Hyperpolarized ¹³C magnetic resonance spectroscopy detects toxin-induced neuroinflammation in mice

Abstract

Lipopolysaccharide (LPS) is a commonly used agent for induction of neuroinflammation in preclinical studies. Upon injection, LPS causes activation of microglia and astrocytes, whose metabolism alters to favor glycolysis. Assessing in vivo neuroinflammation and its modulation following therapy remains challenging, and new non-invasive methods allowing for longitudinal monitoring would be greatly valuable. Hyperpolarized (HP) ¹³C magnetic resonance spectroscopy (MRS) is a promising technique for assessing in vivo metabolism. In addition to applications in oncology, the most commonly used probe of [1-¹³C] pyruvate has shown potential in assessing neuroinflammation-linked metabolism in mouse models of multiple sclerosis and traumatic brain injury. Here, we wished to investigate LPS-induced neuroinflammatory changes using HP [1-¹³C] pyruvate and HP ¹³C urea.

Graphical Abstract

The lipopolysaccharide toxin was injected intracranially into the brains of mice. Hyperpolarized ¹³C MRS data were acquired at 3 Tesla at baseline, 3 days and 7 days after injection. Increased lactate production from hyperpolarized pyruvate was seen at 7 days after injection, along with increased numbers of resting/activated microglia and resting/reactive astrocytes.

2D chemical shift imaging following simultaneous intravenous injection of HP [1-¹³C] pyruvate and HP ¹³C urea was performed at baseline (day 0), day 3 and day 7 post intracranial injection of LPS (n=6) or saline (n=5). Immunofluorescence (IF) analyses were performed for Iba1 (resting and activated microglia/macrophages), GFAP (resting and reactive astrocytes) and CD68 (activated microglia/macrophages).

A significant increase in HP [1-¹³C] lactate production was observed in the injected (ipsilateral) side at 3 and 7 days of the LPS-treated mouse brain, but not in either the contralateral side or saline-injected animals. HP ¹³C lactate/pyruvate ratio, without and with normalization to urea, was also significantly increased in the ipsilateral LPS-injected brain at 7 days compared to baseline. IF analyses showed a significant increase in CD68 and GFAP at 3 days, followed by increased numbers of Iba1 and GFAP positive cells at 7 days post-LPS injection.

In conclusion, we can detect LPS-induced changes in the mouse brain using HP ¹³C MRS, in alignment with increased numbers of microglia/macrophages and astrocytes. This study demonstrates that HP ¹³C spectroscopy holds much potential for providing non-invasive information on neuroinflammation.

Introduction

Lipopolysaccharide (LPS) is a bacterial endotoxin commonly used to induce neuroinflammation and generate preclinical animal models of the inflammatory response alone¹, or to help model complex neurological disorders such as Alzheimer’s disease². Upon in vivo injection (intravenous³, intraperitoneal⁴ or intracranial⁵), LPS causes an inflammatory response, with release of pro-inflammatory cytokines such as tumor necrosis factor alpha, following immune cell recognition of LPS by their cell surface receptors (specifically toll-like receptor 4, TLR4)⁶. Activation of microglia similarly occurs via TLR-4 recognition in Alzheimer’s disease, in that case due to presence of the aberrant protein amyloid-β⁷. Studies have shown that intracranial injection of LPS in the mouse brain leads to an increased number and activation of microglia^2,8 as well as increased astrogliosis^9,10. These cellular traits have been observed in the brains of patients with neurodegenerative diseases such as Parkinson’s disease and Alzheimer’s¹¹. In Parkinson’s disease for example, increased cytokine levels and activation of microglia and astrocytes (as induced in models using LPS) have been associated with the disease’s characteristic loss of dopaminergic neurons^12–14.

Following LPS-induced activation, both activated microglia and astrocytes demonstrate metabolic reprogramming, in particular a shift towards increased glycolysis. This has been shown in BV-2 mouse microglia by Voloboueva et al.¹⁵, where LPS induced an increase in lactate production and a decrease mitochondrial oxygen consumption and ATP production, assessed by both biochemical assays and a Seahorse extracellular flux analyzer. Further, Klimaszewska-Lata et al.¹⁶ showed pyruvate dehydrogenase inhibition in the N9 mouse microglial cell line following LPS treatment. Primary astrocytes in culture were similarly activated (by a combinatory LPS/interferon-gamma treatment) by Bal-Price et al.¹⁷ and were shown to produce more lactic acid, as measured by levels in the growth medium.

