Extending the Scope of 19F Hyperpolarization through Signal Amplification by Reversible Exchange in MRI and NMR Spectroscopy

Abstract Fluorinated ligands have a variety of uses in chemistry and industry, but it is their medical applications as 18F‐labelled positron emission tomography (PET) tracers where they are most visible. In this work, we illustrate the potential of using 19F‐containing ligands as future magnetic resonance imaging (MRI) contrast agents and as probes in magnetic resonance spectroscopy studies by significantly increasing their magnetic resonance detectability through the signal amplification by reversible exchange (SABRE) hyperpolarization method. We achieve 19F SABRE polarization in a wide range of molecules, including those essential to medication, and analyze how their steric bulk, the substrate loading, polarization transfer field, pH, and rate of ligand exchange impact the efficiency of SABRE. We conclude by presenting 19F MRI results in phantoms, which demonstrate that many of these agents show great promise as future 19F MRI contrast agents for diagnostic investigations.


Introduction
Hyperpolarized magnetic resonance spectroscopy( MRS) methods have been successfully used in conjunction with 1 H, 13 C, and 15 N detection. [1] An important nucleust hat could successfully be exploited in asimilarmanner is 19 F. Its detection has already been shown to give remarkable results in traditional (i.e. thermale quilibrium) MRSa pplications. [2] Furthermore, 18 F radio-labelled positron emission tomography( PET) agents are highly successful in the clinicasd iagnostic probes. [3] In the context of MRS applications, 19 Fp resentss everal significant advantageso ver 13 Ca nd 15 N detection, as it is spin-1 = 2 , 100 %a bundant,a nd has 83 %o ft he sensitivity of 1 Hd etection. Moreover,ith as av ery wide and, hence, diagnostic chemical shift range (much larger than those of 13 Ca nd 31 Pa nd comparable to that of 15 N)a nd it is not presente ndogenously in biological tissue, so there is no competing background signal to obscure the corresponding in vivo data. The large chemicals hift range exhibited by hyperpolarized 129 Xe nuclei reflectso ne of the reasons this alternative probe is attracting so much attention. So far, 19 FM RS has been used to aid the probingo fc atabolic and anabolic drug conversionso fa nticancer agents such as 5-fluorouracil [2,4] andt he uptake and elimination of moderna nesthetics used in preclinical magnetic resonance imaging (MRI). [5] It has also been used for the noninvasive pharmacokinetic analysiso fV oriconazole in the brain and plasma. [6] Its large chemical shift range and high sensitivity to the local environmenth ave been exploitedi nt he assessment of cellularm etal-ion concentrations, [7] tissue oxygenation, [8] and pH. [9] Various approaches to improvet he typicallyl ow signal-tonoise ratio (SNR) of 19 FM RI experiments are presented in the literature, such as increasing the number of 19 Fa toms per molecule. Variationsi nt ransverse relaxation times have also been exploited to provide molecular environmentd ata. [10] Hyperpolarization has been shown to be av ery promising way of obtaining significantly stronger 19 Fs ignal compared to conventional methods. So far,t he 19 Fn uclear magnetic resonance (NMR)r esponse of various compounds has been successfully enhanced using both dynamic nuclear polarization [11] and PHIP (para-hydrogen-induced polarization). [12] The hyperpolarization of the 19 Fn ucleus of 3-fluoropyridine using signal amplification by reversible exchange (SABRE), which is also a para-hydrogen (p-H 2 )-based technique, was reportedi n2 009 by Adams et al. [13] More recently,S hchepin et al. [14] presented the first hyperpolarized 19 FS ABRE MRIr esultsu sing the same molecule and achieved signal enhancements of 100-fold at 9.4 Tb y using SABRE-SHEATH( shield enables alignment transfer to heteronuclei). [15] Furthermore, av ery recent report detailed the use of as uperconductingq uantum interference device (SQUID) for the simultaneous