07 Jul 2023

What little scientists know about how human neurons function in vivo comes from resections of brain tissue in people with cancer or intractable epilepsy. Scientists slice these biopsies, keep the slices viable for the day, and characterize the morphology and firing patterns of neurons in real time. In Alzheimer's research, brain biopsies became a coveted resource once it became clear that the computational capacity of human neurons is vastly different than that of rodent neurons. Demand for human tissue for research purposes rose even further when findings in AD rodent models failed to replicate in clinical trials.

Alas, human brain tissue is extremely scarce. And even when scientists can get their hands on the occasional epilepsy or tumor biopsy, neither are ideal for AD research. Epilepsy patients tend to be young adults, and both seizures and tumors confound brain structure, function, and gene expression. The medical needs of the patient at hand naturally dictate what region is resected, hence each biopsy is different, and scientists cannot build standardized sample collections needed to establish group effects.

Enter idiopathic normal-pressure hydrocephalus (iNPH). This age-related illness has become a steady source of tissue from one cortical gyrus for electrophysiologists to probe—at least for the electrophysiologists of the city of Kuopio in eastern Finland. There, research-minded surgeons, working hand-in-glove with Alzheimer's investigators, have devised a system of interlocking research protocols that enable multilevel analyses of a range of different tissues from one and the same deeply phenotyped person. Conveniently for Alzheimerologists, 40 percent of these iNPH patients—being, as they are, in their 60s to 80s—have amyloid; 10 percent have both amyloid and tau pathology. They constitute a natural model of preclinical Alzheimer's disease, with controls. Some carry causative mutations for Alzheimer's or related disorders (see Part 1 of this series).

Every Friday at Kuopio University Hospital, as patients undergo a day surgery called ventriculo-peritoneal shunt placement, the operating surgeon makes four tiny cuts into the spot on the patient's right frontal cortex where the shunt catheter needs to penetrate, gently sliding a pyramid-shaped biopsy into a Falcon tube full of artificial CSF (see Part 2 of this series). Within the hour, the cortical pyramid finds itself getting sliced and prodded with electrodes.

Will its neurons be active? After all, at a diminutive 10 to 20 cubic millimeters, iNPH biopsies measure but a few percent of the typical volume of tumor or epilepsy resections, which are up to 1 cubic centimeter large. Think diamond versus sugar cube. Over the past two years, Antonio Dougalis in Tarja Malm’s lab at the University of Eastern Finland has led efforts to work out a protocol to get these cells to fire.

Now, they are active in a dish. The patients' neurons fire, both spontaneously and when provoked by drugs. They fire bursts both as individual cells, and jointly in oscillation patterns that require coherent synaptic transmission. Application of GABA reduces NMDA-induced activity. These and other basic properties resemble what previous recordings of epilepsy and tumor slices have shown, Malm said.

No Partying on Fridays
Feasibility having been established and technical kinks straightened out, the Kuopio neurophysiology team is now in data-collection mode. Every Friday, they receive one or two pyramids depending on how many iNPH patients came to surgery that day. They record away until the slices give out late at night. As they work, they are blinded to the phenotypic and genotypic information that is being gathered on the donors as part of a longitudinal research program. Without knowing if the slice before them has amyloid or tau pathology, they execute standardized recording scripts Dougalis has developed. They are piling up data in hopes of getting beyond low-N observations and toward robust group effects with statistical power.

In a year or so, when the different types of data being gathered on a given person can be integrated, the scientists are hoping for big insights. "At 50 biopsies per year in Kuopio, to gather sufficient N to see group differences takes time," Malm said. Because the small size of the biopsies precludes extensive tissue sharing for acute-slice neural activity studies, progress would accelerate if other hospital-associated research centers set up similar systems, she added.

That said, Dougalis and Malm have already gleaned early insights from their analysis of the first 55 biopsies over the past year, some with AD pathology, some without. Tantalizingly, it appears that findings at the functional level jibe with findings at the RNA sequence level in the same person.

When fresh brain arrives from the operating room in the Malm lab, postdoc Anssi Pelkonen promptly embeds it in agarose, mounts it in a vibratome, and slowly moves the blade across the block, shaving off 350-micron-thick slices. He sorts them for different uses. The first, irregular bits are frozen for proteomics. One slice gets fixed and returned to the hospital pathologist to support the patient's diagnosis. One slice is for research immunohistochemistry in the Malm lab, one is for EM, one for spatial transcriptomics.

Two adjacent slices are used immediately, one for single-neuron patch-clamp recordings to characterize the behavior of individual neurons, the other for multi-electrode recordings to characterize neural circuits and their higher-order functions. While Pelkonen sorts the slices, their donor in the hospital is about to be wheeled from the OR to the recovery room.

Ph.D. student Mireia Gómez-Budia straps one slice onto a multi-electrode array (MEA). Its 60 electrodes poke into neurons from below, across the entire depth of cortex. Once mounted, the slice accommodates to its new surroundings in oxygenated saline for a half hour before Gómez-Budia puts it to work. She checks activity parameters from action potential burst rates to local field potentials. She records multi-unit activity, i.e., the summed activity of all neurons near a given electrode or multiple electrodes, respectively. These measures characterize more complex circuit functions such as gamma or theta waves.

Some slices fire spontaneously, Gómez-Budia says. Others get active after she stimulates them with NMDA to awaken glutamatergic synapses, or with carbachol to stimulate cholinergic ones.