With the knowledge that neuroinflammation is associated with changes in metabolism of immune and glial cells, one could use non-invasive, clinically-translatable metabolic imaging approaches to enable longitudinal monitoring of neuroinflammatory status, as well as response to anti-inflammatory therapies. Magnetic resonance (MR) techniques include ¹H MR spectroscopy (MRS), and studies using this methodology have suggested that certain ¹H-visible metabolites, such as myoinositol, choline, and total creatine could be linked to inflammation, for example in multiple sclerosis patients¹⁸. Increased lactate as detected by ¹H MRS has also been reported in patients with traumatic brain injury¹⁹, and preclinical work by Lodygensky et al²⁰. showed increased lactate in LPS-injected rat pups. However, ¹H MRS is to date only used in a narrow set of clinical pathologies, at those centers with appropriate expertise to interpret the data. Although optimization of 1H MRS for assessment of neuroinflammation is ongoing²¹, this methodology is not yet standard clinical procedure, and alternative strategies are needed.

¹³C MRS holds great promise for assessing in vivo metabolism. It allows identification of steady-state metabolism in the brain after intravenous administration of ¹³C-labeled substrates over >1h²², and has been applied in the healthy mouse²³, rat²⁴ and human brain²⁵, the healthy aging brain²⁶ and AD²⁷. However, these methods require long infusions of substrate and extended scan times, which limits clinical translation. Hyperpolarized (HP) ¹³C MRS²⁸ is a rapidly expanding alternative imaging technology for visualizing in vivo metabolism^29,30. Acquisition occurs over a matter of minutes, thanks to the 10,000-fold increase in sensitivity over thermal ¹³C MRS. Thus far, HP ¹³C MRS has been particularly informative on the conversion of HP [1-¹³C] pyruvate to [1-¹³C] lactate via the enzyme lactate dehydrogenase, and there has been a focus on applications in cancer^31,32 and cardiovascular disease³³. Applications of HP ¹³C MRS continue to expand, with a wide range of studies including investigations into bacteria³⁴, healthy³⁵ and diseased^36,37 liver, kidneys³⁸, skeletal muscle³⁹, and brain^40,41. Alongside analysis of enzymatic fluxes, HP probes can be used to assess perfusion. For this purpose, metabolically inactive probes such as HP ¹³C urea can be injected, as demonstrated in tumor imaging by von Morze et al.⁴². Further, if co-polarized and administered with HP ¹³C pyruvate, the combination of HP ¹³C pyruvate and HP ¹³C urea can provide a simultaneous readout of metabolism and perfusion, as shown by Lau et al.⁴³ in the rat heart.

Recently, HP ¹³C MRS has been demonstrated to be applicable to the detection of neuroinflammation in models of multiple sclerosis (MS)⁴⁴ and traumatic brain injury (TBI)^45,46, and more broadly in models of inflammation in the liver⁴⁷ and arthritic limb⁴⁸; all studies have taken advantage of the inflammation-related increase in production of HP lactate. In a mouse model of MS, HP [1-¹³C] lactate was increased in the corpus callosum, and was associated with a significant increase in activated microglia in that region. This increase was then no longer observed in transgenic mice with a deficiency in their ability to activate microglia⁴⁴. HP ¹³C MRS has been applied to both a rat⁴⁶ and mouse⁴⁵ model of TBI. Both studies demonstrated an increase in HP ¹³C lactate/pyruvate ratio following injury, and in mice, the depletion of microglia prevented this increase from occurring.

In this study, we hypothesized that HP ¹³C MRS could be used to visualize the effect of intracranially-administered LPS in the in vivo mouse brain. Given the reports that LPS increases glycolysis in brain microglia and astrocytes, we hypothesized that the induced increase in lactate production could be measured by MRS following administration of HP [1-¹³C] pyruvate. We also co-administered HP ¹³C urea in order to control for any changes in perfusion that occurred as a result of the injection, and carried out our study at the clinically-relevant field strength of 3 Tesla.

Our results showed that, following intracranial injection of LPS in the mouse brain, HP [1-¹³C] lactate levels and corresponding HP ¹³C lactate/pyruvate ratios were significantly increased at 3 and 7 days post-injection in the ipsilateral (injected) voxel. Moreover, upon normalization to contralateral side, to HP ¹³C urea, or both, increased HP [1-¹³C] lactate levels and increased HP ¹³C lactate/pyruvate ratios were systematically observed at the 7 day timepoint in the ipsilateral side of LPS-injected animals. Importantly, the 7 day timepoint corresponded to the maximum levels of Iba1 (resting and activated microglia/macrophages), and glial fibrillary acidic protein GFAP (resting and reactive astrocytes), as detected by immunostaining. Overall, our results show that ¹³C MRS of HP [1-¹³C] pyruvate and ¹³C urea can successfully visualize the effect of LPS on the mouse brain.

