Introduction

The kidney is the body’s chief “blood oxygen content sensor”. This contributes to its roles as a regulator of integrated physiologic responses to anemia1 and the major source of systemic erythropoietin (EPO) production.2,3,4 Nevertheless, our understanding of the mechanisms by which it achieves this function remains incomplete. We, and others, have provided evidence that the kidney accurately senses blood oxygen content (CaO2) and converts this signal into a detectable level of microvascular kidney partial pressure of oxygen (PkO2).2,5,6,7 This mechanism allows cells within the kidney to respond to a specific microvascular PkO2 as renal and systemic oxygen delivery change under varying conditions. In addition, this mechanism allows local cells to respond and alter expression of hypoxia-inducible genes, including those that synthesize EPO.3,4 Furthermore, this detection of PkO2 may also trigger afferent signals, via the renal nerves, to initiate additional adaptive integrative physiologic changes that maintain overall oxygen homeostasis.1 One of the important concepts regarding tissue oxygen sensing is that, while oxygen is not the most soluble of biological gases, it nevertheless travels rapidly down concentration gradients across tissues and into cells.8,9 As such, oxygen is not impacted by anatomical divisions within organs; it does not differentiate between renal cortex or medulla, other than by local aspects of metabolic supply (perfusion) and demand (cellular metabolic requirement) and by vascular anatomy.10,11,12

The anatomy and physiology of the kidney reflect its multiple functions including waste removal, fluid and electrolyte homeostasis, and erythrogenesis. To perform these functions, the kidney receives 25% of the cardiac output and consumes about 20% of global oxygen delivery. The functional unit of the kidney is the nephron, largely defined by the filtering glomeruli, the reabsorbing and secreting tubules and vascularization by a complex blood supply. Active resorption of vital nutrients, including sodium, is a function of the cortex (proximal convoluted tubule) and outer medulla (thick ascending loop of Henle). Water resorption is largely a function of the strong sodium concentration gradient established in the medulla. Among the multiple endocrine functions of the kidney, erythrogenesis is regulated by hypoxic induction of EPO production and secretion from cells strategically located near the cortical medullary junction.13,14

In the current study, we utilized a dual light wavelength approach to measuring superficial cortical (blue light) and deeper renal cortical and medullary PkO2 (red light) under different conditions of renal oxygen delivery, using phosphorescence quenching. We hypothesized that the physiology of renal oxygenation is set up in such a way that PkO2 is dependent on CaO2 and that both superficial and deeper renal PkO2 would be different in response to changes in CaO2 induced by hyperoxia (which increases CaO2) and hemodilutional anemia (which decreases CaO2).

Methods

Animal model

All experimental protocols were approved by the Animal Care and Use Committee at St. Michael’s Hospital (Toronto, ON, Canada) and conducted in accordance with the Canadian Council on Animal Care and ARRIVE-2 guidelines. Male Sprague Dawley rats (Charles River) were utilized for the experiments (n = 6; body weight = 400–500 g). Animals were housed in groups of two in standard research cages with species-appropriate enrichment. They had access to food and water ad libitum in a pathogen-free facility with a 12:12 hr light-dark cycle. After arrival to the research facility, the rats were allowed a one-week acclimation period before experiments were performed.

In spontaneously breathing rats, anesthesia was induced within an inhalation chamber with 4% isoflurane. Anesthesia was then maintained via tracheostomy with 1.5–2.0% isoflurane, initially in 21% O2 at a total flow rate of 2 L·min−1 The tail vein and tail artery were cannulated to allow infusion of the hemodilution solutions, measurement of mean arterial pressure (MAP), and drawing of arterial blood for blood oximetry and chemistry (ABL 800 Flex, Radiometer Canada, London, ON, Canada). Heart rate (HR) was measured with electrocardiogram electrodes and the rectal temperature was monitored. A thermoregulatory heating pad was used to maintain rectal temperatures near 37 °C. A computerized data-acquisition system (PowerLab AD Instruments Inc., Colorado Springs, CO, USA) was used to continually monitor MAP, HR, and rectal temperature.

