Vitamin B12 is a limiting factor for induced cellular plasticity and tissue repair

Abstract

Transient reprogramming by the expression of OCT4, SOX2, KLF4 and MYC (OSKM) is a therapeutic strategy for tissue regeneration and rejuvenation, but little is known about its metabolic requirements. Here we show that OSKM reprogramming in mice causes a global depletion of vitamin B₁₂ and molecular hallmarks of methionine starvation. Supplementation with vitamin B₁₂ increases the efficiency of reprogramming both in mice and in cultured cells, the latter indicating a cell-intrinsic effect. We show that the epigenetic mark H3K36me3, which prevents illegitimate initiation of transcription outside promoters (cryptic transcription), is sensitive to vitamin B₁₂ levels, providing evidence for a link between B₁₂ levels, H3K36 methylation, transcriptional fidelity and efficient reprogramming. Vitamin B₁₂ supplementation also accelerates tissue repair in a model of ulcerative colitis. We conclude that vitamin B₁₂, through its key role in one-carbon metabolism and epigenetic dynamics, improves the efficiency of in vivo reprogramming and tissue repair.

Main

Cellular reprogramming consists of the loss of differentiated cell identity followed by the acquisition of embryonic stem pluripotency, which can be achieved by the simultaneous expression of the transcription factors OCT4, SOX2, KLF4 and MYC (OSKM; in mice encoded by Pou5f1, Sox2, Klf4 and Myc, respectively)¹. During recent years, it has become evident that this process involves intermediate states in which cells acquire various degrees of plasticity and differentiation potential, which may have broad implications in regenerative medicine and organ repair (reviewed in ref. ²). Continuous expression of OSKM in mice can recapitulate full reprogramming to pluripotency, a process that culminates with the generation of teratomas³. Interestingly, transient expression of OSKM leads to molecular and physiological features of rejuvenation, including an enhanced capacity for tissue regeneration^4,5,6,7,8,9. Nevertheless, in vivo reprogramming via OSKM remains a poorly understood process, with low efficiency and high risks, including teratoma and cancer development^3,10,11. Thus, we sought to unravel new molecular mechanisms of in vivo reprogramming that could be harnessed to manipulate cell plasticity and tissue repair.

Given the unique metabolic requirements of in vitro reprogramming^12,13, we hypothesized that unique metabolic requirements may also operate during in vivo reprogramming. As a new approach, we considered the gut microbiota as a commensal community in metabolic equilibrium with its host. Indeed, the microbiota is sensitive to perturbations in host physiology, capable of adapting and rewiring itself based on nutrient availability and depletion¹⁴, a process known as the host–gut microbiota metabolic interaction¹⁵. We reasoned that analysis and manipulation of the microbiota could provide new insights into the metabolic requirements of in vivo reprogramming.

In vivo reprogramming is dependent on the microbiota

To study modulators of in vivo reprogramming, we used a previously described mouse model in which doxycycline drives systemic, inducible OSKM expression^3,9,16,17. On a short timescale (7 days), OSKM induction causes focal regions of abnormal tissue architecture, correlating with the appearance of rare NANOG-positive cells (a marker of embryonic pluripotency) predominantly in the pancreas, colon and stomach³. We first asked whether the microbiota was important for in vivo reprogramming by disrupting it with a commonly used, broad-spectrum cocktail of antibiotics (ABX): ampicillin, metronidazole, neomycin and vancomycin¹⁸. We administered ABX for 3 weeks before and during the 7 days of OSKM induction (Fig. 1a). We noted that mice treated with ABX had very low levels of serum doxycycline (Extended Data Fig. 1a), therefore precluding the induction of OSKM in organs beyond the gastrointestinal tract (Extended Data Fig. 1b). Nevertheless, doxycycline efficiently induced OSKM in the colon and stomach in the presence of ABX (Extended Data Fig. 1b). Strikingly, despite strong transgene induction, reprogramming was significantly reduced in the colon and stomach of ABX-treated mice (Fig. 1b and Extended Data Fig. 1c). Reduction in reprogramming was also reflected in the reduced abundance of SCA1-positive and KRT14-positive cells (Fig. 1b and Extended Data Fig. 1c), markers of early and advanced stages of intermediate in vivo reprogramming, respectively¹⁹. Consistent with low levels of reprogramming, ABX-treated mice lost significantly less weight than mice with normal levels of reprogramming (Extended Data Fig. 1d). These results indicate that the microbiota is critical for the successful reprogramming of tissues in vivo.

In vivo reprogramming causes microbial dysbiosis

Given the profound impact that disruption of the microbiota had on in vivo reprogramming, we reasoned that a functional analysis of microbial changes during this process could illuminate previously unknown requirements for reprogramming. To this end, we isolated bacterial DNA from paired stool samples of both OSKM-expressing mice and wild-type (WT) littermate control mice before and after 7 days of doxycycline treatment, and performed shotgun metagenome sequencing²⁰ (Extended Data Fig. 2a–c and Supplementary Tables 1 and 2). In both WT and OSKM mice, the microbial diversity as measured by the Shannon index decreased following 7 days of doxycycline treatment, with the most profound loss of diversity occurring in reprogrammed mice (Extended Data Fig. 2a). At a genus level, reprogrammed mice were characterized by a relative expansion of Chlamydia, Bacteriodes and Alistipes spp. and a relative contraction of Muribaculaceae spp. (Extended Data Fig. 2b). Muribaculaceae have been reported to contract during inflammatory colonic injury²¹, which shares features with in vivo reprogramming including inflammation and loss of differentiated cell identity²². Alistipes on the other hand, have been reported to promote colonic interleukin (IL)-6 production²³, which is an important mediator of in vivo reprogramming¹⁶.

