It is well known that obesity increases the risk of insulin resistance and type 2 diabetes. It is also well known that some people who gain weight suffer from the disease and others do not. The reasons for these differences are not clear, but they are related to the function of the adipose tissue rather than the amount of body fat.

The current study, published in the journal PNAS, is mainly based on experiments in mice, but the research indicates that the newly discovered mechanism also applies to humans.

Weight gain increases the breakdown of the structural protein collagen to make room for the growing fat cells within adipose tissue. Collagen is a natural building block in the body that provides strength to cartilage, muscles, and skin.

The breakdown of collagen is handled by macrophages, a type of white blood cell that is part of the immune system. Macrophages are involved in the destruction of invading bacteria, but they also engulf and digest damaged cells and debris such as degraded collagen in adipose tissue during weight gain.

The macrophage function is impaired in obesity

The collagen is fragmented by enzymatic degradation outside the fat cells, and the collagen fragments are then engulfed by macrophages for complete degradation. What this study shows and describes is how highly regulated this uptake of collagen fragments is.

And it is fast when it works properly. However, this function of macrophages was found to be deactivated in obesity and insulin resistance, leading to the accumulation of collagen fragments in adipose tissue.

While this has not been considered a problem until now, the study shows that collagen fragments are not just debris, but actively influence various cellular processes such as inflammation and cell division.

The process thus goes from maintaining normal adipose tissue function during weight gain, to becoming pathogenic in some cases. When samples of human macrophages were exposed to diabetes-like conditions, they also lost their ability to “clean up” collagen.

Identification and prevention

The study was carried out by a research team at the Sahlgrenska Academy, University of Gothenburg, led by Ingrid Wernstedt Asterholm, Professor in Physiology.

“When adipose tissue grows, macrophages help the remodel the tissue in a controlled way. Exactly why this mechanism is sometimes deactivated is difficult to say, but perhaps it happens when there is at a certain, genetically determined, degree of adiposity” she says.

“It is our hope that these results ultimately lead to new strategies for preventing or treating type 2 diabetes. It is also conceivable that certain collagen fragments could serve as measurable biological markers, for example to identify individuals at higher risk of developing type 2 diabetes.”

It’s an achievement with important implications for scientists studying the brain and working on treatments for a broad range of neurological and neurodevelopmental disorders, such as Alzheimer’s and Parkinson’s disease.

“This could be a hugely powerful model to help us understand how brain cells and parts of the brain communicate in humans,” says Su-Chun Zhang, professor of neuroscience and neurology at UW-Madison’s Waisman Center. “It could change the way we look at stem cell biology, neuroscience, and the pathogenesis of many neurological and psychiatric disorders.”

Printing methods have limited the success of previous attempts to print brain tissue, according to Zhang and Yuanwei Yan, a scientist in Zhang’s lab. The group behind the new 3D-printing process described their method today in the journal Cell Stem Cell.

Instead of using the traditional 3D-printing approach, stacking layers vertically, the researchers went horizontally. They situated brain cells, neurons grown from induced pluripotent stem cells, in a softer “bio-ink” gel than previous attempts had employed.

“The tissue still has enough structure to hold together but it is soft enough to allow the neurons to grow into each other and start talking to each other,” Zhang says.

The cells are laid next to each other like pencils laid next to each other on a tabletop.

“Our tissue stays relatively thin and this makes it easy for the neurons to get enough oxygen and enough nutrients from the growth media,” Yan says.

The results speak for themselves — which is to say, the cells can speak to each other. The printed cells reach through the medium to form connections inside each printed layer as well as across layers, forming networks comparable to human brains. The neurons communicate, send signals, interact with each other through neurotransmitters, and even form proper networks with support cells that were added to the printed tissue.

“We printed the cerebral cortex and the striatum and what we found was quite striking,” Zhang says. “Even when we printed different cells belonging to different parts of the brain, they were still able to talk to each other in a very special and specific way.”

The printing technique offers precision — control over the types and arrangement of cells — not found in brain organoids, miniature organs used to study brains. The organoids grow with less organization and control.

“Our lab is very special in that we are able to produce pretty much any type of neurons at any time. Then we can piece them together at almost any time and in whatever way we like,” Zhang says. “Because we can print the tissue by design, we can have a defined system to look at how our human brain network operates. We can look very specifically at how the nerve cells talk to each other under certain conditions because we can print exactly what we want.”

That specificity provides flexibility. The printed brain tissue could be used to study signaling between cells in Down syndrome, interactions between healthy tissue and neighboring tissue affected by Alzheimer’s, testing new drug candidates, or even watching the brain grow.

