Protein Nanostructures Produce Self-Adjusting ... - ACS Publications

Subscriber access provided by University of South Dakota

Article

Protein Nanostructures Produce Self-Adjusting Hyperpolarized Magnetic Resonance Imaging Contrast through Physical Gas Partitioning Martin Kunth, George J. Lu, Christopher Witte, Mikhail G Shapiro, and Leif Schröder ACS Nano, Just Accepted Manuscript • DOI: 10.1021/acsnano.8b04222 • Publication Date (Web): 11 Sep 2018 Downloaded from http://pubs.acs.org on September 11, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Protein Nanostructures Produce Self-Adjusting Hyperpolarized Magnetic Resonance Imaging Contrast through Physical Gas Partitioning Martin Kunth†‡, George J. Lu†, Christopher Witte‡, Mikhail G. Shapiro†*, Leif Schröder‡* †California Institute of Technology, Division of Chemistry and Chemical Engineering, Pasadena, California 91125, USA ‡Leibniz-Forschungsinstitut für Molekulare Pharmarkologie (FMP), 13125 Berlin, Germany *Address correspondence to [email protected] or [email protected].

ACS Paragon Plus Environment

1

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 33

ABSTRACT: Signal amplification strategies are critical for overcoming the intrinsically poor sensitivity of nuclear magnetic resonance (NMR) reporters in non-invasive molecular detection. A mechanism widely used for signal enhancement is chemical exchange saturation transfer (CEST) of nuclei between a dilute sensing pool and an abundant detection pool. However, the dependence of CEST amplification on the relative size of these spin pools confounds quantitative molecular detection, with a larger detection pool typically making saturation transfer less efficient. Here we show that a recently discovered class of genetically encoded nanoscale reporters for 129Xe magnetic resonance overcomes this fundamental limitation through an elastic binding capacity for NMR-active nuclei. This approach pairs high signal amplification from hyperpolarized spins with ideal, self-adjusting saturation transfer behavior as the overall spin ensemble changes in size. These reporters are based on gas vesicles — microbe-derived, gasfilled protein nanostructures. We show that the xenon fraction that partitions into gas vesicles follows the ideal gas law, allowing the signal transfer under hyperpolarized xenon chemical exchange saturation transfer (Hyper-CEST) imaging to scale linearly with the total xenon ensemble. This conceptually distinct elastic response allows the production of quantitative signal contrast that is robust to variability in the concentration of xenon, enabling virtually unlimited improvement in absolute contrast with increased xenon delivery, and establishing a unique principle of operation for contrast agent development in emerging biochemical and in vivo applications of hyperpolarized NMR and MRI.

KEYWORDS: nanocarriers, genetic engineering, xenon, hyperpolarization, chemical exchange saturation transfer, magnetic resonance.


2

Page 3 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Magnetic resonance imaging (MRI) is a valuable tomographic technique for visualizing the chemical composition of opaque specimens, but faces significant sensitivity limitations in detecting low-concentration analytes due to the small macroscopic magnetization of nuclear spins at room temperature. Solving this sensitivity issue is the focus of many approaches, with the common aspect that a spin pool of detection nuclei is read out after being in contact with a more dilute molecular reporter pool, which serves as a label for some aspect of the biological or physico-chemical environment. The total nuclear signal and the efficiency of nuclear interaction with the reporter agent determine the molecular detection limit. For example, reporters acting on aqueous 1H through chemical exchange saturation transfer (CEST1,2) can be detected at µM concentrations. Typical CEST agents have a fixed number of spin binding sites which are populated according to their binding constant. Spins at these sites can be saturated with RF irradiation due to their distinct chemical shift and are released into the bulk pool, resulting in a loss in the bulk signal. Though this provides improved sensitivity compared to conventional direct detection, 1H CEST is fairly inefficient since the water detection pool is rather large to quickly pick up significant changes from the dilute reporter pool. Moreover, the dynamic range of induced signal loss is limited by a steady state level due to spin-lattice relaxation that is counter-acting the spin saturation.3 A key question is how to increase the number of spins per reporter unit, particularly without causing an over proportional increase of the free spin pool. A conventional CEST agent based on chemical affinity loses efficiency because the binding constant causes the pool of free spins to grow stronger than the pool of bound spins when aiming to increase the host occupancy. A concept different from chemical binding is simple physical


3

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 33

partitioning into a bulk (detection) pool and an ensemble of nano-compartments that is CESTresponsive. Such compartments should simply fill up according to a given partitioning coefficient. This simple approach would allow these nanostructures to behave as ideal reporters for exchange-coupled spin pools and circumvent a fundamental limitation of typical saturation transfer reporters that are based on chemical affinity. The ideal scenario comprises a detection spin pool with sufficient signal for encoding spectral or spatial information, but comprising a relatively small concentration of nuclei to enable efficient saturation transfer from the accompanying dilute reporter. As the latter needs to be in the concentration range of nM – pM to sense a wide range of biochemical analytes, a detection pool in the µM range would be more suitable than the ~110 M of aqueous 1H. With thermal, Boltzmann distributed polarization being insufficient to provide enough NMR signal at µM concentrations, this is best achieved with hyperpolarized nuclei. For example, 129

Xe can be prepared in a non-equilibrium state of 104-fold enhanced spin polarization,4 and

introduced into the sample by bubbling, injection or inhalation.5 Xe inclusion complexes can then reversibly bind the hyperpolarized Xe.6,7 The CEST technique is the detection method of choice for dissolved hyperpolarized Xe for the following two reasons: 1) As a noble gas, Xe is not part of any biologically relevant molecule. Thus, to bring the hyperpolarized nuclei in contact with any target of interest, they need to be captured in a molecular host with xenon affinity (the latter preferably being coupled to a binding moiety for a certain analyte).8 2) This host guest encapsulation is a reversible process with large changes in chemical shift for Xe and is therefore ideal to be combined with CEST as each host can act on hundreds to thousands of Xe atoms during the RF saturation period. Whereas direct detection requires high host concentrations, selective RF saturation is very efficient in causing a loss in the abundant free Xe signal. The


