(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Differences in temporal processing speeds between the right and left auditory cortex reflect the strength of recurrent synaptic connectivity [1]

['Demetrios Neophytou', 'Cuny Graduate Center', 'New York', 'United States Of America', 'Diego M. Arribas', 'Department Of Neurobiology', 'Behavior', 'Stony Brook University', 'Stony Brook', 'Biomedicine Research Institute Of Buenos Aires']

Date: 2022-11 

Brain asymmetry in the sensitivity to spectrotemporal modulation is an established functional feature that underlies the perception of speech and music. The left auditory cortex (ACx) is believed to specialize in processing fast temporal components of speech sounds, and the right ACx slower components. However, the circuit features and neural computations behind these lateralized spectrotemporal processes are poorly understood. To answer these mechanistic questions we use mice, an animal model that captures some relevant features of human communication systems. In this study, we screened for circuit features that could subserve temporal integration differences between the left and right ACx. We mapped excitatory input to principal neurons in all cortical layers and found significantly stronger recurrent connections in the superficial layers of the right ACx compared to the left. We hypothesized that the underlying recurrent neural dynamics would exhibit differential characteristic timescales corresponding to their hemispheric specialization. To investigate, we recorded spike trains from awake mice and estimated the network time constants using a statistical method to combine evidence from multiple weak signal-to-noise ratio neurons. We found longer temporal integration windows in the superficial layers of the right ACx compared to the left as predicted by stronger recurrent excitation. Our study shows substantial evidence linking stronger recurrent synaptic connections to longer network timescales. These findings support speech processing theories that purport asymmetry in temporal integration is a crucial feature of lateralization in auditory processing.

Here, we tested directly whether there are differences in the temporal integration properties between the auditory cortices and dissected the underlying circuit dynamics. To assess cortical circuit mechanisms that could underlie hemispheric asymmetry in temporal processing, we used circuit-mapping techniques to screen for connectivity differences in excitatory pathways. We show novel differences in recurrent connectivity between the hemispheres, particularly in the superficial layers. To examine how these asymmetries in recurrence translate into differences in temporal integration, we recorded spontaneous spikes from awake mice and developed a statistical method to estimate time constants and their uncertainty from spike trains. Our method utilizes the dichotomized Gaussian (DG) spiking model to generate surrogate spike trains with a data matching autocorrelation function. The estimated time constants from each neuron recorded in superficial layers of the left and right ACx were aggregated to form a distribution over the network time constants. We found significant differences in temporal integration consistent with the timescale of hypothesized lateralized auditory signal processing. Numerous models of neural architectures have been proposed to account for observed differences in integration timescales throughout the brain. Here, we show, for the first time, compelling evidence of differences in synaptic circuit organization that translate into distinct temporal integration windows.

Studies of animal models can provide more mechanistic insight regarding circuit function. Circuit mapping of the mouse auditory cortex (ACx) has shown that the synaptic organization of Layer 3 (L3) differs between the 2 hemispheres. In the left ACx, principal neurons in L3 receive out-of-column excitatory input from L6 cells located in higher frequency bands, whereas in the right ACx, the same pathway has balanced frequency projections across the tonotopic axis. This lateralized synaptic organization is in turn associated with differences in sound-evoked activity in L3 [ 6 ]. Behavioral studies in gerbils have reported asymmetries in temporal processing. Lesions in the right ACx impact discrimination of frequency sweep direction, suggesting it plays a role in processing global temporal cues. Whereas lesions in the left ACx impact discrimination of gap durations, implying a role in processing local temporal cues [ 7 ].

Social communication calls have myriads of constituent sounds that are temporally and spectrally dynamic. The auditory system must have processes in place to quickly decode and encode ethologically relevant features of an auditory signal to elicit an appropriate response. Hemispheric asymmetry in sound processing (i.e., lateralization) has long been proposed to be critical in the dynamic processing of speech sounds. Human studies have shown that the left superior temporal gyrus (STG) is more capable of integrating information over a shorter timescale and plays a greater role in speech perception and phonological processing than the right [ 1 – 3 ]. On the other hand, the right STG has longer integration windows to potentially subserve the processing of suprasegmental information [ 4 , 5 ]. The unanswered question remains: What neural mechanisms underlie these differences in temporal integration? One possible mechanism we dissect in this study is interhemispheric differences in excitatory recurrent connections. This would introduce positive feedback that would counteract the exponential decay of individual neurons within a neural population, leading to effectively slower temporal dynamics with a longer time constant.