Materials and Methods

Animals

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of California, San Francisco.

A total of 26 mice were used in this study. Eleven mice (male C57BL/6J, 10–12 weeks old, Jackson Laboratories) underwent the imaging protocol described below prior to surgery (Baseline group). These animals were then injected intracranially with either LPS (n=6) or saline (n=5), and the imaging protocol was repeated at 3 and 7 days post-surgery. Separate groups of animals were euthanized at the experimental timepoints (n=5 baseline, n=6 at 3 days, n=6 at 7 days) for histological tissue analysis.

Intracranial injections

Mice were anesthetized with isoflurane in oxygen (isoflurane: 3% induction, 2% maintenance), and fur shaved from the top of the head. After placement in a stereotactic frame (Stoelting, IL, USA) and on a heating pad, a small incision was made in the skin to expose the skull. Following orientation using the bregma and lambda to ensure the head was flat, a small hole was made in the skull (1.55mm to right and 1mm front of the bregma, depth of 2.8mm), and 5μl of either saline or 1μg/ml LPS in saline (0.005mg; from E.coli O111:B4, Sigma Aldrich) injected automatically into the striatum using a Hamilton syringe (Figure 1A). The syringe was then removed and the wound sutured closed. On removal from the anesthesia, animals were ambulatory after 5–10 minutes.

Figure 1.

(A) MRS grid used for acquisition of HP ¹³C MRSI data overlaid on a T₂-weighted MR image, and voxels of interest post-voxel shift used for analysis (C=contralateral; I=ipsilateral). Stack of dynamic HP ¹³C spectra (grey), with summed spectrum at the rear (black), showing the resonances of HP [1-¹³C] pyruvate, HP [1-¹³C] lactate and HP ¹³C urea. (B) T₂-weighted MR images and summed HP ¹³C spectra from ipsilateral voxel at each timepoint, for both LPS (red) and saline (blue) animals.

MR acquisitions

MR acquisitions were performed on a 3 Tesla horizontal MR system (Bruker Biospec) with a dual-tuned ¹H/¹³C mouse head volume coil (2cm diameter, Doty Scientific, South Carolina). Animals were anesthetized with isoflurane (2% in O₂) and a 27G tail vein catheter placed before the animal was positioned in a home-built cradle inside the MR system. T₂-weighted images of the brain were acquired for co-registration of the spectral voxels (field of view (FOV)=20×20mm, matrix 192×192, number of averages (NA)=4, repetition time (TR)=2s, echo time (TE)=60ms, 9 slices of thickness 1mm; total acquisition time 10min 8s). 24μl [1-¹³C] pyruvate sample (pyruvic acid, 15mM OX63 trityl radical (Oxford Instruments), and 1.5mM Gad-DOTA) and 55μl ¹³C urea (6.4M urea in glycerol, with 23mM OX63 trityl radical) were co-polarized for ~1 hour in a Hypersense polarizer (Oxford Instruments, 3.35T), and rapidly dissolved in 4.5ml heated buffer (80mM NaOH in Tris with EDTA) to give a pH7 solution of 80mM pyruvate and 78mM urea. Immediately following dissolution, 300μl of this HP solution was injected via the tail vein catheter over 14 seconds. Data were acquired from the start of intravenous injection every 4.2s using a dynamic 2D chemical shift imaging (CSI) sequence (flip angle = 10°, slice thickness 5mm, FOV=24×24mm, matrix 8×8, TR=66.4ms, TE=1.24ms; total acquisition time 1min 25s).

Histological analysis

Animals were euthanized using an overdose of ketamine/xylazine, and then perfused with ice-cold phosphate-buffered saline solution (0.9%), followed by an ice-cold paraformaldehyde (PFA, 4%) solution. The brains were removed and submerged in PFA for 2 hours, before being transferred to a sucrose gradient (5% for 2 hours, 10% for 2 hours, 20% overnight). Brains were then frozen in liquid nitrogen and stored at −80°C. Cryosections (10μm) from the imaging voxels were obtained using a microtome (Leica Biosystems, Germany). Immunofluorescence staining was carried out using the following antibodies: a primary rabbit anti-GFAP (resting and reactive astrocytes, 1:500 dilution, Z0334, Dako), and a primary rabbit anti-Iba1 (resting and activated microglia/macrophages, 1:500 dilution, 019–19741, Wako), both with secondary antibodies goat anti-rabbit fluorescein isothiocyanate (1:500 dilution, 111-096-144, Jackson Immunoresearch Lab), and a primary rat anti-mouse CD68 antibody (activated microglia/macrophages, 1:200 dilution, MCA1957, Biorad) with secondary goat anti-rat Alexa-Fluor 555 (1:200 dilution, A21434, Invitrogen). Slides were counterstained using Hoechst 33342 (H3570, 1:2000 dilution; Invitrogen), then sections were mounted using Prolong Gold Antifade (P36930; Invitrogen).

