Validation of Dexamethasone-Enhanced Continuous-Online Microdialysis for Monitoring Glucose for 10 Days After Brain Injury

Abstract

Traumatic brain injury (TBI) can induce a pathophysiologic state that is worsened by secondary injury. Monitoring brain metabolism with intracranial microdialysis can provide clinical insights to limit secondary injury in the days following TBI. Recent enhancements to microdialysis include the implementation of continuously-operating electrochemical biosensors for monitoring the dialysate sample stream in real time and dexamethasone retrodialysis to mitigate the tissue response to probe insertion. Dexamethasone-enhanced continuous-online microdialysis (Dex-enhanced coMD) records long-lasting declines of glucose after controlled cortical impact in rats and TBI in patients. The present study employs retrodialysis and fluorescence microscopy to investigate the mechanism responsible for the decline of dialysate glucose after injury of the rat cortex. The findings confirm the long-term functionality of Dex-enhanced coMD for monitoring brain glucose after injury, demonstrate that glucose microdialysis is coupled to glucose utilization in the tissues surrounding the probes, and validate the conclusion that aberrant glucose utilization drives the post-injury glucose decline.

Graphical Abstract

INTRODUCTION

Traumatic brain injury (TBI) has a global incidence of 69 million cases per year and represents a major public health crisis in many countries.¹ Causes of TBI include falls, vehicle accidents, sports injuries, military action, blasts, and assaults. Negative TBI outcomes include long-lasting disabilities, a persistent vegetative state, and death.^2,3 The pathophysiology of severe traumatic brain injury includes at-risk tissue or penumbra vulnerable to secondary injury, which can worsen the extent of injury.^4–6 Improved detection and monitoring of secondary injury following TBI will lead to more effective clinical care.

Intracranial microdialysis was first translated to the human brain by Ungerstedt and colleagues^7,8 and is now approved for use in human patients.⁹ Prior efforts with clinical microdialysis have established correlations between brain metabolite levels and outcomes for TBI patients^10–13 suggesting that microdialysis has the potential to diagnose secondary injury. This potential motivates on-going efforts to enhance the technical capability and clinical compatibility of the microdialysis approach.

Recent enhancements include the introduction of continuous-online microdialysis (coMD), in which the dialysate stream is analyzed continuously with electrochemical biosensors mounted in a 3D-printed electrode holder.¹⁴ By eliminating the need to collect, store, and analyze discrete dialysate samples, coMD delivers information in real time and with sufficient temporal resolution to monitor chemical transients associated with spreading depolarization (SD) in the injured brain.^5,15,16 coMD systems have a small physical footprint, operate autonomously for extended periods, and can be placed at a patient’s bedside, all features that are highly desired in the intensive care setting.

Strategies to mitigate the foreign body response to the insertion of microdialysis probes into brain tissue have also advanced.^17,18 Without such mitigation, probe insertion leads to focal ischemia, gliosis, and cell loss in the vicinity of the probe track.¹⁹ Over time, these processes degrade the functionality of microdialysis. ²⁰ Retrodialysis of low concentrations of dexamethasone (Dex), a powerful anti-inflammatory agent, is showing promise as a simple yet effective mitigation strategy.^21–23

Robbins et al used Dex-enhanced microdialysis to monitor K⁺ and glucose in the rat cortex for 10 days after controlled cortical impact (CCI),²⁴ a widely studied animal model of brain injury.^25–27 Consistent with other studies,^23,28 we observed both K⁺ and glucose transients associated with SD in the injured brain. We also observed long-term changes in K⁺ and glucose levels in dialysate. In 80% of injured rats, glucose levels declined to levels too low measure, possibly indicating the onset of metabolic abnormality in the injured tissue. Below, we report that a glucose decline has also been observed during the first use of Dex-enhanced coMD in a small group of TBI patients.

The present study employs retrodialysis and fluorescence microscopy to investigate the mechanism responsible for the decline of dialysate glucose after injury. The findings confirm the long-term functionality of Dex-enhanced coMD for monitoring brain glucose after injury, demonstrate that glucose microdialysis is coupled to glucose utilization in the tissues surrounding the probes, and validate the conclusion that aberrant glucose utilization drives the glucose decline.

RESULTS AND DISCUSSION

Concordance between Dex-enhanced microdialysis after CCI in animals and TBI in patients.

