Abstract
Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract It is generally assumed that human intelligence relies on efficient processing by neurons in our brain. Although grey matter thickness and activity of temporal and frontal cortical areas correlate with IQ scores, no direct evidence exists that links structural and physiological properties of neurons to human intelligence. Here, we find that high IQ scores and large temporal cortical thickness associate with larger, more complex dendrites of human pyramidal neurons. We show in silico that larger dendritic trees enable pyramidal neurons to track activity of synaptic inputs with higher temporal precision, due to fast action potential kinetics. Indeed, we find that human pyramidal neurons of individuals with higher IQ scores sustain fast action potential kinetics during repeated firing. These findings provide the first evidence that human intelligence is associated with neuronal complexity, action potential kinetics and efficient information transfer from inputs to output within cortical neurons. https://doi.org/10.7554/eLife.41714.001 eLife digest Our brains are made up of almost 100 billion brain cells. Each of them acts like a small chip: they collect, process and pass on information in the form of electrical signals. In brain areas that integrate different types of information, such as frontal and temporal lobes, brain cells have larger dendrites – long projections specialized to collect signals. Theoretical studies predict that larger dendrites help cells to initiate electrical signals faster. Because of difficulty in accessing human neurons, it has been unknown whether any of these features also relate to human intelligence. Previous studies have revealed that people with a higher IQ have a thicker outer layer (the cortex) in areas such as the frontal and temporal lobes. But does a thicker cortex also contain cells with larger dendrites and is their role different? To test whether smarter brains are equipped with faster and larger cells, Goriounova et al. studied 46 people who needed surgery for brain tumors or epilepsy. Each took an IQ test before the operation. To access the diseased tissue deep in the brain, the surgeon also removed small, undamaged samples of temporal lobe. These samples still contained living cells and their electrical signals were measured in the lab. The experiments showed that cells from people with a higher IQ had larger dendrites that transported information more quickly, especially when they are very active. Computer models were then used to understand how these findings can lead to more efficient information transfer in human neurons. Traditionally, research on human intelligence has focused on three main strategies: to study brain structure and function, to find genes associated with intelligence and to study the connection between our mind and behavior. Goriounova et al. are the first to take the single-cell perspective and link cell properties to human intelligence. The findings could help connect these separate approaches, and explain how genes for intelligence lead to thicker cortices and faster reaction times in people with higher IQ. https://doi.org/10.7554/eLife.41714.002 Introduction A fundamental question in neuroscience is what properties of neurons lie at the heart of human intelligence and underlie individual differences in mental ability. Thus far, experimental research on the neurobiological basis of intelligence has largely ignored the neuronal level and has not directly tested what role human neurons play in cognitive ability, mainly due to the inaccessibility of human neurons. Instead, research has either been focused on finding genetic loci that can explain part of the variance in intelligence (Spearman’s g) in large cohorts (Lam et al., 2017; Sniekers, 2017; Trampush et al., 2017; Coleman et al., 2018) or on identifying brain regions in whole brain imaging studies of which structure or function correlate with IQ scores (Karama et al., 2009; Hulshoff Pol et al., 2006; Narr et al., 2007; McDaniel, 2005; Deary et al., 2010). Some studies have highlighted that variability in brain volume and intelligence may share a common genetic origin (Hulshoff Pol et al., 2006; Posthuma et al., 2002; Sniekers, 2017), and individual genes that were identified as associated with IQ scores