neurones, much of the difference in current amplitudes derives from differences in Gmax (shown as grey diamonds superimposed on the vertical axis), with variation in V½ and k (grey lines), contributing some increase only at more positive values. In CA3 neurones, the non-linear increases in Ih amplitude reflect differences in V½ and k in addition to the larger Gmax.

To examine kinetic differences illustrated in Fig. 2D13 in more detail, traces were fitted with the sum of two exponential functions: Ih=Afast exp (−tfast) +Aslow exp(−tslow), where Afast and Aslow are the amplitudes of two distinct kinetic components with time constants τfast and τslow. This function provided a good description of the time course of Ih activation, as illustrated in Fig. 5A, which shows pairs of representative traces (capacity- and leak-subtracted) from naive and vitronectin-exposed CA3 pyramidal neurones, scaled to match their maximum amplitudes, and with fitted functions superimposed in red.

Shown below are data summarizing the activation properties of Ih: the activation time constants τfast and τslow in Fig. 5B13, and the amplitudes Afast and Aslow in Fig. 5C13. The time constants τfast and τslow were voltage dependent, becoming more rapid at increasingly negative voltages, and differed by about an order of magnitude at all voltages.

Consider the data for naive CA1 and CA3 neurones, which indicate that two factors contribute to the slower activation of Ih in CA3 pyramids. First, the intrinsic time constants of CA3 neurones are slower. For example, in naive (black curves in Fig. 5B1 and 3) Swiss-Webster CA3 neurones at −90 mV, τfast was 552.5 ± 174.3 ms versus 323.1 ± 96.6 ms in CA1 (P < 0.05), and τslow was 5353.5 ± 1997.6 versus 2423.8 ± 513.2 in CA1 (P < 0.01). Second, the relative contribution of the fast component to total Ih is smaller in CA3 neurones, where Afast was often smaller than Aslow (compare • and ▪ between Figs 5C1 and 3) and represented an average of 47.3 ± 8.4% (pooled across −90 to −120 mV) of total Ih. In CA1 neurones, Afast was always larger than Aslow and contributed an average of 63.1 ± 5.7% (pooled across the same voltages) to the total (P < 0.05 as compared to CA3).

Vitronectin, in contrast, did not appear to affect the values of time constants underlying Ih in either CA1 or CA3. As illustrated in Fig. 5B1–3 (compare black with red or green curves as appropriate), τfast and τslow did not differ between naive and vitronectin-exposed neurones, or between neurones derived from wild-type or vitronectin-deficient mice. Comparisons of mean values for τfast and τslow at each voltage did not reveal statistically significant differences (P > 0.05 at all voltages), nor did the slopes of τ–V relations for τfast and τslow (evaluated as mV per e-fold change) differ significantly (P > 0.05).

Rather, only the amplitudes of Afast and Aslow, and their relative contributions to total Ih, differed with vitronectin exposure or deficiency (Fig. 5C13). The average percentage difference in Afast (pooled values between −90 and −120 mV) in vitronectin-exposed neurones was +34.7 ± 10.4% in CA1 and +74.2 ± 44.9% in CA3, and in vitronectin-deficient neurones was −23.0 ± 5.2% in CA1 (P < 0.05 at virtually all voltages). Over these same voltages, Aslow was also consistently larger in vitronectin-exposed as compared to naive neurones (+18.7 ± 5.3% in CA1 and +7.8 ± 22.2% in CA3) and smaller in vitronectin-deficient as compared to wild-type neurones (−15.0 ± 9.8% in CA1), although these measurements were very variable and differences did not reach statistical significance. Nevertheless, at every voltage between −90 and −120 mV, percentage differences in Afast were larger than those of Aslow. Thus, embedded in differences in Ih amplitude seen with vitronectin exposure or deficiency is a consistent preferential sensitivity of Afast to vitronectin, and thus variation in its relative contribution to total Ih.

Immunochemical examination of HCN1 and HCN2

The potential molecular underpinnings of these variations in Ih include the four members of the HCN channel gene family expressed in mammalian neurones (HCN1, HCN2, HCN3 and HCN4; Santoro & Tibbs, 1999). HCN1 and HCN2 are highly expressed in hippocampal pyramidal neurones, and native hippocampal Ih channels are probably heteromultimers incorporating HCN1 and/or HCN2 in varying combinations with other subunits (see Discussion).

We therefore examined expression of HCN1 and HCN2 proteins in cultured naive and vitronectin-exposed slices by immunofluorescence. Shown in Fig. 6A are representative images of HCN1 immunoreactivity; grey-scale images are shown above and pseudocoloured images below. Slices were grown and processed in parallel, and images were acquired using identical parameters. Visual comparison suggests a generalized luminance increase throughout the vitronectin-exposed slice, with particular accumulations of HCN1 immunoreactivity in entorhinal cortex adjacent to the hippocampus (region a in Fig. 6A), in the striatum radiatum and striatum lacunosum-moleculare of CA1 (region b), and in an area of the slice adjacent to the fornix (region c).