Hyperoxia protocol

Rats underwent spontaneous ventilation at three levels of fractional inspired oxygen (FIO2): 21% (baseline), 30%, and 50%, for 20 min each. At each level of inspired oxygen, three one-minute periods of measurement of microvascular PkO2 were performed in sequence utilizing: 1) red wavelength light-emitting diode (LED, λmax = 635 nm); 2) blue wavelength LED (λmax = 450 nm) and finally; 3) red laser light (λmax = 635 nm), according to the methods outlined below. Arterial blood gases (ABGs) were measured at the end of each period prior to changing the FIO2. At the end of this experiment, the FIO2 was returned to baseline (21%) and animals were allowed to re-equilibrate for 30 min before initiating the hemodilution protocol.

Hemodilution protocol

After re-establishing baseline conditions, a baseline ABG was taken prior to the first stage of hemodilution. Rats underwent hemodilution by replacing 25% of the estimated blood volume with hydroxyethyl starch 6% (Voluven 130/0.4, Fresnius Kabi, Toronto, ON, Canada). Hemodilution was performed with a push-pull infusion pump (PHD2000, Harvard Apparatus, St. Laurent, QC, Canada) to maintain a steady hemodilution rate over the ten-minute period. All fluids were warmed to 37 °C before administration. After hemodilution, arterial blood was collected for analysis of oximetry and chemistry (ABG, radiometer) into heparinized syringes. Following hemodilution, microvascular PkO2 was measured by alternating sequentially to three different light sources in the following order: red LED then blue LED then red laser. Arterial blood oximetry and chemistry were assessed as described for the hyperoxia protocol. Subsequently, a second 25% hemodilution was performed (total 50%) and then PkO2 and arterial blood oximetry and chemistry were re-assessed. At the end of each experiment, animals were euthanized by isoflurane overdose and intravascular injection of T-61 euthanasia solution.

Microvascular kidney oxygen measurements (PkO2)

Microvascular kidney PkO2 was measured in the left kidney using the phosphorescence quenching method. The intravascular oxygen probe Oxyphor PdG4 was deployed in combination with a time-domain OxyLED phosphorometer (Oxygen Enterprises Ltd., Philadelphia, PA, USA).8,15 We utilized an oximeter (OxyLED) capable of generating light at 450 nm (blue) and 635 nm (red) to excite the Soret and the Q-bands of the tetrabenzoporphyrin at the core of the probe, respectively. A common light detector was utilized to detect emitted phosphorescence following probe excitation with either blue or red light. This methodology was initially developed by Rumsey et al.7 and adapted by Johannes et al.5 to utilize excitation of phosphorescence of PdG2 (spectroscopically identical to PdG4)15 using blue and red light to discriminate between the PkO2 measurements in the superficial renal cortex (blue light) and deeper renal cortical and medullary tissue (red light) (Fig. 1). The left kidney was exposed via a dorsal flank incision and the subjacent soft tissue and peri-renal fat were retracted. The kidney was kept within the retroperitoneal space and in the body cavity to maintain the renal temperature comparable to the core temperature. The animal was kept in a hooded experimental chamber to prevent ambient light from interfering with the measurements.

Fig. 1
figure 1

A graphic representation of the method of oxygen measurement by quenching of phosphorescence of an intravascular oxyphor. The depth of penetration of blue and red light within the kidney shows the expected paths of oxyphor excitation in the superficial renal cortex (blue light) and cortex and outer medullary regions (red light).