Recently, we used Dex-enhanced microdialysis to monitor K⁺ and glucose from the rat cortex for 10 or more days after CCI.²⁴ There were 4 key findings: 1) K⁺ transients indicative of SD; 2) glucose transients associated with some, but not all, K⁺ transients; 3) long term increases in K⁺ to supra-normal levels; and 4) declines in glucose to sub-normal levels. The glucose decline, which is the main focus of the present study, was observed in 80% of the injured rats. The decline began at various times between two and 11 days after CCI, followed various time courses, and was of sufficient magnitude that glucose levels fell below the detection limit of the glucose biosensor (near 20 μM: see Methods and SI Figure 1). After declining, glucose levels did not recover within the time frame of the experiments.

Figure 1 reports our initial experience of monitoring K⁺ and glucose with bedside Dex-enhanced coMD in a small group of patients with severe TBI. There are several differences between the protocols for microdialysis in animals and patients (see Methods). The probes are different, as the animal probes are not suitable or approved for human use, and vice versa. The procedures for probe placement are different. In animals, we surgically placed the probes adjacent to the CCI site. In patients, we inserted the probes via a cranial bolt, along with a Licox probe for monitoring brain tissue oxygenation (Integra Life Sciences, Princeton, NJ) and a Camino probe for monitoring intracranial pressure (Natus Medical, Pleasanton, CA).

Figure 1.

A sample recording of K+(purple) and glucose (red) obtained by bedside Dex-enhanced coMD in a patient 6 days after TBI. The recordings exhibit K+ transients (circles), a long-term glucose decline (red arrow), and a long-term rise in K+ (purple arrow). Monitoring was interrupted near hour 6 for system maintenance.

Bedside Dex-enhanced coMD in TBI patients reproduced 3 of the 4 key outcomes of the CCI model. Figure 1 shows a clinical recording that exhibits K⁺ transients (highlighted by circles), a long-term increase in K⁺, and a decline in glucose to essentially zero (defined as a glucose level below the detection limit of the biosensor). These recordings were obtained six days after probe insertion. Prior to the sixth day, the K⁺ and glucose levels had been steady and unremarkable. The recording window in Figure 1 contains a 1-hr break, during which regularly-scheduled system maintenance, including removal, calibration, and replacement of the coMD biosensors, was performed. The K⁺ and glucose levels were similar before and after system maintenance, confirming that changes in biosensor performance are not responsible for the rise of the K⁺ signal or decline of the glucose signal.

To date, we have performed bedside Dex-enhanced coMD for 4–6 days in four patients with severe TBI. Microdialysis was continued until the patients were moved from the Intensive Care Unit to a rehabilitation ward, according to routine clinical practice (standard of care). In three patients, glucose levels were initially normal and then declined to near essentially zero. In the fourth patient, glucose levels were persistently low, but medical imaging was unable to confirm placement of the probe in parenchyma. Because of the limited number of cases thus far, we make no attempt to correlate microdialysis findings with patient outcomes (clinical microdialysis is on-hold at this time, due to pandemic-related restrictions). However, the similarities in the outcomes of Dex-enhanced coMD in injured animals and in patients motivates a better understanding of the outcomes from the CCI model.

Experimental design for coMD in rats.

The studies reported below were designed to provide deeper insight into the post-injury glucose decline by means of retrodialysis and histological analyses of the probe tracks. Our objectives were to 1) confirm the functionality of the glucose monitoring system and 2) investigate the mechanism driving the glucose decline after injury.

Figure 2 explains the time-line of our studies. On day 0, experimental rats (n=10) received a CCI injury and a microdialysis probe, while control rats (n=10) received a microdialysis probe without CCI. The probes were perfused continuously for 10 days. Glucose declined to essentially zero by day 5 in some, and by day 7 in all, injured rats but none of the control rats. Retrodialysis procedures were performed in sub-groups of the experimental and control rats on days 1–3 and days 5–10. The following substances were delivered by retrodialysis: a diffusion marker, d-YASFL; glucose; a glycolysis inhibitor, 3-bromopyruvate (3-BP); a fluorescently-tagged analog of 2-deoxyglucose (2-NBDG). Each rat underwent multiple retrodialysis procedures (n values stated in the following sections), but never more than one per day: retrodialysis of 3-BP and of 2-NBDG were performed in separate rats. After microdialysis, brain tissue was collected for immunohistochemistry and fluorescence microscopy.