Fluorescence image acquisition was performed using an inverted microscope (Ti, Nikon). The images were recorded with an Andor Zyla 5.5 sCMOS camera at 20x magnification. Quantitative analyses of immunofluorescence images from the striatum posterior to the injection site within the MRS imaging voxel were performed using NIH ImageJ (v2.0.0). Quantification was executed based on coverage, and expressed as percentage of the total area.

Analysis of hyperpolarized ¹³C data

For each animal at each timepoint, HP ¹³C 2D CSI data were analyzed using in-house MATLAB code. First, voxel shifts were applied to the raw data to generate data from ipsilateral (injection site) and contralateral voxels (symmetrical from midline, Figure 1A). Spatial voxel shifting was achieved by applying a linear phase shift to the k-space data prior to Fourier transform. Dynamic spectra over time (shown in grey) were summed to generate a summed spectrum (shown in black, Figure 1A). HP [1-¹³C] pyruvate, HP [1-¹³C] lactate, and HP ¹³C urea levels were estimated using a Lorentzian fit. HP ¹³C lactate/pyruvate and HP ¹³C pyruvate/urea ratios were calculated, and subsequently HP ¹³C lactate/(pyruvate/urea) (Figure S1). Heatmaps of individual metabolites SNR (HP [1-¹³C] pyruvate, HP [1-¹³C] lactate, and HP ¹³C urea) were generated at each timepoint using in-house MATLAB code and displayed using SIVIC (Figure 2). In Figure 3, all values (individual metabolites and all ratios) have been normalized to data from the contralateral side of the brain to eliminate intra-animal variability.

Figure 2:

Heat maps showing HP [1-¹³C] lactate, HP [1-¹³C] pyruvate, and HP [1-¹³C] urea SNR following injection of HP [1-¹³C] pyruvate in an LPS-injected animal. Baseline, 3 day and 7 day data are shown.

Figure 3:

All data are shown for baseline (●), and at 3 days (■) and 7 days (▲) after surgery, in LPS and saline-treated groups. Further, all data were normalized to the contralateral side of the brain. (A) Data over time for individual metabolites assessed following injection of hyperpolarized [1-¹³C] pyruvate. HP [1-¹³C] lactate levels were significantly increased in LPS-treated animals at 7 days when compared to baseline levels and 3 day data (7 days: 164 ± 19% of baseline, p=0.0.0007; 136 ± 19% of 3 day data, p=0.018). (B) Ratios for ¹³C lactate/pyruvate, ¹³C pyruvate/urea and ¹³C lactate normalized to ¹³C pyruvate/urea. HP ¹³C lactate/pyruvate ratios were significantly increased in the LPS brains at 7 days when compared to baseline (150 ± 28% of baseline, p=0.0097). HP ¹³C lactate/(pyruvate/urea) ratios were significantly increased at 7 days compared to both baseline and 3 day data (7 days: 169 ± 23% of baseline p=0.026, 151 ± 23% of 3 day data, p=0.019).

*p<0.05, **p<0.01, ***p<0.001

Statistical analyses

For all CSI data, a repeated measures one-way analysis of variance (ANOVA) with a Tukey multiple comparisons test was used to establish differences within treatment groups over time, for each side of the brain. Histological data were analyzed using two-way ANOVAs with Tukey multiple comparisons tests to establish significant differences between timepoints and sides of the brain. Graphs show standard deviation, and statistical significance was considered when p≤0.05 (denoted *). Further significances are denoted **p≤0.01, ***p≤0.001, and ****p≤0.0001.

Results

Figure 1B shows summed HP ¹³C spectra at each timepoint of interest (baseline, 3 days and 7 days) for an LPS (red) and a saline-treated (blue) mouse. The resonances of HP ¹³C urea (163ppm), HP [1-¹³C] pyruvate (171ppm) and its metabolic product HP [1-¹³C] lactate (183ppm) were detected at all experimental timepoints, with increased lactate visible at 7 days in the LPS-treated animal. Figure 2 shows heatmaps of the individual metabolites of HP [1-¹³C] lactate, HP [1-¹³C] pyruvate and HP [1-¹³C] urea from an LPS-injected animal at each timepoint of interest (baseline, 3 days and 7 days).

