Recently, a multi-institute research team led by Yosuke Togashi reported on a study that showed that cancer cells transfer mitochondria with mutant DNA and inhibitory molecules to immune cells to reduce the antitumor response. This previously unknown mechanism also shows that cancer cells with mutated mitochondrial (mt) DNA indicate a poor prognosis for certain patients.
Cancers use many strategies to avoid immune surveillance. Among those are the acquisition of mitochondria from T cells that infiltrate the tumor. Those mitochondria enhance the cancer cells energy production. But the transfer of mitochondria also works in the other direction. Mitochondria with mutations from the cancer cells are transferred to the immune cells. Those defective organelles degrade the antitumor response. Thus, the metabolic reprogramming resulting from the bidirectional exchange of mitochondria creates an environment that encourages tumor growth.
The Togashi research team sought to determine the mechanisms that promote that exchange. They examined clinical samples of immune cells for mtDNA mutations from the cancer cells. They found similar mutations in mtDNA in the immune cells. In most cases, mitochondria with mutated mtDNA are destroyed by mitophagy. However, the team identified proteins that attach to the mitochondria with mutated mtDNA and are transferred with them to the immune cells. One of these was the mitophagy-inhibiting protein USP30. Those recipient immune cells then failed to undergo mitophagy. In addition, the T cells with the mutant mitochondria became senescent.
This work has valuable clinical implications. Finding of mutant mtDNA in the cancer cells indicates that immune checkpoint inhibitors may be less effective in patients with melanoma and non-small-cell lung cancer.
A conversation with Dr. Togashi
MitoWorld: What directions do you foresee your research taking to advance this work?
Togashi: Our next steps are to clarify the mechanistic basis of mitochondrial replacement in addition to just transfer, the key molecules on donor and recipient cells, and the tumor microenvironmental conditions that promote it. We will then expand the work across additional cancer types and other recipient cell populations (beyond T cells) to build a broader map of who transfers mitochondria to whom and with what functional consequences. We are also interested in whether similar mitochondrial transfer in non-cancer diseases, such as chronic inflammation disease. In parallel, we will determine how transfer intersects with mitochondrial dynamics (fusion-fission, mitophagy, and quality control) that may determine whether transferred mitochondria persist. Ultimately, we aim to translate these findings into biomarkers that predict therapy response and therapeutic strategies that restore mitochondrial quality control to improve immunotherapy efficacy.
MitoWorld: Do you have any thoughts on the mechanism by which the immune cells receiving the mutated mtDNA become senescent?
Togashi: We think senescence is likely driven by a combination of mitochondrial dysfunction and stress signaling. Mutant mitochondria can impair respiratory capacity and elevate reactive oxygen species, which may trigger DNA-damage responses and activate pathways, such as p53–p21 and/or p16. Importantly, if mitophagy is suppressed, these damaged organelles may persist, making the dysfunction more durable.
MitoWorld: As you noted, mitochondria might be mediated by various mechanisms. Do you have a favorite candidate for this process?
Togashi: Several mechanisms could plausibly mediate mitochondrial transfer, including tunneling nanotubes, extracellular vesicles, and direct cell-cell contact-dependent processes. From our observations so far, a leading candidate is tunneling nanotubes and/or EVs. At the same time, we have the impression that the dominant route may differ, depending on the cell type and context. So these mechanisms are not mutually exclusive and may operate in parallel in different tumor microenvironments.
MitoWorld: Your work here and that of others have provided good evidence for the concept of mitochondrial transfer. However, some researchers have been slow to warm to this idea. Can you speculate on why there has been this reluctance to accept this?
Togashi: I think the hesitation is understandable. Demonstrating mitochondrial transfer convincingly in vivo is technically challenging: mitochondria are abundant, dynamic, and easy to mis-attribute due to imaging limitations, labeling artifacts, or contamination during cell isolation. In addition, the field has long emphasized mitochondria as strictly cell-autonomous organelles, so the conceptual shift takes time. In our case, we view the evidence as particularly strong because we can identify transferred mitochondria by shared genetic mutations in both clinical specimens and in vivo models, providing an orthogonal line of support beyond imaging alone. As methods improve, particularly lineage tracing, spatial approaches, and rigorous controls, I expect the evidence base will continue to strengthen and the community will converge.