Figure 2.

Timeline of studies in rats. CCI and probe insertion occurred on day 0. Retrodialysis of Dex occurred on days 0–5. Retrodialysis of glucose occurred on days 1–3 (pre-decline) and 5–8 (post-decline), similarly retrodialysis of d-YASFL occurred on days 1 and 7. Retrodialysis of 3-BP occurred after days 5–6 (post-decline). Retrodialysis of 2-NBDG occurred on days 1 and 10.

Retrodialysis of d-YASFL.

d-YASFL is a non-native pentapeptide used recently as an internal standard for the determination of neuropeptides in brain dialysates (Wilson et al).^29,30 We are aware of only one known peptidase, found in the gut of a marine mollusk, that can hydrolyze a peptide containing a D-amino acid.³¹ Thus, to the best of our knowledge, d-YASFL is neither a substrate for any metabolic enzyme present in the brain nor a ligand for any receptor or transporter. Therefore, it is a suitable solute for examining the effects of diffusion in the dialysis membrane and the surrounding tissue on microdialysis-based measurements.

The concentration of a solute in the dialysate at the outlet of a microdialysis probe may have two contributions, one from solute recovered from the external medium (the probe’s surroundings) and one from solute delivered by retrodialysis but not extracted into the external medium:³²

$${\text{C}}_{\text{out}}=\text{R}\cdot {\text{C}}_{\text{ext}}+(1-\text{E})\cdot {\text{C}}_{\text{in}}$$ Eq 1

where C_out is the solute concentration at the probe outlet, C_in is the solute concentration at the probe inlet (also the retrodialysis concentration), C_ext in the solute concentration in the external medium, R is the recovery fraction, and E is the extraction fraction. Rearranging Eq 1 gives the equation for the concentration-differences plot:

$${\text{C}}_{\text{in}}-{\text{C}}_{\text{out}}=\text{E}\cdot {\text{C}}_{\text{in}}-\text{R}\cdot {\text{C}}_{\text{ext}}$$ Eq 2

which has a slope of E and a y-intercept of - R · C_ext, the negative of the solute concentration recovered in the dialysate when C_in is zero, so-called conventional microdialysis. Equations 1 and 2 apply to the simple case that E is a constant, as expected for a diffusion-controlled solute such as d-YASFL.

Figure 3 reports concentration-differences plots for d-YASFL in injured and control rats on days 1 and 7. The E values were determined by linear regression. There were no significant differences in d-YASFL’s E values among the four experimental conditions. The consistency of d-YASFL’s E values confirms that solute diffusion in the dialysis membrane and surrounding tissues undergo no detectable changes due either to the elapsed time since probe insertion or the presence of CCI injury. Thus, Figure 3 confirms that potential confounders, such as blockage of the pores of the membrane and surrounding tissue, do not contribute to the glucose decline after CCI.

Figure 3.

Concentration-difference plots for d-YASFL in A) control rats (n=3) and B) rats injured by CCI (n=5) on day 1 (in blue) and day 7 (in red). Data points are means ± SEM. There were no significant differences among the slopes of the lines as determined by linear regression (ANOVA, p = .9872).

Retrodialysis of glucose.

Unlike d-YASFL, glucose is a substrate for the enzymes involved in its utilization and is a ligand for glucose transporters.^33–35 Figure 4 compares glucose concentration-differences plots from days 1–3 and days 5–8 after CCI. The plots are nonlinear, indicating that glucose extraction is concentration-dependent, and confirming that glucose microdialysis is not purely diffusion controlled. For this reason, Equations 1 and 2, which are written for the case that E is concentration-independent, do not precisely describe these glucose results. However, linear regression of the plots in the range 0 < C_in < 1250 μM gives r² values close to 1. Thus, as a purely practical matter, we take the slope of this almost-linear region of the plots as a measure of glucose’s effective E value (E_eff). There is a clear and significant increase in glucose’s E_eff between days 1–3 and days 5–8, i.e. before and after the post-injury glucose decline.

Figure 4.

Concentration-differences plots for glucose from rats injured by CCI (n=5) on days 1–3 (in blue) and days 5–8 (in red). Data points presented as mean ± SEM. The inset reports the magnitudes of the slope of the semi-linear portions of the plots: for days 1–3, E_eff is 0.29 ± 0.01, for days 5–8, E_eff is 0.72 ± 0.05 (also mean ± SEM). * indicates p<0.05 in a paired t-test.