Figure 6.

Analyses of HCN1 and HCN2 immunoreactivity
A, fluorescence images of HCN1 immunoreactivity in cultured hippocampal slices. Images are shown above using a greyscale, and below using a pseudocolour scale to better illustrate luminance differences between the naive and vitronectin-exposed slices. B and C, cumulative probability histograms of HCN1 and HCN2 immunoreactivity, quantified as pixel luminance normalized to the average pixel luminance of the naive slices in each set of cultures (see Methods). Shown in each case are background luminance (black; determined by antigen absorption of the primary antibody or use of the secondary antibody only), and luminance of naive (blue) and vitronectin-exposed (red) slices. The vertical broken lines indicate median values of relative luminance; the naive values are not exactly 1.00 because these are median values of luminance normalized to appropriate means. The HCN1 luminance histogram is clearly shifted towards brighter values in vitronectin-exposed slices, while the HCN2 luminance histograms essentially superimpose. nHCN1= 8 for naive, 7 for + vitronectin, nHNC2= 8 for naive, 8 for + vitronectin; each analysed in two independent sets of cultures. See Results for further details.

Semi-quantitative comparisons of immunoreactivity in naive and vitronectin-exposed slices suggest a selective increase in HCN1 immunoreactivity, with little change seen in HCN2 levels. In each experiment, we measured the luminance of each pixel in groups of naive and vitronectin-exposed slices, and normalized these values against the mean luminance of the naive slices (see Methods). Viewed as cumulative probability histograms of relative pixel luminance (Figs 6B and C), the rightward shift in the histogram for HCN1 immunoreactivity suggests higher HCN1 levels in vitronectin-exposed slices. In contrast, the histograms for HCN2 immunoreactivity virtually superimpose, suggesting little difference in HCN2 levels. Increases in mean HCN1 luminance were consistent with this interpretation, but were trends that did not reach statistical significance: relative HCN1 pixel luminance increased ∼15% from 1.00 ± 0.22–1.15 ± 0.21, while for HCN2, relative pixel luminance showed virtually no change (from 1.00 ± 0.18–1.03 ± 0.14). Median relative pixel luminance values showed a similar pattern (vertical lines to baseline in Figs 6B and C): increase of ∼15% from 0.96 ± 0.23–1.11 ± 0.21 for HCN1 and little or no change between 1.02 ± 0.19 and 1.04 ± 0.15 for HCN2.


We observed that over a period of days, Ih in pyramidal neurones of Ammon's horn is regulated by differences in vitronectin levels, with elevated vitronectin increasing Ih amplitude and speeding its activation. As a consequence, by altering the properties of Ih and affecting the balance between Ih and other currents such as IK(ir), vitronectin may influence the excitability and integrative properties of pyramidal neurones. We suggest that Ih regulation may occur, at least in part, by increases in the contributions of rapidly activating channel populations (probably incorporating HCN1 subunits) to total Ih, although additional actions of other mechanisms are not excluded. In the context of these experiments, vitronectin could either act directly on neurones, or indirectly by triggering signals originating in other cells, an issue awaiting further investigations.

Observations suggest that the larger Ih amplitudes seen in vitronectin-exposed neurones in vitro probably represent true enhancements of Ih. First, Ih development in naive neurones in vitro was nearly identical to in vivoIh development, suggesting that culture per se does not result in slower Ih development or its progressive decline. Drawing on data from our previous study of Ih maturation (Vasilyev & Barish, 2002; note that voltages in this publication were not corrected for junction potentials, but have been for this comparison), Ih amplitude at −120 mV in neurones of acutely cut P5 hippocampal slices was 233.5 ± 75.3 pA in CA1 n= 8; increase from 93.7 ± 16.8 pA on P1, P < 0.001, n= 9) and 92.1 ± 32.3 pA in CA3 (n= 9; increase from 59.5 ± 15.5 pA on P1, P < 0.05, n= 7). Ih amplitudes in neurones of naive cultured slices at approximately the equivalent time (3–4 days in culture from P0-P1) were almost identical: 227.0 ± 96.2 pA for CA1 (P > 0.05 as compared to in vivo values, n= 13) and 87.9 ± 20.0 pA for CA3 (P > 0.05, n= 19). Also, if vitronectin exposure resulted in increased Ih amplitude by a simple acceleration of development, then one might expect the intrinsic time constants of Ih activation to accelerate as they do during Ih maturation (Fig. 6 of Vasilyev & Barish, 2002). However, in the present experiments these time constants were not sensitive to the presence or absence of vitronectin (Fig. 5B13).