MitoWorld: Do you have any speculation on how your findings might be translated to clinical applications?
Togashi: We see a few translational opportunities. First, tumor mtDNA mutation profiles could serve as biomarkers to stratify patients who are less likely to benefit from immune checkpoint inhibitors and, thus, help to guide treatment selection. Second, therapeutically targeting the transfer process or restoring mitochondrial quality control in recipient immune cells may help to preserve anti-tumor immunity. Third, these insights may inform combination strategies (e.g., pairing immunotherapy with mitochondria-targeted agents), thereby converting non-responders into responders.
MitoWorld: How did you become interested in mitochondria?
Togashi: My original training is in pulmonary medicine and thoracic oncology. From there, I became increasingly interested in cancer immunology, and during my postdoctoral period, I worked primarily on regulatory T cells. I was already aware of the accumulating evidence that mitochondrial dysfunction is an important feature of exhausted T cells, but when I was preparing to start my own laboratory, I was searching for an original research direction. Around that time, I learned about the phenomenon of mitochondrial transfer, and it immediately struck me as a fascinating and potentially transformative concept for understanding tumor-immune interactions. That was the starting point that led me to pursue this line of work.
Reference
Ikeda H, Kawase K, Nishi T, Watanabe T, Takenaga K, Inozume T, Ishino T, Aki S, Lin J, Kawashima S, Nagasaki J, Ueda Y, Suzuki S, Makinoshima H, Itami M, Nakamura Y, Tatsumi Y, Suenaga Y, Morinaga T, Honobe-Tabuchi A, Ohnuma T, Kawamura T, Umeda Y, Nakamura Y, Togashi Y (2025) Immune evasion through mitochondrial transfer in the tumour microenvironment. Nature 638: 225-236.
A paper in Nature Cancer by a multi-institute research team, led by Michael Cangkrama and Sabine Werner at the ETH Zurich, describes how cancer cells use the transfer of mitochondria to highjack fibroblasts to support the cancer. The study also implicates a protein as critical for the transfer that may be useful in novel therapeutic strategies.
The transfer of mitochondria from various cell types into cancer cells has been reported to enhance cancer cell proliferation, motility and more. Fibroblasts associated with tumors are known to adopt certain characteristics, and they frequently support cancer development. The Cangkrama-Werner team sought to determine if the transfer of mitochondria might work in the opposite direction. In other words, do cancer cells transfer mitochondria to the fibroblasts?
To answer this question, the team examined cancer cells and fibroblasts in co-cultures and xenograft tumors. Intriguingly, mitochondria were transferred to the fibroblasts where they caused changes in the metabolism of the recipient cells. The fibroblasts began to express markers associated with cancer-associated fibroblasts and to release secreted soluble and extracellular matrix proteins that support tumor growth. The team also found that the transfer of mitochondria depends on the mitochondrial trafficking protein MIRO2. Without MIRO2, the transfer and the other changes failed. These findings are consistent with the overexpression of MIRO2 at the leading edge of epithelial skin cancers and other malignancies. Finally, they found that the mitochondria are transferred through thin membranous structures called tunneling nanotubes.
This study demonstrates that mitochondrial transfer from cancer cells to fibroblasts is a key regulator of CAF differentiation. The findings also implicate MIRO2 and mitochondrial transfer as potential targets for strategies to treat cancers.
A conversation with Drs. Cangkrama and Werner
MitoWorld: Can you give us an idea of what studies you might do to advance the findings presented here?
Werner: There are several open issues, including the signals that induce the transfer, the genes and proteins that promote the transfer and the targets of the transferred mitochondria, which induce the pro-tumorigenic phenotype in fibroblasts.
MitoWorld: You showed that oxidative phosphorylation is the main activity that is influenced by the mitochondrial transfers. Can you speculate on how that process translates into the changes in gene expression in the fibroblasts? Might there be other mitochondrial activities involved in activating the interferon-response genes?
Werner: Oxidative phosphorylation is highly relevant in this context, but it is probably not the only relevant factor. Oxidative phosphorylation results in the production of various metabolites, which are likely to regulate transcription factors that control gene expression in fibroblasts. We found that the interferon-response genes are not regulated by transferred mitochondria, but likely by transfer of other molecules from cancer cells to fibroblasts in co-cultures, which remain to be identified.