measurement of 1 Ha nd 19 Fs pectra using SABRE. [16] 3-Fluoropyridine, ethyl-5-fluoronicotinic Fluorinated ligands have av ariety of uses in chemistry and industry,b ut it is their medicala pplications as 18 F-labelled positron emission tomography( PET) tracers where they are most visible. In this work, we illustrate the potentialo fu sing 19 F-containing ligands as future magneticr esonance imaging (MRI) contrastagents and as probes in magnetic resonance spectroscopy studies by significantly increasing their magnetic resonance detectability through the signala mplification by reversi-ble exchange( SABRE)h yperpolarization method. We achieve 19 FS ABRE polarization in aw ide range of molecules, including those essential to medication, and analyze how their steric bulk, the substratel oading, polarization transfer field, pH, and rate of ligand exchange impact the efficiency of SABRE. We conclude by presenting 19 FM RI results in phantoms, which demonstrate that many of these agentss how great promise as future 19 FMRI contrast agents for diagnostic investigations. acid and 3,5-bis(trifluoromethyl)pyridine were investigated. In addition, af urther report described an attemptt ou tilize continuoush yperpolarization to hyperpolarize pentafluropyridine, [17] but it was unsuccessful as the latter did not polarize. We seek to add to these early studies by screening aw ide range of substrates to increase the applicability of this approach and suggest optimalroutes to increasing 19 Fhyperpolarization by SABRE. SABRE utilizes p-H 2 , [13] which is as pini somer of H 2 .A sanuclear singlet, it has no net spin and, so, it is unobservable from an NMR perspective.H owever,o nce p-H 2 has come into contact with am etal center, therebyf orming p-H 2 -derived hydride ligands, this previously NMR invisible singlet state can be accessed.N ow,p olarization is transferred from the p-H 2 -derived nuclei to spin-1 = 2 nuclei that are located in ligandsa ttached to the same metal center through the J-coupling network. [18] The ligandst hat lie trans to the p-H 2 -derived hydrides receive polarizationo ptimally,a st hey exhibit the largestd ifference in cis and trans couplings. Ther eversible nature of both H 2 addition and analyte ligation at the metal centere nables hyperpolarization to buildup in non-ligated analyte molecules in ap rocess that has been described as being catalytic in the transfer of polarization.T he hyperpolarization that is created in this way can be readily read-outi na nN MR experiment and, hence,t he signal intensityofthe resulting NMR spectrum or magnetic resonance(MR) image is suitably enhanced (Scheme 1).
The transfer of polarization through SABRE typically occurs at low magnetic field and is maximized when alevel anti-crossing (LAC) condition is met. [19] This is associated with the spin system becoming strongly coupled as ac onsequence of the chemicals hift differenceb etween the interacting groups collapsing to values that match the shared spin-spin couplings, thus resulting in coherent polarization transfer.C onsequently, the efficiency of polarization transfer is field dependent. However,m ethods exist to transfer such polarization at high magnetic field in the presenceo fr adio frequency excitation, [20] such that hyperpolarized NMR spectra can be collected. [21] Polarization transfer through SABREh as, in fact, been shown to be successfully transferred to 1 H, [13,22] 13 C, [22a, 23] [13,14,16] 31 P, [25] 29 Si, and 119 Sn. [26] We further note that there are al arge number of drugs that contain 19 Fn uclei;i n2 010, it wasc alculated that about 20 %o f administered drugs contained fluorine atoms or fluoroalkyl groups. [27] In this work, we test ar ange of 19 F-containing substrates and drug molecules with ap yridyl arrangement (as depicted in Scheme2)t od emonstrate the possibility of applying SABRE in MR material and clinicali nvestigations that rely on the response of 19 Fn uclei. Furthermore, we seek to determine the conditions required for maximizing the 19 FNMR response of different ligands.