“In the past, we have often looked at one thing at a time, which means we often miss some critical components. Our brain operates in networks. We want to print brain tissue this way because cells do not operate by themselves. They talk to each other. This is how our brain works and it has to be studied all together like this to truly understand it,” Zhang says. “Our brain tissue could be used to study almost every major aspect of what many people at the Waisman Center are working on. It can be used to look at the molecular mechanisms underlying brain development, human development, developmental disabilities, neurodegenerative disorders, and more.”

The new printing technique should also be accessible to many labs. It does not require special bio-printing equipment or culturing methods to keep the tissue healthy, and can be studied in depth with microscopes, standard imaging techniques and electrodes already common in the field.

The researchers would like to explore the potential of specialization, though, further improving their bio-ink and refining their equipment to allow for specific orientations of cells within their printed tissue..

“Right now, our printer is a benchtop commercialized one,” Yan says. “We can make some specialized improvements to help us print specific types of brain tissue on-demand.”

This study was supported in part by NIH-NINDS (NS096282, NS076352, NS086604), NICHD (HD106197, HD090256), the National Medical Research Council of Singapore (MOH-000212, MOH-000207), Ministry of Education of Singapore (MOE2018-T2-2-103), Aligning Science Across Parkinson’s (ASAP-000301), the Bleser Family Foundation, and the Busta Foundation.

Due to this enormous data gap, researchers have been skeptical that recent AI advances can tell us much about human learning and development. An ideal test for demonstrating a connection would involve training an AI model, not on massive data from the web, but on only the input that a single child receives. What would the model be able to learn then?

A team of New York University researchers ran this exact experiment. They trained a multimodal AI system through the eyes and ears of a single child, using headcam video recordings from when the child was six months and through their second birthday. They examined if the AI model could learn words and concepts present in a child’s everyday experience.

Their findings, reported in the latest issue of the journal Science, showed that the model, or neural network, could, in fact, learn a substantial number of words and concepts using limited slices of what the child experienced. That is, the video only captured about 1% of the child’s waking hours, but that was sufficient for genuine language learning.

In this video, the researchers describe their work in greater detail.

“We show, for the first time, that a neural network trained on this developmentally realistic input from a single child can learn to link words to their visual counterparts,” says Wai Keen Vong, a research scientist at NYU’s Center for Data Science and the paper’s first author. “Our results demonstrate how recent algorithmic advances paired with one child’s naturalistic experience has the potential to reshape our understanding of early language and concept acquisition.”

“By using AI models to study the real language-learning problem faced by children, we can address classic debates about what ingredients children need to learn words — whether they need language-specific biases, innate knowledge, or just associative learning to get going,” adds Brenden Lake, an assistant professor in NYU’s Center for Data Science and Department of Psychology and the paper’s senior author. “It seems we can get more with just learning than commonly thought.”

Vong, Lake, and their NYU colleagues, Wentao Wang and Emin Orhan, analyzed a child’s learning process captured on first-person video — via a light, head-mounted camera — on a weekly basis beginning at six months and through 25 months, using more than 60 hours of footage. The footage contained approximately a quarter of a million word instances (i.e., the number of words communicated, many of them repeatedly) that are linked with video frames of what the child saw when those words were spoken and included a wide range of different activities across development, including mealtimes, reading books, and the child playing.

The NYU researchers then trained a multimodal neural network with two separate modules: one that takes in single video frames (the vision encoder) and another that takes in the transcribed child-directed speech (the language encoder). These two encoders were combined and trained using an algorithm called contrastive learning, which aims to learn useful input features and their cross-modal associations. For instance, when a parent says something in view of the child, it is likely that some of the words used are likely referring to something that the child can see, meaning comprehension is instilled by linking visual and linguistic cues.

“This provides the model a clue as to which words should be associated with which objects,” explains Vong. “Combining these cues is what enables contrastive learning to gradually determine which words belong with which visuals and to capture the learning of a child’s first words.”

After training the model, the researchers tested it using the same kinds of evaluations used to measure word learning in infants — presenting the model with the target word and an array of four different image options and asking it to select the image that matches the target word. Their results showed that the model was able to learn a substantial number of the words and concepts present in the child’s everyday experience. Furthermore, for some of the words the model learned, it could generalize them to very different visual instances than those seen at training, reflecting an aspect of generalization also seen in children when they are tested in the lab.

“These findings suggest that this aspect of word learning is feasible from the kind of naturalistic data that children receive while using relatively generic learning mechanisms such as those found in neural networks,” observes Lake.

The work was supported by the U.S. Department of Defense’s Defense Advanced Research Projects Agency (N6600119C4030) and the National Science Foundation (1922658). Participation of the child was approved by the parents and the methodology was approved by NYU’s Institutional Review Board.