HP [1-¹³C] lactate levels, HP ¹³C lactate/pyruvate ratio and HP ¹³C lactate/(pyruvate/urea) ratio are increased in the ipsilateral voxel following LPS injection

On full quantification, our results show that HP [1-¹³C] lactate levels were significantly increased in LPS-treated animals in the injected ipsilateral voxel at both 3 days and 7 days when compared to baseline levels (Figure S1A; 3 days: 175 ± 33% of baseline levels, p=0.042; 7 days:195 ± 32% of baseline levels, p=0.036). In contrast, no changes in HP [1-¹³C] lactate levels were observed in the contralateral voxel. In the saline-injected group, HP [1-¹³C] lactate levels remained unchanged in both contralateral and ipsilateral voxels throughout the experimental period. When looking at HP ¹³C urea or HP [1-¹³C] pyruvate levels, no significant changes in the levels of either metabolite were detected over time in either saline- or LPS-injected groups in either contralateral or ipsilateral voxels (Figure S1A).

As shown in Figure S1B, HP ¹³C lactate/pyruvate ratios were significantly increased in the ipsilateral side of the LPS brains at 3 days and 7 days when compared to baseline (3 days : 165 ± 28% of baseline, p=0.046; 7 days: 263 ± 16% of baseline, p=0.001). This ratio was also significantly increased between 3 days and 7 days (Figure S1B, 7 days: 159 ± 27% of 3 day data, p=0.043). Further, HP ¹³C lactate/pyruvate ratios were significantly increased in the contralateral side of the LPS brains at 7 days compared to baseline data (Figure S1B, 167 ± 33% of baseline, p=0.004). In saline-injected animals, no significant changes in HP ¹³C lactate/pyruvate ratios were detected in either ipsilateral or contralateral voxels at any time points.

No significant changes were observed in HP ¹³C pyruvate/urea ratios either in the LPS or saline-treated animals (Figure S1B).

Finally, our results show that the HP ¹³C lactate/(pyruvate/urea) ratios were significantly increased at 7 days compared to baseline in the ipsilateral side of the LPS-injected animals (Figure S1B, 347 ± 30% of baseline, p=0.012). No significant changes in this ratio could be observed in either the contralateral side of LPS-injected animals, or the ipsilateral and contralateral voxels in the saline-injected animals.

HP [1-¹³C] lactate levels, HP ¹³C lactate/pyruvate ratio and HP ¹³C lactate/(pyruvate/urea) ratio are increased following LPS injection, upon normalization to contralateral data

Following normalization to contralateral data, our results show that HP [1-¹³C] lactate levels were significantly increased in LPS-treated animals at 7 days when compared to baseline levels and levels at 3 days (Figure 3A: 7 days: 164 ± 19% of baseline, p=0.0007; 136 ± 19% of 3 days, p=0.018). In the saline-injected group, HP [1-¹³C] lactate levels remained unchanged throughout the experimental period. When looking at HP ¹³C urea or HP [1-¹³C] pyruvate levels, no significant changes in the levels of either metabolite were detected over time in either saline- or LPS-injected groups (Figure 3A).

As shown in Figure 3B, HP ¹³C lactate/pyruvate ratios were significantly increased in the LPS brains at 7 days when compared to baseline (7 days: 150 ± 28% of baseline, p=0.0097). In saline-injected animals, no significant changes in HP ¹³C lactate/pyruvate ratios were detected at any time point. Figure 3B also shows that we observed no significant changes in HP ¹³C pyruvate/urea ratios in either LPS or saline-treated animals.

Finally, our results show that the HP ¹³C lactate/(pyruvate/urea) ratios were significantly increased at 7 days compared to both baseline and 3 day data (Figure 3B, 7 days: 169 ± 23% of baseline p=0.026, 151 ± 23% of 3 day data, p=0.019). No significant changes in this ratio could be observed in the saline-injected animals.

Iba1, CD68 and GFAP levels are modulated following LPS injection

Figure 4 shows the results for Iba1 (resting and activated microglia/macrophages), CD68 (activated microglia/macrophages), and GFAP (astrogliosis) immunostaining performed in the ipsilateral (I, large square) and the contralateral side (C, small insert) on LPS-injected animals at baseline, 3 days and 7 days. All values for immunostaining from saline-injected animals were below 0.16% coverage (data not shown).