Figure 4 confirms that the microdialysis membranes and surrounding tissues maintain their permeability to glucose throughout these experiments. If glucose were, for some reason, unable to diffuse through the membrane or surrounding tissue, extraction would be prevented, and E would be zero.

The increase in glucose’s E_eff after the glucose decline is a potentially significant finding. It reveals an increase in the capacity of the tissue surrounding the probe to remove interstitial glucose. It appears that glucose removal serves to maintain a steep concentration gradient across the dialysis membrane, thereby promoting diffusion of glucose out of the probe. To the best of our knowledge, glucose removal occurs via the blood flow and via glucose utilization by cells for the production of ATP, etc., both of which involve glucose transporters.³⁶ Thus, Figure 4 indicates that the microdialysis of glucose operates under the influence of glucose removal.

The concentration-differences plots for glucose (Figure 4) flatten out at high glucose concentrations, especially on days 5–8. Thus, glucose extraction approaches a maximum, indicating that an enzyme or transporter involved in glucose removal approaches saturation. Jaquins-Gerstl et al²² gave a similar explanation for the nonlinearity of dopamine concentration-differences plots obtained by Dex-enhanced microdialysis. Also, there is a difference in the y-intercepts of the plots, which is non-zero on days 1–3 and zero on days 5–8. This is consistent with Equation 2, which equates the y-intercept with R · C_ext, the concentration of glucose recovered from the brain.

Retrodialysis of a fluorescent analog of 2-DG.

We used retrodialysis of a fluorescent analog of 2-deoxyglucose, 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino)-2-deoxyglucose (2-NBDG), to test the hypothesis that glucose extraction operates under the influence of glucose utilization in the tissue surrounding the probe. Similar to 2-DG itself, 2-NBDG is taken up by cells via glucose transporters, undergoes phosphorylation, and accumulates inside cells.^37,38 2-NBDG is used widely to visualize glucose utilization in astrocytes and neurons in culture and in vivo.^39–41

Figure 5 demonstrates intracellular fluorescence surrounding probe tracks after retrodialysis of 2-NBDG on day 1 and day 10 in control and injured rats (see SI Figure 2 for the quantitative assessment of these images). These images provide unequivocal evidence for the extraction of 2-NBDG from the probes into the surrounding tissue, further confirming their permeability, as well as 2-NBDG uptake by glucose transporters and intracellular accumulation after phosphorylation. Thus, Figure 5 confirms on-going glucose utilization in the vicinity of the probes, independent of the elapsed time since probe implantation and CCI.

Figure 5.

Representative images of intracellular fluorescence after retrodialysis of 2-NBDG in a control rat (A), on day 1 in an injured rat injured (B), and on day 10 in an injured rat (C). The center of each probe track is marked with an asterisk. From left to right, columns show magnifications of 20, 40, 60, and 100×, with scale bars indicating 100 μm for 20× and 40×, and 50 μm for 60× and 100×.

Retrodialysis of a glycolysis inhibitor.

We used retrodialysis of 3-bromopyruvate (3-BP), a well-known inhibitor of glycolysis,⁴² to test the hypothesis that glucose utilization likewise affects glucose recovery. 3-BP targets hexokinase II and has been used widely to study glycolysis in cancer cells.^43–46 We performed retrodialysis of 3-BP (5 mM) for periods of 30 min in injured rats on days 5–9, which was >24 hr after the glucose decline. Calibration studies confirm that the glucose sensors do not detect 3-BP (SI Figure 3). However, calibration studies also show that 3-BP reduces the glucose sensitivity of coMD (<30%): this effect is reversible, as the sensitivity returns to normal 30 min after the termination of 3-BP retrodialysis (SI Figures 4 and 5).

Retrodialysis of 3-BP caused increases in dialysate glucose (Figure 6). When retrodialysis was stopped, glucose levels returned to essentially zero. As reported in Figure 6, we performed multiple 3-BP retrodialysis deliveries in some individual rats. We delivered 3-BP 17 times (n=5 CCI rats) and observed an increase in glucose every time (summarized in SI Figure 6). These data confirm 1) that glucose recovery is utilization-dependent, and 2) that the probes maintain their functionality for glucose recovery after a CCI injury.

