Changes in Oxygen Delivery during experimental models of Cerebral Malaria

Cerebral malaria (CM) is a severe manifestation of malaria that commonly occurs in children and is hallmarked by neurologic symptoms and signi�cant Plasmodium falciparum parasitemia. It is currently hypothesized that cerebral hypoperfusion from impaired microvascular oxygen transport secondary to parasitic occlusion of the microvasculature is responsible for cerebral ischemia and thus disease severity. Animal models to study CM, are known as experimental cerebral malaria (ECM), and include the C57BL/6J infected with Plasmodium berghei ANKA (PbA), which is ECM-susceptible, and BALB/c infected with PbA, which is ECM-resistant. Here we sought to investigate whether changes in oxygen (O 2 ) delivery, O 2 �ux, and O 2 utilization are altered in both these models of ECM using phosphorescence quenching microscopy (PQM) and direct measurement of microvascular hemodynamics using the cranial window preparation. Animal groups used for investigation consisted of ECM-susceptible C57BL/6 (Infected, n = 14) and ECM-resistant BALB/c (Infected, n = 9) mice. Uninfected C57BL/6 (n = 6) and BALB/c (n = 6) mice were included as uninfected controls. Control animals were manipulated in the exact same way as the infected mice (except for the infection itself). C57BL/6 ECM animals at day 6 of infection were divided into two cohorts: Early-stage ECM, presenting mild to moderate drops in body temperature (> 34 < 36°C) and Late-stage ECM, showing marked drops in body temperature (< 33°C). Data were analyzed using a general linear mixed model. We constructed three general linear mixed models, one for total O 2 content, another for total O 2 delivery, and the third for total O 2 content as a function of convective �ow. We found that in both the ECM-susceptible C57BL/6J model and ECM-resistant BALB/c model of CM, convective and diffusive O 2 �ux along with pial hemodynamics are impaired. We further show that concomitant changes in p50 (oxygen partial pressure for 50% hemoglobin saturation), only 5 mmHg in the case of late-stage CM C57BL/6J mice, and O 2 diffusion result in insu�cient O 2 transport by the pial microcirculation, and that both these changes are required for late-stage disease. In summary, we found impaired O 2 transport and O 2 a�nity in late-stage ECM, but only the former in either early-stage ECM and ECM-resistant strains.


Introduction
Cerebral malaria (CM) is one manifestation of malaria in the pediatric populace in areas where the parasite is endemic 1 . Often the diagnosis is clinical and involves deep coma and signi cant P. falciparum parasitemia. Symptoms accompanying CM include headache, stiff neck, drowsiness, disorientation, and seizures. CM also often presents with several metabolic acidosis, accompanied by abnormal respiratory patterns.
Morbidity and mortality are high with 11% experiencing neurologic sequelae and between 13 and 25% of cases leading to death [1][2][3] . The exact pathology of CM is unknown, though postmortem CM studies have revealed that parasitized red blood cells (pRBCs), particularly sequestrations of trophozoites and schizonts, adhere to the capillary endothelium thereby resulting in obstruction of the pial microvasculature 4,5 . Many pathophysiological aspects of CM are still not completely understood. Possible contributing mechanisms involve microcirculatory dysfunction, vasoconstriction, hemolysis, and reduced deformability of red blood cell (RBC). These mechanisms are not mutually exclusive, but they all result in impaired blood ow. It is hypothesized that these microvascular obstructions in conjunction with anemia result in depressed cerebral oxygenation and subsequently cerebral ischemia, coma, and death, if left untreated 4,5 . Clinically cerebral hypoperfusion and ischemia are assessed indirectly by elevated cerebrospinal uid lactate and retinal hypoperfusion 6-8 . Susceptible murine models of experimental cerebral malaria (ECM) by Plasmodium berghei ANKA (PbA) have further supported the hypothesis of cerebral hypoperfusion and ischemia by demonstrated increased expression of hypoxia-inducible factor-1α (HIF-1α), and positive staining by the hypoxia probe (pimonidazole) 9 . However, these studies have not yet directly measured oxygen (O 2 ) delivery, O 2 uxes, and O 2 consumption in these models.