Figure 4:

Histological analysis of ipsi- and contra- lateral brain slices at each timepoint in the LPS-treated animals. Panels show Iba1, CD68 and GFAP staining examples (contralateral inset), with quantification of percentage coverage. (A) Iba1 staining was significantly increased in the ipsilateral side of LPS-treated animals at 7 days compared to baseline and 3 day levels, and at the 7 day timepoint compared to contralateral (7 days: 1096% of baseline, p<0.0001; 363% of 3 day data, p=0.0006, 835% of contralateral p=0.0001). (B) GFAP staining was significantly increased in the ipsilateral side of LPS-treated animals at 7 days compared to ipsilateral baseline and 3 day levels (7 days: 48550% of baseline, p<0.0001, 254% of 3 days, p=0.0002). Ipsilateral GFAP staining was also significantly higher at 3 days compared to baseline data (3 days: 19130% of baseline, p=0.022 (C) CD68 staining was elevated in the ipsilateral side at 3 days post-injection compared to the ipsilateral side at baseline and 7 days (3 day data: 1752% of baseline p=0.0008, 557% of 7 days p=0.003). At 3 days, CD68 was also significantly increased in the ipsilateral side as compared to contralateral (2769% of contralateral, p=0.005).

*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

As shown in Figure 4A, Iba1 staining was significantly increased in the ipsilateral side of LPS-treated animals at 7 days compared to baseline and 3 day levels, indicating a large number of microglia/macrophages present at that time point (7 days: 1096% of baseline, p<0.0001; 363% of 3 day data, p=0.0006). At the 7 day timepoint, Iba1 staining was also significantly increased in the ipsilateral side as compared to contralateral (835% of contralateral p=0.0001). No significant differences between ipsilateral and contralateral sides were observed at baseline or 3 days.

As depicted in Figure 4B, GFAP staining was significantly increased in the ipsilateral side of LPS-treated animals at 7 days compared to ipsilateral baseline and 3 day levels (7 days: 48550% of baseline, p<0.0001, 254% of 3 days, p=0.0002). Ipsilateral GFAP staining was also significantly higher at 3 days as compared to baseline data (3 days: 19130% of baseline, p=0.022). Overall we observed a maximum level of astrogliosis at the 7 day timepoint. When comparing brain sides, our results show that GFAP levels were significantly higher in the ipsilateral voxel than in the contralateral voxel at both 3 days and 7 days (3 days ipsilateral: 1733% of contralateral, p<0.0001; 7 days ipsilateral: 2209% of contralateral, p=0.0003). Figure 4C shows that CD68 staining was elevated in the ipsilateral side at 3 days post-injection compared to the ipsilateral side at baseline and 7 days, indicating that monocyte activation reached a maximum level at that time point (3 day data: 1752% of baseline p=0.0008, 557% of 7 days p=0.003). At the 3 day timepoint, CD68 staining was also significantly increased in the ipsilateral side as compared to contralateral (2769% of contralateral, p=0.005). No significant differences between ipsilateral and contralateral sides were observed at baseline or 7 days.

Discussion

In this study, we demonstrated our ability to image altered brain metabolism following intracranial injection of LPS. Metabolism was assessed using ¹³C MRS following an intravenous injection of HP [1-¹³C] pyruvate and HP ¹³C urea at the clinicially-relevant field strength of 3 Tesla.

Overall, we observed significant changes over time in the detected HP [1-¹³C] lactate, HP ¹³C lactate/pyruvate and HP ¹³C lactate/(pyruvate/urea) only in the LPS-treated group and not in the saline-injected group. This result is indicative of the fact that the metabolic changes observed at 7 days by HP ¹³C MRS were linked to the presence of LPS rather than being a side effect of the surgical procedure.

We did not observe any significant changes in either HP [1-¹³C] pyruvate or HP ¹³C urea at any timepoint at the group level, which indicates that the in situ delivery of HP substrates is not significantly different between groups. Recent work by Miller et al. has questioned our understanding of cerebral HP signal, and the authors demonstrated a vast increase in metabolic signal when the blood-brain barrier (BBB) is opened by mannitol⁴⁹, allowing more HP [1-¹³C] pyruvate to enter the brain. The literature discussing BBB leakiness as a result of LPS administration is inconclusive, with a review by Varatharaj et al. observing BBB disruption in only 60% of studies considered⁵⁰. Nevertheless, although we have not carried out contrast-enhanced imaging in this study, the fact that we did not observe any significant changes in urea or pyruvate between control and LPS animals suggests that the BBB was not significantly affected by LPS injections⁵¹.