Figure 6.

Representative coMD recording of glucose from an injured rat, after the post-injury glucose decline, during 3 consecutive deliveries of 3-BP by retrodialysis.

Immunofluorescence of glucose transporters.

Figures 4–6 strongly suggest that glucose transporters play a role in glucose extraction and recovery. In the brain, glucose transporter 1 is associated with blood vessels, whereas glucose transporter 3 is associated with parenchyma.^35,36,47 Immunofluorescence confirms the presence of glucose transporters 1 and 3 in the vicinity of probe tracks on day 10 after CCI (Figure 7A–D). Quantitative assessment of multiple images (n=9 images from n=3 rats, Figure 7E and 7F) found no significant differences between immunolabeling of regions of interest surrounding the probe tracks and ipsilateral comparative tissue.

Figure 7.

Immunofluorescence of glucose transporters 1 and 3 (red and blue, respectively) in the vicinity of a probe track (left) on day 10 after CCI and comparative tissue (right, B and D). The probe track is in the center of the images and marked with an asterisk (A and C). Scale bars are 200 μm. Quantitative assessment (E and F) of images of probe tracks and ipsilateral comparative tissue, error bars are SEM assessment of n=9 images per group, as detailed in the Methods.

Dex-enhanced coMD of glucose for 10 days in control rats.

Dialysate glucose levels fall to essentially zero by day 7 after CCI. Figure 8 confirms that this event is unique to injured animals and does not occur in rats in the non-injured control group.

Figure 8.

Recordings of glucose by coMD in control rats on day 1 (red) and day 7 (blue) (solid lines are the means, dotted lines are the ±SEMs, n=4).

CONCLUSION

The current study (1) establishes concordance between the outcomes of Dex-enhanced continuous-online microdialysis after CCI in animals and TBI in patients; (2) confirms the functionality of Dex-enhanced glucose microdialysis for 10 days after CCI injury in experimental animals; and (3) demonstrates that Dex-enhanced microdialysis extraction and recovery of glucose are coupled to glucose utilization in the tissue surrounding the probes. From this, we conclude that glucose utilization is involved in the glucose decline after CCI.

We have previously documented the functionality of long-term Dex-enhanced microdialysis. However, some of our prior studies were performed in the rat striatum and focused on monitoring dopamine.^20,22 We monitored K⁺ and glucose in the rat cortex but without CCI.²³ Thus, the use of Dex-enhanced microdialysis for multiple days after injury represents a new paradigm for intracranial monitoring. In Robbins et al.,²⁴ we presented two lines of evidence for the functionality of microdialysis in rats after CCI. First, we reported the continued recovery of K⁺ after the glucose decline. Second, we used mass spectrometry to show the presence of higher molecular weight components in dialysate samples. In retrospect, however, these data do not confirm the functionality of microdialysis specifically for glucose, nor do they provide insight as to the mechanism for the glucose decline after injury.

Our study detected no evidence of microdialysis system failure that might explain the glucose decline after injury. We consistently used newly prepared and calibrated glucose biosensors. Likewise, we consistently monitored the perfusion flow. The decline was specific to injured rats and did not occur in non-injured control rats (Figure 8). The consistency of d-YASFL’s E values (Figure 3) confirms the absence of overt changes in solute diffusion in the membrane or surrounding tissues. We confirmed the glucose permeability of the microdialysis probes and the surrounding tissues (Figure 4). We further confirmed the permeability for K⁺,²⁴ d-YASFL, 3-BP, and 2-NBDG. The results of 3-BP retrodialysis (Figure 6) confirm the ability of the probes to recover glucose from injured tissues up to 9 days after implantation. Collectively, these findings confirm the functionality of the Dex-enhanced coMD for long-term glucose monitoring after CCI.

In addition, our findings confirm that microdialysis is coupled to glucose utilization in brain tissue. Figure 5 shows that 2-NBDG molecules diffused out of the probe and into the surrounding tissues, then entered cells via glucose transporters and accumulated after phosphorylation. Likewise, Figure 6 shows that 3-BP caused glucose to accumulate in the interstitial space and then diffuse into the probe. These processes depend on glucose transporters. Figure 7 documents the presence of glucose transporters 1 and 3 in the vicinity of the probe tracks 10 days after CCI.