4

Page 5 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

combination of hyperpolarized xenon and detection with CEST amplification (Hyper-CEST9) allows such host structures to be detected at concentrations at the pM to nM level. The typical Hyper-CEST agent has a hydrophobic xenon binding site in a cage-like molecule that can be functionalized to bind molecular targets. Prominent examples for imaging applications are based on cryptophane-A (CrA)10–12 or cucurbit[6,7]uril (CB6,7).13–15 (Hyper-)CEST efficiency scales with the concentration of bound nuclei, [B] (Figure 1). To increase this number for a fixed concentration of hosts, one must increase the host occupancy, β. This parameter is typically a function of the binding constant, KB, and the concentration of free nuclei in pool [A]: β = x/(x + 1) with x = KB[A].16 The following quantitative consideration illustrates the dilemma with chemical affinity of the host-guest complex. Commonly observed Xe binding constants are in the range of 102 – 104 M-1 and typical Xe concentrations in aqueous solutions are 10-4 M.13 Thus, the achieved occupancy is usually β ≤ 50%. Due to the very nature of chemical binding with 1:1 stoichiometry, an increase to β ~ 90 % requires [A] to increase ca. 10-fold (see Supporting Information Figure S1). Both pools [A] and [B] grow disproportionately, resulting in a fraction of bound nuclei, fB, that changes to an unfavorable low value (ca. 5-fold reduction) which makes the CEST performance as a relative signal loss worse (see Figure 1A). On the other hand, reducing the free nuclei concentration [A] to improve fB is usually detrimental as this impedes the overall detectability of unbound atoms to encode any CEST information. The starting magnetization provided by spin pool A is of particular importance for hyperpolarized nuclei because it does not self-renew and must provide sufficient transverse magnetization for efficient recycling in echo trains to minimize the number of nucleus deliveries and improve the signal-to-noise ratio (SNR).17 In this regard, increasing the overall spin concentration to make the echo trains last longer above the noise level is a method of


5

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 33

choice. Overall, CEST agents based on reversible chemical binding (both for 1H and other nuclei) come with an intrinsic dilemma: the best signal transfer efficiency is achieved as the magnetization of the detection pool approaches the noise level. An ideal CEST agent would operate under a different mechanism, wherein it would not lose efficiency when [A] is increased to gain SNR, but rather provide self-adjusting saturation transfer into pool A (see Figure 1B) by proportionally expanding β. Here we present a class of host structures that enable such a mechanism. The solution is a system of exchange-coupled spin pools where both pools grow proportionately to keep their size ratio, fB, and, thus, the CEST efficiency constant. This approach is conceptually different from 1:1 Xe complex formation, protein-based sensors18,19 or reversible chemical binding in 1H CEST. We demonstrate ideal CEST performance through simple physical partitioning of a dissolved gas into a nano-sized volume (Figure 1B), thereby introducing the concept of an “elastic” contrast agent for HyperCEST. Our proof of concept is based on genetically-encoded gas-filled protein-shelled nanostructures, or gas vesicle (GV) nanostructures, evolved in prokaryotes as a means to achieve buoyancy. These GV nanostructures, which were previously shown to act as Hyper-CEST contrast agents,20 can be extended to many biological and diagnostic applications through surface functionalization,21,22 and genetic encoding.23,24 Due to their volume, GVs are expected to bind thousands of gas atoms/molecules, and they can provide a very efficient flux of depolarized nuclei into the detection spin pool. This is a critical aspect for the emerging applications because in in vitro experiments, Xe can be detected immediately after dissolution, but for biological applications it is already undergoing relaxation at the time it will encounter the host structure. Together with the reduced life time of the hyperpolarization, this concept requires the ability to


6

Page 7 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

generate a rapid build-up of the CEST response. These critical quantitative aspects of GVs’ interactions with xenon have not been investigated.

RESULTS AND DISCUSSION Gas exchange behavior of “elastic” Xe hosts. To test our hypothesis about “elastic” CEST reporters, we chose bacterial GVs as our xenon hosts. GVs comprise 2 nm-thick protein shells enclosing hollow, gas-filled compartments with dimensions on the order of 200 nm, varying in their exact size and shape based on which bacteria they come from.25 GVs have previously been developed as contrast agents for ultrasound22–24,26 and susceptibility-based 1HNMR/MRI.24 They can be engineered at the genetic level to alter their surface and targeting properties,21 and can be expressed heterologously as genetically encoded reporters.23,24 In addition, it was previously shown that several varieties of GVs could produce Hyper-CEST contrast at pM concentrations.20 However, the exchange properties underlying their spin transfer functionality have not been studied. Here we hypothesized that the gas-filled interior of GVs would allow them to accommodate increasing numbers of 129Xe atoms as the concentration of the noble gas increases in the surrounding solvent, allowing them to produce a steady level of Hyper-CEST contrast. We reasoned that the equilibrium between dissolved gas from the surrounding media and the hollow space inside these nanostructures should follow the Ostwald coefficient of the gas/solvent combination. This would allow these contrast agents to produce a constant intensity of the CEST effect independent of the total number of participating nuclei (Figure 1B). Transmission electron microscopy (TEM) images of GVs derived from Anabaena flosaquae (Ana), Halobacterium salinarum (Halo) and Bacillus megaterium (Mega) (Figure 2)


7

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 33

showed that these nanostructures have volumes in the range of 0.65 to 6.25 attoliter (aL; Table 1), and could thus be expected to accommodate ~104-105 atoms of an ideal gas under standard conditions, depending on the species. To characterize the interactions of each type of GV with xenon, we performed NMR on these nanostructures dissolved in phosphate buffered saline at 9.4 T in the presence of hyperpolarized

129

Xe, and analyzed the results using a quantitative Hyper-CEST (qHyper-

CEST16) MRI framework (Figure 3). In this approach, a series of MR images is acquired where the frequency of the RF saturation pulse prior to the image encoding is incrementally swept along a large frequency range where both the response of free Xe (assigned to 0 ppm offset) and the response from encapsulated Xe (between -150…-180 ppm offset) is expected to be. While direct saturation around 0 ppm always results in complete cancelation of the detected signal of dissolved Xe, it represents the unspecific response that is not related to the host structure which will be eventually used as a targeted reporter. However, z-spectra with both signals need to be acquired since only the combined information from both exchanging pools allows a complete quantification of the exchange kinetics and involved spin pool sizes. Otherwise, changes that occur, e.g., in the fitted transverse relaxation time of free xenon can compensate for the quantity of the exchange rates.16 Although our sample comprised only one homogeneous compartment, we decided to employ an MR imaging protocol (instead of simple saturation – FID or saturation – Carr-PurcellMeiboom-Gill acquisition) to have more realistic conditions for future applications where the spectra should also be derived from spatially resolved data. In fact, one important aspect was to find out if exchange-related T2 relaxation would be prohibitive of maintaining enough signal for the phase encoding steps.


