Contrast-enhanced magnetic resonance imaging (MRI) has become an indispensable tool in medical imaging (Rogers et al., 2006; Park et al., 2017). Gadolinium (Gd)-based small molecules and iron oxide nanoparticles have been mainly used as MRI-T₁ and -T₂ contrast agents, respectively, to provide high contrast sensitivity against background signals and have been approved by the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMEA) (Park et al., 2017; Wahsner et al., 2019). Nevertheless, there are still some challenges facing these contrast agents in practical applications, especially for long-term tracking in vivo. The main issue lies in their biosafety and the exocytosis of inorganic substances over a long period (Wahsner et al., 2019). For example, Gd was recently reported to be somewhat toxic to the liver and kidney (Calvin et al., 2010), and iron (Fe) overload has also been proposed to be involved in degenerative diseases (Zecca et al., 2004; Stephenson et al., 2014). Thus, it has inspired a new wave to develop new contrast agents, where control of iron accumulation in cells could work as a means to alter longitudinal or transverse relaxation times, generating contrast, probably derived endogenously from the organism itself.

Recently, protein-based MRI contrast agents have attracted increasing attention, such as transferrin and ferritin (Jutz et al., 2015; Schilling et al., 2017). Transferrin and ferritin are critical proteins that regulate iron metabolism in vivo, which are capable of binding Fe to exhibit paramagnetic properties (Yang et al., 2016). Although the two proteins, especially ferritin, have relatively large magnetic moments and influence the transverse relaxation of the proton, the effect appears to be too weak to be used directly as contrast agents for MRI in vivo (Liu and Theil, 2005; Jutz et al., 2015). Generally, ferritin is used as a natural reactor in vitro to transform hydrated paramagnetic ferric oxide into superparamagnetic iron oxide nanoparticles (Liu and Theil, 2005; Turano et al., 2010). These transformed protein molecules have been proposed as a new type of nanoparticle and partly improve the biosafety of contrast agents; however, they are essentially an exogenous substance.

Xie et al. discovered a novel magnetic protein biocompass in Columba livia, which integrated both magnetoreceptor (MagR) and type IV cryptochrome (clCry4) (Qin et al., 2016). MagR is the homologue of the iron–sulfur cluster assembly protein (A-type ISC protein, IscA). The IscA protein is an iron chaperon that can bind to intracellular iron to form Fe–S clusters for electron transfer (Ding and Clark, 2004; Holm and Lo, 2016). The Cry4 protein is considered an electron donor excited by light that can form long-lived radical pairs to activate downstream pathways (Maeda et al., 2012; Zoltowski et al., 2019; Xu et al., 2021). Because clMagR/clCry4 protein is involved with the process of electron transfer, we hypothesized it would be suitable for MRI contrast as a “live” agent.

We investigated the MRI contrast performance of clMagR/clCry4 gene in a living organism. Unlike the strategy used for ferritin, the clMagR/clCry4 gene was simply transfected into prokaryotic E. coli, rather than administering clMagR/clCry4 protein molecules. The transfected E. coli showed significant T₂ contrast on MRI when cultured in an iron-supplemented medium, while the protein itself was unable to show this effect. The transfection of the clMagR/clCry4 gene led to the formation of iron oxide nanoparticles within E. coli, which could mediate the alteration of the MRI transverse relaxation rate.

Transfection of the clMagR/clCry4 gene into E. coli is shown schematically in Figure 1A. Similar to that of our previous report (Xue et al., 2020), transfection was shown to be successful based on the sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) pattern of the expressed clMagR/clCry4 protein (Figure 1B). Because the d electrons in Fe atom have been demonstrated to contribute to the net spins of the Fe-loaded IscA monomer (Beinert et al., 1997; Xiao et al., 2020), the characteristic signal at g' ≈ 4.3 (g-factor) in the electronic spin resonance (ESR) spectrum proved the presence of the protein–Fe(III) complex (Figure 1C) (Sun and Chasteen, 1994; Wajnberg et al., 2018). Furthermore, the broad peak of the ESR also indicated the heterogeneity of the protein clMagR/clCry4 protein expressed inside E. coli. The bacteria were resuspended in PBS buffer solution for cell suspensions (Supplementary Figure S1). Then, the clMagR/clCry4-transfected E. coli were tested by qualitative MRI-T₁ and -T₂ scans. As shown from the images, there were no significant differences for either mode nor were there any differences in the relaxation rates R₁ and R₂ (Figure 1D). Furthermore, MRI-T₁ and -T₂ mapping modes were used to quantitatively evaluate the imaging contrast of transfected E. coli. The mean gray values acquired from the mapping images were measured using the Python language and two-way ANOVA statistics, followed by the Bonferroni and Tukey’s HSD post-hoc test. Both the images and the statistical values of the mean gray areas also showed minimal differences (Figures 1E, F). In addition, only the clMagR-transfected E. coli also showed a little imaging contrast with the control (Supplementary Figure S2). We tested the MRI contrast properties for the purified clMagR/clCry4 protein. The protein itself showed a little MRI contrast effect (Supplementary Figure S3). Based on these cases, the transfection of the clMagR/clCry4 gene was unable to influence MRI spin relaxation signals. Due to its presence in a living organism, the clMagR/clCry4 protein could perceive a magnetic field; thus, we hypothesized that the magnetism of the clMagR/clCry4 protein would associate with the electron transfer, that is, binding with Fe. Because there was insufficient Fe in the medium for E. coli, the clMagR/clCry4 complex was incapable of producing magnetic behaviors.