Potential mechanism(s) of Ih modification

In principle these long exposures to vitronectin might modify the properties of existing Ih channels, or stimulate increases in the synthesis and/or membrane insertion of additional functional channels, or some combination of the two.

In the mechanism we wish to suggest, one consistent with physiological and immunochemical evidence, vitronectin would, over a period of days regulate the numbers and relative contributions of rapidly activating Ih channels to total Ih without affecting the physiological properties of these channels. This class of mechanism can account for the differences in Gmax and variations in V½ and half-activation times that were seen in vitronectin-exposed or vitronectin-deficient neurones without requiring parallel changes in the time constants of underlying kinetic processes. This is illustrated by the modelling exercise shown in Fig. 7. Here we represented fast and slow kinetic processes as two independent two-state conductances with differing V½ values (see Methods), a simplification permitting us to manipulate each process separately. Kinetic parameters were chosen to approximate the values of τfast and τslow shown in Fig. 5 (symbols shown in Fig. 7A are averages of naive and + vitronectin values), and gave rise to the τfastV and τslowV relations shown by the lines. The amplitudes Afast and Aslow used in the model (lines in Fig. 7B) were chosen by fitting linear relations to the experimental values (symbols). Varying the magnitudes of Afast and Aslow as appropriate for naive and vitronectin-exposed CA3 neurones resulted in two predicted GV relations differing in V½ by +2.4 mV (Fig. 7C), and predicted currents whose half-activation times were smaller by 26–31% between −80 and −120 mV (Fig. 7D). These behaviours are qualitatively similar to those seen experimentally, and thus indicate that one need not, of necessity, invoke changes in intrinsic channel voltage dependence to account for apparent differences in V½ and activation rate.

As suggested by the immunochemical data, the channels underlying vitronectin-linked conductance probably incorporate HCN1 subunits. They may be similar to the heteromeric HCN1-containing (Santoro & Tibbs, 1999) subunits that are strongly expressed in CA1 neurones (Santoro et al. 1997; Moosmang et al. 1999; Santoro et al. 2000; Bender et al. 2001; Vasilyev & Barish, 2002). HCN1 is implicated in these changes because, in vivo, the presence of HCN1 subunits correlates with the fast activation phenotype of CA1 neurones (Franz et al. 2000), and, in comparison to wild-type CA1 neurones, Ih activates more slowly in HCN1-deficient neurones (Morozov et al. 2000; Nolan et al. 2000) and more rapidly in HCN2-deficient neurones (Ludwig et al. 2003). Further, in heterologous systems, the presence of HCN1 subunits determines the rapid activation kinetics of heteromeric channels (Chen et al. 2001b; Ulens & Tytgat, 2001).

In the simplest model, changes in the faster component (Afast) would represent alterations in the contributions of HCN1-dominated channels to total Ih, and any changes in the slower component (Aslow) would reflect differences in other subunits such as HCN2. While appealing, this scenario is unlikely because for both HCN1 and HCN2, heterologously expressed channels display both fast and slow kinetic components with values separated by factors of 5–10, and with absolute values about 10-fold larger for HCN2 channels (Santoro et al. 2000; Chen et al. 2001b; Ulens & Tytgat, 2001). As a consequence, time constants for the slower component of HCN1 channels and the faster component of HCN2 channels are similar (but not overlapping), and Aslow in hippocampal neurones may thus reflect contributions of both HCN1- and HCN2-dominated channels. The small but consistent differences in Aslow seen with vitronectin exposure and vitronectin deficiency are consistent with this interpretation, since the slow component comprises about 20% of total HCN1 channel current (Santoro et al. 2000; Chen et al. 2001b). Therefore, the relatively smaller vitronectin-sensitive differences in Aslow could reflect variation in HCN1 subunit levels and need not reflect additional differences in other subunits.

At the same time, our data do not exclude combination of cyclic nucleotides with Ih channels as contributing to the physiological changes observed. This is a potent and commonly observed mechanism by which Ih activation is shifted towards more positive voltages and Gmax is increased (DiFrancesco & Tortora, 1991; Pedarzani & Storm, 1995). Ih channels of CA1 pyramidal neurones are thought to incorporate both HCN1 and HCN2 subunits (see above) and are expected to have cyclic nucleotide responses dominated by the higher sensitivity of HCN2 (Chen et al. 2001b; Ulens & Tytgat, 2001). In principle, vitronectin could chronically regulate cAMP or other nucleotide levels, and over the long-term influence maximum Ih conductance and its voltage dependence (Tokimasa & Akasu, 1990; Chen et al. 2001b; Ulens & Tytgat, 2001). However, while our experiments did not directly address this question, when the cyclic nucleotide sensitivity of Ih in CA1 neurones has been directly examined in acute experiments, the changes in Gmax have been smaller, and the shifts in V½ larger, then we observed here. In the study of Gasparini & DiFrancesco (1999), for example, application of 5-HT and elevation of cAMP resulted in a +3.2 mV shift in V½ and increase in Gmax of 13.6%. Other studies, demonstrating cAMP-induced Gmax increases have invariably also reported significant shifts in V½ (e.g. Tokimasa & Akasu, 1990). Further, cyclic nucleotide-induced shifts in GV relations (and V½) have also been accompanied by parallel shifts in τ–V (e.g. DiFrancesco, 1999), a behaviour not evident in our present measurements but detectable under other circumstances (i.e. during development; Vasilyev & Barish, 2002). Thus while we do not favour this mechanism as the dominant contributor to the vitronectin-dependent differences in Ih observed, the potential contributions of cyclic nucleotide combination with HCN channels to the differences in Ih associated with vitronectin, as well as additional potential mechanisms of channel modification such as channel phosphorylation (e.g. Accili et al. 1997), should be examined in future experiments.