Beyond microvascular occlusion, O 2 transport is mediated by blood pH (Bohr effect), hemoglobin (Hb) concentration (anemia), and core body temperature, all of which are altered systemically during malarial infection. Particularly in CM, intra-erythrocytic parasitic digestion of Hb into hemozoin further contributes to anemia. We have previously reported oxygen tension in the cerebral pial circulation of two strains of mice, namely the C57BL/6J, which is ECM-susceptible, and BALB/c, which is ECM-resistant 9 . Speci cally, we found impaired pial microvascular hemodynamics and depressed oxygen tension from phosphorescence quenching microscopy (PQM) [9][10][11][12][13][14] in the ECM-susceptible C57BL/6J strain compared to the BALB/c ECMresistance strain 15

Methods
Closed cranial window animal preparation. Animal handling and care followed the NIH Guide for Care and Use of Laboratory Animals. All protocols were approved by the La Jolla Bioengineering Institutional Animal Care and Use Committee. 8 to 10-week old C57Bl/6J and Balb/cJ mice (Jackson Laboratories, ME) were implanted with a closed cranial window model as described elsewhere 16 . Brie y, mice were anesthetized with ketamine-xylazine and were administered dexamethasone (0.2mg/Kg), carprofen (5mg/Kg) and ampicillin (6mg/kg) subcutaneously, to prevent post-surgical swelling of the brain, in ammatory response, and infection. After shaving the head and cleansing with ethanol 70% and betadine, the mouse was placed on a stereotaxic frame and the head immobilized using ear bars. The scalp was removed with sterilized surgical instruments and lidocaine-epinephrine was applied on the periosteum, which was then retracted to expose the skull. A 3-4mm diameter skull opening was made in the left parietal bone using a surgical drill. Under a drop of saline, the craniotomy was lifted away from the skull with very thin-tip forceps and gelfoam previously soaked in saline was applied to the dura mater to stop any eventual small bleeding. The exposed area was covered with a 5mm glass cover slip secured with cyanocrylate-based glue and dental acrylic.
Carprofen and ampicillin were given daily for 3-5 days after recovery from surgery. Mice presenting signs of pain or discomfort were euthanized with 100mg/kg of euthasol IP. Two to three weeks after surgery, mice ful lling the inclusion criteria (see below) were inoculated with P. berghei ANKA and, on day 6 of infection, they were lightly anesthetized with iso urane (4% for induction, 1-2% for maintenance) and held on a stereotaxic frame for measurements of pH and pO 2 .
Inclusion criteria. Animals were suitable for the experiments if: 1) animal behavior was normal and 2) microscopic (x350 magni cation) examination of the cranial window did not reveal signs of edema or bleeding.
Parasite infection. Animals were inoculated with an IP injection of 1 x 106 Plasmodium berghei ANKA parasites expressing the green uorescent protein (PbA-GFP, a donation from the Malaria Research and Reference Reagent Resource Center -MR4, Manassas, VA; deposited by CJ Janse and AP Waters; MR4 number: MRA-865). Parasitemia, body weight, rectal temperature and clinical status (using six simple tests adapted from the SHIRPA protocol, as previously described) were monitored daily from day 4 of the infection 17 . Parasitemia was checked using ow cytometry by detecting the number of uorescent GFPexpressing pRBCs in relation to 10,000 RBCs. ECM was diagnosed when one or more of the following clinical signs of neurological involvement were observed: ataxia, limb paralysis, poor righting re ex, seizures, rollover, or coma.
Physiological ranges of the variables measured for the animal species used. Two groups of animals, C57BL/6 (n = 6) and BALB/c (n = 6), instrumented with the closed cranial window were used to characterize normal microhemodynamic (vessel diameter and blood ow), intravascular and perivascular PO2s and pH in the pial microenvironment.