Our findings show that the glucose decline following brain injury involves glucose transporters and utilization in the tissues surrounding the probe. Glucose utilization acts to lower the glucose concentration in the interstitial spaces surrounding the probe. During glucose retrodialysis (Figure 4), this maintains a steep concentration gradient between the probe and surrounding tissue, which promotes glucose extraction. But, during conventional microdialysis, this diminishes the concentration gradient that drives diffusion into the probe, which lowers glucose recovery. Thus, we conclude that the increase in glucose extraction and decrease in glucose recovery after CCI are both due to glucose utilization.

Although our data establish that glucose microdialysis is coupled to glucose utilization, tissue glucose levels reflect the balance between utilization and vascular delivery.⁴⁸ Thus, our data show that glucose utilization increases after injury relative to glucose delivery, which leaves open the possibility that ischemia or hypoglycemia may be contributing factors in the post-injury glucose decline.⁴⁹ Based on its design, however, our study is not able to delineate all these possible contributing factors.

This study was motivated by the concordance between the outcome of Dex-enhanced microdialysis in animals and patients (Figure 1). In both cases, dialysate glucose levels decline to essentially zero after injury. This appears to be a novel finding since, as far as we are aware, similar observations have not been reported before. There are relatively few prior microdialysis studies in the CCI animal model.⁵⁰ Hashemi et al⁵¹ reported declines in glucose associated with SD clusters in patients: the present study contains no clear evidence for a role of SD clusters. Vespa and colleagues used microdialysis to monitor glucose for 10 days in TBI patients and reported glucose declines in the event of terminal herniation,⁹ which did not occur during the present studies. Our findings confirm that the post-CCI glucose decline is a sign of a metabolic abnormality in injured tissue, although it remains to be ascertained if the same conclusion applies to the outcome of clinical microdialysis.

We were intrigued to find that cells in injured tissues near the probe tracks utilize molecules delivered via the probe itself (Figures 4 and 5). This might have therapeutic implications, although it remains to be seen if it will be possible to reestablish the balance between glucose supply and demand in injured tissues. To be clear, we doubt that retrodialysis would be effective for this, as 2-NBDG only labelled cells within about 200 μm of the probe, which is a very small region of tissue.

Our preliminary experience with bedside Dex-enhanced coMD in a small group of TBI patients reproduced 3 of the 4 key findings from the CCI model. Thus far, we have not observed glucose transients during SD, possibly because the probes, being inserted via cranial bolts, may not have accessed the pericontusional space: furthermore, cortical electrophysiology to detect SD was not performed here. However, glucose transients have been observed by clinical coMD with surgical placement of the probes into the pericontusional space.²⁸ Overall, however, our study validates the conclusion that the long-term glucose decline detected by Dex-enhanced coMD after brain injury indicates a metabolic abnormality, possibly a form of metabolic crisis in injured brain tissue.

METHODS

Bedside Dex-enhanced clinical microdialysis.

Bedside Dex-enhanced microdialysis was conducted in patients with approval of the Institutional Review Board of the University of Pittsburgh. Patients admitted to UPMC Presbyterian Hospital with severe TBI and who received intracranial pressure and brain tissue oxygen catheters as a standard of care were screened for participation. Inclusion criteria included ages of 18–70 years and an initial Glasgow Coma Scale score of 3–10, motor score not following commands. Written, informed proxy consent was obtained for placement of a microdialysis catheter (70 Microdialysis Bolt Catheter, M Dialysis AB, Johannesshov, Sweden) through an open port of the cranial bolt. These catheters have a polyamide membrane with a molecular weight cut-off of 20 kDa and an active sampling length of 10 mm. The catheters were primed with CNS perfusion fluid (M Dialysis AB) premixed by a hospital pharmacist with dexamethasone (10 μM) before sterile catheter insertion. The catheters were perfused at 2 μL/min with a 107 Microdialysis Pump (M Dialysis AB). Monitoring of K⁺ and glucose in the dialysate stream was performed by continuous-online microdialysis (coMD), as described below.

Dex-enhanced microdialysis in rats.