8

Page 9 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

The qHyper-CEST approach allowed us to individually obtain the bound xenon fraction, the exchange rate and the relative chemical shift of contrast agent-bound xenon with respect to free xenon in solution for each type of GV (see Table 2). By relating the bound xenon fraction to the known nanostructure concentration, we found that at a total pressure of 1.2 atm and 5% xenon in the gas mixture, ca. 5,800 and ca. 4,760 xenon atoms were hosted by each Halo and Ana GV, respectively, and Mega hosted ca. 890. This result is intuitively validated by comparing the individual sizes of gas vesicles (Figure 2), and reflects the direct link between the bound magnetization pool size and the physical size of the nanoparticle. Combined with the fast rate of exchange, ca. 20,000 s-1, this provides an explanation for the extraordinary molar efficiency of GVs as a CEST agent. However, this fast exchange also manifests directly in the broad CEST responses and caused rather short apparent spin-spin relaxation times T2,app (ca. 2.2 ms; see Supporting Information Figure S2) as reflected in fast decaying echo trains in the RARE acquisition protocol. CEST signal stability at variable Xe concentration. Next, we tested Mega GVs using qHyper-CEST MRI at different xenon gas total pressures (Figure 4A; Supporting Information S3) to observe the hypothesized elastic gas filling properties. Mega was chosen because these are the most stable GVs for extended data acquisition times and high pressures (stability test for Ana and Mega GVs given in the Supporting Information S4). To achieve an unambiguous quantification of all exchange-related parameters of the system, including the pool size of GVbound Xe (see next paragraph in more detail), the spin system was investigated for each pressure with three different RF saturation conditions.16 This allowed us to verify that the magnetization pool inside the GVs scales with the applied atmosphere conditions in the headspace above the solution, and thus the amount of detected, freely dissolved xenon. Whereas other methods of


9

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 33

increasing the dissolved xenon concentration are available, such as changing the composition of the gas mixture, changing the pressure was most convenient in our experimental apparatus. The qualitative result obtained from this set of spectra yields a clear picture: Comparing z-spectra with identical RF saturation conditions, e.g., 15 µT for 10 s (Figure 4A; red squares) for different xenon gas total pressures revealed excellent stability of the CEST amplitude throughout the data set. We therefore conclude that the pool of GV-bound Xe increases simultaneously with the pool of free dissolved Xe. Pressure-dependent spin pool population following the ideal gas law. Next, we quantitatively examined the interactions of Mega GVs with

129

Xe to demonstrate that the gas

binding capacity of these dissolved containers occurs according to the ideal gas law. The available Hyper-CEST data was used to perform absolute quantitative analysis of the spin pool size. To this end, each diagram with three plots in Fig. 4A was used to derive one quantification of the number of GV-bound Xe atoms. Plotting the on-average number of xenon atoms per Mega-derived GV, obtained from qHyper-CEST fitting, with respect to the total gas pressure (Figure 4B) revealed a linear relationship with a slope of (890 ± 20) bar-1. The gas volume per Mega GV nanostructure was ca. 0.65 aL, such that each nanostructure should be able to host ca. 17,410 atoms of an ideal gas at standard conditions. Thus, 890 xenon atoms per bar corresponded to a xenon filling factor of ~5 % per nanostructure, which matched exactly the 5 % xenon fraction of the gas mix used to bubble xenon into the liquid (see Supporting Information S5; similar data for Ana GVs in Supporting Information S6). Thus, partitioning into the nano-sized volume directly reflected the ideal gas law and conditions in the macroscopic head space therefore remotely control the magnetization pool in the nanosized GV via the liquid as the transfer medium. Although such partitioning can be somewhat expected to follow the ideal gas


10

Page 11 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

law, it is not a self-evident result as the protein structure might include (un)specific binding sites for hydrophobic guests like Xe that could cause a deviation from a linear behavior with respect to the binding capacity. Our proposed mechanism explaining this behavior is that in thermodynamic equilibrium, the partitioning is driven by the Ostwald solubility coefficient (from the dissolved gas into the gas vesicle and vice versa), the partial pressure and the temperature. If the free xenon concentration is tripled, then the number of xenon atoms inside the GVs is also tripled, thus keeping the pool size ratio constant (Figure 1B). This partitioning makes the normalized signal loss due to saturation transfer provided by GVs invariant to the free xenon concentration, manifesting as a stable contrast in the z-spectrum throughout different total gas pressures (such as each group of plots in Figure 4A for one particular RF saturation condition with the same color). We expect this concept of elastic binding also to be applicable to other classes of contrast agents, including nano-droplets and bacterial spores.27–29 As a consequence of the behavior represented by the data in Fig. 4B, the self-adjusting spin pool inside the GVs provides an elastic amount of saturation transfer that scales directly with the bulk pool. Hence, the overall observable CEST effect as a normalized loss in signal remains constant. This can be seen by comparing all plots for one particular RF saturation condition (same color) in Figure 4A.

Signal improvements with increased starting magnetization. Host structures with (relative) fast Xe exchange provide high CEST efficiency (see Supporting Information S7).13 However, the high turnover in combination with the large chemical shift separations (~10 kHz at 9.4 T) can cause a significant exchange-mediated spin-spin relaxation effect that manifests as a


11

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 33

shortened apparent relaxation time, T2,app.30 As mentioned above, this is of particular importance for hyperpolarized nuclei because they are preferably detected with echo trains to improve SNR from each batch of transverse magnetization17 or to encode as much spatial information as possible. Exchange effects thus cause an unwanted signal loss which can be compensated for by increasing the overall Xe concentration to make the bulk pool echo trains last longer above the noise level. The standard deviation of the signal (plotted as error bars derived as described in Supporting Information S8) of image-derived z-spectra were indeed large at low gas pressures (as seen in Supporting Information S9) and confirmed that the fast exchange which we saw in the broad responses in the z-spectra also makes the echoes quickly disappear into the noise level. This was also reflected by the noise level in the reconstructed images. As the pressure was raised to 4.5 bar, the noise level became less significant and thus the standard deviation in the (normalized) z-spectra decreased continuously. This illustrates the advantage of increasing the size of the overall spin ensemble. The stable CEST contrast is therefore paired with an increasing signal quality and our data demonstrates the improved behavior of the exchange-coupled spin pools. In other words, the feature of the partitioning into the GVs enables a virtually unlimited scaling of the magnetization in the CEST pool and yields a self-compensating elasticity of the saturation transfer, i.e., a stable absolute contrast with variable xenon delivery.

Comparison with a fixed-stoichiometry Hyper-CEST agent. The “ideal” response of the GVs was compared with that of xenon binding to the commonly used synthetic contrast agent CrA as a function of the xenon gas total pressure. This system follows the principle of chemical


12

Page 13 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

affinity determined by a binding constant KB and it can also be characterized through qHyperCEST analysis at different gas pressures. The fitted z-spectra with different combinations of RF cw saturation power and time (see Supporting Information S3) allowed us to derive the host occupancy β when the total cage concentration was known. The host occupancy, β, was approximately 30% for 10 µM CrA at 1.2 bar Xe (Table 2). We observed that β changed only moderately in the pressure regime under investigation: a three-fold increase of the gas pressure from 1.5 bar to 4.5 bar yielded a change in the CrA occupancy from 40% to only 55% (Table 3). Related to this moderate increase in β, we observed from the set of z-spectra that the relative CEST amplitude decreased with increasing gas pressure (Figure 5A). As expected, the pool of free Xe grew faster than that of bound Xe and fB became unfavorably small. In total, the CEST effect for CrA dropped from 75% to 25% signal depletion. The comparison of both host systems clearly indicated that the increase in starting magnetization (provided by the gradual pressure increase from 1.2 to 4.5 bar) yielded a constant CEST contrast for Mega GVs but a decrease of ca. 50% for contrast generated by CrA (Figure 5C). This clearly demonstrates that a CEST agent based on chemical binding becomes less efficient as the bulk pool is increased. The lack of elasticity in the CEST pool for self-adjusting saturation transfer into the detection pool is a clear disadvantage: While the signal-to-noise ratio is better with a larger spin pool, the induced contrast falls off. The GVs thus show a superior behavior as a CEST agent when optimizing signal quality This loss of CEST performance for CrA is explained by the xenon-CrA interaction being a 1:1 complex formation. In chemical equilibrium, the xenon exchange to CrA is driven by the following two mechanisms:30,15,31 first, when xenon meets a xenon-free CrA molecule, then pseudo-first order exchange with different forward kf and backward kb exchange rates occurs: Xe