Results

Lateralized connectivity motifs are found in every layer of the auditory cortices To screen for hemispheric differences in the organization of excitatory pathways, we used glutamate uncaging-based laser scanning photostimulation (LSPS; [8]). We performed voltage-clamp recordings on principal neurons in Layers 2 to 6 of the left (n = 229) and right (n = 209) ACx, (Fig 1A and 1B). The uncaging stimulus grid covered the entire primary ACx and all cortical layers (total area of 1.125*1.125 mm, 256 stimulation sites/map pixels). Hence, we specifically measured only intracortical sources of synaptic input. We focused on excitatory input by holding the membrane potential at the reversal for inhibition (-70 mV). To assess potential synaptic connectivity differences in each cortical layer, we performed a statistical comparison of the population data underlying corresponding map pixels from each hemisphere. We found statistically significant differences in the strength and organization of synaptic input between most layers of the auditory cortices (Fig 1C). To assess the likelihood that these differences were by chance, we randomly assigned each pixel to the left or right ACx. We largely found fewer significant differences between the hemispheres when the pixels were randomized (Fig 1D). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Summary of excitatory pathways in the left and right ACx across all cortical layers. (A, B) Patterns of excitatory synaptic input were recorded using LSPS as described in the text. Maps were averaged over all cells in each layer and then interpolated for clarity. Masked areas indicate direct hits in >50% of cells for that region. White triangles denote location of somata. Laminar boundaries were defined with respect to the fractional distance from the L1/L2 boundary to the white matter. L5A and L5B were defined as the upper and lower 50% of L5, respectively. N indicates number of recorded cells for each panel. (C) Pixel-wise significance maps (p < 0.05, unpaired 2-tailed t tests) for response amplitude in left vs. right ACx. Red and green denote significantly higher average response in the left and right ACx, respectively. Gray denotes no significant difference. (D) Same as C but maps from both hemispheres were pooled and assigned to 2 groups at random with the same total N shown in A, B. (E) Graphs of the significant pixel counts for measured (C) and random (D) comparisons. (F) Models depicting potential synaptic mechanisms underlying the right ACx’s higher excitatory connectivity in superficial layers compared to the left ACx. The left panel shows both cortices with a similar pool of presynaptic sources of EPSCs (traces) projecting onto postsynaptic targets (triangles), but the distribution of event amplitudes differs. The right panel shows the left and right ACx differ in their pool of presynaptic inputs but have a similar event amplitude distribution. (G) Observed frequency of event amplitudes in superficial layers of the left and right ACx. Inset shows empirical cumulative distribution of event amplitudes. The data underlying all the plots in this figure are included in S1 Data. ACx, auditory cortex; LSPS, laser scanning photostimulation. https://doi.org/10.1371/journal.pbio.3001803.g001 A large source of presynaptic input to L2 in the right ACx arose from intralaminar and other superficial layers (Fig 1A–1C). In contrast, in the left ACx, the most influential presynaptic pathway to L2 arose from deep layers (L5/6), as was reported previously [9]. There were also hemispheric differences in the organization of input to L3, which were also previously described in detail [6]. Intracortical input to L4 was stronger in the left ACx compared to the right (Fig 1A–1C). Layers 5A and 5B are functionally distinct in the ACx [10,11], and the organization of their intracortical synaptic input appears to be lateralized. In the left ACx, there was greater ascending input to L5A compared to the right ACx. Conversely, there was greater descending input to L5B in the right ACx compared to the left ACx (Fig 1A–1C). Finally, L6 had complementary patterns of intralaminar synaptic input along the tonotopic axis: The left ACx had greater input arising from higher frequency bands, and the right ACx from lower frequency bands (Fig 1A–1C). A comparison of the number of measured and random pixels for each layer revealed that the most abundant differences in synaptic input between the hemispheres arose in the superficial layers of the left and right ACx (Fig 1E). Several synaptic mechanisms could underlie the observation of more significant hotspots in superficial layers of the right ACx compared to the left. One possibility is that each hemisphere has a similar pool of presynaptic inputs in these layers, but individual synaptic events were larger in the right ACx compared to the left (Fig 1F left). Another possibility is that the distribution of the synaptic events’ amplitudes is similar between the hemispheres, but there is a larger pool of presynaptic inputs in the right ACx compared to the left (Fig 1F right). To disentangle these possibilities, we analyzed the distribution of event amplitudes in superficial layers (L2-4). We randomly chose the same number of cells to analyze from each hemisphere (n = 96), and the threshold for synaptic events was set to above the average baseline events. Fewer synaptic events met the threshold criteria in the left ACx (n = 1,090) compared to the right (n = 1,761, Fig 1G). Additionally, the distribution of event amplitudes was similar between the auditory cortices (Fig 1G inset). This supports the prediction that there is a larger pool of presynaptic inputs in superficial layers of the right ACx.