A2 56-scan 19 FNMR spectrum of 20 equivalents of L 1 or L 5 in the presence of 1 (5 mm)y ielded only as ingle set of peaks, that of free L 1 or L 5 .Asecond set of peaks, owing to the formation of [Ir(IMes)(COD)L] + ,w as not observed. This contrasts strongly with the data obtainedf or L 2 under the same conditions, in which two peaks are observedi naratio of 0.84:19.16 that correspond to [Ir(IMes)(COD)L 2 ] + and free L 2 ,r espectively. These two peaks are separated by 362 Hz in a1 .4 TN MR field. The same observation was made when L 2 was exchanged for L 4 ;i nt his case, the two peaks weres eparated by 335 Hz in a 1.4 TN MR field,a lthough the ratio was smaller (0.25:19.75 for L 4 /[Ir(IMes)(COD)L 4 ] + ). Conversely,w hen L 5 was used, the 19 FNMR spectrum comprised of only three peaks, representative of the three different 19 FNMR environments that L 5 possesses.J ust like L 1 ,t here were no other observable peaks for bound L 5 .A sb oth L 1 and L 5 possess ortho-fluorine nuclei, there could be apossibility that steric hindrance around the nitrogen used to ligate to the metal centero f1 is preventing effective ligation. Compared to hydrogen, 19 Fh as ag reater van der Waals radius by 0.27 ,b ut is still smaller than a methyl group (van der Waals radius 2.00 ). [31] Shchepin et al. reportedt hat lutidines and picolines possessing am ethyl group in the ortho-position yielded no detectable 15 N hyperpolarization through SABRE-SHEATH. [15] In addition, the pK a of L 1 (À0.44) is such that it mayn ot be an effective ligand in comparison to pyridine (pK a = 5.23) or 3-fluoropyridine (pK a = 2.97). To the besto fo ur knowledge,n or eported pK a data are availablef or L 4 ,a lthough 3,5-dichloropyridine has ar eported pK a of 0.51. [32] L 4 would be expected to have as imilar pK a and, thus, is more basic than L 1 but less so than L 2 .W ang et al. indicated that the pK a of L 4 was < 2, as the hydrolysis rate of their ruthenium complexesi ncreased relative to when pyridineo r3picoline weree mployed. [33] L 5 is reported to not react with hot aqueous hydrogen iodide or hydrochlorica cid, owing to the base-weakening effect of the two ortho-fluorines. [34] The combination of steric as well as binding affinity to the metal center may preclude L 1 and L 5 acting as efficient ligandst o1. L 5 has previously been reported as not being SABREactive. [17]

1 HS ABRE Hyperpolarization of F-Substituted Ligands
We then sought to evaluate the ability of 1 to form efficient SABRE polarization-transfer catalysts with the ligandss hown in Scheme2.T his is normally associated with the formation of [Ir(H) 2 (IMes)(L) 3 ]Cl. L 1 ,w hen examined under SABRE,p roduced as mall enhancement (below unity) for both ligand loadings tested here (4 and 20 equivalents of L relative to 1). Furthermore, as trong hydrides ignal for the corresponding complex [Ir(H) 2 (IMes)(L 1 ) 3 ]Cl was not observed. Attempts to increaset he binding strength by using as maller carbene (IMe) were unsuccessful( see Section2.1 of the Supporting Information). Hence, L 1 is poorly suited to SABRE.
In contrast, L 2 proved to provideasubstantial 1 HS ABRE response. When the substrate loading is 1:4, 1 HS ABRE NMR spectra recorded after adding3bar of p-H 2 and shaking the sample for 10 si nafield of approximately 65 Gs how that the protons H-2 and H-6 of L 2 exhibit a2 103-fold summed signal gain, whichc ompares to an enhancemento fÀ2397 reported for pyridine by Lloyd et al. under analogousc onditions. [29] When the metal/ligand ratio is 1:20, which corresponds to 17fold excess of ligand to catalyst( spectra presented in Figure 1), the corresponding values are 393-fold for pyridine and 1296fold for L 2 ,avalue approximately six times highert han the one reported by Shchepin et al. for the same sample at 9.4 T. [14] We attribute this significant differencet ot he difference in the purity of the p-H 2 gas used for sample polarization( 96 %i n this work and 50 %i nR ef. [14]). This suggestst hat L 2 is ag ood agent forSABRE in accordance with early observations [13,14] and confirms that the p-H 2 concentrationh as as ignificant effect on the efficiency of the polarization transfer process. 1 HNMR spectra recorded after H 2 addition and subsequent catalysta ctivation show that [Ir(IMes)(H 2 )(L 2 ) 3 ]i sf ormed and is the dominant SABRE polarization-transfer catalysti ns olution, with ad iagnostic hydride signal at dÀ23.11ppm. As econd complex, indicated by ap air of hydride resonances d at 23.87 and À25.15 ppm, was found to be [Ir(IMes)(H 2 )(L 2 ) 2 (MeOH)] (see the Supporting Information for further details and characterization data). The possibility of Cl rather than methanol binding was excludedb ya cquirings pectra of the same reaction in dichloromethane and comparing the results. As exemplified elsewhere, [35] methanol can be an active participant in the polarization-transferc atalyst and, when employingm ild acidic conditions, solvent polarization can be observed and quantified by measuring the enhancement of the OH resonance. Although evidence for methanolb inding was obtained, no OH or residual CD 2 HOD signal enhancementw as observed. Shchepin et al. presented 1 HNMR spectra in their Supporting Information, which possessu nattributed signals that may reflect the observation of species of this type. [14] The activation parameters corresponding to the process of free ligand buildup in solutionf ollowing dissociation from the dominant complexw ere calculated by using data obtained from as eries of variable-temperature exchange spectroscopy (EXSY) measurements (see Section 2.4 of the Supporting Information).T he L 2 buildup rate at 300 Kw as extrapolated through Eyring analysis and was found to be 65.1 s À1 ,t hat is, approximately 2.8 times larger than the corresponding rate of pyridine measured using the same conditions. [29] Furthermore, the enthalpy value fort he buildupp rocess of 98 kJ mol À1 suggests that the binding energy is slightly higher than that of pyridine (for which the site trans to hydrideh as a DH°( build-up) value of 95 kJ mol À1 ).