MitoWorld: The transfer of mitochondria from cell to cell has been somewhat controversial. Can you comment on the seeming reluctance of some researchers to embrace this concept?
Werner: Many results were obtained with MitoTracker, which is known to be leaky. Several studies did not include appropriate controls to address this problem, which might have contributed to skepticism about mitochondrial transfer. The concept was also unexpected when first proposed. Even when transfer occurs, it is difficult to predict and appears to depend on specific conditions, such as cellular stress or high metabolic demand, further fueling doubts about its physiological relevance. However, many exciting and well-controlled studies have recently been published, which show the relevance of mitochondrial transfer under various conditions. This makes mitochondrial transfer a promising area of research, with the potential to uncover new mechanisms of intercellular communication and to develop innovative therapeutic strategies.
MitoWorld: Can you speculate how your findings might translate to therapies for cancer?
Werner: There is a long way from this discovery to a cancer drug. Nevertheless, our work suggests that inhibition of mitochondrial transfer (e.g., by blocking MIRO2 or other proteins important for the transfer) could be a novel approach for cancer treatment. In addition, identification of the relevant targets of the mitochondrial transfer in recipient fibroblasts could provide new therapeutic opportunities.
MitoWorld: How did you first become interested in mitochondria?
Werner: Several previous studies of our laboratory revealed the important role of mitochondria in the pro-tumorigenic cancer-associated fibroblast phenotype. This raised our interest in these exciting organelles.
Reference
Cangkrama M, Liu H, Wu X, Yates J, Whipman J, Gäbelein CG, Matsushita M, Ferrarese L, Sander S, Castro-Giner F, Asawa S, Sznurkowska MK, Kopf M, Dengjel J, Boeva V, Aceto N, Vorholt JA, Werner S (2025) MIRO2-mediated mitochondrial transfer from cancer cells induces cancer-associated fibroblast differentiation. Nature Cancer 6: 1714–1733.
A recent paper in Science Advances describes the findings of a research team led by Mondira Kundu, MD, PhD, at St. Jude Children’s Research Hospital. The study investigates how mitochondrial (mt) DNA mutations influence leukemia pathology. Unexpectedly, their research showed that a moderate burden of mtDNA mutations can enhance the development of leukemia. They also show that cancer can be re-initiated by inhibiting a specific enzyme in cells carrying a high burden of mtDNA mutations.
Cancers are highly energy dependent, with the mitochondria serving as the cell’s main energy producers. While most research into this connection has focused on mutations in nuclear DNA that affect mitochondrial function, Dr. Kundu’s team explored whether mutations directly in the mtDNA contribute to tumor development.
The researchers began with three lines of mice expressing a mutant exonuclease-inactive mitochondrial DNA polymerase (Polgmut) that lacks accurate proofreading ability. These lines possess either zero (Polgwt/wt), one (Polgwt/mut), or two (Polgmut/mut) copies of the mutated allele, resulting in a graded accumulation of mtDNA mutations. Hematopoietic progenitor cells (HPCs) were isolated from these mice and engineered to express NMyc, a member of the MYC family of transcription factors. Members of the MYC family are commonly dysregulated in many blood cancers, such as leukemia. By transplanting these HPCs into irradiated recipient mice, the team assessed the impact of different levels of mtDNA mutations on cellular metabolism and cancer development.
The findings were unexpected: while metabolism was reduced in mice with either heterozygous or homozygous mutant cells, mice with a moderate mutation load (Polgmut/wt) were more prone to tumor formation than those with a high mutation load (Polgmut/mut). Metabolic plasticity was affected by the number of mtDNA mutations and was critical to the tumorigenic potential of the HPCs. In essence, a moderate number of mutations makes cells more metabolically flexible and more carcinogenic, whereas extensive mutations diminish both their metabolic flexibility and cancerous potential.
Conversation with Dr. Kundu
MitoWorld. Your results are unexpected. Do you have any thoughts on why the partially damaged mitochondria would be beneficial to the cancer? Yes, it appears that partially damaged mitochondria create a unique metabolic environment that cancer cells can exploit. Moderate mitochondrial dysfunction may allow for metabolic reprogramming—enhancing glycolysis and other pathways that support rapid cell growth—without fully compromising energy production. This balance may give cancer cells a survival and growth advantage.