might aid intelligence by facilitating neuron growth (Sniekers, 2017; Coleman et al., 2018) and directly influencing neuronal firing (Lam et al., 2017). Intelligence is a distributed function that depends on activity of multiple brain regions (Deary et al., 2010). Structural and functional magnetic resonance imaging studies in hundreds of healthy subjects revealed that cortical volume and function of specific areas correlate with g (Karama et al., 2009; Choi et al., 2008; Narr et al., 2007). In particular, areas located in the frontal and temporal cortices show multiple correlations of grey matter thickness and functional activation with IQ scores: individuals with high IQ show larger grey matter volume of, for instance, Brodmann areas 21 and 38 (Choi et al., 2008; Deary et al., 2010; Karama et al., 2009; Narr et al., 2007). Cortical grey matter consists for a substantial part of dendrites (Chklovskii et al., 2002; Ikari and Hayashi, 1981), which receive and integrate synaptic information and strongly affect functional properties of neurons (Bekkers and Häusser, 2007; Eyal et al., 2014; Vetter et al., 2001). Especially higher order association areas in temporal and frontal lobes in humans harbor pyramidal neurons of extraordinary dendritic size and complexity (Elston, 2003; Mohan et al., 2015) that may constitute variation in cortical thickness, neuronal function, and ultimately IQ. These neurons and their connections form the principal building blocks for coding, processing, and information storage in the brain and give rise to cognition (Salinas and Sejnowski, 2001). Given their vast number in the human neocortex, even the slightest change in efficiency of information transfer by neurons may translate into large differences in mental ability. However, whether and how the activity and dendritic structure of single human neurons support human intelligence has not been tested. To investigate whether structural and functional properties of neurons of the human temporal cortex associate with general intelligence, we collected a unique multimodal data set from 46 human subjects containing single cell physiology (31 subjects, 129 neurons), neuronal morphology (25 subjects, 72 neurons), pre-surgical MRI scans and IQ test scores (35 subjects, Figure 1, data available at the Dryad Digital Repository: https://doi.org/10.5061/dryad.83dv5j7). Figure 1 with 2 supplements see all Download asset Open asset Summary of the approach: multidimensional data set from human subjects contained single cell physiology, neuronal morphology, MRI and IQ test scores (WAIS FSIQ). The area of the brain highlighted in blue indicates the location of cortical thickness measurements, black square indicates the typical origin of resected cortical tissue. https://doi.org/10.7554/eLife.41714.003 Human cortical brain tissue was removed as a part of surgical treatment for epilepsy or tumor (Table 1). The tissue almost exclusively originated from middle temporal gyrus, approximately 4 cm posterior to the temporal pole (Figure 2b) as a block of ~1–1.5 cm in diameter and was removed to gain access to the disease focus in deeper lying structures such as hippocampus or amygdala. In all patients, the resected neocortical tissue was not part of the epileptic focus or tumor and displayed no structural/functional abnormalities in preoperative MRI investigation, electrophysiological whole-cell recordings or microscopic investigation of histochemically stained tissue (Mohan et al., 2015; Testa-Silva et al., 2014; Testa-Silva et al., 2010; Verhoog et al., 2016; Verhoog et al., 2013). In line with the non-pathological status of tissue, we observed no correlations of cellular parameters or IQ scores with the subject’s disease history and age (Figure 1—figure supplements 1–2). After resection the tissue was immediately placed in ice-cold artificial cerebro-spinal fluid (aCSF) and within 15 min transported to the lab, sliced and maintained to enable single cell physiological recordings and biocytin filling. Figure 2 Download asset Open asset IQ scores positively correlate with cortical thickness of the temporal lobe. (a) MRI analysis pipeline: (1) Presurgical MRI T1-weighted scans; (2) Morphometric analysis; (3) Detection of cortical thickness from pial and white-grey matter boundaries; (b) Typical resection location for tissue used in this study