Quantitative Tissue Ph Measurement during Cerebral Ischemia Using Amine and Amide Concentration-Independent Detection (AACID) with MRI

Abstract

Tissue pH is an indicator of altered cellular metabolism in diseases including stroke and cancer. Ischemic tissue often becomes acidic due to increased anaerobic respiration leading to irreversible cellular damage. Chemical exchange saturation transfer (CEST) effects can be used to generate pH-weighted magnetic resonance imaging (MRI) contrast, which has been used to delineate the ischemic penumbra after ischemic stroke. In the current study, a novel MRI ratiometric technique is presented to measure absolute pH using the ratio of CEST-mediated contrast from amine and amide protons: amine/amide concentration-independent detection (AACID). Effects of CEST were observed at 2.75 parts per million (p.p.m.) for amine protons and at 3.50 p.p.m. for amide protons downfield (i.e., higher frequency) from bulk water. Using numerical simulations and in vitro MRI experiments, we showed that pH measured using AACID was independent of tissue relaxation time constants, macromolecular magnetization transfer effects, protein concentration, and temperature within the physiologic range. After in vivo pH calibration using phosphorus (³¹P) magnetic resonance spectroscopy (³¹P-MRS), local acidosis is detected in mouse brain after focal permanent middle cerebral artery occlusion. In summary, our results suggest that AACID represents a noninvasive method to directly measure the spatial distribution of absolute pH in vivo using CEST MRI.

Keywords

acute stroke brain ischemia energy metabolism focal ischemia imaging MRI MR Spectroscopy pH

INTRODUCTION

Cerebral tissue pH is normally regulated at ∼7.0 to 7.2 pH units by the cooperative action of a multitude of sensors and regulators within cellular compartments.¹ Deviation from pH homeostasis is an indicator of disease and occurs in stroke when metabolic demand for oxygen exceeds supply. Lactic acidosis leads to irreversible cellular damage in ischemic tissue (IT).² Therefore, pH is a potentially important and sensitive fundamental measure of metabolic state and disease progression. However, few methods measure absolute pH in vivo.³ Even noninvasive magnetic resonance imaging (MRI) methods, which have had the greatest success to-date, either have low spatial and temporal resolution⁴ or require assumptions that limit their clinical application.⁵

Several MRI observable parameters are sensitive to in vivo pH.³ For example, the chemical shift between inorganic phosphate (P_i) and phosphocreatine (PCr) observed with phosphorus (³¹P) magnetic resonance spectroscopy (³¹P-MRS) is directly related to pH.⁴ However, ³¹P-MRS is limited by low signal-to-noise ratio resulting in low spatial and temporal resolution. Recently, Zhou et al⁵ proposed a novel technique called amide proton transfer (APT) to measure pH using the chemical exchange saturation transfer (CEST) effects from amide protons associated with mobile peptides and proteins. Amide protons freely exchange with bulk water protons at a rate that is pH dependent. However, APT contrast is also affected by cellular water content, amide content, the bulk water spin-lattice (T₁) relaxation time constant, and macromolecular exchange-relayed nuclear Overhauser effects (NOEs).^{5, 6, 7} Therefore, APT pH measurement is susceptible to error in conditions that cause appreciable changes in protein concentration and/or water content (Supplementary Information). Consequently, groups have successfully employed APT contrast to acquire pH-weighted magnetic resonance (MR) images that correlate well with lactic acidosis.^{8, 9} The CEST effects derived from various metabolites including glutamate¹⁰ and creatine¹¹ are also sensitive to pH; however, quantitative pH measurement using these CEST effects is limited by concurrent concentration and T₁ dependence.