13

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 33

+ CrA ⇌ Xe@CrA; second, when xenon meets a CrA molecule that is occupied by one xenon atom, then the unbound xenon atom can kick out the bound xenon atom and replaces it instantly with identical on and off rates κ (known as “kick-out model”): Xe*aq + Xe@CrA ⇌ Xeaq + Xe*CrA.16,31,32 With this kick-out exchange becoming dominant at high xenon concentration, the contrast evolution is expected to level off – in our case at a level of about 60 % at 4.5 bar total pressure compared to the starting value (Figure 5C). In general, the narrow CEST responses of CrA (Figure 5A) reflected slow Xe exchange. The T2-related noise level for the CrA data was dominated by a constant, intrinsic noise of the acquisition protocol (Figure S9). One might be tempted to argue that the narrower spectral CEST response of CrA is an advantage over that produced by GVs. The partial overlap of the GV CEST responses with free xenon (at 0 ppm) in Figs. 3 and 4 is qualitatively different from the spectral separation observed for CrA (Figure 5A). The spectral signature around 0 … -150 ppm is, however, uncritical because the offresonant control measurement for a CEST image is done with a symmetric offset at the opposite frequency relative to detection spin pool of free Xe at 0 ppm (i.e., on the mirrored frequency at around ca. +155 … + 180 ppm offset, depending on the type of GV). Since the spectral response clearly recovers to baseline level within 100 ppm downfield from direct saturation at 0 ppm, even additional line broadening should not be considered a limitation. The Xe exchange to GVs is still in a regime where it broadens CEST responses in z-spectra, but not significantly unless excessive RF saturation powers are applied. Overall, the faster Xe exchange provided by all type of GVs is a clear advantage compared to CrA.

Absolute GV spin turnover quantification. In addition to the number of enclosed xenon atoms, another important parameter for Hyper-CEST contrast efficiency is the exchange


14

Page 15 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

rate for xenon between the bound compartment and the bulk pool. Our qHyper-CEST fits provided this parameter for all three GV types (Table 2). The product of the number of xenon atoms hosted by each contrast agent with the exchange rate provides a fundamental measure of contrast amplification.13 By this accounting, GVs derived from Ana are the most efficient contrast agents on a per particle consideration with a turnover of ca. 92 × 106 s-1, closely followed by Halo with ca. 89 × 106 s-1, followed by Mega with ca. 20 × 106 s-1 at 1.2 atm and 5 % xenon (Table 2). This ordering is expected from the respective GV volumes. Notably, all the xenon gas turnover values are at least ca. 4

×

106-fold faster than those for CrA (Table 2), with this

difference becoming even more pronounced at higher xenon concentrations.

Despite the

different surface-to-volume ratios (Table 1) and different individual gas turnover values of the three GV types, their gas exchange rate was similar, suggesting that the CEST-derived gas exchange rate (i.e., the xenon off rate into the bulk pool) is determined by the protein shells, which have high sequence homology between species, rather than geometric factors influencing spin diffusion. The much higher turnover rate of GVs, and other fast-exchanging xenon hosts such as cucurbituril,13,33 compared to CrA, suggests a proportionally lower detection limit for these agents. However, other limitations come into play in approaching such dilute detection limits. First, taking full advantage of the high exchange rate requires saturation pulse amplitudes that may be above the tolerable limits for biological samples and MRI hardware, and may also cause spectral broadening leading to a loss of contrast specificity. For example, for the measured Ana GV exchange rate of 19,300 s-1, the ideal saturating field would be ca. 1-2 mT (Supplementary Fig. S6). Second, the exchange of xenon in and out of the contrast agent may be limited by diffusion. This is not the case for GVs under the conditions used in our study. Using the Ana GV


15

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 33

exchange rate of 19,300 s-1, which represents the rate of xenon exit from the contrast agent,16 and a bound xenon fraction of 6.9 x 10-4, results in a rate of xenon entry from the solution into the GV compartment of 13.3 s-1 under steady-state conditions. This timescale corresponds to a xenon diffusion distance in water of approximately 29.3 µm, or much larger than the mean expected inter-GV distance in solution of 3.5 µm at our experimental concentration. The impact of diffusion must be considered for each contrast agent at each concentration being utilized. The significantly higher gas turnover values compared to CrA are an important improvement for future biological applications. They demonstrate that GVs are prime candidates for a rapid CEST build-up to capture image contrast with the available limited magnetization. Previous simulations20 have shown that these structures should be capable of inducing a measurable image contrast. Those predictions were even based on insights from a setup with less efficient CEST performance than the one used here with high polarizer signal stability.34 Taking into account recent promising studies regarding

129

Xe MRI of the brain35 or kidney,36 it can be

assumed that the in vivo concentration of Xe is sufficient to yield measurable contrast with high gas turnover hosts like these GVs.

CONCLUSION In summary, our results revealed that physical partitioning of dissolved xenon into genetically-encodable GV nanostructures follows the ideal gas law, with their scalable xenon capacity allowing these nanostructures to produce a constant CEST effect (as a normalized signal loss) with varying xenon concentration. This is in contrast with the fixed 1:1 stoichiometry of discrete xenon binders such as CrA and other synthetic and protein-based compounds. Though a


16

Page 17 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

partitioning concept also applies to 1H liposomal CEST agents or labeled red blood cells (called LipoCEST/cellCEST,37–39 performed in water with fixed ~110 M proton concentration), the key concept in this work is that the concentrations of NMR-active spins in both exchange-connected pools scale simultaneously because Henry’s law applies at both liquid interfaces with the gas head space and with the volume of the GVs. This conceptually different approach combines significant advantages over CEST agents based on chemical affinity. First, in a classic CEST agent, the CEST pool and the detection pool do not scale proportionally with increased total nuclei. In this approach, however, loading of the CEST pool through partitioning populates both pools in a fixed proportion governed by Ostwald solubility. Second, these GVs are high gas turnover hosts, a generally favorable aspect. Unlike other high Xe gas turnover complexes such as Xe@CB6/7, the CEST performance of the GV nanostructures does not shrink when the pool of free Xe is increased. Consequently, their positive aspects can be combined with the preferred NMR encoding based on long lasting echo trains of the bulk pool at sufficient concentration. This superior performance of physical partitioning over chemical affinity of the inclusion complexes should enable GV-based contrast agents to take advantage of future improvements in xenon polarization and delivery into biochemical and in vivo specimens by providing a corresponding improvement in contrast-to-noise ratio. This suggests GVs as ideal contrast agents for the noble gas xenon and its detection with Hyper-CEST MRI. We expect the linear scaling observed for the partitioning between free bulk xenon and GV-bound xenon to apply to other contrast agents that rely on partitioning into different physical-chemical compartments, including nanodroplets and bacterial spores. In addition, we anticipate that similar principles would apply to contrast agents acting on other hyperpolarized nuclei, such as 13C.