is marked by a black circle; average total resected area from the patient is shown in red and maximum resected area in orange; (c) selection of temporal cortical area for correlations with IQ in b (red). (d) Average cortical thickness in temporal lobe (from area highlighted in red in c) positively correlates with IQ scores from the same subjects (n subjects = 35). Here and in figures below, Pearson correlation coefficients and p-values are reported in graph insets, the solid line represents linear regression (R2 = 0.13), shaded area indicates 95% confidence bounds of the fit. https://doi.org/10.7554/eLife.41714.006 Table 1 Subject details. https://doi.org/10.7554/eLife.41714.007 Patient numberIQAgeDiagnosisGenderAntiepileptic drugs18841TumorMCBZ27821OtherFLEV; VPA311966TumorFNone48831TumorFCBZ; LEV58151OtherFCLB; LTG; OXC66958MTSFCZP710728TumorMLTG; LEV811529MTSFLTG; TPM912520TumorMCBZ; LEV108427TumorFCBZ, LTG1111041TumorMCBZ; LTG128718MTSMOXC136723MTSFLEV; OXC147253MTSMCBZ; CLB159725TumorMNone1610419OtherMCLB; OXC178848OtherFCBZ186538MTSFCBZ; LEV196240OtherFNone2084.531OtherFNone218835OtherFCZP; LCS; LTG; LEV227754TumorMVPA239125OtherMCLB; LCS; LEV247031MTSFCBZ; CLB2511449OtherMCBZ; CLB; LEV268325TumorMNone2710945OtherFCBZ; CLB; LTG2810247TumorFCBZ296722OtherMCLB; LTG; LEV309738MTSMCBZ317940MTSFCBZ, CLB, LTG, LEV3211744OtherMLCS; VPA339930TumorFCLB; OXC347244MTSMLTG; LEV358241OtherFCBZ, LEV, TPM369529OtherMCBZ; PB379120OtherFCBZ; LEV388221TumorMCBZ; LCS; LTG; LEV3911540MTSMCBZ; LEV409748MTSFCBZ; ZNS419440MTSFCLB; LTG; ZNS428144MTSMCBZ; LTG437033MTSFCBZ; CLB; LEV448251OtherMCBZ4511418TumorFOXC469023OtherMOXC M = male; F = female; Antiepileptic drugs specified: Carbamazepine (CBZ); Lamotrigine (LTG); Levetiracetam (LEV); Topiramate (TPM); Clobazam (CLB); Oxcarbazepine (OXC); Clonazepam (CZP); Phenobarbital (PB); Phenytoin (PHT); Lacosamide (LCS); Sodium valproate (VPA); Zonisamide (ZNS) We recorded action potentials (APs) from human pyramidal neurons in superficial layers of temporal cortex and digitally reconstructed their complete dendritic structures. We tested the hypothesis that variation in neuronal morphology can lead to functional differences in AP speed and information transfer and explain variation in IQ scores. In addition to our experimental results, we used computational modelling to understand underlying principles of efficient information transfer in human cortical neurons. Results IQ scores positively correlate with cortical thickness of the temporal lobe Cortical thickness of the temporal lobe has been associated with IQ scores in hundreds of healthy subjects (Choi et al., 2008; Deary et al., 2010; Hulshoff Pol et al., 2006;Karama et al., 2009; Narr et al., 2007), and we first asked whether this applies to the subjects in our study as well. From T1-weighted MRI scans obtained prior to surgery, we determined temporal cortical thickness in 35 subjects using voxel-based morphometry of temporal cortical areas. These areas included the surgically resected cortical tissue (Figure 2b) used for cellular recordings and neuronal reconstructions, which typically came from locations at 4 cm from temporal pole and was 1–1.5 cm in diameter (black circle in Figure 2b). The total resected cortical area varied for each patient, but consisted of a larger part of the temporal lobe (Figure 2b; average resected area in red, maximum in orange). The mean distance of resection boundaries from temporal pole was 4.2 ± 1.7 cm on superior temporal gyrus, 4.8 ± 1.5 cm on middle temporal gyrus, and 4.9 ± 1.5 cm on inferior temporal gyrus for the 46 subjects in this study. In MRI images, cortical thickness was measured in temporal lobe that included the resection areas and corresponded to the areas identified to associate with IQ in healthy subjects (Choi et al., 2008; Deary et al., 2010; Hulshoff Pol et al., 2006; Karama et al., 2009; Narr et al., 2007) (Figure 2c; in red). The superior temporal gyrus was excluded from this analysis as it contains areas for auditory, gustatory and language processing that are spared during resection. Cortical thickness measurements were collapsed to one mean value for cortical thickness for each subject. In line with findings in healthy subjects (Choi et al., 2008; Deary et al., 2010; Hulshoff Pol et al., 2006; Narr et al., 2007;Karama et al., 2009) mean cortical