MitoWorld. Can you speculate on how the damaged mitochondria are able to increase their metabolic support for the tumors? Damaged mitochondria may trigger adaptive responses in cancer cells, activating alternative metabolic pathways and stress responses. For instance, partial mitochondrial dysfunction can increase the reliance on glycolysis (the Warburg effect) and other biosynthetic pathways, thereby supporting both energy needs and the synthesis of cellular building blocks required for proliferation. The cells remain metabolically flexible, which is crucial for tumor growth.
MitoWorld. You note both similar and different results with this mutation and different cancers. To what would you attribute those varying results? The impact of mtDNA mutations likely depends on the tissue context and the specific metabolic requirements of different cancers. Some tumors may be more resilient to mitochondrial dysfunction, while others are more dependent on intact mitochondrial metabolism. Additionally, the interplay between mtDNA mutations and nuclear gene mutations can vary, influencing how cells adapt metabolically and whether they become more or less tumorigenic.
MitoWorld. Your results show that inhibiting pyruvate dehydrogenase kinase improved the ability of the homozygous mutant to cause tumors. Does this observation have any therapeutic implications? While our findings are not immediately translatable to therapy, they do highlight the critical role of metabolic plasticity in leukemogenesis. The fact that inhibiting pyruvate dehydrogenase kinase (PDK) restored the proliferative capacity of homozygous mutant HPCs suggests that metabolic pathways can profoundly influence tumor development. This raises the possibility that targeting metabolic enzymes like PDK could one day be leveraged to modulate cancer cell growth, either by restricting the metabolic flexibility of cancer cells or by exploiting specific metabolic vulnerabilities. However, more research is needed, especially since the mutation burden in human leukemias typically comes from single mtDNA mutations, unlike the mouse model, where the burden comes from the cumulative effect of many mutations.
MitoWorld. What do you see as the next steps to follow up on this work? The next step is to introduce specific mtDNA mutations into primary cells and then add oncogenes. This will allow us to examine how individual mtDNA mutations—and their allele fractions—impact tumorigenesis. By better modeling the mutation patterns seen in human cancers, we can determine whether particular mtDNA mutations or metabolic states make cancer cells more susceptible to targeted metabolic therapies. Ultimately, this line of research could guide the development of new therapeutic strategies that exploit the metabolic dependencies created by mtDNA mutations in cancer cells.
MitoWorld. What attracted you to the study of mitochondria? As a hematopathologist, I have a strong interest in hematologic diseases, but my research has always spanned a variety of biological systems. My interest in mitochondria began during my postdoctoral fellowship in Craig Thompson’s lab, where I used red blood cell maturation as a model to study mitophagy—the process by which cells selectively remove damaged or superfluous mitochondria. That experience highlighted for me how central mitochondrial function and dysfunction are to cell biology and disease, including cancer. I am particularly intrigued by how cells sense and respond to mitochondrial dysfunction—especially when it’s caused by mtDNA mutations—and how these adaptive pathways can contribute to cancer development. Understanding the complex interplay between mitochondrial function and cellular responses not only deepens our knowledge of cancer biology but also points to new possibilities for therapeutic intervention.
Reference
Li-Harms X, Lu J, Fukuda Y, Lynch J, Sheth A, Pareek G, Kaminski MM, Ross HS, Wright CW, Smith AL, Wu H, Wang Y-D, Valentine M, Neale G, Vogel P, Pounds S, Schuetz JD, Ni M, Kundu M (2025) Somatic mtDNA mutation burden shapes metabolic plasticity in leukemogenesis. Science Advances 11(1): eads8489.
Do neurons help cancer spread? In a paper published in Nature, a multi-institution research team, led by Simon Grelet at the University of South Alabama, provides strong evidence for a key relationship between cancer cells and neurons. They showed that mitochondria migrate from neurons to cancer cells to increase the metabolism of the cancer cells.
Dr. Grelet’s team focused on the mitochondria in cancer cells. The spread of tumors requires a significant amount of energy, and mitochondria produce energy for the cells. Cancer cells are well-known for adapting their metabolism to fit changing conditions. However, this metabolic plasticity was thought to be due to changes within the tumor cells themselves. For example, they could obtain energy by glycolysis (a less efficient process for producing ATP in the cytoplasm) or by oxidative phosphorylation (a more efficient process in the mitochondria). However, cancer cells also receive outside help in the form of metabolites, growth factors and cytokines from other cells.