17

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 33

MATERIALS AND METHODS Sample preparation. The gas vesicle nanostructures were prepared according to previous publications.22,24 Briefly, Ana and Halo GVs were expressed in their native host bacteria: Anabaena flos-aquae (CCAP strain 1403/13F) and Halobacteria NRC-1 (Carolina Biological Supply). The details of culturing condition were described previously.22 Mega GVs were heterologously expressed in E. coli RosettaTM2(DE3)pLysS (EMD Millipore). Mega GV gene cluster40 was cloned into the pST39 plasmid for expression under the control of the T7 promoter, and the transformed cells were grown at 30 °C in LB media supplemented with 0.2% glucose. 20 µM isopropyl β-D-1- thiogalactopyranoside (IPTG) was added at OD600 between 0.4 and 0.6 to induce the expression of GVs for overnight. After harvesting, the culture was transferred to sterile separating funnels, and the buoyant cells were allowed for 48 h to float to the top. The subnatant was discarded and the floating cells were collected. A. flos-aquae and E. coli cells were lysed with 500 mM sorbitol and 10% Solulyse solution (Genlantis), whereas Halobacteria cells were hypotonically lysed with the addition of excess low-salt TMC buffer (10 mM Tris, pH = 7.5, 2.5 mM MgCl2, 2 mM CaCl2). Additionally, 250 µg ml-1 lysozyme and 10 µg ml-1 DNaseI was added to E. coli cells. GVs were then separated from cell debris by repeated centrifugally assisted floatation followed by resuspension in 1 × PBS (Teknova). The centrifugation speed was carefully controlled to avoid the hydrostatic collapse of GVs. Specifically for Mega GVs, the natively clustered GVs required the treatment of 6M urea, 2 additional rounds of centrifugally assisted floatation and overnight dialysis in PBS to achieve unclustered form. The unclustered Mega GVs were used for all subsequent experiments. Purified GVs were stored at 4 °C until the day of the experiment.


18

Page 19 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Each sample was diluted in Dubecco’s phosphate-buffered saline (dPBS) separately and adjusted to an optical density at 500 nm (OD500; using an ultraviolet-visible spectrophotometry UV/Vis; JASCO V-550) corresponding to the following concentrations (with the following conversion: Ana: 114.35 pM/OD500; Halo: 47.35 pM/OD500; Mega: 2,030 pM/OD500): Mega in Figure 3A and Figure 4: OD500 = (0.09 ± 0.01), corresponding to (183 ± 20) pM; Ana in Figure 3B: OD500 = (0.33 ± 0.02), corresponding to (37.7 ± 2.3) pM; Halo in Figure 3C: OD500 = (0.33 ± 0.02), corresponding to (15.6 ± 0.1) pM. The 1.03 mg of Cryptophane-A monoacid (obtained from Kang Zhao, Tianjin University, China) were ultrasonically dissolved in 109.79 mL of dPBS to obtain a 5 µM solution. To ensure full solvation, the mixture was ultrasonically heated to 40 °C for 2 hours. For transmission electron microscope (TEM), GVs were diluted to OD500 ~ 0.2 and spotted on Formvar/Carbon 200 mesh grids (Ted Pella) that were rendered hydrophilic by glow discharging (Emitek K100X). The samples were then negatively stained using 2 % uranyl acetate. Images were acquired using a Tecnai T12 LaB6 120 kV TEM. Hyperpolarization, Xenon Delivery and Xenon Concentration. A xenon gas mixture of {5, 10, 85}-vol. % of {Xe, N2, He} (with 26.4 % of natural abundant

129

Xe) was hyperpolarized by

spin exchange optical pumping under continuous flow at a flow rate of 350 mL/minute using a custom-designed polarizer.34,41 The xenon polarization was about 20 %. The hyperpolarized xenon gas mixture was then guided to the NMR spectrometer and gently dispersed into the sample at a flow rate of 20 mL/min through 5 silica-fused glass capillaries (outer diameter 350 nm) for 7 seconds. Another waiting period of 1.5 seconds was added to allow remaining bubbles to collapse. The xenon gas total pressures used were {1.2, 1.5, 2.5, 3.5, 4.5} bar. The achieved xenon concentration in solution was estimated using Henry’s law and the Ostwald solubility


19

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 33

coefficient (assumed to be similar to water of 0.11 L/bar; which is equivalent to 4.9 mM/bar, if 1 mole of an ideal gas occupies 22.41 L). The xenon concentration was then calculated by Henry’s law to: [Xe] = 0.05 × total pressure in bar × 4.9 mM/bar = {294, 367.5, 612.5, 857.5, 1102.5} µM, using the total pressure values as given above. MRI Setup. The NMR and MRI experiments were done on a Bruker Biospin 400 MHz wide bore magnet, which is additionally equipped with a gradient coil system for MRI. A 10 mm dualchannel probe (1H/129Xe) was used and tuned to 400.150 MHz and 110.7066 MHz, respectively. Prior to each session, the sample was shimmed (on 1H) and flip angle calibrated (for both 1H and 129

Xe). A single-shot rapid pulse sequence was set up to saturation transfer using the

magnetization transfer module from Bruker’s ParaVision: version 5 for gas vesicle data, version 6 for cryptophane-A data. The parameters for GV data were the following: RARE sequence with effective echo time: 20 ms, 90° hermite excitation pulse (length: 3.75 ms; bandwidth: 1,600 Hz), 180° Mao refocusing pulse (length: 3.105 ms; bandwidth: 2,000 Hz), field of view: 20 × 20 mm2, matrix size: 32 × 32, in plane resolution: (625 µm)2, slice thickness: 20 mm, centric encoding, RARE factor: 32 (single-shot). For preparing the saturation offset calibration, a simple FID spectrum with 4 acquisitions was taken prior to each imaging series to determine the on-resonant frequency of dissolved Xe (near 110.7065 MHz at 9.4 T). This was used as the transmitter frequency for imaging and as the reference “0 ppm” for all saturation offsets (i.e., 0 ppm was set to the frequency of the detection spin pool of free Xe; similar to 1H CEST). The RF saturation parameters (pulse duration and amplitude) are given in the respective figures. The z-spectra contained in total 41 saturation offsets covering the frequency range of -290 ppm ...,+160 ppm for Mega GVs and 30 data points covering

-235 … +100 ppm for the other two GV types.