HP [1-¹³C] lactate alone was significantly increased at 7 days and 3 days compared to baseline in the ipsilateral side of the LPS group. On normalization to HP [1-¹³C] pyruvate, which is typically performed to account for variability in pyruvate injection³¹, the change at 7 days was exacerbated (p=0.001 compared to p=0.036). Not only was ipsilateral HP ¹³C lactate/pyruvate significantly increased at 7 days and 3 days compared to baseline, but further, 7 days was increased compared to 3 days. On normalization to HP ¹³C urea, the 3 day increase in HP ¹³C lactate/pyruvate was no longer significant compared to baseline data. This result may indicate that, although no significant changes were measured when looking at HP ¹³C urea or [1-¹³C] pyruvate/urea data in group analyses, animal specific changes in delivery due to experimental technical variabilities (e.g. injection rate, or percentage polarization of the pyruvate) may have contributed to the increase HP ¹³C lactate/pyruvate ratio observed at the 3 day timepoint. However, at the 7 day timepoint, the observed significant increase in HP [1-¹³C] lactate production remained following normalization, indicating that the increased HP [1-¹³C] pyruvate to HP [1-¹³C] lactate conversion was robust enough to be detected by ¹³C MRS at 3 Tesla.

HP [1-¹³C] pyruvate to HP [1-¹³C] lactate conversion was not expected to be altered in the contralateral side of the brain. We observed no significant changes except for a significantly increased HP ¹³C lactate/pyruvate at 7 days compared to baseline. This unforeseen increase may be due to partial volume effects, as the HP ¹³C lactate/pyruvate ratio is dramatically increased in the adjacent ipsilateral side at that timepoint (263% increase), or experimental variability between injections. On normalization to HP ¹³C urea, HP ¹³C lactate/(pyruvate/urea) ratio was no longer different between baseline and 7 days, which demonstrates the value of co-injecting HP ¹³C urea to specifically elucidate metabolic changes, independent of experimental variation in intravenous injections.

To provide an internal normalization, and account for biological variability between animals, in vivo data were next normalized to the contralateral side of the brain. Once again, significant differences were only observed for HP [1-¹³C] lactate, HP ¹³C lactate/pyruvate and HP ¹³C lactate/(pyruvate/urea) data in the LPS-treated group, not in saline-injected animals, indicating the surgical procedure did not affect brain metabolism. Specifically, upon normalization to contralateral data, differences in HP [1-¹³C] lactate, HP ¹³C lactate/pyruvate and HP ¹³C lactate/(pyruvate/urea) between 7 day and baseline data remained highly significant. Differences in HP [1-¹³C] lactate and HP ¹³C lactate/pyruvate between 3 day and baseline data were no longer observed. For HP [1-¹³C] lactate and HP ¹³C lactate/(pyruvate/urea) parameters, data were significantly different between 3 and 7 days. Overall, data normalized to contralateral brain strengthen the results from the ipsilateral analyses, confirming that HP [1-¹³C] pyruvate to HP [1-¹³C] lactate conversion was increased independently of delivery at 7 days compared to baseline.

To confirm the inflammatory response induced by LPS, we carried out histological analysis of animals at each timepoint studied by HP ¹³C MRS. Iba1 and GFAP staining confirmed a substantial increase in microglia/macrophages and astrocytes at 7 days following LPS injection, respectively (compared to both 7 day contralateral data and baseline/3 day ipsilateral data). This result was in line with several previous studies^{1,2,8,9,52,53}; Go et al⁸ demonstrated a significant increase in microglia at 7 days following an intrahippocampal LPS injection, and Sharma et al⁵³ showed a similar increase, this time following an intracerebroventricular injection, and additionally observed an increase in GFAP-positive astrocytes. Future studies could clarify the homogeneity and extent of the LPS response throughout the brain relative to the injection site. However, such experiments are beyond the scope of the current project, which focuses on comparing ipsi vs. contralateral voxel for validation of the imaging methods.

When considering the histology alongside our in vivo data, our results show that the time of maximum HP ¹³C lactate increase as observed by MRS (7 days), coincided with the time of maximal Iba1 and GFAP staining. An increase in number or a change in activation status of these cell types may contribute to the increased HP [1-¹³C] pyruvate to [1-¹³C] lactate conversion.

Increased numbers of microglia/macrophages and astrocytes in the ipsilateral LPS-injected brain may be responsible for this increased conversion, given the 1096% and 48550% increase in cell types (respectively), compared to baseline values. Resting microglia express the necessary genes for glycolysis⁵⁴, resting macrophages use both glycolysis and oxidative phosphorylation to produce ATP⁵⁵, and resting astrocytes have been shown to produce lactate in culture⁵⁶. Therefore the increase in number alone may be sufficient to be responsible for the increased MRS signal, irrespective of activation status.