In vivo reprogramming reduces systemic vitamin B₁₂ levels

Our whole-genome approach allowed us to investigate changes not only in bacterial species abundance, but also in gene composition and ontology groups, which could uncover pathways relevant to reprogramming. Remarkably, we found that microbial gene modules related to the biosynthesis and metabolism of cobalamin (vitamin B₁₂) dominated the bacterial Gene Ontology (GO) groups altered during reprogramming (Fig. 1c and Supplementary Table 2). Under conditions of disrupted cobalamin bioavailability, competition for vitamins can shift the relative abundance of cobalamin-producing and cobalamin-utilizing bacteria in a process referred to as ‘corrinoid remodelling’^14,24. We found microbial changes consistent with this phenomenon in reprogramming: the few genera of bacteria able to synthesize B₁₂ (~20 genera)²⁵ were generally enriched in OSKM mice after 7 days of doxycycline, with Proteus, Escherichia and Salmonella being most significantly enriched among the B₁₂ synthesizers (Extended Data Fig. 2c and Supplementary Table 2).

The observed changes in the gut microbiota could be indicative of a systemic deficit in B₁₂, affecting not only the microbiota but also the entire physiology of the host. To test this, we examined systemic vitamin B₁₂ levels in the serum during reprogramming, which were significantly reduced in OSKM mice after 7 days of doxycycline administration (Fig. 1d). The liver is one of the organs with the greatest demand for vitamin B₁₂ (ref. ²⁶) and, as such, is sensitive to B₁₂ deficiency²⁷. In rodents, this manifests as depletion of phosphatidylcholines (PCs)²⁸, which are produced in large quantities by the liver in a B₁₂-dependent manner. We saw that PCs were significantly reduced in the serum of reprogrammed mice as compared to WT mice treated with doxycycline (Extended Data Fig. 3a). Importantly, the liver of OSKM mice does not exhibit histological changes after 7 days of doxycycline⁹, making the reduction in PCs unlikely to reflect liver dysfunction as a result of reprogramming. The kidney is another organ that is refractory to reprogramming in our mouse model³; however, we did observe a significant depletion of vitamin B₁₂ from the proximal tubules during reprogramming (Extended Data Fig. 3b). The kidney is the primary site of B₁₂ concentration and storage in rodents, from where it is released for use by other organs upon systemic deficiency^27,29,30,31. Collectively, these results suggest that vitamin B₁₂ becomes systemically depleted during in vivo reprogramming, affecting both the colonic microbiota and the host.

Vitamin B₁₂ supplementation improves in vivo reprogramming

Given the systemic reduction of vitamin B₁₂ during in vivo reprogramming, we hypothesized that B₁₂ supplementation could enhance reprogramming under normal conditions (that is, in the absence of ABX). Indeed, vitamin B₁₂ supplementation significantly improved in vivo reprogramming in the pancreas, colon and stomach, as evaluated by the extent of histological dysplasia and SCA1 or KRT14 levels (Fig. 1e and Extended Data Fig. 3c–e). B₁₂ also increased the number of NANOG⁺ cells, a marker of full pluripotency, in the pancreas (Fig. 1e and Extended Data Fig. 3f). B₁₂ administration did not affect transgene expression (Extended Data Fig. 3g). Even after B₁₂ supplementation, we could not detect histological evidence of reprogramming in the kidney (Extended Data Fig. 3h). However, we did observe a significant increase of vitamin B₁₂ stores within the kidney after supplementation (Extended Data Fig. 3b), indicating that B₁₂ absorption, distribution and storage were occurring normally in the reprogrammed mice.

We also wondered if B₁₂ supplementation could rescue the reprogramming defect of ABX-treated mice. Interestingly, B₁₂ supplementation was able to partially rescue reprogramming in the colon (Extended Data Fig. 3c–e). This supports the concept that an important role of the microbiota during murine reprogramming is to increase the dietary supply of B₁₂ through coprophagy. Another B vitamin that is partly supplied by the microbiota in rodents and humans is vitamin B₉ (folate)³², which is functionally related to B₁₂ (ref. ³³). However, co-supplementation of B₁₂ and B₉ was indistinguishable from B₁₂ alone (Fig. 1e and Extended Data Fig. 3c,f,g). Collectively, these results demonstrate that vitamin B₁₂ is a limiting factor for in vivo reprogramming.

One-carbon metabolism drives vitamin B₁₂ demand during reprogramming

In both humans and mice, vitamin B₁₂ is used as a cofactor by only two enzymes: methionine synthase (MS) and methylmalonyl-CoA mutase (MUT)²⁶. MS uses B₁₂ as a cofactor to regenerate methionine (Met) from homocysteine (Hcy), forming an integral part of one-carbon (1C) metabolism (Fig. 2a). Met is used to synthesize S-adenosylmethionine (SAM), the universal methyl donor for all methylation reactions³³. The nuclear-encoded mitochondrial enzyme MUT uses B₁₂ as a cofactor for the catabolism of branched-chain amino acids via isomerization of methylmalonyl-CoA to succinyl-CoA, for use in the tricarboxylic acid cycle (Extended Data Fig. 4a)²⁶.