20

Page 21 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Another 12 offsets at +3,000,000 Hz with 0 s saturation were used for estimating the T1 relaxation in the presence of different GV types that occurs during the otherwise used 10 s saturation pulse. The parameters for CrA data were the following: RARE sequence with effective echo time: 20 ms, 90° calculated excitation pulse (length: 2.1 ms; bandwidth: 2000 Hz, sharpness: 3), 180° calculated refocusing pulse (length: 1.7 ms; bandwidth: 2000 Hz, sharpness: 3), field of view: 20 × 20 mm2, matrix size: 32 × 32, in plane resolution: (625 µm)2, slice thickness: 20 mm, centric encoding, RARE factor: 32 (single-shot). For preparing the saturation offset calibration, a simple FID spectrum with 4 acquisitions was taken prior to each imaging series to determine the onresonant frequency of dissolved Xe (near 110.7065 MHz at 9.4 T). This was used as the transmitter frequency for imaging and as the reference “0 ppm” for all saturation offsets (i.e., 0 ppm was set to the frequency of the detection spin pool of free Xe; similar to 1H CEST). The RF saturation parameters (pulse duration and amplitude) are given in the respective figures. The zspectra contained in total 63 saturation offsets: sparse sampling with 21 offsets covering the wide frequency range of -19930,...,4430 Hz for the baseline of the z-spectra; dense spectral sampling with 17 offsets covering the narrow frequency range of -14170,...,-15277 Hz for the CrA response; medium dense sampling with 13 offsets covering -1000,...,1000 Hz for the direct saturation; 12 offsets at +3,000,000 Hz with 0 s saturation were used for estimating the T1 relaxation that occurs during the otherwise used 10 s saturation pulse. The workflow obtaining zspectra with hyperpolarized xenon is illustrated in Figure S7. All measurements were done at room temperature (T = 295 K); controlled by a variable temperature unit from Bruker. B1-field inhomogeneities, which potentially could affect the exchange parameter quantification, were not significant for our micro imaging system.16


21

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 33

Data Processing. All z-spectra were derived automatically from two region-of-interests within the signal and noise areas of the MR images (Figure S7). Z-Spectra fitting was done in Matlab 7 (The Mathworks, Natick, MA, USA) as described in Ref. 16. Quantitative Saturation Transfer with Hyperpolarized Xenon. The quantitative saturation transfer with hyperpolarized xenon is an analysis that fits Hyper-CEST z-spectra directly to both the Bloch-McConnell equations and an analytical solution of these (the full Hyper-CEST solution42) globally for different RF cw saturation pulse conditions.16 The change in the data thus must be induced solely from the RF cw saturation pulse conditions, but not by a loss of nanostructure concentration itself during the measurement, e.g., by collapsing GVs. We therefore performed stability tests for Ana, and Mega nanostructures to ensure that the acquisition of three consecutive z-spectra did not cause collapse of these gas vesicles (see Figure S4). In contrast, gas vesicles derived from Halo were rather fragile. For the qHyper-CEST experiments done with Halo, we therefore used a fresh sample for each z-spectrum.


22

Page 23 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

FIGURES.

Figure 1. Illustration of xenon magnetization pools with molecular containers. (A) Cryptophane-A (CrA) is able to bind a single Xe atom (green sphere, pool B). Thus, its capacity is limited and the impact of RF cw saturation of pool B on pool A of free Xe does not change proportionally with the overall spin ensemble size. (B) Gas vesicle (GV) nanostructures can host an increasing number of Xe atoms in proportion to the surrounding xenon concentration. The physical partitioning causes the relative signal loss in pool A to grow proportionally as pool A and pool B increase simultaneously.


23

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 33

Figure 2. Size and shape of gas vesicle (GV) nanostructures. (A)-(C) Representative TEM images of GVs encoded by (A) Bacillus megaterium (Mega), (B) Halobacterium salinarum (Halo) and (C) Anabaena flos-aquae (Ana). (D) Typical dimensions of each GV type derived from TEM images.

Figure 3. z-Spectra for quantitative chemical exchange saturation transfer with hyperpolarized xenon (qHyper-CEST) analysis of GVs from (A) Mega (OD500 = (0.09 ± 0.01); corresponding to (180 ± 20) pM), (B) Ana (OD500 = (0.33 ± 0.02); corresponding to (38 ± 2) pM), and (C) Halo (OD500 = (0.33 ± 0.02); corresponding to (15.6 ± 0.1) pM).


24

Page 25 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Figure 4. Ideal gas behavior of Mega GVs. (A) z-Spectra (for various RF cw saturation conditions; see legend) of Mega nanostructures at OD500 = 0.09 ± 0.01 (corresponding to (180 ± 20) pM) for different xenon gas total pressures. (B) Number of xenon atoms on average per Mega gas vesicle.


25

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 33

Figure 5. Change in Hyper-CEST contrast as a function of the xenon gas pressure. Z-Spectra of (A) cryptophane-A (RF cw saturation: 15 µT for 10 s; 5 µM of CrA) and (B) nanostructures from Mega (RF cw saturation: 20 µT for 20 s; same data as in Figure 4). (C) Change in Hyper-CEST contrast (normalized to the contrast at an initial pressure of 1.2 bar) calculated at each Larmor frequency (i.e., CrA: ca. – 133 ppm; GVs from Mega: ca. – 150 ppm).

TOC-Figure:


26

Page 27 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Table 1. Typical dimensions of gas vesicle nanostructures (extracted from TEM data) and resulting quantities.

GV type

diameter / nm

length / nm

approx. surface area / nm2

approx. volume / aL

surface-to-volume ratio / (nm2/aL)

Ana

136

519

210,000

6.05

34,710

Halo

251

400

190,000

6.25

30,400

Mega

72

250

47,000

0.65

72,300

Table 2. Xenon-host interaction parameters obtained from qHyper-CEST at a total gas pressure of 1.2 bar for three types of gas vesicles (all at OD500 = 0.33) and for Cryptophane-A at 10 µM. xenon host

[host]a

bound xenon fractiona / 10-4

xenon capacity per hostb

exchange rate / s-1

gas turnover rate per hostc / 106 s-1

Ana

37.7 pM

6.9 ± 0.2

4,760 ± 30

19,300 ± 1,800

91.868

- 176. 2 ± 0.1

Halo

15.6 pM

3.5 ± 0.1

5,800 ± 20

15,300 ± 1,500

88.740

-181.7 ± 0.5

Mega

669.5 pM

8.1 ± 0.5

890 ± 40

22,600 ± 2,800

20.114

-155.7 ± 2.1

CrA

10 µM

58 ± 6

0.29 ± 0.01

16 ± 3

4.6 × 10-6

-132.14 ± 0.02

relative chemical shift / ppm

a

Directly related to each other. A constant, similar to a binding constant, and independent on the host concentration [host] and bound xenon fraction. c Given by the product of the xenon capacity and exchange rate. b


27

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 33

ASSOCIATED CONTENT The authors declare no competing financial interest.