thickness in temporal lobes positively correlated with IQ scores of the subjects (Figure 2d). IQ scores positively correlate with dendritic structure of temporal cortical pyramidal neurons Cortical association areas in temporal lobes play a key role in high-level integrative neuronal processes and its superficial layers harbor neurons of increased neuronal complexity (DeFelipe et al., 2002; Elston, 2003; Scholtens et al., 2014; van den Heuvel et al., 2015). In rodents, the neuropil of cortical association areas consists for over 30% of dendritic structures (Ikari and Hayashi, 1981). To test the hypothesis that human temporal cortical thickness is associated with dendrite size, we used 72 full reconstructions of biocytin-labelled temporal cortical pyramidal neurons from layers 2, 3 and 4 (median number of neurons per subject = 2; average 2.8; ranging from 1 to 10) part of which was previously reported (Mohan et al., 2015). We calculated total dendritic length (TDL) that included all basal and apical dendrites without apparent slice artifacts for each neuron. We computed TDL from multiple neurons for each subject and correlated these mean TDL values to mean temporal cortical thickness from the same subject. We found that dendritic length positively correlated with mean temporal lobe cortical thickness (Pearson correlation coefficient r = 0.5, explained variance R2 = 0.25), indicating that dendritic structure of individual neurons contributes to the overall cytoarchitecture of temporal cortex (Figure 3a). Figure 3 Download asset Open asset IQ scores positively correlate with dendritic structure of temporal cortical pyramidal cells. (a) Average total dendritic length in pyramidal cells in superficial layers of temporal cortex positively correlates with cortical thickness in temporal lobe from the same hemisphere (area shaded in a, n subjects = 20; n neurons = 57, R2 = 0.25). Inset shows a scheme of cortical tissue with a digitally reconstructed neuron and the brain area for cortical thickness estimation (red) (b) Cortical depth of pyramidal neurons, relative to cortical thickness in temporal cortex from the same hemisphere, does not correlate with IQ score (n subjects = 21, R2 = 0.03). Inset represents the cortical tissue, blue lines indicate the depth of neuron and cortical thickness (c) Total dendritic length (TDL) and (d) number of dendritic branches positively correlate with IQ scores from the same individuals (n subjects = 25, n neurons = 72, TDL R2 = 0.26, Branch points R2 = 0.22). Symbols highlighted in blue were shifted along the x axis for display purposes. Data are mean per subject ±standard deviation. https://doi.org/10.7554/eLife.41714.008 TDL is in part determined by the soma location within cortical layers: cell bodies of pyramidal neurons with larger dendrites typically lie deeper, at larger distance from pia (Mohan et al., 2015). To exclude a systematic bias in sampling, we determined the cortical depth of each neuron relative to the subject’s temporal cortical thickness in the same hemisphere. There was no correlation between IQ score and relative cortical depth of pyramidal neurons indicating that we sampled neurons at similar depths across subjects (Figure 3b). Finally, we tested whether mean TDL and complexity of pyramidal neurons relates to subjects’ IQ scores. We found a strong positive correlation between individual’s pyramidal neuron TDL and IQ scores (Pearson correlation coefficient r = 0.51, explained variance R2 = 0.26; Figure 3c) as well as between number of dendritic branch points and IQ scores (r = 0.46, R2 = 0.22; Figure 3d). Thus, larger and more complex pyramidal neurons in temporal association area may partly contribute to thicker cortex and link to higher intelligence. Larger dendrites lead to faster AP onset and improved encoding properties Dendrites not only receive most synapses in neurons, but dendritic morphology and ionic conductances act in concert to regulate neuronal excitability (Bekkers and Häusser, 2007; Eyal et al., 2014; Vetter et al., 2001). In model simulations where neurons are reduced to balls and sticks, increasing the dendritic membrane surface area, that is the dendritic impedance load, speeds up the onset phase of APs. This is a consequence of the decrease in the effective time constants of the neuron with increasing dendritic size and dendritic