Member publications

Intercellular communication in the brain through a dendritic nanotubular network

10/02/2025 Science

KLAM author: Minhyeok Chang PhD, Research Associate at Johns Hopkins University

Intercellular nanotubular networks mediate material exchange, but their existence in neurons remains to be explored in detail. We identified long, thin dendritic filopodia forming direct dendriteโ€“dendrite nanotubes (DNTs) in mammalian cortex. Super-resolution microscopy in dissociated neurons revealed DNTsโ€™ actin-rich composition and dynamics, enabling long-range calcium ion (Caยฒโบ) propagation. Imaging and machine learningโ€“based analysis validated in situ DNTs as anatomically distinct from synaptic spines. DNTs actively transported small molecules and human amyloid-ฮฒ (Aฮฒ); DNT density increased before plaque formation in the medial prefrontal cortex of APP/PS1 mice (APP, Aฮฒ precursor protein; PS1, presenilin-1), suggesting that the dendrite-DNT network might play a role in Alzheimerโ€™s disease pathology. Computational models of DNT-mediated Aฮฒ propagation recapitulated early amyloidosis, predicting selective intracellular accumulation. These findings uncover a nanotubular connectivity layer in the brain, extending neuronal communication beyond classical synapses.

์„ธํฌ ๊ฐ„ ๋ฌผ์งˆ ๊ตํ™˜์„ ๋งค๊ฐœํ•˜๋Š” ์„ธํฌ๊ฐ„ ๋‚˜๋…ธํŠœ๋ธŒ ๋„คํŠธ์›Œํฌ๊ฐ€ ๋‰ด๋Ÿฐ ์‚ฌ์ด์—๋„ ์กด์žฌํ•˜๋Š”์ง€๋Š” ์•„์ง ์ถฉ๋ถ„ํžˆ ๊ทœ๋ช…๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ์ฅ ๋Œ€๋‡Œํ”ผ์งˆ์—์„œ ์ˆ˜์ƒ๋Œ๊ธฐ ํ•„๋กœํฌ๋””์•„๊ฐ€ ์ธ์ ‘ํ•œ ์ˆ˜์ƒ๋Œ๊ธฐ์™€ ์ง์ ‘ ์—ฐ๊ฒฐ๋˜์–ด ์ˆ˜์ƒ๋Œ๊ธฐ ๊ฐ„ ๋‚˜๋…ธํŠœ๋ธŒ(dendritic nanotube; DNT)๋ฅผ ํ˜•์„ฑํ•˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•˜์˜€์Šต๋‹ˆ๋‹ค. ๋ถ„๋ฆฌ ๋ฐฐ์–‘๋œ ๋‰ด๋Ÿฐ์—์„œ์˜ ์ดˆ๊ณ ํ•ด์ƒ๋„ ํ˜„๋ฏธ๊ฒฝ ๋ถ„์„์„ ํ†ตํ•ด DNT๊ฐ€ ์•กํ‹ด์ด ํ’๋ถ€ํ•œ ์—ญ๋™์  ๊ตฌ์กฐ์ด๋ฉฐ ์žฅ๊ฑฐ๋ฆฌ ์นผ์Š˜ ์ „๋‹ฌ์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•จ์„ ๋ฐํ˜”์œผ๋ฉฐ, ์˜์ƒ ๋ถ„์„ ๋ฐ ๊ธฐ๊ณ„ํ•™์Šต ๊ธฐ๋ฐ˜ ๋ถ„์„์„ ํ†ตํ•ด ์ƒ์ฒด ๋‚ด DNT๊ฐ€ ์‹œ๋ƒ…์Šค ๊ฐ€์‹œ๋Œ๊ธฐ์™€ ํ•ด๋ถ€ํ•™์ ์œผ๋กœ ๋ช…ํ™•ํžˆ ๊ตฌ๋ณ„๋จ์„ ๊ฒ€์ฆํ•˜์˜€์Šต๋‹ˆ๋‹ค. ๋˜ํ•œ, DNT๋Š” ๋ฒ ํƒ€-์•„๋ฐ€๋กœ์ด๋“œ๋ฅผ ๋น„๋กฏํ•œ ์ž‘์€ ๋ถ„์ž๋“ค์˜ ์ˆ˜์†ก ํ†ต๋กœ๋กœ ๊ธฐ๋Šฅํ–ˆ์Šต๋‹ˆ๋‹ค. ์•Œ์ธ ํ•˜์ด๋จธ ์งˆํ™˜ ๋ชจ๋ธ ์ฅ์˜ ์ „์ „๋‘ํ”ผ์งˆ์—์„œ ์•„๋ฐ€๋กœ์ด๋“œ ํ”Œ๋ผํฌ ํ˜•์„ฑ ์ด์ „์— DNT ๋ฐ€๋„๊ฐ€ ์ฆ๊ฐ€ํ•จ์„ ๊ด€์ฐฐํ•˜์˜€๋Š”๋ฐ, ์ด๋Š” ์ˆ˜์ƒ๋Œ๊ธฐ-DNT ๋„คํŠธ์›Œํฌ๊ฐ€ ์•Œ์ธ ํ•˜์ด๋จธ๋ณ‘ ๋ณ‘๋ฆฌ์— ๊ธฐ์—ฌํ•  ๊ฐ€๋Šฅ์„ฑ์„ ์ œ์‹œํ•ฉ๋‹ˆ๋‹ค. ๋‚˜์•„๊ฐ€, ์ปดํ“จํ„ฐ ๋ชจ๋ธ๋ง์„ ํ†ตํ•ด DNT๋ฅผ ๋งค๊ฐœ๋กœ ํ•œ ๋ฒ ํƒ€-์•„๋ฐ€๋กœ์ด๋“œ์˜ ์ „ํŒŒ๊ฐ€ ํŠน์ • ๋‰ด๋Ÿฐ ๋‚ด ์ถ•์ ์„ ๊ฐ€์†ํ™”ํ•˜๋Š” ๊ฒƒ์„ ์˜ˆ์ธก, ์žฌํ˜„ํ–ˆ์Šต๋‹ˆ๋‹ค. ๋ณธ ๊ฒฐ๊ณผ๋Š” ๊ธฐ์กด์˜ ์‹œ๋ƒ…์Šค ๊ธฐ๋ฐ˜ ์†Œํ†ต์„ ๋„˜์–ด, ๋‡Œ ๋‚ด์— ๋‚˜๋…ธํŠœ๋ธŒ๋ฅผ ํ†ตํ•œ ์ƒˆ๋กœ์šด ์„ธํฌ ์—ฐ๊ฒฐ๋ง์ด ์กด์žฌํ•จ์„ ์‹œ์‚ฌํ•ฉ๋‹ˆ๋‹ค.