Supporting Information. The Supporting Information is available free of charge on the ACS Publications website at DOI: xxx [to be filled out by the journal]. Classic CEST site occupancy; Xenon gas pressure dependence for Mega GVs and CrA; Gas vesicle stability test; Ideal gas law considerations for GVs; Elastic scaling for GVs derived from Ana; Classification of contrast build-up; Workflow of data quantification; Xenon gas pressure dependence z-spectra.

AUTHOR INFORMATION Corresponding Author *[email protected] *[email protected] ORCID Martin Kunth: 0000-0002-9741-9881 George J. Lu: 0000-0002-4689-9686 Christopher Witte: 0000-0002-4098-6623 Mikhail G. Shapiro: 0000-0002-0291-4215 Leif Schröder: 0000-0003-4901-0325t


28

Page 29 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

ACKNOWLEDGMENT Support by the Human Frontiers Science Program (HFSP Program Grant No. RGP0050/2016 to M.S. and L.S.) and by the German Research Foundation (Koselleck Grant SCHR 995/5-1 to L.S.) is gratefully acknowledged.


29

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 33

REFERENCES: (1) (2) (3)

(4) (5) (6) (7) (8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

Castelli, D. D.; Terreno, E.; Longo, D.; Aime, S. Nanoparticle-Based Chemical Exchange Saturation Transfer (CEST) Agents. NMR Biomed. 2013, 26, 839–849. Bar-Shir, A.; Bulte, J. W. M.; Gilad, A. A. Molecular Engineering of Nonmetallic Biosensors for CEST MRI. ACS Chem. Biol. 2015, 10, 1160–1170. Zhou, J.; Wilson, D. A.; Sun, P. Z.; Klaus, J. A.; Van Zijl, P. C. M. Quantitative Description of Proton Exchange Processes between Water and Endogenous and Exogenous Agents for WEX, CEST, and APT Experiments. Magn. Reson. Med. 2004, 51, 945–952. Walker, T. G.; Happer, W. Spin-Exchange Optical Pumping of Noble-Gas Nuclei. Rev. Mod. Phys. 1997, 69, 629. Goodson, B. M. Nuclear Magnetic Resonance of Laser-Polarized Noble Gases in Molecules, Materials, and Organisms. J. Magn. Reson. 2002, 155, 157–216. Berthault, P.; Huber, G.; Desvaux, H. Biosensing Using Laser-Polarized Xenon NMR/MRI. Prog. Nucl. Magn. Reson. Spectrosc. 2009, 55, 35–60. Schröder, L. Chapter 8 HyperCEST Imaging. In Chemical Exchange Saturation Transfer Imaging; Pan Stanford Publishing, 2017; pp 121–158. Spence, M. M.; Rubin, S. M.; Dimitrov, I. E.; Ruiz, E. J.; Wemmer, D. E.; Pines, A.; Yao, S. Q.; Tian, F.; Schultz, P. G. Functionalized Xenon as a Biosensor. Proc. Natl. Acad. Sci. 2001, 98, 10654–10657. Schroder, L.; Lowery, T. J.; Hilty, C.; Wemmer, D. E.; Pines, A. Molecular Imaging Using a Targeted Magnetic Resonance Hyperpolarized Biosensor. Science 2006, 314, 446–449. Klippel, S.; Döpfert, J.; Jayapaul, J.; Kunth, M.; Rossella, F.; Schnurr, M.; Witte, C.; Freund, C.; Schröder, L. Cell Tracking with Caged Xenon: Using Cryptophanes as MRI Reporters upon Cellular Internalization. Angew. Chem. Int. Ed. 2014, 53, 493–496. Rose, H. M.; Witte, C.; Rossella, F.; Klippel, S.; Freund, C.; Schröder, L. Development of an Antibody-Based, Modular Biosensor for 129Xe NMR Molecular Imaging of Cells at Nanomolar Concentrations. Proc Natl Acad Sci U A 2014, 111, 11697–11702. Witte, C.; Martos, V.; Rose, H. M.; Reinke, S.; Klippel, S.; Schröder, L.; Hackenberger, C. P. R. Live-Cell MRI with Xenon Hyper-CEST Biosensors Targeted to Metabolically Labeled Cell-Surface Glycans. Angew Chem Int Ed Engl 2015, 54, 2806–2810. Kunth, M.; Witte, C.; Hennig, A.; Schröder, L. Identification, Classification, and Signal Amplification Capabilities of High-Turnover Gas Binding Hosts in Ultra-Sensitive NMR. Chem Sci 2015, 6, 6069–6075. Schnurr, M.; Sloniec-Myszk, J.; Döpfert, J.; Schröder, L.; Hennig, A. Supramolecular Assays for Mapping Enzyme Activity by Displacement-Triggered Change in Hyperpolarized (129)Xe Magnetization Transfer NMR Spectroscopy. Angew Chem Int Ed Engl 2015, 54, 13444–13447. Hane, F. T.; Li, T.; Smylie, P.; Pellizzari, R. M.; Plata, J. A.; DeBoef, B.; Albert, M. S. In Vivo Detection of Cucurbit[6]Uril, a Hyperpolarized Xenon Contrast Agent for a Xenon Magnetic Resonance Imaging Biosensor. Sci. Rep. 2017, 7, 41027. Kunth, M.; Witte, C.; Schröder, L. Quantitative Chemical Exchange Saturation Transfer with Hyperpolarized Nuclei (QHyper-CEST): Sensing Xenon-Host Exchange Dynamics and Binding Affinities by NMR. J. Chem. Phys. 2014, 141, 194202.