Activation of these glial cells may also play a role. CD68, one of the most commonly-used markers for microglial activation, showed a significant but small increase in the ipsilateral side of the LPS-treated brain at 3 days compared to baseline and 7 days. Interestingly, this timepoint of maximum activation does not coincide with the timepoint of maximum HP [1-¹³C] lactate signal. However, it is recognized that assessing microglial/macrophage activation is not trivial, and likely necessitates measurement of a wide range of markers; such biological characterization is largely beyond the scope of this manuscript. Investigation of the association between CD68 levels and glycolysis would be required to conclude further on the contribution of CD68-activated microglia to the detected in vivo HP signal. When considering the contribution of astrocytes, maximum astrogliosis was observed at 7 days (the timepoint of maximum HP [1-¹³C] lactate). Studies have reported that GFAP levels as measured by IF are linked to reactive astrocytes^57–59. It is thus plausible that astrogliosis as observed at 7 days may contribute to the increased HP ¹³C pyruvate to ¹³C lactate conversion.

Considering activation and cell number, our data suggest that the large increase in number of both cell types (as assessed by percentage coverage in IF) is likely the driving factor for the increased HP ¹³C lactate. However, one cannot distinguish the relative contribution of each cell type to the detected HP [1-¹³C] lactate signal. Studies using modulations of microglia or astrocytes levels could help elucidate cell-specific contributions to the HP signal. Guglielmetti et al.⁴⁴ showed increased mononuclear phagocytes in the brain alongside increased HP ¹³C lactate/pyruvate in a multiple sclerosis model, and in a traumatic brain injury model⁴⁵ demonstrated that increased HP ¹³C lactate/pyruvate following injury was no longer observed on microglial depletion. Similarly, Lewis et al.⁶⁰ showed increased lactate production in the heart following a myocardial infarction, which was normalized following monocyte/macrophage depletion. Future bioreactor studies⁶¹ of the individual cell types could also be valuable to decipher the contributions of specific cell types.

Only a few HP studies have been carried out on a preclinical, murine-dedicated 3 Tesla system^62,63. This study contributes to the validation of HP acquisitions at this clinically relevant field strength, while taking advantage of the improved gradient strength leading to higher spatial resolution compared to a clinical 3 Tesla. We opted for a 2D CSI dynamic acquisition instead of a single timepoint to remove potential bias from variability in inter-user injection procedures, which might affect the kinetics of HP [1-¹³C] pyruvate and HP [1-¹³C] lactate detected by MRS. In future studies, calculations of rates of conversion (k_pl) could be carried out, providing sufficient SNR is achieved, such as in work by Park et al. in the non-human primate brain⁶⁴.

Conclusion

In conclusion, we have shown that we can successfully detect increased HP [1-¹³C] lactate production in vivo following LPS injection into the mouse brain, coinciding with an increased number of microglia/macrophages and astrocytes as visualized by histology. Many diseases have inflammatory components, and HP ¹³C MRS could be a valuable additional tool with which to assess inflammatory status non-invasively and longitudinally.

Supplementary Material

Supp FigS1

Figure S1: All data are shown for both ipsi- and contra-lateral sides of the brain, in LPS and saline-treated groups, at baseline (●), and at 3 days (■) and 7 days (▲) after surgery. (A) Data over time for individual metabolites assessed following injection of HP [1-¹³C] pyruvate. HP [1-¹³C] lactate levels were significantly increased in LPS-treated animals in the injected ipsilateral voxel at both 3 days and 7 days when compared to baseline levels (3 days: 175 ± 33% of baseline levels, p=0.042; 7 days:195 ± 32% of baseline levels, p=0.036) (B) ¹³C Lactate/pyruvate, ¹³C pyruvate/urea, and ¹³C lactate normalized to ¹³C pyruvate/urea ratios. HP ¹³C lactate/pyruvate ratios were significantly increased in the ipsilateral side of the LPS brains at 3 days and 7 days when compared to baseline (3 days : 165 ± 28% of baseline, p=0.046; 7 days: 263 ± 16% of baseline, p=0.001). This ratio was also significantly increased between 3 days and 7 days (7 days: 159 ± 27% of 3 day data, p=0.043). Further, ¹³C HP lactate/pyruvate ratios were significantly increased in the contralateral side of the LPS brains at 7 days compared to baseline data (167 ± 33% of baseline, p=0.004). HP ¹³C lactate/(pyruvate/urea) ratios were significantly increased at 7 days compared to baseline in the ipsilateral side of the LPS-injected animals (347 ± 30% of baseline, p=0.012). *p<0.05, **p<0.01

Abbreviations:

LPS

lipopolysaccharide

HP

hyperpolarized

IF

immunofluorescence

Iba 1

ionized calcium binding adaptor molecule 1

GFAP

glial fibrillary acidic protein

CD68

Cluster of Differentiation 68

MS

multiple sclerosis

TBI

traumatic brain injury

NA

number of averages

CSI

chemical shift imaging

PFA

paraformaldehyde