When considering the related molecules L 4 , L 6 ,a nd L 7 ,v ery good 1 HS ABRE polarization (of the order of hundreds up to thousands) was obtained in all cases. The SABRE polarizationtransfer catalysts formed with 1 in solution are the tris-substituted species, [Ir(IMes)(H) 2 (L) 3 ] + ,a nd the analogous methanol complex, [Ir(IMes)(H) 2 (L) 2 (MeOH)] + (see the Supporting Information for enhancementv alues, thermodynamic parameters, and characterization data).
As eries of one-shot 1 HNMR spectra were collected on samples containing one-fold and 17-fold excesses of L 7 in the presence of 5mm of 1 in MeOD solution. In the case of the sample containing ao ne-fold excess, substantial signal enhancements were observed for all three of the non-exchangeable resonances of the free substrate, with the largest 1 HNMR signal enhancement being observed forH-2 (À272 AE 24), followedb yH -4a nd H-6 (À181 AE16 and À121 AE 9a t4 00 MHz, respectively). The corresponding values for the solution containing a1 7-fold excess of ligand were 59 AE 2, 50 AE 2, and 55 AE 2. These values are much lower than those seen for L 2 and reflect the fact that the protonated form of the free ligand (L 7 a)d ominates in solution.
Evidence for methanol binding to the iridium centerw as again seen in the hydride region of the associated 1 HNMR spectra,w hich, in the case of L 7 ,c ontains ar esonance at dÀ23.45 ppm, corresponding to the tris-substituted complex. Ap airo fr esonances,l ocated at dÀ23.84 and À24.04 ppm, arise from ac omplex where the equatorial sites are occupied by one substrate molecule, one methanol molecule, and two hydrides.
The corresponding 1 HS ABRE NMR experiments now result in ap olarized OH resonance, exhibiting average enhancement values of 6.65 AE 0.5 (one-fold excess) and 51 AE 2( 17-fold excess). We have shown, in our previous work, [35a] that the de-protonation of the Ncenterofthe conjugate acid form of nicotinic acid and, subsequently,m ore efficient binding to 1,c an be achieved by adding am ild base to the solution. When analyzing samples containing 20 and 100 mm of L 7 after adding Cs 2 CO 3 in equal amounts to the substrate (one-fold and 17fold excess to 1,r espectively), the enhancemento ft he freeligand resonances increases considerably in both cases, as a result of the molecule being deprotonated (L 7 b)a nd increasing the ligand's probability of binding to the catalyst. We note, however,t hat in the case of low ligand loading, the presence of base promotes and accelerates H-D exchange, ap henomenon that occurs immediatelya fter activation and on the timescale of the experiments, resulting in ap rogressive decrease of the total enhancement with each addition of fresh p-H 2 .D ata shows that at least 40 %d euteration of the three sites takes place in the first 15 minutes (see Section 2.2 of the Supporting Information).
For the sample prepared using ao ne-fold excesso fl igand and ao ne-fold excesso fb ase, relative to the catalyst,t he total signal enhancemento btained across the three sites was 2.5times higher than that seen for asimilar sample prepared without base;t he correspondingd ifference increased to 4.5 times for the 17-fold loading (spectra presented in Figure 2). Furthermore, neither methanolb inding nor OH signal enhancement was observed when the base was present.T hese results relate to those reported forn iacin [35a] and confirmt he importance that the pH can play in manipulating substrate binding.