Full size image

Light excitation and detection

Excitation light was delivered from three different sources: 1) a red LED (λmax=635 nm), the output of which was focused on an area of ~3.0 mm diameter on the cortical surface; 2) a blue LED, (λmax = 450 nm), the output of which was focused on an area of ~ 3.0 mm diameter, and 3) a red laser (λmax =635 nm), the output of which was focused on an area ~0.1 mm in diameter. At the red wavelength, light penetration is expected to exert its maximal energy at a depth of 1–2 mm and have some degree of excitation as deep as 3–4 mm with significant signal attenuation in proportion to the depth of tissue penetration. Thus, light measurements are taken from the microvasculature within the entire renal cortex (~ 2 mm) and the underlying superficial medulla. The intensity of the red laser is higher than that of the red LED and its beam is more tightly focused. Nevertheless, as the light wavelength and O2 detection methods are similar, these methods are expected to produce comparable values. By contrast, at the blue wavelength, the depth of light penetration is expected to be attenuated to less than 1 mm, thus providing a measurement that is essentially contained within the microcirculation of the superficial renal cortex (Fig. 1). We performed sequential measurement over the same kidney region utilizing the red LED followed by the blue LED by switching the light source through the same excitation light guide, and then conducted measurements with the red laser, with the beam targeted to the centre of the broader LED light beam. The second red laser measurements served as a positive control for the first red LED measure and bracketed the blue light measurement.

For all three excitation lights, the emission detecting light guide was positioned 1–2 mm directly above the focus of excitation and received input from the tissue at a wavelength of 813 nm. Typically, 200 data collection cycles were averaged in a single measurement (two milliseconds), and the measurements were performed at one-second intervals. Experimental conditions were deemed acceptable with a signal to noise ratio of phosphorescence decay of greater than 2.

Intravascular oxygen sensing probe

Prior to commencing PkO2 measurements, Oxyphor PdG4 was injected intravenously (450 µL of ~ 25 µM solution in saline) to reach a final concentration in the blood plasma of ~2 µM. The excitation light at 635 and 450 nm is relatively weakly absorbed by the endogenous tissue chromophores (e.g., hemoglobin, cytochromes, etc.). The phosphorescence of Oxyphor PdG4 (λmax = 813 nm) is attenuated by endogenous absorption even less than the excitation light, but the emitted photons generated at depths have to diffuse back to the surface to be sensed by the detector. As a result, the volume of tissue probed in such a surface excitation/emission scheme does not have strictly defined borders, although it does vary with the light transport coefficients at the wavelengths specified. It is reasonable to assume that the recorded signals are dominated by the phosphorescence originating in the upper 1–4 mm of tissue with tissue layers closer to the surface giving larger contribution to the signal (red light). By contrast, the recorded signals are predominantly derived from phosphorescence originating in the upper < 1 mm of tissue for the blue light measurements. Thus, the measurements performed in this study carry information chiefly about microvascular PkO2 within the combined renal cortex and superficial medullary tissue (red light) or conversely from the superficial renal cortex alone (blue light).

Sample size calculation

The sample size was calculated for the primary outcome of change in PkO2 based on preliminary data from previous experiments2 and in accordance with the ARRIVE-2 guidelines emphasizing the goal of refining the experimental protocol to reduce animal numbers required. We estimated an expected PkO2 difference between high and low O2 (21% and 50% FIO2) and baseline vs 50% hemodilution FIO2 to be ~ 20 mmHg in each case from baseline (increased in the former and decreased in the latter). Estimating a standard deviation of 5.0, and assuming a power of 0.8 and an alpha of 0.05, our predicted sample size was six animals for each study. As a brief exposure to 30 and 50% FIO2 was not expected to have a long-term impact on oxygen delivery or tissue oxygenation, we utilized the same animal for both protocols to minimize the number of animals required for these studies. Because there was a single experimental protocol, no blinding was performed.

Outcome measures

Measured outcomes in all animals include PkO2 (measured with red LED, blue LED, and red laser), MAP, rectal temperature, ABG analyses and co-oximetry (pH, arterial pressure of carbon dioxide (PaCO2), arterial partial pressure of oxygen (PaO2), bicarbonate (HCO3), hemoglobin (Hb), arterial blood oxygen saturation (SaO2), and CaO2), and arterial electrolyte concentration (sodium, chloride, potassium, calcium, glucose, and lactate), all during hyperoxia and hemodilution.