Searching for a memory mechanism of sorts, Gómez-Budia tries to induce LTP in the slices. Via the MEA, she stimulates cortical layers 4 or 5 and records in layers 2/3. "We have data of slices that are potentiating and inducing LTP," she told Alzforum.

Five meters away, postdoc Polina Abushik tries to get a single electrode to form a seal with the cell membrane of a layer 2/3 pyramidal neuron for her patch-clamping session with the cortical slice adjacent to Gómez-Budia's. She hopes her electrode will penetrate an individual neuron so she can record the goings-on, ranging from its resting membrane potential, action potential threshold, frequency, and other electrophysiological properties.

"You can hear how one neuron works," Abushik says. First she listens in on the cell's intrinsic activity, then she documents evoked responses to current she injects via a stimulating electrode placed on the neuron's “cheek,” aka the side of its soma. To build homogeneous data sets that allow conclusions about specific cell types, the scientists started in layers 2/3, patching dozens of pyramidal and interneurons. The latter are quite different morphologically and functionally, and tricky to record, Malm said.

Snce a slice is “exhausted” and the recording session ends, hopefully after many hours of data collection, the scientists don't toss it. It's still precious tissue. They fix MEA slices for histology, to relate the presence of, for example, amyloid or tau pathology, to the neurons' documented activity. They fill patched neurons with the neuron stain biocytin and then visualize their arborization and dendritic spines. Already, the group has documented a far greater variety of morphological differentiation than previously seen in mouse cortex.

Shapes and Sizes. Post-recording morphometric reconstructions of cortical neurons from living shunt recipients. Cortical layers, and cortical depth in millimeters, are indicated on left. Pyramidal neurons are shown with their apical dendrites in blue and basal dendrites in red. On the right, two interneurons. [Courtesy of Dougalis et al.]

Overall, Dougalis' feasibility study found that just under half the iNPH biopsies yield slices that appear to have normal intrinsic, synaptic, morphometric, and network properties. They exhibit intact excitatory and inhibitory synaptic transmission, but no abnormal epileptiform hyperexcitation. The circuitry to support high-power theta- and gamma-band oscillations is preserved. Curiously, across the age range available in this collection of samples, the capacity of cholinergic transmission to induce gamma-band activity was weaker in older than in younger people. In toto, the scientists conclude, this protocol is suitable for detailed functional exploration of adult human neurons.

How They Fire. Simplified properties of example interneuron (top) and pyramidal neuron (bottom) from a living older person's frontal cortex. The interneuron fired faster and narrower action potentials than the pyramidal neuron (column B). In voltage clamp, the interneuron's electrophysiological trace sustained a higher frequency (column C). Both neurons had spontaneous synaptic activity, and synaptic depression after repetitive stimulation (column D).

Regarding those deeper explorations, analyses are at early stages, but Malm noted some budding trends. One comes from her collaborators Evan Macosko and Beth Stevens at the Broad Institute, Cambridge, Massachusetts. They receive small bits of tissue from the same cortical location from the same Kuopio shunt recipients for transcriptomics analyses. When sorting the biopsy's cell types based on expression signatures, the Broad group noticed that a particular population of interneurons was mysteriously absent from cortical layer 1 in people who have amyloid pathology there.

To learn more about this, Malm’s team stimulates layer 1 in the biopsies. They want to know if there is inhibitory input from layer 1 into deeper layers such as 2/3, and whether that might weaken in people who have amyloid pathology nearby. In theory, if a set of interneurons, which are usually inhibitory, is missing, then their target neurons can become disinhibited and hyperexcitable. "Based on the RNA-Seq finding that a type of L1 interneuron is lost in brains with Aβ burden, we would expect to see a difference in L1 operational characteristics, and in L2/3 drug-induced responses," Malm said.

Importantly, Dougalis is already starting to see changes in markers of connectivity in people with Aβ pathology. By this he means connectivity within a slice, not the brain-wide connectivity as measured commonly by fMRI. Dougalis deduces connectivity based on analyses of spike cross-correlation and local field potential coherence probability. These analyses enable a comparative interpretation of the signals coming from each of the MEA's electrodes. When Dougalis adds NMDA or carbachol, he sees drug-specific and band frequency-specific changes in coherence probability and a reduction in the electrodes with cross-correlated spike trains. Both were related to AD pathology. "This indicates some degree of loss of connectivity in amyloid-positive brains," Malm said.

This increased excitability at the transcriptional level of layer 2/3 pyramidal cells with Aβ pathology seen Macosko’s lab is echoed by the data in Malm's lab. Specifically, the loss of L1 interneurons led to changes in the oscillatory local field potentials in that layer, and to a hyperexcitable phenotype in layer 2/3 based on spiking activity.

This implies that in the superficial layers of the cortex of a given person, Macosko and Malm see the same thing, he with RNA, she with physiology. How the added presence of tau pathology may affect all this is a focus of intense curiosity in the lab these days; thus far, Dougalis and colleagues have recorded from about a dozen such patients. 

To see how layer 1 interneurons and layer 2/3 excitatory neurons feature in Macosko and Steven's first big study of these biopsies, see Part 4.—Gabrielle Strobel

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

1. Fresh Brain Every Friday: Biopsies Transform Alzheimer's Science 3 Jul 2023
2. A Day's Work: Cortex Biopsy Comes Out. Shunt Goes In. Patient Goes Home. 3 Jul 2023
3. Brain Tissue From Living People with Amyloid Plaques Can Fire in a Dish 6 Jul 2023

Further Reading

News

1. Meta-Analysis of RNA-Seq Data is Robust, Finds Key Changes in AD 21 Sep 2022