Knockdown of the dyslexia-associated gene Kiaa0319 impairs temporal responses to speech stimuli in rat primary auditory cortex

Abstract

One in 15 school age children have dyslexia, which is characterized by phoneme-processing problems and difficulty learning to read. Dyslexia is associated with mutations in the gene KIAA0319. It is not known whether reduced expression of KIAA0319 can degrade the brain's ability to process phonemes. In the current study, we used RNA interference (RNAi) to reduce expression of Kiaa0319 (the rat homolog of the human gene KIAA0319) and evaluate the effect in a rat model of phoneme discrimination. Speech discrimination thresholds in normal rats are nearly identical to human thresholds. We recorded multiunit neural responses to isolated speech sounds in primary auditory cortex (A1) of rats that received in utero RNAi of Kiaa0319. Reduced expression of Kiaa0319 increased the trial-by-trial variability of speech responses and reduced the neural discrimination ability of speech sounds. Intracellular recordings from affected neurons revealed that reduced expression of Kiaa0319 increased neural excitability and input resistance. These results provide the first evidence that decreased expression of the dyslexia-associated gene Kiaa0319 can alter cortical responses and impair phoneme processing in auditory cortex.

Keywords: reading; specific language impairment; temporal processing; timing.

Figure 1.

In utero RNAi of Kiaa0319 (KIA−) caused delayed speech-evoked LFPs in both awake and anesthetized rats. LFPs in panels (A) and (C) were created by averaging across 47 sites from 9 awake KIA− animals and 6 sites from 3 awake control animals. LFPs in panels (B) and (D) were created by averaging across responses are averaged across all recording sites 247 sites from 5 anesthetized KIA− animals and 255 sites from 6 anesthetized control animals. Latency of the LFP was calculated by milliseconds to the peak of each component. (A) LFPs in response to the speech sound “dad” had a lower amplitude in KIA− neurons recorded from awake animals compared with controls. (B) These abnormal response properties were not affected by pentobarbital anesthesia. Speech-evoked LFPs to the sound “dad” had longer (but not significantly longer) latency and lower amplitude in KIA− neurons recorded from animals anesthetized with dilute pentobarbital compared with controls. (C) LFPs in response to the speech sound “bad” had a longer P1 latency than controls. (D) These abnormal properties were more obvious in the presence of pentobarbital anesthesia. Though the LFP response to the sound “dad” (A and B) were slightly different from the LFP response to the speech sound “bad” (C and D), the average response to these sounds was extremely similar.

Figure 2.

In utero RNAi of Kiaa0319 caused degraded spatiotemporal response patterns to consonant speech sounds. (A) Consonant sounds evoked unique patterns of activation across the A1 of anesthetized control rats. Each multiunit site's average response to the speech sound over 20 repeats is organized by the characteristic frequency of the site. The average response across all sites is shown on top of each subpanel. These patterns are similar to previous studies using unaffected rats (Engineer et al. 2008; Perez et al. 2012; Ranasinghe, Vrana, et al. 2012; Shetake et al. 2011). (B) Response patterns of multiunit recordings taken from primary auditory cortex of anesthetized rats that had undergone in utero RNAi. In utero RNAi of Kiaa0319 caused delayed response to speech sounds, as well as reduced precision in firing latency.

Figure 3.

In utero RNAi of Kiaa0319 caused delayed activity to speech sounds. (A) Onset latency and peak latency in response to the consonant sound “d” were longer in multiunit recordings from anesthetized rats transfected with Kiaa0319 shRNA in utero (*P < 0.01). (B) Average onset latency and peak latency to consonant stimuli were later in multiunit recordings from anesthetized KIA− rats as compared to anesthetized control sites (*P < 0.01).

Figure 4.

In utero RNAi of Kiaa0319 caused increased variability in neural responses to speech stimuli. (A) Onset latency to speech sounds was more variable across trials in multiunit responses from anesthetized KIA− rats (light bar) compared with sites from anesthetized controls (dark bar; *P < 0.01). (B) The number of action potentials fired during speech sounds was more variable trial to trial in KIA− sites (light bar) during vowels (400-ms window; *P < 0.01). (C) The average number of evoked action potentials to speech sounds was not significantly different between control and KIA− sites for either a 40-ms or a 400-ms analysis window (P = 0.10). This result suggests that the increased trial-by-trial variability is not due to an increased firing rate.

Figure 5.