impedance load (Eyal et al., 2014). Larger dendrites act as a larger sink for currents generated in the axon initial segment during AP onset and result in faster membrane potential changes. Furthermore, we found previously that human neocortical pyramidal neurons, which are three times larger than rodent pyramidal neurons (Mohan et al., 2015), have faster AP onsets compared to rodent neurons and are able to track and encode fast synaptic inputs and sub-threshold changes in membrane potential with high temporal precision (Testa-Silva et al., 2014). We asked whether the observed differences in TDL between human pyramidal neurons affected their encoding properties and ability to transfer information. To this end, we incorporated the 3-dimensional dendritic reconstructions of the 72 human pyramidal neurons into in silico models, equipped them with excitable properties (see Materials and methods) and tested whether their APs have faster onset. We found that TDL of model neurons with realistic dendritic trees positively correlated with the steepness of AP onsets (r = 0.4, R2 = 0.16; Figure 4a,b) and larger dendrites enabled neurons to generate faster APs. Figure 4 Download asset Open asset Larger dendrites lead to faster AP onset and improved encoding properties. (a,b) Higher TDL results in faster onsets of model-generated APs: (a) example phase plot of an AP is shown with a red line representing onset rapidity - slope of AP derivative at 10 mV/ms (grey dashed line); (b) onset rapidity values of simulated APs positively correlate with TDL (R2 = 0.36). (c) Model neurons received simulated sinusoidal current-clamp inputs and generated spiking responses of different magnitudes and frequencies. Red and blue traces are response magnitudes of example neurons with low (blue) and large (red) TDLs; inset shows examples of morphological reconstructions with their TDLs in mm shown above. Cut-off frequencies are defined within the frequency range (shaded area) at which the model neuron can still track the inputs reliably (produce response of 0.7 response magnitude, dashed line). (d) Cut-off frequencies positively correlate with TDL (R2 = 0.16; example neurons from panel (c) are highlighted by the same colors). (e) Responses to the same in example neurons from (b) and firing frequency of the model neuron with large TDL (red) the with higher temporal precision than the model neuron with TDL The of action potential firing cortical neurons to pass on temporal information by synaptic inputs et al., 2008; et al., Testa-Silva et al., 2014; et al., pyramidal neurons not sustain high frequency firing and generally not encode high frequency synaptic in Instead, the precision in of AP does these neurons to encode high frequency information in their In to rodent neurons, human neurons can encode sub-threshold membrane potential changes on a by of APs (Testa-Silva et al., 2014). This synaptic strongly relies on the rapidity of AP onset et al., 2013). APs neurons to to fast synaptic which AP is neurons with faster APs can translate higher frequencies of synaptic membrane potential into AP and ultimately encode more information. The (Eyal et al., using neuron models showed that neurons with larger dendritic not only have faster AP onset but could also time AP to faster changes in membrane increasing the frequency of and the frequency of information encoding three However, it is not whether the same for the human cortical pyramidal neurons we recorded and whether the range of dendritic we might lead to We tested this by sinusoidal inputs of increasing frequencies into in silico of the neurons we recorded and and studied how the of AP firing of these neurons sub-threshold membrane potential changes. We find that human neurons with larger TDL can reliably time their APs to faster membrane potential with frequencies up to neurons had their frequencies at (Figure Furthermore, was a positive correlation between the dendritic length and the frequency (Figure Finally, the same - of the of three of increasing frequencies - larger neurons were able to encode temporal information into of AP compared to neurons (Figure Thus, we find that differences in dendritic length of human neurons lead to faster APs and to frequency of encoding synaptic inputs into of AP Higher IQ scores associate with faster APs cortical pyramidal neurons