All procedures involving animals were approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Studies were performed in male Sprague-Dawley rats (Charles River, Raleigh, NC) weighing 250–350g. Concentric-style microdialysis probes were constructed in-house with a 4 mm active sampling length using cellulose membranes with 13 kDa molecular weight cutoff (Spectra/Por). The inlet and outlet lines were fused silica (75 μm ID, 150 μm OD). Prior to use the probes were flushed with ethanol followed by sterile filtered (Nalgene sterile filters, 0.2 μm pore size, Fisher, Pittsburgh, PA) artificial cerebrospinal fluid (aCSF; 142 mM NaCl, 1.2 mM CaCl₂, 2.7 mM KCl, 1.0 mM MgCl₂, and 2.0 mM NaH₂PO₄ (all salts from Sigma-Aldrich) adjusted to a pH of 7.4). In all procedures, including those involving retrodialysis, the probes were perfused at 1.67 μL/min with a syringe pump (Harvard Apparatus). Probes were inserted at an angle of 51° from vertical, to ensure that the entire active length was within the cortex. The probes were perfused with aCSF containing 10-μM Dex for the first 24 hrs, 2-μM Dex for the next 96 hrs, and no Dex thereafter.

Controlled cortical impact.

A controlled cortical impact (CCI) device (Leica Impact One, Leica Biosystems, Buffalo Grove, IL)) was used to model a moderate to severe TBI. The CCI site was accessed via a craniotomy posterior to bregma on the right hemisphere. The exposed dura was impacted to a depth of 2.4 mm, at an average velocity of 3.99 m/s ± .03, with a 100 ms dwell time. After CCI, a microdialysis probe was inserted 3 mm anterior to the CCI site. Control rats received neither a CCI nor a sham craniotomy.

Glucose detection.

To maintain consistency with our prior work, glucose extraction measurements were performed with the rapid-sampling microdialysis (rsMD) system described previously.⁵² All other glucose measurements were performed with the continuous online microdialysis (coMD) system recently developed by Boutelle and coworkers (SI Figure 1 presents example calibration data for both systems).⁵³ Briefly, needle-style microelectrodes for K⁺ and glucose were interfaced to the outlet line of the microdialysis probe with a 3D-printed microelectrode holder (SomosWaterShed XC 11122). As the glucose sensors are hand-made inhouse, there is variability in their performance. Only those sensors meeting practical performance metrics were used in these studies. The sensors used herein were those exhibiting a linear response (r²>0.99) over the concentration range of 50–1000 μM glucose with a sensitivity exceeding 0.5 pA/μM. In the case of microdialysis in animals, the fused silica outlet line was attached to the microelectrode holder with a 5-cm section of polyurethane tubing (0.18 mm ID × 0.36 mm OD: Instech Laboratories Inc.) to provide oxygen for glucose sensing. The clinical microdialysis catheters are constructed with polyurethane lines that were connected directly to the microelectrode holder.

Retrodialysis and analysis of d-YASFL.

A 1 mM stock solution of the pentapeptide, d-YASFL (each of the five amino acids in “d-YASFL” are D-stereoisomers, GL Biochem, Shanghai, China), was prepared by diluting the solid in purified water (Millipore Milli-Q Synthesis A10, city, state). The stock solution was stored frozen at −20 °C. Solutions for retrodialysis were prepared through serial dilution of the stock solution in aCSF containing 2-uM Dex (on days 1–3) to final concentrations of 100, 250, and 400 nM. These solutions were filtered using 0.2 μm PES syringe filters (Corning Inc., Corning, NY). Retrodialysis was performed in random order. Solutions were perfused through the microdialysis probe for 30 minutes each: after a 10-min wait period, 20-min dialysate samples were collected at each d-YASFL concentration and stored at −20 °C. Then, 20-μL aliquots of the retrodialysis and dialysate solutions were individually diluted to a total volume of 200 μL with Optima LC-MS grade water (Fisher Chemical) in autosampler vials (MS Certified Autosampler Vials with 350 μL inserts, Thermo Scientific, Rockwood, TN) and placed in a refrigerated autosampler at 5 °C (Thermo Fisher Dionex UltiMate 3000, Thermo Scientific). The peptide labeling method, along with the HPLC and MS instrumental details, were as described in detail by Wilson et al., with the following minor modification. A sample desalting step was added by incorporating a precolumn (30 × 0.10 mm, Acquity 5 μm BEH C18, Waters, Milford, MA).

Retrodialysis of glucose.