Previous experiments had shown that somehow the neurons aid tumor progress, but what really interested the Grelet team was a relatively new concept that mitochondria can move from cell to cell. When neurons were co-cultured with tumor cells, the neurons experience changes to their metabolism. The number of their mitochondria increases, and some are transferred to the tumor cells. How could this work?
The team first looked at a co-culture of an aggressive breast cancer cells and neurons. The neurons underwent a morphological change from globular to tubular structures, which indicated the cancer-induced differentiation of the neurons. Neuronal differentiation is associated with changes of metabolic programming from glycolysis to oxidative phosphorylation. These changes were also associated with the establishment of long neuron protrusions, forming a neural network in the culture in vitro, which reflects the establishment of a nerve-cancer crosstalk and suggests the communication or sharing of biological materials.
Next, the team used fluorescent-labeled mitochondria to show that the mitochondria migrated from neurons to cancer cells. These methods revealed additional insights into the transfers, but they had significant limitations. To overcome those limitations, the team developed MitoTRACER. This system used genetic engineering where the donor cell mitochondria carry a tag protein that, once transferred to the recipient cells, triggers an enzymatic reaction activating the permanent expression of a fluorescent protein. The trick is that, when mitochondria move from the neurons to the cancer cells, the red fluorescence is lost, and a green signal is permanently activated. Thus, it is easy to monitor cells with mitochondria that have migrated into a new cell.
Using this unique approach, the researchers have been able to investigate the fate of the recipient cells during the cancer progression cascade, and by fate mapping of the recipient cells from the primary tumor during metastasis progression in vivo. The researchers noted enrichment of acquired mitochondria in brain metastases that might indicate enhanced metabolic plasticity from those extra mitochondria. Neuronal mitochondria are more metabolically active and might better enable tumor cells to thrive in the brain environment. This approach allowed us to define the role of mitochondrial transfer in the primary tumor environment. This process generates a subset of highly efficient metastasis cells that succeed through the complex stops and that are resilient to the metastatic barriers to ultimately form distant metastases, which is the main driver of cancer-associated mortality.
In summary, this fascinating paper might have significant implications for future cancer therapies, and beyond the cancer context, the mitoTRACER approach could have broader applications in studying cell-cell transfer of mitochondria in health and disease to understand the physiopathological implications of these transfers better.
Discussion with Dr. Grelet
MitoWorld: Your paper brings to mind the work of Judah Folkman who showed that tumors need and encourage the formation of a blood supply to progress. What do you believe the tumor may be getting from the nerve cells?
Interesting comparison. Yes, there are definitely commonalities between cancer angiogenesis and cancer innervation. In fact, cancer can actively “call” neurons to innervate them. In the context of cancer, we recently showed that aggressive subtypes of breast cancer cells can secrete axon guidance molecules, including Semaphorin-4F, to promote cancer innervation. We also found that this increased nerve density is associated with enhanced metastasis (PMID: 34810279).
The role of axon guidance molecules in promoting cancer innervation was demonstrated years ago by Dr. Gustavo Ayala (PMID: 11746267; PMID: 24097862), who pioneered the field of cancer innervation and collaborated with us on this study. Since then, many studies, including our own, have confirmed the contribution of these molecules to cancer innervation and cancer aggressiveness. While increased nerve density is often associated with poor clinical outcomes, the underlying mechanisms have remained poorly understood. This gap in understanding is precisely what motivated us to conduct this study.
MitoWorld: The movement of mitochondria from cell to cell is still controversial. What do you think it will take to solidify the concept?
The idea of intercellular mitochondrial transfer remains controversial, largely due to the lack of robust tools and clear in vivo evidence. To solidify the concept, new rigorous and physiologically relevant approaches are needed. For example, genetic systems, such as our MitoTRACER approach, which allow for precise and conditional signaling of transfer events, open the avenue for lineage-based tracking of transferred cells in vivo. Such models are critical to move the field forward and to convincingly demonstrate that mitochondrial transfer is not merely an artifact, but a biologically meaningful process.
MitoWorld: How do you think the cells signal to each other to initiate the transfer of mitochondria?