Epigenetic control of topoisomerase 1 activity presents a cancer vulnerability

08/12/2025 Nature Communications

KLAM author: Tae-Hee Lee PhD, Postdoctoral Research Fellow at Johns Hopkins University

DNA transactions introduce torsional constraints that pose an inherent risk to genome integrity. While topoisomerase 1 (TOP1) activity is essential for DNA supercoil removal, the aberrant stabilization of TOP1:DNA cleavage complexes (TOP1ccs) can result in cytotoxic DNA lesions. What protects genomic hot spots of topological stress from excessive TOP1cc accumulation remains unknown. Here, we identify chromatin context as an essential means to coordinate TOP1cc resolution. Through its ability to bind poly(ADP-ribose) (PAR), the histone variant macroH2A1.1 facilitates TOP1cc repair factor recruitment and lesion turnover, thereby preventing DNA damage in response to transcription-associated topological stress. The alternatively spliced macroH2A1.2 isoform is unable to bind PAR or protect from TOP1ccs. Impaired macroH2A1.1 splicing, a frequent cancer feature, was predictive of increased sensitivity to TOP1 poisons in a pharmaco-genomic screen in breast cancer cells, and macroH2A1.1 inactivation mirrored this effect. We propose macroH2A1 alternative splicing as an epigenetic modulator of TOP1-associated genome maintenance and a potential cancer vulnerability.