We have also analyzed the fluorine-substituted nucleobase used widely used in cancer therapy,5 -fluorouracil (L 8 ,c ommonly sold under the commercial name of Adrucil), and the anti-fungald rug Vorincazole (L 9 ), both of which are listed on the World Health Organisation's essential medicines list. As L 8 is present in solution in its fully protonated form (thus preventing the Nc enters from binding to 1), we adopted as trategy similart ot he one used for L 7 to promote the formation of the SABRE polarization-transfer catalyst andt oi mprove the signal enhancements. In the hydrider egion of this 1 HNMR spectrum, we initially see two signals at dÀ15.4 and À18.1 ppm, which are in emission and absorption, respectively,o wing to the formation under ALTADENA( adiabatic longitudinal transport after dissociation engendersn ucleara lignment) [36] conditions ( Figure 3). We note that these observations are consistentw ith other reports that have considered the activation of SABRE catalysts. [37] In addition, we see two SABRE enhanced CH resonances belonging to ab ound COE (cyclooctene) ligand at d 3.57 and 3.35 ppm, which is consistent with ap rior report, [37a] as well as aw eakly polarized peak corresponding to the mesityl protons of the carbene.
Hence, the initial product, [Ir(H) 2 (IMes)(COE)(L 8 ) 2 ]Cl, exhibits significant SABREa ctivity.T he signal enhancement reduces over the next 5minutes, as this complex is converted into multiple species (see the Supporting Information for detailed information). Under these conditions, when the excess of L 8 is fourfold and 0.5-fold excesso fC s 2 CO 3 is present,w es ee as ignal at d 7.56 ppm, corresponding to the free resonanceo fH -5. This signal possesses an initial enhancement of 53-fold and a second peak at d 5.97 ppm, corresponding to the H-5r esonance of the ligand when bound to the Ir center, shows an initial 26-folde nhancement. As et of optimization experiments were undertakent oi dentify the bestb ase and ligand concentrations, and ultimately a1 00-fold 1 HNMR signal enhancement was obtained for the free H-5 resonance of L 8 and approximately a5 7-fold enhancement was observedf or the peak corresponding to the bound ligand. Unfortunately,t hese values decrease rapidlyw ith time, owing to ap rocess of deactivation that quenches the activity of the catalyst. Attempts to block this cyclometalationp rocess using as maller carbene were unsuccessful under the conditions employed.W ea lso note that rapid H/D exchange leads to the formation of CD 3 OH alongside the deuterationo ft he iridium hydride.
Polarization transfer of four equivalents of L 9 using 1 in the presence of p-H 2 in earth's magnetic field was also probed. The pyrimidine ring protons were enhanced twofold. Compared with the other ligandss tudied, L 9 possesses the greatest steric bulk, and so this could be the main factor for the low enhancement obtained. When the ratio of L 9 /1 was increasedt o2 0:1, SABRE enhancement of the pyrimidine ring protons was not detectable.
Hence, we can concludet hat very high levels of 1 Hh yperpolarization can be achieved in 19 F-containing materials under SABRE,with the exceptionofL 9 .