A variety of signals could potentially be involved in initiating mitochondrial transfer, but these remain to be better defined. In the context of cancer innervation, axon guidance molecules may contribute; more generally, transfer events may occur as part of a stress or help response, or by hijacking existing physiological communication routes. These signals could include reactive oxygen species, cytokines, or changes in membrane potential or lipid composition. Cancer cells, for instance, can upregulate molecules involved in cell-cell contact to facilitate mitochondrial transfer, as elegantly demonstrated by Watson et al. (PMID: 37169842). Deciphering the molecular language of this intercellular request remains a critical next step for the field.
MitoWorld: These experiments seem to have ramifications for how the nuclear and mitochondrial genomes coordinate their activities. Do you think the communication between the donated mitochondria and the recipient cell also involves the donor genes?
Absolutely, and this is a topic that warrants further investigation. Mitochondrial transfer introduces not only a new metabolic organelle with all its associated machinery, but also foreign mitochondrial DNA into the recipient cell, raising important questions about nuclear–mitochondrial coordination. While the nuclear genome of the recipient cell continues to govern most mitochondrial functions, the donor mitochondria carry their own genome and organelle machinery, which in our case is fine-tuned for neuronal metabolism and may not be fully compatible. This mismatch could influence mitochondrial gene expression, protein stoichiometry, and ultimately, the bioenergetic output. Whether donor mitochondrial DNA or its transcripts actively modulate recipient cell behavior remains an open and fascinating question. Clarifying the extent and mechanisms of this intergenomic crosstalk is essential to fully understand the consequences of mitochondrial transfer.
MitoWorld: You note that more work is needed to determine if the effects of the donated mitochondria are related to metabolic efficiency or simply that there are more of them. Do you have any sense of which it might be?
In the context of neurons, it is likely a combination of both abundance and quality. On one hand, the increase in mitochondrial mass may help support basic energy demands, particularly under stress. On the other hand, the quality and origin of the donated mitochondria, such as their neuronal bioenergetic efficiency, may actively contribute to metabolic reprogramming in the recipient cell. In our system, the integration of neuronal mitochondria appears to enhance not only energy production but also metabolic plasticity, enabling cancer cells to better adapt to new microenvironments. Dissecting the relative roles of mitochondrial abundance versus functional impact will be key to understanding how mitochondrial transfer influences cell fate and behavior.
MitoWorld: Folkman used his findings to suggest that cancers could be treated by preventing angiogenesis, and that is now widely accepted. Do you have any thoughts on how your findings with neurons might be used to develop therapeutic strategies for cancer?
Yes, absolutely, that is the ultimate goal. There are several possible strategies. First, we can target tumor innervation itself, which supports cancer progression. Second, we can aim to block the intercellular transfer of mitochondria, which fuels cancer cell adaptation. Third, and perhaps most compelling, we can selectively target recipient cancer cells that have acquired unique metabolic or phenotypic traits as a result of mitochondrial uptake. These cells may represent a vulnerable subpopulation that could be eliminated through precision therapies. Together, these approaches offer a new framework for tackling cancer’s adaptability.
MitoWorld: You looked at breast cancer and melanoma cells, both highly metastatic cancer types. Is the neuronal connection equally valid in less metastatic cancers?
We also previously reported a link between cancer plasticity and cancer innervation (PMID:37046688). Our findings suggest that more aggressive cancer cells are more likely to establish neuronal connections and engage in mitochondrial transfers. This, in turn, may further amplify their aggressive behavior, creating a vicious cycle. The ability of these cells to co-opt neural signaling and reshape their metabolic and phenotypic programs reflects a high degree of plasticity that likely contributes to disease progression.
MitoWorld: Also your work focused on solid tumors. Is there anything to be learned about blood tumors?
Interesting point. Blood cancers reside in richly innervated niches, such as the bone marrow, where neurons regulate hematopoiesis and immune responses. So, similar mechanisms might shape the behavior of leukemia or lymphoma cells. However, this remains largely unexplored, and adapting lineage-tracing tools, such as MitoTRACER, to hematological models could open exciting new directions in the field.
Reference
Hoover G, Gilbert S, Curley O, Obellianne C, Lin MT, Hixson W, Pierce TW, Andrews JF, Alexeyev MF. Ding Y, Bu P, Behbod F, Medina D, Chang JT, Ayala G, Grelet S (2025) Nerve-to-cancer transfer of mitochondria during cancer metastasis. Nature