Lateralized recurrent connectivity in superficial layers of the auditory cortices Neural circuits composed of neurons with short time constants can effectively have long temporal memory and computation by forming a long feedforward chain or concise recurrent feedback loops [12]. Therefore, we investigated whether the synaptic connectivity differences translate into systematic differences in the recurrent interlaminar feedback. To compare the relative strength of interlaminar pathways between the hemispheres, we computed connectivity matrices for the left and right ACx. These input–output matrices (presynaptic-postsynaptic) summarize the organization of local excitatory networks [13,14]. We ordered maps according to the cortical depth of the soma and summed the 2D LSPS-derived input maps for each cell over the tonotopic axis (across the horizontal dimension), which produced vectors of input strength as a function of cortical depth (Fig 2A). Therefore, each neuron’s input vector represents presynaptic input from different laminar locations. Each row in the laminar connectivity matrix represents input to that specific laminar location, and each column represents synaptic output from that laminar location. Very local connections (<50 μm) within each layer lie along the diagonal and were under-sampled due to direct excitation [13]. In a previous study, we determined that neuronal density and photoexcitability of individual cells did not significantly differ between the hemispheres; therefore, we did not normalize maps by these factors [6]. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Recurrent connections are significantly stronger in superficial layers of the right ACx. (A) Schematic diagram of the generation of laminar input–output maps [14]. Left, 2D LSPS-derived input maps for each cell were summed over the anterior-posterior axis to produce 1D maps of input strength vs. cortical depth (vertical strips). Note that anteroposterior information is discarded, such that the 2 L6 input hotspots for the L3 cell (top) collapse to a single spot. Triangles denote soma location. Middle, the 1D maps were rotated 90 degrees (for graphic clarity), sorted by cortical depth of the soma, and binned (bin size = 80 μm). Right, the binned maps were combined into a single 2D map of presynaptic input location (x axis) vs. binned postsynaptic soma location (y axis). Maps were interpolated for display. (B) Input–output map for the left ACx, constructed as shown in A. Diagonal line indicates x = y with respect to cortical depth. The diagonal does not span the full x axis because recorded cell bodies (y axis) were confined between L2 and L6, whereas the stimulation grid (x axis) extended more broadly from L1 into the white matter. (C) Same as B but for the right ACx. (D) Same map as in B but showing only pathways in the 75th percentile and above in the left ACx. (E) Same as D but for the right ACx. (F) Summary of loops and pathways in the 75th percentile and above in the left ACx and right ACx (G). In F and G, the arrow thickness indicates strength of the pathway (thickest in the 99th percentile to thinnest in the 75th percentile), ascending pathways are shown in black, descending in gray, recurrent in violet, and open loop in dashed line. The data underlying all the plots in this figure are included in S1 Data. ACx, auditory cortex; LSPS, laser scanning photostimulation. https://doi.org/10.1371/journal.pbio.3001803.g002 The most striking asymmetry in the laminar connectivity matrices was the stronger synaptic connections in superficial layers of the right ACx compared to the left (Fig 2B and 2C). To quantify the strengths of the pathways, we calculated fractional input and output. In this analysis, we summed along rows and columns and normalized by the total [14]. The fractional input and output of Layers 2 to 4 was significantly greater in the right ACx (p = 0.0079 for input, p = 0.0411 for output). In the deeper layers, input to L6 was significantly greater in the left ACx (p<<0.001), but there was no hemispheric difference in the output. Significant hemispheric differences in intralaminar and interlaminar loops were a major theme in the organization of auditory circuits. Using the input/output laminar connectivity matrices of the left and right ACx, we examined pathways in the 75th percentile to capture the most significant trends. In the left ACx, we observed strong intralaminar recurrent connectivity only in L6 and to a lesser extent in Layers 2 and 5B (Fig 2D). The strongest interlaminar recurrent connections were observed between deep layers, where L5B and 6 form nested loops (intralaminar loops within L5B and 6, which are in turn connected to each other). In contrast to the sparser prevalence of recurrent connections and loops in the left ACx, these were the dominant motifs in the right ACx. All cortical layers (except L5) in the right ACx were part of nested loops: At the innermost level, there were intralaminar loops (input returning to the same layer), followed by local interlaminar loops between neighboring layers, and transcortical loops that coupled superficial and deep layers (Fig 2E and 2G). On the whole, we identified 5 recurrent loops in the left ACx and 10 in the right ACx. Taken together, these widespread loops of recurrent connections suggest the right ACx may have different temporal filtering properties compared to the left ACx.