19 FSABRE Hyperpolarizationo fF-Substituted Ligands
We have investigated the 19 Fh yperpolarization of the ligands analyzed in this work by using a5 00 MHz spectrometer (11.74 T) equipped with dual capacity 1 H- 19 Fh igh-resolution probe and acquired 19 Fh yperpolarized spectra using p/2 pulses immediately after shaking the samples in the stray field of the magnet. As was the case for 1 H, L 2 provided the best response, with enhancements of approximately 60 for the free resonance (d À127.5 ppm) and 36 for the bound equatorial signal (d À124.1 ppm).W en ote that the signal is anti-phase when polarization is conducted in the fringe field of the magnet, as al ongitudinalt wo-spino rder 1 H- 19 Ft erm is created, which is analogous to the homonuclear 1 H-1 Ht erm that is createdu nder in-field PHIP, [38] and so enhancements are calculated by using magnitude mode. Lower values were obtained for the other substrates considered( see Section4 of the Supporting Information for 19 FNMR spectra and full enhancement data). Interestingly,t he 19 Fs ignalo fL 4 is entirely in-phase, whereas, when SABRE-SHEATH is employed, the signal becomes anti-phasei nc haracter. Conversely,t he opposite is true for L 2 .A na verage enhancement for L 8 could not be calculated, owing to the fast evolution of the cyclometalation process, but both the free and the bound resonancea re enhanced and display an antiphase behavior (Figure 4). L 9 resulted in af ivefold enhancement for the 19 Fn ucleus of the pyrimidine ring when ar atio of 4:1 L 9 to 1 was employed.  This was observed as an anti-phase signal. However, no polarization was observed for the other two 19 Fn uclei, owing to insufficient J coupling to propagate polarization to these two nuclei. SABRE-SHEATHm ethods did not yield any substantial increasei nt he polarization observed, although they did convert the signal from anti-phase to in phase (see the Supporting Information). Despite the low enhancement observed, the T 1 of the polarized 19 Fr esonance is 5.15 si nam easurement field of 1.4 T. This is comparable to the T 1 observed for ar atio of 20:1 L 9 /1 at 11.74 Tfor the other substrates studied (see Table 2).
Althought he results of the 19 Fh yperpolarization studies are very promising and exhibit enhancement values high enough to allow forh igh-resolution images (with the exception of L 9 ) to be recorded (see below), they are two orders of magnitude lower than those obtained for 1 H. This is aremarkable decrease considering the high sensitivity of 19 F, whichi sv ery close to that of proton. In order to fully rationalize this difference, one must take into account the relaxation and exchange rates for the ligands analyzed, as well as the dependence of the polarization transfer on the value of the magnetic fielda tw hich the transfer takes place.

1 Ha nd 19 FR elaxation Times
We have measureda nd compared the 1 Ha nd 19 F T 1 values for each ligand, as it is widelyk nown that relaxation is one of the main factors affecting the efficiency of the SABREh yperpolarization transfer process. When examining the 1 Hl ongitudinal relaxation times of the ligandsp repared in MeOD solutioni n the presence of the polarization-transfer catalyst (17-fold excess of substrate to Ir), we have found that the T 1 values range between 15 and 45 s, most of them being above 20 s (measured at 9.4 T). The data, presented in Ta ble 1, show that all substrates analyzed exhibit relatively long relaxation times, higher or comparable to those of pyridine. The lowest values measured correspond to L 7 in its protonated form, probably owing to the presence of ah ydrogen atom on the binding center.
As shown in our previous work, [35a] addition of basea nd the subsequentc hange in pH has af avorable effect on the T 1 of H-2 of L 7 ,w hich,i nt his case, increases remarkably from 14.5 to 46.1 s. Smaller increases can also be noted for protons H-4 and H-6.
When performing similare xperimentst od eterminet he corresponding 19 F T 1 values, we have found that they lie between 3.3 s(L 7 b)and 5.5 s(L 4 )when measured at 11.74 Tunder Boltzmann equilibrium conditions. This significant differencei nr elaxation rates betweenp roton and fluorine can at least partially account for the much lower enhancements obtained for 19 F in comparison to 1 H. We also note that these values are smaller than those reported by Shchepin et al. for hyperpolarized 3-fluoropyridine at 9.4 T. However,asecond set of T 1 data collected at 1.4 Tf or L 4 and L 2 reports that the T 1 values of the 19 Fn uclei are far longert han at 11.74 Tb yafactor of 3-5 times. This is attributed to the relaxation process being dominated by chemical shift anisotropy rather than dipolar relaxation and can, therefore, be expected to scale with B 0 .I nt he contexto f SABRE-based experiments, as imilare ffect hasp reviously been observed for 15 N, [39] and more recently 19 F. [14] Consequently, these data motivated the collection of 19 Fh yperpolarized NMR spectra of L 2 (4 equiv relative to 1)a t1 .4 T. Relative to at hermal trace, this spectrum revealed that the enhancement for L 2 was now 244-fold (calculated using magnitude data), an improvement of four times over those collected at 11.74 T. Furthermore, employmento fS ABRE-SHEATH converted the antiphase signal into a9 7% in-phase signal, but reduced the signale nhancement to 93-fold. From an imaging perspective, the appearance of the signal is important because broadened lines can lead to partial signal cancellation.