Statistical analysis

Normality was tested with the Shapiro–Wilk test. Data that did not significantly violate normality were subjected to two-way analysis of variance (ANOVA) for repeated measures and post-hoc Tukey test for pairwise comparisons. Data that failed the Shapiro–Wilk test were analyzed using the Kruskal–Wallis one-way ANOVA on ranks followed by post-hoc pairwise comparisons using Dunn’s method. Lines of best fit were generated by ordinary least products regression analysis. Analysis of covariance was used to determine whether these relationships differed according to the intervention (hyperoxia or hemodilution). A two-tailed P < 0.05 was considered statistically significant. Analyses were performed using either Sigmaplot or Systat software (Systat Software Inc., San Jose, CA, USA).

Results

Physiologic parameters

Following stepwise increases in FIO2 (hyperoxia), there were no changes in HR or MAP (Fig. 2, panels A and B). Following hemodilution, there was a significant increase in HR after 25% and 50% hemodilution, relative to baseline (Fig. 3, panel A). No statistically significant changes were observed in the MAP; however, there was a trend for MAP to decrease after 50% hemodilution (Fig. 3, panel B). Rectal temperature remained stable, near 37 °C throughout the experimental period.

Fig. 2
figure 2

Hemodynamic and microvascular kidney PO2 (PkO2) measurements after exposure to hypoxia. A) Heart rate, and B) mean arterial pressure (MAP) were maintained throughout the stepwise increase in the fraction of inspired oxygen (FIO2). C) At all three FIO2 levels, the blue light-emitting diode (LED) measured PkO2 significantly higher than both the red LED and red laser measurements (#P < 0.001). The red LED and red laser measurements were similar at 21% and 30% FIO2. The stepwise increase in FIO2 showed expected trending of increased PkO2. For all sources of light, the PkO2 at 50% was significantly higher than both 21% and 30% (*P < 0.002). No significant difference was observed between FIO2 at 21% and 30% (n = 6).

Full size image
Fig. 3
figure 3

Hemodynamic and PkO2 measurements after exposure to hemodilution. A) Heart rate and B) mean arterial pressure (MAP) were maintained throughout the hemodilutions. C) At control and the two hemodilution levels, the blue light-emitting diode (LED) measured significantly higher kidney tissue oxygen tension (PkO2) than both the red LED and red laser measurements (#P < 0.001). The red LED and red laser measurements were similar during control and 25% hemodilution. At 50% hemodilution, the red laser PkO2 was significantly lower than the red LED ($P = 0.023). For all light sources there was a progressive decrease in PkO2 after 25% and 50% hemodilution relative to baseline (*P < 0.029) (n = 6).

Full size image

Arterial blood gas analysis

The blood oximetry data showed the expected changes in PaO2, Hb, and CaO2 during hyperoxia and hemodilution (Tables 1 and 2). With increasing FIO2, there was an expected stepwise increase in PaO2 and arterial blood Hb saturation (SaO2), and a slight increase in CaO2, as dictated by the formula: CaO2 = (1.34 * Hb * SaO2/100) + (0.0031 * PaO2) (P < 0.05). All other parameters did not change as FIO2 increased from 21% to 30% to 50% (Table 1). Following hemodilution, there was an expectant stepwise decrease in blood hemoglobin concentration and CaO2 (P < 0.001). As hydroxyethyl starch is constituted in 0.9% w/v NaCl, an increase in plasma chloride concentration was observed (P = 0.025); plasma HCO3 decreased (P = 0.03) and lactate increased slightly after 50% hemodilution (P = 0.006) (Table 2). Serum glucose decreased after hyperoxia and 50% hemodilution (P < 0.05) There were no statistically significant differences in other parameters over time (Table 2).