Topoisomerase 1 ํ™œ์„ฑ์˜ ํ›„์„ฑ์œ ์ „ํ•™์  ์กฐ์ ˆ๊ณผ ์•” ์ทจ์•ฝ์„ฑ์˜ ๋ถ„์ž์  ๊ธฐ์ „

DNA์˜ ๋™์  ๊ณผ์ • (์ „์‚ฌ, ๋ณต์ œ, ์ˆ˜์„ )์€ ์œ„์ƒ์  ์ œ์•ฝ์„ ์œ ๋ฐœํ•˜์—ฌ ์œ ์ „์ฒด ์•ˆ์ •์„ฑ์„ ์œ„ํ˜‘ํ•ฉ๋‹ˆ๋‹ค. Topoisomerase 1 (TOP1)์€ ์ด๋Ÿฌํ•œ DNA ์ดˆ๋‚˜์„ ์„ ํ•ด์†Œํ•˜๋Š”๋ฐ ํ•„์ˆ˜์ ์ด์ง€๋งŒ, ๋น„์ •์ƒ์ ์œผ๋กœ ์•ˆ์ •ํ™”๋œ TOP1 ์ ˆ๋‹จ ๋ณตํ•ฉ์ฒด (TOP1cc)๋Š” DNA ์†์ƒ์„ ์ดˆ๋ž˜ํ•ฉ๋‹ˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ๋Š” ํžˆ์Šคํ†ค ๋ณ€์ด์ฒด macroH2A1์˜ ์Šคํ”Œ๋ผ์ด์‹ฑ ๋ณ€์ด์ฒด ์ค‘ ํ•˜๋‚˜์ธ macroH2A1.1์ด PARP1 ํ™œ์„ฑ์„ ๋งค๊ฐœ๋กœ TOP1cc ์œ ๋ฐœ DNA ์†์ƒ์˜ ํšŒ๋ณต์„ ์ด‰์ง„ํ•˜์—ฌ ์ „์‚ฌ ๊ด€๋ จ DNA ์œ„์ƒ์  ์ŠคํŠธ๋ ˆ์Šค์— ์˜ํ•œ ์†์ƒ์„ ์–ต์ œํ•จ์„ ๊ทœ๋ช…ํ•˜์˜€์Šต๋‹ˆ๋‹ค. ๋ฐ˜๋ฉด ๋˜ ๋‹ค๋ฅธ ๋ณ€์ด์ฒด macroH2A1.2๋Š” poly(ADP-ribose) (PAR) ๊ฒฐํ•ฉํ•˜์ง€ ๋ชปํ•ด TOP1cc๋กœ๋ถ€ํ„ฐ ์œ ์ „์ฒด๋ฅผ ๋ณดํ˜ธํ•˜์ง€ ๋ชปํ–ˆ์Šต๋‹ˆ๋‹ค. ๋˜ํ•œ ์•”์—์„œ ํ”ํžˆ ๊ด€์ฐฐ๋˜๋Š” macroH2A1.1 ์Šคํ”Œ๋ผ์ด์‹ฑ ์žฅ์• ๋Š” TOP1 ์–ต์ œ์ œ์— ๋Œ€ํ•œ ๋ฏผ๊ฐ์„ฑ์„ ๋†’์˜€์œผ๋ฉฐ, ์ด๋Š” ํ›„์„ฑ์œ ์ „์  ์œ ์ „์ฒด ์•ˆ์ •์„ฑ ์กฐ์ ˆ๊ณผ ์•” ์ทจ์•ฝ์„ฑ์˜ ๋ถ„์ž์  ๊ธฐ์ „์„ ์ œ์‹œํ•ฉ๋‹ˆ๋‹ค.


Sleep need-dependent plasticity of a thalamic circuit promotes homeostatic recovery sleep

06/19/2025 Science

KLAM author: Sang Soo Lee PhD, Research Associate at Johns Hopkins University

Sleep is a universal and essential behavior regulated by a homeostatic process: the longer we stay awake, the stronger our drive to sleep. What mechanisms allow the sleep drive to persist even after extended wakefulness? In this study, we identified a thalamic neural population, the nucleus reuniens (RE), projecting to the zona incerta (ZI) that plays a causal and essential role in homeostatic sleep. The REโ€“ZI pathway undergoes sleep-need-dependent synaptic plasticity and is modulated by a CaMKII signaling cascade. RE activation not only induces deep, prolonged sleep but also initiates naturalistic, sleep-preparatory behaviors, indicating that this circuit encodes an internal state of โ€œsleepinessโ€ rather than simply triggering the sleep state. The REโ€“ZI pathway represents the first identified โ€œsleep driveโ€ module: a dedicated, plastic circuit that integrates internal sleep pressure and orchestrates both behavioral and physiological components of recovery sleep.