In afurther investigation, we also probed L 4 in as imilar fashion, but also utilized our knowledge of the kinetic exchange rates to inform our experimental protocol. Thus, the solution was cooled to 0 8Cp rior to hyperpolarizationi nam-magnetic shield or at earth's magnetic field. No change was evident for the measurements that were polarizeda te arth's magnetic field, whereas employing SABRE-SHEATH at 0 8Cy ielded an inefold improvement over those at room temperature. Furthermore, the 19 Fs ignal is now completely in-phase,w hereas the analogouss pectrum collected following polarization transfer at earth's magnetic field displays anti-phasec haracter.
The relatively long 1 Hl ongitudinal relaxation times (Table 2) were exploited in INEPT experiments, in which the 1 Hp olarization was transferred to 13 Ca nd to 19 Fn uclei, respectively.I n the case of 13 Cn uclei, this led to as ignificant increase in the enhancement values compared to the results obtained in Boltzmann equilibrium conditions, allowing us to record highquality 13 Cs pectra in less than one second (see Section 3o f the Supporting Information). We also note that, in the case of L 2 and L 7 b,t he use of an INEPT sequence for the acquisition of hyperpolarized 19 Fs pectra led to as ignal gain of approximately 230-and1 00-fold, respectively,w hen the resultso btained [a] As reportedb yL loyd et al. [29]

19 FMRI Results
To assess the potential of the ligands studied in this work as MR contrasta gents for medicali nvestigations,w eh ave undertaken as eries of MRIe xperimentsp erformedo np hantoms containing 5mm of 1 and1 00 mm of substrate dissolved in MeOD. Following p-H 2 addition, the samples were shaken in the stray field of the magneta nd, after being in the imaging probe head, were quickly (after ca. 10 s) inserted into the spectrometer. One-shot 19  The increase in SNR obtained by using SABRE was evaluated by comparing the values obtained for the hyperpolarized image with the SNR obtained for images acquired in Boltzmann equilibrium conditions (see Section5 of the Supporting Information). The bestr esults were obtained when using L 2 , for whicht he hyperpolarized results exhibited a1 04-fold maximum increasei nS NR anda na verage gain of 100-fold when comparedw ith the image acquired in Boltzmann equilibrium conditions (see Section 5o ft he Supporting Information for details).
Furthermore, the MRI experiments emphasize the effect of deprotonation on the polarization-transfer efficiency:w hile the hyperpolarized images of L 7 a have av ery low intensity and exhibit no increase in SNR compared to the thermalc ounterpart, the values obtainedf or L 7 b show that SABRE hyperpolarization lead to a7 5-foldS NR gain compared to the result acquired in Boltzmann equilibrium conditions ( Figure 5).
We believe thatt heser esults would be significantly improvedi ft he time taken to transfer the sample into to the magnetf or observation could be reduced( i.e. the amounto f time between polarization and detection), as as ignificant amount of signal is lost during the 10 si nterval (ca. 2 T 1 )i mposed herebyour hardware limitations.

Conclusions
We have evaluated the capacity of aw ide range of fluorinated molecules to reactw ith [IrCl(IMes)(COD)] (1)a nd H 2 to successfully form aS ABRE-active catalyst that facilitates polarization transfer to 19 F. This builds on the earlier reports of Adamse tal. and Shchepin et al. [13,14] We have found that 2-fluoropyridine (L 1 )a nd pentafluoropyridine (L 5 )r eact weakly,i fa ta ll, with the catalystp recursor.A sb oth molecules possess ortho-fluorine nuclei, there could be ap ossibility that steric hindrance aroundt he nitrogen used to ligate to the metal center of 1 is preventing effective ligation. In the case of 4-fluoropyridine.HCl (L 3 .HCl), attempts to isolate the base wereu nsuccessful and this ligand was not investigated in detail. SABRE polarizationtransfer catalysts were formed by using L 2 , L 4 , L 6 , L 7 , L 7 ,a nd L 9 through reaction with 1 in the presence of H 2 ;i nt he case of L 7 and L 8 ,t he formation of the activatedc atalystw as promoted by deprotonating the Nc enter of the conjugate acid using am ild base (Cs 2 CO 3 ). Thep erformance of the new SABRE catalysts wasa nalyzed as af unctiono fl igand loading, polarization transfer field, temperature (L 4 ), and pH (L 7 ).