Table 1 Arterial blood gas and co-oximetry data
Full size table
Table 2 Arterial blood electrolyte data
Full size table

Kidney microvascular PkO2 values

At baseline, the PkO2 values obtained with the red light (LED, laser) were comparable to those reported previously1 (Figs 2 and 3, panel C). In all conditions, the PkO2 values obtained with the blue light were higher than those generated using red light (Figs 2 and 3). In all cases, the microvascular renal PkO2 values were comparable when taken using the red LED, prior to blue LED measurement, and those of the red laser which were taken immediately after the blue LED measurement. This sequence showed the stability of the red light measurements and that the PkO2 values were comparable when measured using excitation red light delivered by differing light intensities and beam widths.

During hyperoxia exposure, renal kidney tissue PkO2 increased with each increase in the level of FIO2 as measured following oxyphor excitation by both red and blue light (Fig. 2, panel C). During sequential hemodilution, the PkO2 decreased in proportion to the reduction in blood hemoglobin concentration and CaO2, in a stepwise manner, despite a relatively constant PaO2 (Fig. 3, panel C). The red LED and laser values were comparable when measured before and after the blue LED values.

Assessment of different correlations between physiologic parameters showed that renal microvascular PkO2, measured with blue or red light, did not correlate with MAP (r2 = 0.08 and 0.005, respectively) (Fig. 4, panels A). Nevertheless, values for microvascular renal PkO2 obtained with blue vs red LED were closely correlated (r2 = 0.78) (Fig. 4, panel B).

Fig. 4
figure 4

Scatterplot of the relationships between PkO2 measured using the red and blue light emitting diodes (r2 = 0.78). Data are from experiments in five rats. The x-intercept of the line of best fit did not differ significantly from zero [point estimate (95% confidence interval) = 1.04 (-8.36 to 10.44) but the slope was significantly less than unity (0.71 (0.56 to 0.86)]. In analysis of covariance, neither the intercept (P = 0.85) nor the slope (P = 0.98) varied significantly according to the intervention (hyperoxia or hemodilution).

Full size image

Assessment of the relationships between microvascular renal PkO2 utilizing blue and red light showed significant correlations with CaO2 (Fig. 5). PkO2 was closely correlated with CaO2 both for blue (r2 = 0.56) and red (r2 = 0.69) light, representing measurements from the superficial renal cortex and deep renal cortex and outer medulla, respectively. The relationships between these variables did not vary significantly between the two interventions (Fig. 5).

Fig. 5
figure 5

Scatterplots of the relationships between arterial blood oxygen content (CaO2) and PkO2 measured using either (A) blue (r2 = 0.56) or (B) red (r2 = 0.69) light emitting diodes (LEDs). Data are from experiments in five rats. In analysis of covariance, neither the intercepts nor the slopes of these relationships differed significantly according to the intervention (hyperoxia or hemodilution). P values for differences in intercept and slope between the two interventions were ≥ 0.13 for both blue and red LEDs

Full size image

We hypothesized that a change in CaO2 with hemodilution would follow a different relationship than hyperoxia. Nevertheless, our analysis failed to detect any significant differences. Therefore, while there is a difference between red and blue light, we found a linear relationship between hyperoxia and hemodilution. We ran the regression analyses for the individual interventions, and found the following results for hyperoxia (red LED r2 = 0.232, P = 0.07, blue LED r2 = 0.343, P = 0.02, red laser r2 = 0.051, P = 0.42) and for hemodilution (red LED r2 = 0.687, P < 0.001, blue LED r2 = 0.469, P = 0.005, red laser r2 = 0.847, P < 0.001). Thus, as predicted by the data in the scatterplots, the relationships were more robust for hemodilution than for hyperoxia. Nevertheless, a significant relationship was observed for hyperoxia with the blue LED and a strong tendency was seen with the red LED.