In utero transfection with Kiaa0319 shRNA caused impaired neural discrimination of consonant and vowel stimuli. (A) Single-trial neural responses from a typical high-frequency control site. Action potentials are plotted over the course of the consonant sound and organized by repeat (each row is a different repeat). To the sound “d,” this site fired quickly after the sound onset, while this site fired with a 10-ms delay to the sound “b.” Repeats during which the classifier guessed incorrectly as to the stimulus identity are marked by an “x” on the right. Total percent correct for this pair is on top of the panel. (B) Single-trial neural responses from a typical high-frequency KIA− site. To the sound “d,” this site fired later than the control site, and with greater variability across repeats. To the sound “b,” this site fired later than the control site, and with greater variability across repeats. This site made many more errors in identifying the source stimulus as compared to the control, due to the amount of variability across repeats. (C) Average percent correct across 10 neural consonant discrimination tasks. A 2-alternative forced-choice classifier compared the Euclidean distance between a single-trial response and the average response to 2 consonant stimuli. Using activity from a single site in 1 ms bins, the classifier guesses which sound evoked the single-trial activity by picking the comparison yielding the smallest Euclidean distance. KIA− neurons were significantly impaired at discriminating between consonant sounds (*P < 0.01). (D) Single-trial neural responses from a typical low-frequency control site. Average number of spikes fired per repeat is plotted by repeat. To the vowel sound “a” (as in “dad”), this site fired more spikes than in response to the vowel sound “u” (as in “dud”). Vowels were more difficult than consonants to discriminate, as the classifier compared each trial's spike count with the average spike count for that sound (vertical lines) and this is reflected by a greater number of classifier errors, as marked by an “x” on the right. (E) Single-trial neural responses from a typical low-frequency KIA− site. The average number of spikes per vowel was more similar in KIA− sites (vertical lines), which makes the classifier guess incorrectly more often. (F) Average percent correct across 6 neural vowel discrimination tasks. The same neural classifier used single site activity in 400-ms bins to identify which vowel sound evoked the single-trial activity using Euclidean distance. KIA− neurons were significantly impaired at discriminating between vowel sounds (*P < 0.01).

Figure 6.

In utero RNAi of Kiaa0319 impaired neural firing properties to repetitive broadband stimuli. (A) Example of single site responses to the first broadband noise burst. Each row represents a repeat of the stimulus, and each dot shows the location of an action potential with respect to time. The control site (top) responded consistently across repeats, while the KIA− site (bottom) responded later and with more variability across repeats. (B) Onset latency to the first broadband noise burst was later in KIA− neurons compared with control neurons (*P = 0.01). (C) Variability in peak latency across repeats was significantly longer in KIA− neurons compared with control neurons (*P < 0.01). (D) Firing rate to each broadband burst was lower in KIA− neurons as compared to controls at all 4 presentation rates (*P < 0.01). (E) VS was significantly lower in KIA− neurons at all 4 presentation rates (*P < 0.01).

Figure 7.

In utero RNAi of Kiaa0319 caused impaired neural firing to tonal stimuli. (A) Spontaneous levels in RNAi rats were higher than controls (*P < 0.01). Similar to speech and noise-burst stimuli, KIA− neurons had a longer peak latency to tone stimuli compared with control neurons (*P < 0.01). We used only tones louder than 60 dB to plot the average response. This cutoff ensured that the narrower bandwidths observed in KIA− recordings did not bias the comparison. (B) Average number of spikes to tones at each intensity tested. Number of spikes for each site was counted within the 10 sites surrounding that site's CF to account for bandwidth difference. KIA− neurons fired more spikes to tones than control neurons at most intensities tested (*P < 0.01).

Figure 8.

Basic neural firing impairments may contribute to poor neural classification of phonemes. We tested groups of control neurons that mimicked the distribution of KIA− neurons with respect to several neural firing properties. All * are P < 0.01. (A) Several abnormal firing properties contributed to poor consonant classification. Control neurons with a distribution of spontaneous firing (third bar; 68 ± 1%; P < 0.01), peak latencies (fourth bar; 68 ± 1%; P < 0.01), and bandwidths (at 20 dB above threshold, fifth bar; 68 ± 1%; P < 0.01) that match the KIA− sites were significantly different from the full set of control sites on neural discrimination of consonants. (B) Abnormal spontaneous firing levels seemed to contribute to poor vowel performance. Control neurons with a distribution of spontaneous firing rates that match the KIA− sites were significantly different from the full set of control sites on neural discrimination of vowels (57 ± 1%; P < 0.01).

Figure 9.

Neurons with transfection of Kiaa0319 shRNA had higher membrane resistance than controls. Four experimental groups were used in this experiment: KIAA RNAi neurons expressed the Kiaa0319 shRNA transgene, KIAA OE (over expression) neurons expressed a transgene which increased Kiaa0319 expression, scramble neurons expressed a scrambled shRNA, and rescue neurons expressed both the Kiaa0319 shRNA (which reduced expression) and a transgene to increase expression. (A) KIAA RNAi neurons fired significantly more action potentials per pÅ of current injection than scramble controls (at 200 pÅ of current, KIAA RNAi neurons fire 5.5 ± 1 spike vs. 0.5 ± 0.5 spikes in controls; *P < 0.01). (B) Example voltage traces from 1 KIAA RNAi neuron and 2 control neurons. In response to current injection, membrane potential of KIAA RNAi neurons changed more drastically and fired more action potentials than scramble controls. (C) Input resistance function for all 4 experimental groups tested. For each pÅ of current injected, the membrane potential of KIAA RNAi neurons changed significantly more than control groups (at 100 pÅ of current, KIAA RNAi membrane potential changed by 18.4 ± 3.6 vs. 8.1 ± 1.1 mV in controls; *P = 0.02). (D) KIAA RNAi neurons had a higher membrane resistance than scramble control neurons (193.7 ± 25.3 MΩ in KIA− cells vs. 103.6 ± 21.4 MΩ in scramble control; *P = 0.01). KIAA OE and rescue controls did not have increased membrane resistance (*P = 0.05).