30

Page 31 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

(17) Döpfert, J.; Witte, C.; Schröder, L. Fast Gradient-Encoded CEST Spectroscopy of Hyperpolarized Xenon. ChemPhysChem 2014, 15, 261–264. (18) Wang, Y.; Roose, B. W.; Philbin, J. P.; Doman, J. L.; Dmochowski, I. J. Programming A Molecular Relay for Ultrasensitive Biodetection through 129 Xe NMR. Angew. Chem. Int. Ed. 2016, 55, 1733–1736. (19) Roose, B. W.; Zemerov, S. D.; Dmochowski, I. J. Nanomolar Small-Molecule Detection Using a Genetically Encoded 129Xe NMR Contrast Agent. Chem. Sci. 2017, 8, 7631– 7636. (20) Shapiro, M. G.; Ramirez, R. M.; Sperling, L. J.; Sun, G.; Sun, J.; Pines, A.; Schaffer, D. V.; Bajaj, V. S. Genetically Encoded Reporters for Hyperpolarized Xenon Magnetic Resonance Imaging. Nat. Chem. 2014, 6, 629–634. (21) Lakshmanan, A.; Farhadi, A.; Nety, S. P.; Lee-Gosselin, A.; Bourdeau, R. W.; Maresca, D.; Shapiro, M. G. Molecular Engineering of Acoustic Protein Nanostructures. ACS Nano 2016, 10, 7314–7322. (22) Lakshmanan, A.; Lu, G. J.; Farhadi, A.; Nety, S. P.; Kunth, M.; Lee-Gosselin, A.; Maresca, D.; Bourdeau, R. W.; Yin, M.; Yan, J.; Witte, C.; Foster, F. C.; Schröder, L.; Shapiro, M. G. Preparation of Biogenic Gas Vesicle Nanostructures for Use as Contrast Agents for Ultrasound and MRI. Nat. Protoc. 2017, 12, 2050–2080. (23) Bourdeau, R. W.; Lee-Gosselin, A.; Lakshmanan, A.; Farhadi, A.; Kumar, S. R.; Nety, S. P.; Shapiro, M. G. Acoustic Reporter Genes for Noninvasive Imaging of Microorganisms in Mammalian Hosts. Nature 2018, 553, 86–90. (24) Lu, G. J.; Farhadi, A.; Szablowski, J. O.; Lee-Gosselin, A.; Barnes, S. R.; Lakshmanan, A.; Bourdeau, R. W.; Shapiro, M. G. Acoustically Modulated Magnetic Resonance Imaging of Gas-Filled Protein Nanostructures. Nat. Mater. 2018. (25) Walsby, A. E.; Hayes, P. K. Gas Vesicle Proteins. Biochem. J. 1989, 264, 313. (26) Farhadi, A.; Ho, G.; Kunth, M.; Ling, B.; Lakshmanan, A.; Lu, G. J.; Bourdeau, R. W.; Schröder, L.; Shapiro, M. G. Recombinantly Expressed Gas Vesicles as Nanoscale Contrast Agents for Ultrasound and Hyperpolarized MRI. AIChE J. in print. (27) Stevens, T. K.; Ramirez, R. M.; Pines, A. Nanoemulsion Contrast Agents with SubPicomolar Sensitivity for Xenon NMR. J. Am. Chem. Soc. 2013, 135, 9576–9579. (28) Klippel, S.; Freund, C.; Schröder, L. Multichannel MRI Labeling of Mammalian Cells by Switchable Nanocarriers for Hyperpolarized Xenon. Nano Lett. 2014, 14, 5721–5726. (29) Bai, Y.; Wang, Y.; Goulian, M.; Driks, A.; Dmochowski, I. J. Bacterial Spore Detection and Analysis Using Hyperpolarized 129Xe Chemical Exchange Saturation Transfer (Hyper-CEST) NMR. Chem. Sci. 2014, 5, 3197–3203. (30) Döpfert, J.; Schnurr, M.; Kunth, M.; Rose, H. M.; Hennig, A.; Schröder, L. TimeResolved Monitoring of Enzyme Activity with Ultrafast Hyper-CEST Spectroscopy. Magn. Reson. Chem. 2018, 56, 679–688. (31) Bartik, K.; Luhmer, M.; Dutasta, J.-P.; Collet, A.; Reisse, J. 129Xe and 1H NMR Study of the Reversible Trapping of Xenon by Cryptophane-A in Organic Solution. J. Am. Chem. Soc. 1998, 120, 784–791. (32) Korchak, S.; Kilian, W.; Mitschang, L. Degeneracy in Cryptophane–Xenon Complex Formation in Aqueous Solution. Chem. Commun. 2015, 51, 1721–1724. (33) Wang, Y.; Dmochowski, I. J. Cucurbit[6]Uril Is an Ultrasensitive 129Xe NMR Contrast Agent. Chem. Commun. 2015, 51, 8982–8985.


31

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 33

(34) Witte, C.; Kunth, M.; Rossella, F.; Schröder, L. Observing and Preventing Rubidium Runaway in a Direct-Infusion Xenon-Spin Hyperpolarizer Optimized for High-Resolution Hyper-CEST (Chemical Exchange Saturation Transfer Using Hyperpolarized Nuclei) NMR. J. Chem. Phys. 2014, 140, 084203. (35) Rao, M.; Stewart, N. J.; Norquay, G.; Griffiths, P. D.; Wild, J. M. High Resolution Spectroscopy and Chemical Shift Imaging of Hyperpolarized 129Xe Dissolved in the Human Brain in Vivo at 1.5 Tesla. Magn. Reson. Med. 2016, 75, 2227-2234. (36) Chacon-Caldera, J.; Maunder, A.; Rao, M.; Norquay, G.; Rodgers, O. I.; Clemence, M.; Puddu, C.; Schad, L. R.; Wild, J. M. Dynamic MRI of Hyperpolarized Xenon-129 Uptake in the Human Kidney Using a Dedicated Transmission-Only-Reception-Only Array at 3 Tesla. In Proc. Intl. Soc. Mag. Reson. Med.; International Society for Magnetic Resonance in Medicine, 2018; Vol. 26, p 4470. (37) Ferrauto, G.; Gregorio, E. D.; Baroni, S.; Aime, S. Frequency-Encoded MRI-CEST Agents Based on Paramagnetic Liposomes/RBC Aggregates. Nano Lett. 2014, 141110105732007. (38) Di Gregorio, E.; Ferrauto, G.; Gianolio, E.; Lanzardo, S.; Carrera, C.; Fedeli, F.; Aime, S. An MRI Method to Map Tumor Hypoxia Using Red Blood Cells Loaded with a PO2Responsive Gd-Agent. ACS Nano 2015. (39) Ferrauto, G.; Delli Castelli, D.; Di Gregorio, E.; Terreno, E.; Aime, S. LipoCEST and CellCEST Imaging Agents: Opportunities and Challenges. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 2016, 8, 602-618. (40) Li, N.; Cannon, M. C. Gas Vesicle Genes Identified in Bacillus Megaterium and Functional Expression in Escherichia Coli. J. Bacteriol. 1998, 180, 2450–2458. (41) Witte, C.; Kunth, M.; Döpfert, J.; Rossella, F.; Schröder, L. Hyperpolarized Xenon for NMR and MRI Applications. J. Vis. Exp. 2012, No. 67. (42) Zaiss, M.; Schnurr, M.; Bachert, P. Analytical Solution for the Depolarization of Hyperpolarized Nuclei by Chemical Exchange Saturation Transfer between Free and Encapsulated Xenon (HyperCEST). J. Chem. Phys. 2012, 136, 144106–144110.


32

Page 33 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

ACS Nano


Protein Nanostructures Produce Self-Adjusting ... - ACS Publications

Recommend Documents