Discussion

We show that PkO2, in both superficial and deeper layers of the kidney, varies in a linear manner with CaO2. Importantly, a single relationship can adequately explain the variation of PkO2 with CaO2 during hyperoxia and hemodilution. This allows us to attribute the oxygen sensing capacity of the kidney to an ability to translate changes in CaO2 to changes in kidney PkO2. The impact of hyperoxia (relatively high CaO2) and anemia (low CaO2) on changes in PkO2 showed a similar linear relationship, supporting the hypothesis that CaO2 is biologically sensed. The lack of correlation between MAP and arterial PaO2 and renal microvascular PkO2 supports the finding that CaO2 is an important physiologic parameter that regulates renal PkO2.

Our observations are in accord with traditional studies assessing increased expression of renal EPO following exposure to both hypoxia and anemia.3,4 Recent observations in humans regarding the control of EPO release from the kidney support the concept that CaO2 and not PaO2 is the important variable with respect to regulating EPO secretion.16 In a crossover study with human volunteers, Montero and Lundby showed that the increase in systemic EPO levels associated with a reduction of CaO2 induced by subjects breathing low concentrations of carbon monoxide while mainlining PaO2 above 100 mmHg was comparable to that observed by reduced CaO2 induced by breaking a hypoxic gas mixture (FIO2 = 11%). Thus, the mechanisms that regulate kidney oxygenation appear to be set up in such a way that microvascular, and thus tissue PO2, is sensitive to changes in CaO2 no matter how they are induced.16

The dominance of CaO2 in the control of renal tissue oxygenation can be explained by available knowledge regarding oxygen transport to tissue and the incorporation of this knowledge into computational models of renal oxygen transport. It is well established that oxygen delivery to tissue is driven by the gradient between the intravascular and tissue PO2.17 Importantly, in the kidney18 and in other vascular beds,17 oxygen diffuses from the vasculature to the tissue not only from capillaries but also from arterial vessels. Consequently, as blood flows from the main renal artery to the renal microvasculature, its oxygen content falls in proportion to the gradient between the vascular and tissue compartments.19 Some oxygen is also lost via diffusion to the veins, which are often closely associated with the arterial vessels.19,20

Factors driving renal oxygen delivery include CaO2 and dissolved O2. The formula for CaO2 further defines the relationship between Hb-bound O2 and dissolved O2 in the plasma: CaO2 = (1.34 × Hb × SaO2/100) + (0.0031 × PaO2). This formula expresses that PaO2 (i.e., dissolved oxygen in plasma) only represents a minute portion of the total oxygen transported in the blood. Simplistically, oxygen carried by Hb represents the bulk of O2 delivered to tissues while arterial PaO2 and microvascular renal PkO2, represent the component of oxygen progressively released from Hb on route to the mitochondria of metabolically active cells. The complexity of local renal vascular–tissue oxygen gradients are not directly measured by our technique; rather, the sum effect of these local oxygen gradients are measured within a sample of renal microvasculature that directly reflects tissue PO2.2

Simulations derived from computational models of oxygen transport predict that, under conditions where there is an increase in the fraction of inspired O2, arterial PO2 falls rapidly along the arterial tree because this condition is associated with only a small additional quantity of oxygen in the blood. Consequently, the increase in CaO2 (~ 15%) and PkO2 (~ 25%) observed with hyperoxia are relatively small in relation to the increase in PaO2 (~ 250%). Nevertheless, the changes in CaO2 remain in proportion to the changes in renal microvascular PkO2.21,22 On the other hand, when CaO2 is reduced by hemodilution by decreasing the major determinant of CaO2 (i.e., Hb), blood oxygen content is reduced by a larger proportion (~ 60%) as reflected by a more pronounced drop in kidney PkO2 (~ 50%). Under these conditions, oxygen diffusion from the arterial tree results in a more pronounced reduction in the PO2 of blood, as it flows to the microvasculature, compared with normal conditions.23 As we show herein, the change in PkO2 is proportional to the change in CaO2. The consequence of this physiologic system is that the EPO-producing cells in the kidney are in an ideal place to sense changes in CaO2 and thus regulate red cell production in response to acute anemia.2,3,4,13,14

