SummaryEndogenous formaldehyde is produced by numerous biochemical pathways fundamental to life, and it can crosslink both DNA and proteins. However, the consequences of its accumulation are unclear. Here we show that endogenous formaldehyde is removed by the enzyme alcohol dehydrogenase 5 (ADH5/GSNOR), and Adh5−/− mice therefore accumulate formaldehyde adducts in DNA. The repair of this damage is mediated by FANCD2, a DNA crosslink repair protein. Adh5−/−Fancd2−/− mice reveal an essential requirement for these protection mechanisms in hematopoietic stem cells (HSCs), leading to their depletion and precipitating bone marrow failure. More widespread formaldehyde-induced DNA damage also causes karyomegaly and dysfunction of hepatocytes and nephrons. Bone marrow transplantation not only rescued hematopoiesis but, surprisingly, also preserved nephron function. Nevertheless, all of these animals eventually developed fatal malignancies. Formaldehyde is therefore an important source of endogenous DNA damage that is counteracted in mammals by a conserved protection mechanism.

The folate-driven one-carbon (1C) cycle is a fundamental metabolic hub in cells that enables the synthesis of nucleotides and amino acids and epigenetic modifications. This cycle might also release formaldehyde, a potent protein and DNA crosslinking agent that organisms produce in substantial quantities. Here we show that supplementation with tetrahydrofolate, the essential cofactor of this cycle, and other oxidation-prone folate derivatives kills human, mouse and chicken cells that cannot detoxify formaldehyde or that lack DNA crosslink repair. Notably, formaldehyde is generated from oxidative decomposition of the folate backbone. Furthermore, we find that formaldehyde detoxification in human cells generates formate, and thereby promotes nucleotide synthesis. This supply of 1C units is sufficient to sustain the growth of cells that are unable to use serine, which is the predominant source of 1C units. These findings identify an unexpected source of formaldehyde and, more generally, indicate that the detoxification of this ubiquitous endogenous genotoxin creates a benign 1C unit that can sustain essential metabolism.

Formate overflow coupled to mitochondrial oxidative metabolism\ has been observed in cancer cell lines, but whether that takes place in the tumor microenvironment is not known. Here we report the observation of serine catabolism to formate in normal murine tissues, with a relative rate correlating with serine levels and the tissue oxidative state. Yet, serine catabolism to formate is increased in the transformed tissue of in vivo models of intestinal adenomas and mammary carcinomas. The increased serine catabolism to formate is associated with increased serum formate levels. Finally, we show that inhibition of formate production by genetic interference reduces cancer cell invasion and this phenotype can be rescued by exogenous formate. We conclude that increased formate overflow is a hallmark of oxidative cancers and that high formate levels promote invasion via a yet unknown mechanism.

Formaldehyde (FA) is a reactive signaling molecule that is continuously produced through a number of central biological pathways spanning epigenetics to one-carbon metabolism. On the other hand, aberrant, elevated levels of FA are implicated in disease states ranging from asthma to neurodegenerative disorders. In this context, fluorescence-based probes for FA imaging are emerging as potentially powerful chemical tools to help disentangle the complexities of FA homeostasis and its physiological and pathological contributions. Currently available FA indicators require direct modification of the fluorophore backbone through complex synthetic considerations to enable FA detection, often limiting the generalization of designs to other fluorophore classes. To address this challenge, we now present the rational, iterative development of a general reaction-based trigger utilizing 2-aza-Cope reactivity for selective and sensitive detection of FA in living systems. Specifically, we developed a homoallylamine functionality that can undergo a subsequent self-immolative β-elimination, creating a FA-responsive trigger that is capable of masking a phenol on a fluorophore or any other potential chemical scaffold for related imaging and/or therapeutic applications. We demonstrate the utility of this trigger by creating a series of fluorescent probes for FA with excitation and emission wavelengths that span the UV to visible spectral regions through caging of a variety of dye units. In particular, Formaldehyde Probe 573 (FAP573), based on a resorufin scaffold, is the most red-shifted and FA sensitive in this series in terms of signal-to-noise responses and enables identification of alcohol dehydrogenase 5 (ADH5) as an enzyme that regulates FA metabolism in living cells. The results provide a starting point for the broader use of 2-aza-Cope reactivity for probing and manipulating FA biology.

We present the first ratiometric fluorescent probes for formaldehyde, a major reactive species in biology, operating in the visible region.

Aldehyde dehydrogenase class 3, encoded by ADH5 in humans, catalyzes the glutathione dependent detoxification of formaldehyde. Here we show that ADH5 deficient cells turn over formaldehyde using alternative pathways starting from the reaction of formaldehyde with free amino acids. When mammalian cells are exposed to formaldehyde, the levels of the reaction products of formaldehyde with the amino acids cysteine and histidine-timonacic and spinacine-are increased. These reactions take place spontaneously and the formation of timonacic is reversible. The levels of timonacic are higher in the plasma of Adh5 −/− mice relative to controls and they are further increased upon administration of methanol. We conclude that mammals possess pathways of cysteine and histidine dependent formaldehyde metabolism and that timonacic is a formaldehyde reservoir.

In this Article, owing to an error during the production process, the keys to the growth curve graphs in Fig. 3d and 3e were inadvertently missing. The original Article has been corrected.

No abstract