์ˆ˜๋ฉด์€ ํ•ญ์ƒ์„ฑ์— ์˜ํ•ด ์กฐ์ ˆ๋˜๋Š” ๋ณดํŽธ์ ์ด๊ณ  ํ•„์ˆ˜์ ์ธ ํ–‰๋™์ž…๋‹ˆ๋‹ค. ๊นจ์–ด ์žˆ๋Š” ์‹œ๊ฐ„์ด ๊ธธ์–ด์งˆ์ˆ˜๋ก ์ˆ˜๋ฉด ์š•๊ตฌ๋Š” ์ ์  ๊ฐ•ํ•ด์ง‘๋‹ˆ๋‹ค. ๊ทธ๋Ÿฐ๋ฐ ์ด๋ ‡๊ฒŒ ์ƒ๊ธด ์ˆ˜๋ฉด ์š•๊ตฌ๋Š”, ์˜ค๋žœ ๊ฐ์„ฑ ์ƒํƒœ ์ดํ›„์—๋„ ์–ด๋–ป๊ฒŒ ์ง€์†๋  ์ˆ˜ ์žˆ์„๊นŒ์š”? ์ €ํฌ๋Š” ์‹œ์ƒ์˜ ์ž‘์€ ๋ถ€์œ„์ธ โ€˜์žฌ๊ฒฐํ•ฉํ•ต(nucleus reuniens, RE)โ€™์˜ ํŠน์ • ๋‰ด๋Ÿฐ ์ง‘๋‹จ์ด โ€˜๋ถˆํ™•์ •๋Œ€(zona incerta, ZI)โ€™๋กœ ์‹ ํ˜ธ๋ฅผ ๋ณด๋‚ด๋ฉด์„œ, ์ˆ˜๋ฉด ํ•ญ์ƒ์„ฑ์— ํ•ต์‹ฌ์ ์ธ ์—ญํ• ์„ ํ•œ๋‹ค๋Š” ์‚ฌ์‹ค์„ ์•Œ์•„๋‚ด์—ˆ์Šต๋‹ˆ๋‹ค. ์ด REโ€“ZI ๊ฒฝ๋กœ๋Š” ์ˆ˜๋ฉด ํ•„์š”์„ฑ์— ๋”ฐ๋ผ ์‹œ๋ƒ…์Šค ๊ฐ€์†Œ์„ฑ์ด ๋ณ€ํ™”ํ•˜๋ฉฐ, CaMKII ์‹ ํ˜ธ ์ „๋‹ฌ๊ณ„์— ์˜ํ•ด ์กฐ์ ˆ๋ฉ๋‹ˆ๋‹ค. RE๋ฅผ ํ™œ์„ฑํ™”ํ•˜๋ฉด ๊นŠ๊ณ  ์˜ค๋ž˜ ์ง€์†๋˜๋Š” ์ˆ˜๋ฉด์ด ์œ ๋„๋  ๋ฟ ์•„๋‹ˆ๋ผ, ์ž์—ฐ์Šค๋Ÿฌ์šด ์ˆ˜๋ฉด ์ค€๋น„ ํ–‰๋™๊นŒ์ง€ ๋‚˜ํƒ€๋‚ฉ๋‹ˆ๋‹ค. ์ด๋Š” ์ด ํšŒ๋กœ๊ฐ€ ๋‹จ์ˆœํžˆ ์ˆ˜๋ฉด์„ ์ด‰์ง„ํ•˜๋Š” ๊ฒƒ์ด ์•„๋‹ˆ๋ผ, โ€˜์กธ๋ฆผโ€™์ด๋ผ๋Š” ๋‚ด์  ์ƒํƒœ ์ž์ฒด๋ฅผ ๋ถ€ํ˜ธํ™”ํ•˜๊ณ  ์žˆ๋‹ค๋Š” ์ ์„ ๋ณด์—ฌ์ค๋‹ˆ๋‹ค. REโ€“ZI ๊ฒฝ๋กœ๋Š” ์ง€๊ธˆ๊นŒ์ง€ ๋ฐํ˜€์ง„ ์ตœ์ดˆ์˜ โ€˜์ˆ˜๋ฉด ์š•๊ตฌ(sleep drive)โ€™ ๋ชจ๋“ˆ๋กœ, ๋‡Œ๋‚ด ์ˆ˜๋ฉด ์••๋ ฅ์„ ํ†ตํ•ฉํ•˜์—ฌ ํšŒ๋ณต ์ˆ˜๋ฉด์— ํ•„์š”ํ•œ ํ–‰๋™์ ยท์ƒ๋ฆฌ์  ๋ฐ˜์‘์„ ์กฐ์œจํ•˜๋Š” ์œ ์—ฐํ•œ ํšŒ๋กœ ์‹œ์Šคํ…œ์ด๋ผ ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.