When comparingt he 1 H-hyperpolarized NMR resultso btained on the fluorinated compounds studied in this work with the ones obtained using pyridine, we find that when working at low ligand loadings (one-fold excess of substrate), the total enhancement obtainedf or each molecule is considerably lower than for pyridine. However,w hen moving to high ligand loadings (17-folde xcess of substrate), the performance of pyridine is surpassed by L 2 and L 6 (with enhancements which are 3.2 and 1.2 higher), whereas the enhancements obtained for L 4 and L 7 b become comparable to the ones of pyridine. We show that this can be explained by considering the effect of relaxation in these fluorinated ligands, which relax slower,o na verage, than pyridine under the same conditions. We note that, in the case of L 7 ,d eprotonation using Cs 2 CO 3 leads to ar emarkable increase in T 1 of the protonl ocated betweent he Nc enter and the carboxylic group, from 14.5 to 46.1 s. Our work demonstrates that, by varying the ligand loading and p-H 2 concentration,i ti sp ossible to dramatically exceed the previously reported 1 Hh yperpolarization levels. 19 FNMR hyperpolarized spectra were successfully recorded for L 2 , L 4 , L 6 , L 7 a, L 7 b, L 8 , and L 9 ,a nd the average enhancements of the free fluoriner esonance, with thee xception of L 8 and L 9 ,w ere quantified. The highest values were obtained for L 2 and L 7 b (ca. 60-and 39fold enhancements, respectively,a t1 1.74 T), although we note that the enhancement values reported by Shchepin etal. [14] exceedt he ones reported here. However,w es how that, by decreasingthe field strength atwhich experiments are performed and increasing the concentration of p-H 2 ,t he enhancement of the 19 Fr esonance can be significantly improved.
The two orders of magnitude differenceb etween the enhancements obtained for 1 Ha nd 19 Fc an be explained by taking into account the short longitudinal relaxation times of 19 F, which range between 3a nd 5s at 11.74 Tf or all of the ligands investigated here. Although this is apparentlya no bstacle in the way of 19 FS ABREM RI applications, we show that, in some cases, one can circumventt his situationb ye xploiting Furthermore,h yperpolarized 2D 19 FM RI images have been recorded by using as imilar polarization-transfer procedure to that used in NMR spectroscopy experiments. When comparing the resultsw ith their Boltzmann equilibrium counterparts, we find that hyperpolarization facilitates an SNR gain of over one order of magnitude for most of the substrates investigated, particularly L 2 and L 7 b,f or which the maximum SNR gains obtained were approximately 104 and 76, respectively,r epresenting av ery promising for future diagnostic MRI applications.
This work demonstrates the feasibility of using SABRE to hyperpolarize aw ide range of 19 F-substituted ligands, including drugs included on the WHOl ist of essential medicines. We present variousm ethods, through which the polarization of 19 F nuclei can be optimized and we show,u sing NMR and MRI results, that some of the substrates investigated in thisw ork exhibit tremendousp otentialf or future applicationss uch as clinical MRIi nvestigations, molecular imaging, and in vivo pH assessment.

Experimental Section
1 Ha nd 13 CNMR measurements were recorded on aB ruker Avance III series 400 MHz spectrometer,w hereas 19 FNMR measurements as well as T 1 inversion recovery and characterization experiments were performed on aB ruker Avance III series 500 MHz system. Samples concerning L 9 and the T 1 measurements at 1.4 Twere collected by using an Oxford Instruments Pulsar equipped with a 1 H and 19 Fp robe. NMR samples were prepared containing 5mm catalyst precursor in 0.6 mL of [D 4 ]MeOH. NMR measurements were collected using either 4o r20 equivalents of substrate to 5mm of iridium in 0.6 mL MeOD (leading to samples containing one-and 17-fold excesses of ligand relative to iridium, respectively). p-H 2 was prepared by cooling hydrogen gas over charcoal at 20 Kf or all samples, except those of L 9 and measurements made at 1.4 T; these utilized hydrogen gas over charcoal at 77 K. After adding p-H 2 at 3bar pressure, 1 HNMR spectra were recorded by using p/2 excitation pulses immediately after shaking the sample in am agnetic field of 65 G. As imilar procedure was used for the polarization of heteronuclei. For more details about the sample preparation and experimental procedures, please see Section 1o ft he Supporting Information.