with large dendrites have faster APs and can encode more information in AP and large dendrites also associate with higher IQ scores, we asked whether human cortical pyramidal neurons from individuals with higher IQ scores generate faster APs. To test we made whole-cell recordings from pyramidal cells in of temporal cortex (31 subjects, 129 neurons, number of neurons per subject = ranging from 1 to Figure and recorded APs at different firing frequencies in response to We determined AP maximum rise which is correlated with AP onset rapidity (r = n = data not and can more reliably determined from recordings with frequencies between 10 and rise speed of APs on the firing history of the with the first AP in the the AP rise speed and with increasing firing the time between APs (Figure To test whether AP rise speed between IQ we all AP rise speed data into on IQ score – and Although the AP rise speed of the first AP was not different between high and low IQ (Figure the AP in individuals with IQ scores compared to APs of individuals with higher IQ scores (Figure higher firing frequencies the AP rise speed was higher in individuals with IQ scores 100 (Figure AP rise speed high IQ = ± AP rise speed low IQ = ± We calculated the of APs with increasing frequency by rise speed of APs to the rise speed of the first AP in the to first rise speed at showed in subjects with IQ scores and to of the initial AP rise In in neurons from individuals with higher IQ scores, AP rise speed on average at (Figure high IQ = ± low IQ = ± Figure Download asset Open asset Higher IQ scores associate with faster AP (a) of a whole-cell biocytin of a pyramidal neuron from human temporal typical responses to (b) of AP traces at increasing firing frequencies is shown in in recorded from a subject with IQ = and a subject with IQ = AP phase in shaded area is displayed to the (c) APs from subjects with higher IQ are able to their rise speed at increasing frequencies. Average neuron and AP rise speed and (d) relative to first AP rise speeds in neurons from subjects with IQ 100 n subjects = 21, n neurons = and subjects with IQ 100 n subjects = n neurons = are displayed firing data points in shaded area are shown as values for are are mean rise speeds per (e) IQ scores positively correlate with the rise speeds of first AP in the (n subjects n neurons = R2 = AP rise speed at data as panel in R2 = and relative AP rise speeds at data as panel in R2 = Larger neurons show of AP rise speed at higher relative AP rise speeds at for individual neurons are as a function of their TDL (n = 21 neurons, R2 = In data are mean per subject in g data are mean ±standard deviation. We whether these differences at the level correlations between individual IQ scores and AP rise We correlated mean AP rise speeds of the first AP and AP at from all neurons of the same subject to the subject’s IQ The AP rise speed of the first AP in the positively correlated with IQ scores (r = R2 = Figure and this correlation was even for AP rise speeds at frequencies of (r = 0.46, R2 = Figure also relative AP values showed positive correlations with indicating that it is the relative of APs at high frequencies that with intelligence (r = R2 = Figure Finally, we asked whether the of APs relates to the dendritic size of the same neurons, as our model results We find that larger neurons show of AP rise speed relative AP at (r = R2 = Figure These findings that higher IQ scores are by faster APs during repeated AP IQ scores associate with increased AP during neuronal Thus, neurons from individuals with higher IQ scores are equipped to process synaptic signals at high and at faster time which is to encode large of information and Discussion Our findings provide a first into the cellular of human intelligence and explain individual variation in IQ scores on neuronal faster AP rise speed during neuronal activity and more dendrites associate with higher intelligence. AP kinetics have for information In neurons are by high frequency synaptic inputs and the of neurons to track and to these inputs how of this synaptic information can on to neurons (Testa-Silva et al., 2014). The brain at a and even of contain information that can responses et al., Indeed, one of the most and findings in is the association of intelligence scores with of cognitive speed et al., reaction times in provide a