Punctuation Restoration Improves Structure Understanding Without Supervision

03/30/2025 10th Workshop on Representation Learning for NLP

KLAM author: Junghyun Min, PhD student at Georgetown University

Unsupervised learning objectives like autoregressive and masked language modeling constitute a significant part in producing pre-trained representations that perform various downstream applications from natural language understanding to conversational tasks. However, despite impressive generative capabilities of recent large language models, their abilities to capture syntactic or semantic structure within text lag behind. We hypothesize that the mismatch between linguistic performance and competence in machines is attributable to insufficient learning of linguistic structure knowledge via currently popular pre-training objectives. Working with English, we show that punctuation restoration as a learning objective improves performance on structure-related tasks like named entity recognition, open information extraction, chunking, and part-of-speech tagging. Punctuation restoration results in โ–ฒโ‰ฅ 2%p improvement in 16 out of 18 experiments, across 6 out of 7 tasks. Our results show that punctuation restoration is an effective learning objective that can improve structure understanding and yield a more robust structure-aware representations of natural language in base-sized models.

GPT, BERT ๋“ฑ ํ˜„์žฌ ์–ธ์–ด๋ชจ๋ธ์— ์ฃผ๋กœ ์‚ฌ์šฉ๋˜๋Š” ํ•™์Šต ๋ฐฉ๋ฒ•์ธ autoregressive ๋˜๋Š” masked language modeling์€ ๋‹ค์Œ ๋‹จ์–ด ์˜ˆ์ธก์— ๋›ฐ์–ด๋‚œ ์„ฑ๋Šฅ์„ ์•ผ๊ธฐํ•˜์—ฌ ๋Œ€ํ™” ๋˜๋Š” ํ…์ŠคํŠธ ์ƒ์„ฑ ๋ชจ๋ธ ํ•™์Šต์— ์œ ์šฉํ•˜๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ํ•ด๋‹น ๋ชจ๋ธ๋“ค์€ ์ƒ์„ฑ ์„ฑ๋Šฅ์— ๋น„ํ•ด ๊ตฌ๋ฌธ์ , ์˜๋ฏธ์  ๊ตฌ์กฐ์— ๋Œ€ํ•œ ์ดํ•ด๊ฐ€ ๋–จ์–ด์ ธ ํ…์ŠคํŠธ ๋ถ„์„ ๋ฐ ๋ถ„๋ฅ˜์— ์ ํ•ฉํ•˜์ง€ ์•Š๋‹ค. ์ด ๋…ผ๋ฌธ์—์„œ๋Š” ์ธ๊ฐ„์ด ์–ธ์–ด ์ฒ˜๋ฆฌ ์‹œ ์šด์œจ ์ •๋ณด๋ฅผ ํ™œ์šฉํ•˜๋Š” ์ ์— ์ฐฉ์•ˆํ•˜์—ฌ ๋ฌธ์žฅ ๋ถ€ํ˜ธ ๋ณต์›์„ ๋น„์ง€๋„ ํ•™์Šต ๋ฐฉ๋ฒ•์œผ๋กœ ์ œ์•ˆํ•˜๋ฉฐ, ํ•ด๋‹น ํ•™์Šต ๋ฐฉ๋ฒ•์ด ์ •๋ณด ์ถ”์ถœ, ๊ฐœ์ฒด๋ช… ์ธ์‹, ์˜๋ฏธ์—ญ ๋ถ„๋ฅ˜ ๋“ฑ ์–ธ์–ด ๊ตฌ์กฐ ๊ด€๋ จ 6๊ฐœ ํƒœ์Šคํฌ 12๊ฐœ ๋ฐ์ดํ„ฐ์…‹์—์„œ ์„ฑ๋Šฅ ํ–ฅ์ƒ์„ ์ด๋Œ์–ด๋ƒ„์„ ํ™•์ธํ•œ๋‹ค.


KLAM ํšŒ์›๋‹˜๋“ค์˜ ์—…์ ์„ ๊ณต์œ ํ•ฉ๋‹ˆ๋‹ค. ์•„๋ž˜ ์–‘์‹์— ์ ์–ด ๋ณด๋‚ด์ฃผ์„ธ์š”!

โ† Back

Thank you for your response. โœจ