Hemispheric differences in recurrent activity A potential result of stronger recurrent excitatory connectivity in a circuit is the ability to generate longer network events [15,16]. To test whether the greater recurrent excitatory connectivity observed in the right ACx leads to stronger and longer recurrent activity compared to the left ACx, we used experimental conditions to disinhibit slices [14]. The input–output connectivity maps in Fig 1 were obtained in a slice preparation that reduced excitation to prevent polysynaptic activity [8]. Therefore, to promote excitation in the slice, we partially blocked inhibitory circuits using the GABA A receptor antagonist SR95531 (see Methods for more experimental details). We performed LSPS, recorded in the cell-attached configuration from excitatory neurons in L2-4, and compared network events in the left and right ACx in slices from the same animal. Network events were only triggered by photostimulation and were largely initiated by stimulating sites in superficial layers of the ACx. A representative example of events triggered in the left and right ACx of the same animal using the same concentration of SR95531 shows a greater number of action potentials generated per stimulation site and more sites evoking action potential firing in the right ACx (Fig 3A and 3B left and middle panels). This difference in activity was significant across animals (Fig 3A and 3B right panels, n = 16 cells, 4 animals, p < 0.001, Wilcoxon rank sum). We also quantified the duration of these network events: the time from the first photostimulation-evoked action potential to the return of the membrane potential back to the pre-stimulus baseline (Fig 3C). The duration of network events was significantly longer in the right ACx compared to the left (Fig 3D, p<<0.001 Wilcoxon rank sum.) The long duration and variability of these events suggest that the recurrent excitation results from polysynaptic circuits. In summary, disinhibiting the slice revealed the network-wide impact of stronger recurrent excitatory connectivity in the right ACx. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Stronger network events in the right ACx. (A, left) All traces of cell-attached recording from an L3 excitatory neuron in the left ACx with 4 μm concentration of SR95531 in the bath. Black arrow on the x axis marks the onset of photostimulation. (A, middle) Map of action potentials evoked in the cell shown in the left panel and the population of cells recorded in the left ACx (n = 8, right). Triangles show position of the cells recorded. (B, left and middle) Same experimental conditions as in A, but conducted in the right ACx of the same animal. (B, right) Population map of all cells recorded in the right ACx (n = 8). (C) Duration of events was calculated from the onset of the first action potential to the return of the membrane potential back to baseline period (see Methods for details). (D) The duration of network events is shown using the survival function, which demonstrates that duration times are shorter in the left ACx compared to the right (n = 16). All cells were mapped using the same number of stimulus sites (i.e., same stimulus grid). The data underlying all the plots in this figure are included in S1 Data. ACx, auditory cortex. https://doi.org/10.1371/journal.pbio.3001803.g003 Hemispheric differences in membrane properties could also contribute to network activity and time constants. To examine this possibility, we compared the current-voltage and current-firing rate relationship of neurons in superficial layers of the left and right ACx (S1 Fig). We found no hemispheric differences between these input–output relationships (I/V, p = 0.71 n = 10; I/F, p = 0.84 n = 10). Hemispheric differences in resting membrane potential were also not significant (p = 0.86, mean left ACx = −73.4 mV, sem = 1.24, n = 28; mean right ACx = −73.2 mV, sem = 1.23, n = 28).

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001803

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/