Our findings are also consistent with the notion of a gradient in microvascular and tissue PO2 from the renal cortex to medulla,24 at least in anesthetized animals5,25,26 and humans.27,28 This gradient may be less pronounced in the absence of anesthesia.29 In the current study of anesthetized rats, we found PkO2 measured using a red LED was systematically lower than that measured using a blue LED, likely because the red LED penetrates deeper into the tissue.30 Nevertheless, changes in CaO2 resulted in proportionally similar changes in the two measures of microvascular PkO2. We are not aware of any previous direct quantitative comparison of the relative sensitivities of kidney tissue, at various depths below the cortical surface, to changes in CaO2. Nevertheless, our current findings are consistent with the observation of similar changes in renal cortical and medullary PkO2 induced by hemodilution in rats,5 hypoxia in rabbits,31 and hyperoxia in sheep.29

Understanding the relationship between the physiologic function of the kidney with respect to oxygen demand and supply may help explain its role as an oxygen sensor. With respect to metabolic demand, we will focus on sodium (Na) resorption as one of the major components of renal metabolic demand.14 The majority of filtered Na (~ 65%) is recovered in the proximal convoluted tubule within the renal cortex, powered by the sodium potassium ATPase. The next important component for Na resorption (~ 25%) occurs in the loop of Henle within the renal medulla. About half of medullary Na resorption occurs passively, along the generated concentration gradient, while the remaining half is absorbed via active transport in the thick ascending limb of Henle’s loop (Na-K ATPase). The remaining 10% of Na resorption occurs by active transport in the distal convoluted tubule, situated adjacent to the macula densa of the glomerulus.

To optimize renal oxygen delivery, the vasculatures of the renal cortex and medulla are arranged in parallel, starting at the level of the proximal interlobular arteries. Thus, blood flow to these two vascular territories can be independently regulated. In the outer medulla, the blood supply further divides as some vasa recta capillaries form a plexus in the outer medulla while others go on to perfuse the inner medulla. Thus, there is some potential for independent regulation of perfusion of the inner and outer medulla.20

Finally, the balance of oxygen supply and demand dictate how the kidney functions as an oxygen sensor.14 The kidney is the principal regulator of erythrocyte production as a response to vascular oxygen content induced by hypoxia or anemia.2,3,4,32,33 Under resting conditions, the majority of EPO-producing cells are located near the cortico-medullary junction.3,4 Nevertheless, following exposure to conditions that cause renal tissue hypoxia or molecular manipulation to simulate hypoxia, the regions in which EPO-producing cells are identified expands into the superficial renal cortex and deep medulla,4,13,34 thereby showing the capacity for rapid and profound upregulation of EPO transcription and excretion.

By utilizing characteristics of shorter wavelength blue light vs longer wavelength red light, we have observed that measurements of PkO2 are higher in the superficial renal cortex (blue light) than values obtained with more deeply penetrating red light. This finding has been previously reported utilizing stepwise progression of Clarke type microelectrodes6 and utilizing the method of phosphorescence quenching during acute hemodilution.5 Nevertheless, a clear relationship between CaO2 and PkO2 was not derived by these previous studies. The difference in the values measured with blue vs red light suggest that superficial blue light measures more arterialized blood from afferent arterioles as they enter the glomeruli, situated in the cortex. Conversely, red light penetrates more deeply into the renal cortex and outer renal medulla. Furthermore, the renal cortex and outer renal medulla are highly metabolic because of the metabolic activity of the proximal convoluted tubule and thick ascending loop of Henle, which both consume high amounts of oxygen. This is one of the multiple explanations for a decreased PkO2 in the outer medulla compared with the cortex.11 The finding that PkO2 values with blue and red light strongly correlate with one another suggests that measurement of superficial cortical or deeper cortical and medullary microvascular PkO2 both reflect changes in blood CaO2, in balance with local oxygen demand. The net effect reflects the balance between oxygen supply and demand.