Experimental Groups. Group 1 aimed to establish the effects of PbA infection in microhemodynamics, intravascular and perivascular PO2s and pH in the pial microenvironment. The group consisted of ECMsusceptible C57BL/6 (Infected, n = 14) and ECM-resistant BALB/c (Infected, n = 9) mice. Uninfected C57BL/6 (n = 6) and BALB/c (n = 6) mice were included as controls. Control animals were manipulated in the exact same way as the infected mice (except for the infection itself). C57BL/6 ECM animals at day 6 of infection were divided in two cohorts: Early-stage ECM, presenting mild to moderate drops in body temperature (> 34 < 36°C) and Late-stage ECM, showing marked drops in body temperature (< 33°C). Another group, Group 2, was included to establish the relation between vascular in ammation resulting from PbA infection and microhemodynamics and oxygenation in relation to ECM pathophysiological changes. The group consisted of C57BL/6 (Infected, n = 9) mice to which leukocyte adhesion, blood ow and PO2 levels were measured.
Similarly, as in Group 1, the ECM animals at day 6 of infection were divided in two cohorts: Early-stage ECM and Late-stage ECM. All experiments were repeated at least once.
Experimental Setup. Animals were lightly anesthetized with iso urane (4% for induction, 1-2% for maintenance). They were secured to the microscopic stage of an intravital microscope (BX51WI, Olympus, New Hyde Park, NY) on a stereotaxic frame with the head gently held with ear bars for epi-illumination imaging. Body temperature, measured pre-anesthesia, was maintained with a heating pad. The tissue image was projected onto a charge-coupled device camera (COHU 4815) connected to a videocassette recorder and viewed on a monitor. Measurements were carried out using a 40X (LUMPFL-WIR, numerical aperture 0.8, Olympus) water immersion objective. The animals did not recover from anesthesia, as they were euthanized (Euthasol 100mg/kg, IP) right after the intravital microscopy measurements.
Microhemodynamics. A video image-shearing method was used to measure vessel diameter (D) 18 . Changes in arteriolar and venular diameter from baseline were used as indicators of a change in vascular tone. Arteriolar and venular centerline velocities were measured on-line using the photodiode cross-correlation method (Photo Diode/Velocity Tracker Model 102B, Vista Electronics, San Diego, CA). The measured centerline velocity (V) was corrected according to vessel size to obtain the mean RBC velocity 19,20 . Blood ow (Q) was calculated from the measured values as Q = π × V (D/2)^2. This calculation assumes a parabolic velocity pro le and has been found to be applicable to tubes of 15-80 µm internal diameters and for Hcts in the range of 6-60% 19,20 .
Microvascular PO2 distribution. High resolution non-invasive microvascular pO 2 measurements were made using phosphorescence quenching microscopy (PQM) 21 . PQM is based on the relationship between the decay rate of excited Palladium-mesotetra-(4-carboxyphenyl) porphyrin (Frontier Scienti c Porphyrin Products, Logan, UT) bound to albumin and the O 2 concentration according to the Stern-Volmer Eq. 2 1,22 . The method was used previously in microcirculatory studies to determine pO 2 levels in different tissues 22 . pO 2 .
measurements by PQM were obtained following these steps for all groups: 1) the probe was injected (tail injection of 15 mg/kg at a concentration of 10 mg/ml of the phosphorescence complex 10 min before O 2 measurements); 2) the tissue was illuminated (pulsed light at 420 nm wavelength) to excite the probe into its triplet state; 3) the emitted phosphorescence (680 nm wavelength) was collected and analyzed to yield the phosphorescence lifetime; and 4) the phosphorescence lifetime was converted into O 2 concentration, pO 2 .The phosphorescence lifetimes are concentration independent, which permit extravascular uid pO 2 measurements, although the dye albumin complex that extravasates is very small. Extravascular uid pO 2 was measured in regions in between functional capillaries. PQM allows for precise localization of the pO 2 measurements without subjecting the tissue to injury. These measurements provide a detailed understanding of microvascular O 2 distribution and indicate whether O 2 is delivered to the interstitial areas.
Hematocrit and hemoglobin. Blood was collected from the tail in heparinized glass capillaries. Hemoglobin was determined spectrophotometrically from a single drop of blood in a B-Hemoglobin analyzer (Hemocue, Stockholm, Sweden). Hematocrit was estimated by centrifugation.