Physiological consequences of increased Ih

Sensitivity of Ih to vitronectin may be an emerging reaction of cultured hippocampal slices to the trauma of slicing. Vitronectin-deficient mice lack an overt phenotype (Zheng et al. 1995), and we have not observed significant differences in Ih of pyramidal neurones in acute slices from wild-type and vitronectin−/− mice (not shown). However, the hyperexcitability that normally develops in long-term hippocampal slice cultures (McBain et al. 1989; Malouf et al. 1990) is altered in slices derived from vitronectin-deficient mice (Aihara & Barish, 2001, 2002). How might sensitivity to vitronectin be limited to traumatized brain?

One possibility is that vitronectin levels may be elevated in damaged brain. Microglial activation after CNS damage is seen in vivo (Perry & Gordon, 1988) and in vitro (Coltman & Ide, 1996), and microglia emerging in cultured postnatal hippocampus display vitronectin immunoreactivity (Aihara & Barish, 2001, 2002). These observations suggest that activated microglia may normally synthesize and secrete vitronectin. Brain vitronectin levels could also increase after injury by breach of the blood–brain barrier and admission of serum vitronectin (Gingrich & Traynelis, 2000). Conformational lability of vitronectin may also limit its actions; exposure of vitronectin to molecules such as PAI-1 (Minor & Peterson, 2002), enriched at sites of injury (Gingrich & Traynelis, 2000), can alter its conformational state, potentially leading to display of previously hidden combining sites for integrins and other receptors (Seiffert & Smith, 1997). Injury might also alter expression of vitronectin receptors, thereby rendering neurones susceptible to vitronectin-initiated signalling.

Our observations may therefore be most relevant to understanding responses to damage. Persistent hyperexcitability accompanies many forms of neural pathology (Wasterlain & Shirasaka, 1994). Ih is potentially linked to such hyperexcitability since, using CA1 neurones as an example, it contributes to generation and tuning of intrinsic voltage resonance and rhythmicity (Suckling et al. 2000; Hu et al. 2002; Cobb et al. 2003; see also Chen et al. 2001a), and to shaping postsynaptic responses to repetitive excitatory input (Takigawa & Alzheimer, 2003). Thus vitronectin could potentially influence the repetitive activity characteristic of diseased or injured brain. Consistent with this interpretation, in early postnatal development, Ih density is highest during periods of spontaneous activity (Vasilyev & Barish, 2002). This possibility will be addressed in future investigations.

The specific effects of vitronectin-induced increases in Ih on cell excitability are difficult to predict a priori, however (Santoro & Baram, 2003). They will depend on a number of factors including: (a) the cellular context and the relation of Ih to other currents active at similar voltages, including IA, IK(ir), ICa(T) and INa(p), (b) the particular aspect of excitability examined, since synaptic potentials, single action potentials and patterned activities have different temporal and voltage characteristics, and, reciprocally, (c) the precise nature of the change in Ih, since specific properties of HCN subunits may preferentially affect different aspects of excitability.

Of particular interest is the recent study of MacLean et al. (2003) indicating that biosynthesis and insertion of IA and Ih channels is coordinately regulated. We previously described regulation of IA (but not delayed rectifier potassium currents) in pyramidal neurones at an earlier developmental stage (Vasilyev & Barish, 2003), and it will be interesting to determine the extent to which Ih and IA are regulated together by vitronectin in older hippocampal neurones.



We thank Dr David Ginsburg (HHMI, University of Michigan) for the gift of vitronectin-deficient mice, and M. Jill Brantley for assistance with the manuscript. This work was supported by grants from the March of Dimes Birth Defects Foundation (1-FY01-557) and the National Institutes of Health (R01NS23857).

Author's present address

D. V. Vasilyev: Neuroscience Discovery Research, Wyeth Research, Princeton, NJ 08543, USA.