Estimates of renal oxygen consumption and metabolic activity suggest that the proximal tubules and outer medullary thick ascending loop of Henle (mTAL) represent regions of high metabolic oxygen requirement.35,36 Furthermore, the inner renal cortex and outer renal medulla are the locations for EPO synthesis3,4 while the outer medulla is the region of the kidney most susceptible to hypoxia, in part due to relatively limited vascular supply compared with the cortex.11,20 The disposition of the medullary vasculature (the vasa recta) in bundles leads to countercurrent oxygen exchange throughout the medulla. This results in a decreased oxygen supply in the inner renal medulla. The disposition of the vascular bundles, with respect to the mTAL and the collecting duct, results in an oxygen gradient from the bundles to these elements of the nephron, rendering them especially susceptible to hypoxia.11,20 Therefore, the renal outer medulla is at particular risk of hypoxia and hypoxic injury, as the balance of oxygen supply and demand puts this region at risk of inadequate oxygen delivery.35,37 This characteristic is precisely what renders it an efficient sentinel for the oxygenation status of the rest of the body. In fact, the medullary region is known for its capacity to rapidly upregulate hypoxia gene expression,2,4 including EPO, showing the link between low CaO2, microvascular PkO2, and hypoxic biological responses.

Translational implications

We have previously shown that experimental measurement of tissue oxygen tension can inform clinical decision-making in a number of clinical scenarios. For example: 1) anemia is associated with tissue hypoxia,38 possibly explaining the association between preoperative anemia and increased mortality39; 2) systemic β-blockade accentuates anemia-induced cerebral hypoxia,40,41,42 possibly explaining the increased incidence of perioperative stroke43; 3) acute anemia worsens brain tissue hypoxia and greatly enhances the degree of brain injury,44 supporting utilization of brain PO2 monitoring to support clinical decision-making45 and potentially improving patient outcomes46; and 4) assessment of renal PkO2 showed a more profound decrease following hemodilution with starch vs albumin or saline,2 possibly providing a mechanistic explanation for increased renal toxicity associated with starch-based fluid resuscitation in critical care settings.47 The current study shows that microvascular renal PO2 values reflect oxygen delivery to tissues at times of reduced blood oxygen content. Clinical studies have attempted to measure renal tissue hypoxia may predict adverse outcomes, including acute kidney injury.28,48 The common goal of these approaches is to direct therapies to reduce perioperative organ injury and adverse clinical outcomes.

Limitations

There are some limitations to our current study. First, we did not measure cardiac output (CO) in this experimental data set as we have previously showed that CO increases following hemodilution with starch,2,42 but that changes in microvascular renal perfusion are not directly liked to these changes in CO.2 Second, the effect of hyperoxia was limited because of the small increment in CaO2 (PaO2 only contributes to a small proportion of total CaO2). Nevertheless, the slope of the hyperoxia effect was comparable to that of hemodilution, supporting that the impact of both hyperoxia and hemodilution were influenced by CaO2. Third, these measurements were performed under anesthesia, which is known to impact renal oxygen delivery.23 Indeed, increased FIO2 and hyperoxia have been shown to restore renal oxygen tension in experimental models potentially supporting the use of hyperoxia in patients under general anesthesia to preserve renal function.

Perspectives and significance

Utilizing non-invasive methods to measure microvascular renal tissue PkO2, we have shown a linear relationship between blood oxygen content (CaO2) and renal tissue PkO2. The relationship is maintained with two different excitation light wavelengths, confirming the impact of CaO2 in superficial (with blue light) and deeper (with red light) renal tissue. These data provide an experimental means for assessing the impact of tissue hypoxia on renal injury28,48 and the impact of strategies to mitigate such injury.