Thus, we added ammonium ferric citrate (FAC) to the medium as an exogenous iron donor. FAC is a soluble ferric salt that has been approved by the U.S. FDA as a food supplement and clinical drug (Tenne et al., 2015). Hence, it is safe and has clear effects, as confirmed by scanning electron microscopy (SEM) and live/dead BacLight stain tests. As revealed by the SEM micrographs, E. coli retained its typical rod-shaped morphology without any discernible alteration (Supplementary Figure S4). The fluorescence pattern of SYTO 9 and double-staining with propidium iodide (PI) further confirmed the good viability of bacteria as shown in Supplementary Figure S5, where green indicates living cells and red indicates dead cells. Beyond that, the addition of FAC significantly increased the intracellular iron content within bacteria, and E. coli transfected with clMagR/clCry4 gene showed higher levels than the control, which was experimentally verified by ICP-MS (Figure 2A). However, the status of iron within the bacteria was different. Dissociative Fe will produce abundant hydroxyl free radicals resulting from the Fenton reaction, while Fe binding could greatly reduce this production of free radicals (Dixon and Stockwell, 2014). Thus, we used ESR to identify the hydroxyl free radicals with the spin adduct of 5,5-dimethyl-1-pyrroline N-oxide (DMPO)-hydroxyl radicals. For control E. coli, the quartet signal was unambiguously detected, indicating the presence of abundant hydroxyl free radicals (Figure 2B). However, for E. coli transfected with clMagR/clCry4 gene, the intensity of the corresponding ESR signal decreased greatly, meaning that dissociated Fe was bound (Figure 2C). A similar case occurred for E. coli transfected with clMagR gene, while the intensity decrease was slightly lower than that of E. coli transfected with clMagR/clCry4 gene (Supplementary Figure S6). Thus, we assumed that the dissociative Fe within E. coli was bound to the clMagR/clCry4 protein, causing the alteration of bacterial magnetism. The SQUID measurement confirmed this point. As predicted, control E. coli were diamagnetic, even after exposure to FAC (Figure 2D). A similar diamagnetic curve was also observed in E. coli transfected with clMagR/clCry4 gene in the absence of an exogenous iron supply. However, the bacteria exhibited a paramagnetic curve after culture in the presence of media containing an exogenous iron supply, and the intensity of magnetization increased by two orders of magnitude (Figure 2E). As shown by the magnetization loop, there was even a very slight hysteresis, which seemed somewhat superparamagnetic (magnetic susceptibility, 0.00625 emu/g). Surprisingly, this effect was not observed in clMagR-transfected E. coli, where the bacteria retained the diamagnetic behavior (Supplementary Figure S7). Therefore, these findings supported the potential effects on MRI contrast of transfection with clMagR/clCry4 gene in the presence of exogenous iron supply.

As shown by the qualitative MRI scan, the clMagR/clCry4-transfected E. coli exhibited significant MRI-T₂ contrast after culturing with FAC, while there was little influence on the MRI-T₁ mode (Figure 3A); the histograms represent the corresponding statistical R₁ and R₂ values (Figure 3B). Furthermore, to quantitatively confirm the imaging contrast effect, the mean gray values acquired from the mapping of MRI-T₁ and -T₂ images (Figures 3C, D) were calculated. Independently of the time of TE, it can be clearly seen that clMagR/clCry4 gene transfection imparted the contrast effect of MRI-T₂ to E. coli, which was significantly different from that of the control E. coli strains. Moreover, the influence of the FAC concentration (exogenous iron level) in the medium, and the bacterial density on the imaging contrast effect, was evaluated (Supplementary Figures S8, S9). As described in detail, with an increase in the iron level or bacterial density assayed, clMagR/clCry4 protein achieved a more significant influence on the T₂-contrast of the transfected E. coli. Here, the corresponding original gray values were plotted against the TE for different bacterial densities after pseudocolor processing (Supplementary Figure S10). With an increase in the bacterial density, the MRI gray value of E. coli-transfected clMagR/clCry4 gene tended to go from a constant to a high-order power function versus the TE time course of MRI-T₂ mapping.