Oxygen delivery and extraction. Oxygen delivery, DO 2 was approximated as 1 where RBC Hb is the total Hb (g/dL), is the O 2 carrying capacity of saturated Hb, approximated as 1.34 mL O 2 /g Hb, S A is arteriolar blood oxygen saturation, and Q A is arteriolar ow. Similarly, arterio-venous oxygen extraction (O 2 A-V Extraction, or VO 2 ) was approximated as 2 where S A−V is the difference between arteriolar and venular oxygen saturation and Q A−V is the average of arteriolar and venular ow rate. O2 saturations were approximated using the blood O 2 equilibrium curve.

Results
Systemic Blood Gas Parameters. No signi cant differences were observed in hematocrit, arterial blood pH, blood oxygen a nity (p50), and body temperature between C57BL/6J and BALB/c mice at baseline, though were altered with infection at day 6. These results along with detailed statistical analysis are shown in Table 1. diffusive ux, which is the inverse of the slope of the log-log plots shown in Fig. 1, to re ect the absolute difference in total O 2 content between groups, and to represent whether infection alters radial diffusive ux. In arterioles from the C57BL/6J strain (Fig. 1A), we observed a signi cant depression in total oxygen content (p = 1.09x10 − 8 ) and radial diffusive ux (p = 1.0x10 − 15 , interaction p = 0.00135) in both CM groups compared to uninfected controls. We observed a similar result in venules from this strain ( p = 1.0x10 − 15 , p = 1.0x10 − 15 , interaction p = 1.0x10 − 15 ). For the ECM-resistant BALBc strain, in arterioles (Fig. 1C), we also observed a depressed total oxygen content (p = 1.0x10 − 15 ) and depressed radial diffusive ux (p = 1.0x10 − 15 , interaction p = 1.0x10 − 15 ) between the CM group and uninfected controls. We observed similar results for venules ( p = 1.0x10 − 15 , p = 1.0x10 − 15 , interaction p = 8.55x10 − 6 ). Log-log plots for this analysis are shown in Fig. 1D.  Fig. 2A  kinetics. Although there is some evidence of autonomic nervous system dysfunction in malaria 23 , there is no evidence that neural control of vascular tone leads to impaired tissue perfusion. In CM, there seems to be a normal response of the cerebral resistance vessels (small arteries and arterioles) to changes in arterial O 2 and CO 2 24 . Speci cally, we observed that despite similar changes in microvascular diffusive and convective ux in early and late CM models, disease severity was different, likely due to the observed difference in p50.
Our results (Table 1) provide further insight the role of hypothermia in the pathogenesis of CM. Whether hypothermia induced by ECM is a protective mechanism or a strategy to preserve tissue function is still unknown. Hypothermia is known to decrease the metabolic rate of brain tissue, oxygen consumption 9 , and the growth rate of the malaria parasite 25 . Furthermore, our previous studies have demonstrated that hypoperfusion and subsequent ischemia is key to CM pathogenesis 9 . Although cerebral blood ow in CM is within the normal range, it is low in comparison with the arterial O 2 content. Cerebral vascular resistance is increased, whereas cerebral O 2 extraction is diminished. Yet, our results here suggest that concomitant microvascular hypoperfusion and decreased Hb-O 2 a nity are required for severe disease. In this context, our results suggest that hypothermia is a compensatory mechanism to increase Hb-O 2 a nity and is thus bene cial. Importantly, however, this analysis fails to account for the effect of the parasite on temperature regulatory centers in the brain. Thus, future studies are required for investigation of this hypothesis and speci cally the role of hypothermia in CM outcomes. Funding. This work was supported by National Institutes of Health (NIH) grants R01HL162120 and R01HL159862.
Availability of Data and Materials. The raw data and any materials supporting the conclusions of this article will be made available by the authors, without undue reservation, to any quali ed researcher. Venules from BALB/c ESM-resistant mice. Linear regression lines are shown, with the inverse of the slope corresponding to diffusive ux. Data were analyzed using a general linear mixed model.