Retrodialysis solutions with glucose concentrations of 0, 100, 500, 1250, and 2500 μM were prepared in aCSF with concentrations of Dex to match the day of the measurement (2 μM on days 1–3 and 0 μM on days 5–10). The rsMD glucose system was calibrated with each retrodialysis solution in descending order prior to each 30–60 min extraction measurement.

Retrodialysis of 2-NBDG.

2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose (2-NBDG; ab146200, Abcam, Cambridge, MA, Abcam) was dissolved in aCSF to a final concentration of 4 mM. Retrodialysis was performed for 4 hr and followed by a washout with aCSF for an additional 4 hr. Afterwards, rats were anesthetized with isoflurane (5% by volume) and transcardially perfused with 200 mL of phosphate-buffered saline (PBS; 155 mM NaCl, 0.01 M, pH 7.4) followed with 160 mL of 4% paraformaldehyde. The brain and probe were removed and further fixed in 4% paraformaldehyde for 24 hr at 4 °C. The tissue containing the probe track was cut to 35-μm sections in the plane perpendicular to the probe insertion. Sections were imaged with a FluoView 1000 (Olympus, Inc., Tokyo, Japan) at varying magnification.

Retrodialysis of 3-BP.

3-Bromopyruvate (3-BP) was dissolved in aCSF to a concentration of 5 mM. 3-BP was delivered by retrodialysis for 30 min.

Immunohistochemistry and fluorescence microscopy of glucose transporters.

After microdialysis, rats were deeply anesthetized with isoflurane (5%) and perfused transcardially with cold PBS followed by cold 4% paraformaldehyde. Rats were decapitated, and the skull and brain were removed and soaked in cold 4% paraformaldehyde for at least 24 hours. Brain tissues were sectioned to a thickness of 35 μm in the plane perpendicular to the probes using a cryostat. Tissue sections were rehydrated with 5 min washes with PBS, repeated twice, then incubated with Triton X-100 in PBS for 15 min, washed in 0.5 % bovine serum albumin, and soaked in a blocking solution of 2% bovine serum albumin and 2% goat serum for 45 min. Sections were then incubated in primary antibody solution containing 5% goat serum, 0.1% triton X-100, and antibodies for glucose transporters 1 and 3 (1:100 AbCAM) for 18 h at 4 °C. Sections were next washed with PBS (3 × 5 min) and incubated 5% goat serum, 0.1% triton X-100, and secondary antibodies (1:500 goat anti-mouse Alexa 488, Invitrogen, and 1:500 goat anti-rabbit Alexa 568, Invitrogen, Carlsbad CA) for 2 h at room temperature. Sections were washed with PBS (4 × 5 min) and cover-slipped with Fluoromount-G (Southern Biotech, Birmingham AL). Fluorescence microscopy was performed with a 20× objective (Olympus BX61, Olympus; Melville, NY) and Metamorph/Fluor 7.1 software (Universal Imaging Corporation; Molecular Devices).

Data and Statistical Analysis

Data were collected on LabChart 7 and LabChart 8 (AD Instruments), then exported and analyzed in MATLAB R2019a software. Statistical analyses were carried out with GraphPad Prism 9.

Thresholding Images.

We used ImageJ⁵⁴ software to analyze images stained for glucose transporters 1 and 3. Images were opened with ImageJ then channels were spilt into monochrome color (black and white). Threshold limits were applied to each image, which were then converted into a binary image and analyzed with the histogram function. The histogram function consisted of 2 numbers: the total number of black pixels (background) and the total number of white pixels (fluorescent). The white pixels were converted to the percent of fluorescent pixels and graphed in Excel.

Abbreviations

TBI

traumatic brain injury

CCI

controlled cortical impact

SD

spreading depolarization

Dex

dexamethasone

rsMD

rapid sampling microdialysis

coMD

continuous online microdialysis

d-YASFL

a non-native pentapeptide (dYdAdSdFdL)

2-DG

2-deoxyglucose

2-NBDG

2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino)-2-deoxyglucose

3-BP

3-bromopyruvate

LC

liquid chromatography

HPLC

high performance liquid chromatography

MS

mass spectrometry

aCSF

artificial cerebrospinal fluid

PBS

phosphate-buffered saline

C_ext

external concentration

C_out

outlet concentration

C_in

inlet concentration and retrodialysis concentration

E

extraction fraction

E_eff

effective extraction fraction

R

recovery fraction