To explore the underlying mechanism responsible for MRI-T₂ contrast E. coli transfected with clMagR/clCry4 cultured in the presence of FAC, the purified clMagR/clCry4 protein was tested for contrast properties in the MR imaging. As shown in Supplementary Figure S11, the protein itself was unable to achieve MRI-T₂ contrast signals. Thus, an explanation should lie with event results within living organisms. To validate this hypothesis, we first tested the intracellular pH of E. coli after transfection of clMagR/clCry4 gene using the pH-sensitive organic dye BCECF-AM, which has been widely used in elaborating the physiology of prokaryotes and eukaryotes (Burgess and Han, 2010; Chakraborty et al., 2017). As shown in Supplementary Figure S12, the fluorescence alteration indicated that the microenvironment inside E. coli changed from a weakly alkaline pH to acidic pH with the addition of FAC. However, it was found that transfection of clMagR/clCry4 gene inhibited this change in the pH value to maintain a weakly alkaline state, the detailed mechanism of which remains unclear. Apart from that, similar evidence was obtained by means of the conventional glass electrode pH meter measurement (Supplementary Table S1). On the other hand, the Fe ions will bind to biological molecules within the bacteria. It has been known that E. coli secreted the siderophore enterobactin to chelate Fe³⁺ with high affinity, forming soluble Fe (III)-siderophores to transport inside E. coli (Liu et al., 1993; Faraldo-Gomez and Sansom, 2003). Fe ions could then be released either by siderophore hydrolysis or by reduction of flavin oxidoreductase, after which the dissociative Fe would be recruited by the clMagR/clCry4 protein (Coves et al., 1993; Raymond et al., 2003). However, Fe-binding residues in clMagR are flexible and easily disrupted, so the Fe ions can partially dissociate into a free state resulting from the reactive oxygen species (Rouault and Klausner, 1996; Ding and Clark, 2004). In the presence of Fe ions under weakly alkaline conditions, the formation of iron oxide nanoparticles is assumed, which is somewhat similar to a co-precipitation reaction occurring within E. coli. Hence, it should be noted that the FAD group of the clCry4 protein is widely known to generate free radicals under photonic action, which may explain why clMagR/clCry4 transfection is advantageous in the MRI-T₂ contrast of E. coli compared to clMagR transfection alone.

We used TEM to observe ultrathin bacterial sections. Under low magnification, many electron-dense granules were clearly observed in the E. coli transfected with clMagR/clCry4 in the presence of exogenous iron supply (Figures 4A, B). Fe element mapping further indicated that there were abundant iron-based compounds detectable within E. coli (Figure 4C). Hence, we speculated that the electron-dense granules should contain the iron-based nanomaterials. These electron-dense granules were then extracted from the bacteria for TEM characterization by repeating magnetic separation and washing. Interestingly, a magnet could be used to easily attract the extracted granules (Supplementary Video S1), indicating that the magnetic iron oxide nanoparticles were present. In the TEM micrographs, aggregates of tiny nanoparticles can be clearly observed (Figure 4D). The exact matching of Fe and O elements in the XEDS mapping confirmed that the composition of the nanoparticles was iron oxide.

In addition, STEM micrographs were acquired together with XEDS elemental mappings to give complementary information on the particles being imaged, and the extracted particles demonstrated a small size range (averaged 20.62 ± 3.5 nm) (Supplementary Figure S13). Furthermore, SAED and HRTEM demonstrated the coexistence of the Fe₂O₃ and Fe₃O₄ polycrystal phases in the tiny nanoparticles (Figures 4E, F), which provides evidence supporting the presence of a magnetization loop and the MRI-T₂ contrast effect of clMagR/clCry4 transfected in E. coli. Nonetheless, it should be mentioned that aggregates of tabular-like ferric oxide nanoparticles with poor crystalline features were also detected (Supplementary Figure S14). These amorphous precursors demonstrated, in part, that the formation process of such nanoparticles was somewhat similar to that achieved by a chemical co-precipitation method.

This study was the first to exploit the transfection of the clMagR/clCry4 gene to produce an MRI-T₂ contrast agent effect on a living organism. This innovative phenomenon has neither been observed nor hypothesized previously. Using E. coli as the model organism, an exogenous iron supply was shown to be a critical factor for this phenomenon. Exogenous Fe could be internalized in E. coli and form a dynamic “bind-release” process with the clMagR/clCry4 protein. During the cycle, a biosynthesis process occurred in the presence of dissociative Fe under weakly alkaline intracellular conditions, which is somewhat similar to the chemical co-precipitation reaction, leading to the formation of tiny iron oxide nanoparticles. These tiny iron oxide particles influence the transverse relaxation observed on MRI. Because this strategy depends on the activity of the associated protein complex, it has a wide range of potential applications that are dependent on the presence of an exogenous iron supply. We believe that this novel technique will play a key role in future biomedical imaging and tracking applications. Furthermore, due to the key role of clMagR/clCry4 protein in magnetoreception, our findings are useful for harmonizing the long-term controversy over the existence of magnetoreceptors in organisms, ranging from prokaryotes to animals.