To overcome limitations associated with concentration dependence, Ward and Balaban¹² developed a ratiometric approach that was applied to exogenous compounds using the CEST effects at two distinct frequencies to measure quantitative pH independent of bulk water T₁ and exchange site concentration. Desmond et al¹³ recently proposed a similar ratiometric approach to measure pH using endogenous CEST effects from amide (3.5 parts per million (p.p.m.)) and amine (2.0 p.p.m.) protons. Here, we describe a novel method to measure absolute pH using the CEST effects of both amine (2.75 p.p.m.) and amide (3.50 p.p.m.) protons associated with endogenous mobile peptides and proteins: amine/amide concentration-independent detection (AACID). Amine/amide concentration-independent detection produces quantitative pH maps with markedly higher spatial resolution compared with ³¹P-MRS, and eliminates the assumptions of constant water content and protein concentration that limit APT. It is also independent of NOE and requires a simple pH-calibration step for in vivo application.

Amine and amide proton CEST effects are often considered to originate from mobile intracellular proteins and peptides.^{5, 13, 14, 15} Recently, pH changes measured using endogenous amide CEST effects were considered to be of intracellular origin⁵ because nearly 90% of total protein content exists in the intracellular space.¹⁶ However, it is more accurate to consider the pH to be an average cerebral tissue pH that is heavily weighted to the intracellular compartment. Regardless, the pH resolution achieved using CEST effects (∼0.2 pH units)⁵ does not appear to be sufficient to distinguish the physiologic differences between intracellular and extracellular spaces (∼0.1 pH unit). Therefore, the measured pH is described as the cerebral tissue pH.

In the current study, numerical simulations were used to investigate the dependence of AACID measurements on pH, macromolecular magnetization transfer (MT), direct saturation and relaxation times. In addition, the pH, protein concentration, and temperature dependencies of the AACID method were characterized in vitro. The AACID pH measurement was then calibrated in mouse brain using ³¹P-MRS pH as the standard measurement. Finally, high-resolution in vivo pH maps were acquired in the mouse brain after permanent middle cerebral artery occlusion (MCAO) and compared with cellular damage identified histologically. The assumptions required for absolute pH measurement using AACID, the impact of CEST effects from other metabolites on AACID, and potential applications of AACID at lower magnetic field strengths are discussed.

MATERIALS AND METHODS

Chemical Exchange Saturation Transfer Theory

Chemical exchange saturation transfer MRI contrast can be detected after the selective saturation of solute protons that exchange with bulk water protons. Solute protons resonate at specific frequencies that are chemically shifted (Δω) from bulk water. Figure 1A illustrates the distribution of different solute proton pools based on their resonant frequencies relative to bulk water. To observe CEST contrast, the chemical shift (Δω) of the exchanging solute protons must satisfy the slow-intermediate exchange condition (Δω>k), where k is the exchange rate between solute and bulk water.¹⁷ During frequency-selective saturation of the solute protons, a net transfer of irradiated high energy protons to the bulk water pool ensues by chemical exchange producing a reduction in the net magnetization of the bulk water leading to negative contrast in MRI. Figure 1B shows the CEST effects on the bulk water after frequency-selective radio frequency (RF) saturation of the amide protons (3.50 p.p.m.) and the amine protons (2.75 p.p.m.). During saturation, longitudinal (T₁) relaxation simultaneously drives both pools toward their equilibrium population (Boltzmann) distribution leading to a steady-state condition.¹⁷ After a long saturation pulse of frequency ω, the bulk water steady-state magnetization ( $M_{SS}^{W} (ω)$ ) is described by equation (1), where $M_{0}^{W}$ is the bulk water magnetization without saturation preparation, k is the solute proton exchange rate, and f is the mole fraction of solute protons as described by equation (2).^{12, 17} A plot of $M_{SS}^{W} (ω) ∕ M_{0}^{W}$ as a function of the saturation pulse frequency is called a CEST spectrum (Figure 1C). Effects of CEST are often quantified using the magnetization transfer ratio (MTR) defined by equation (3).

Figure 1.

Illustration of chemical exchange saturation transfer (CEST) principles and the effect of pH on amine and amide CEST signal: (A) Schematic diagram of relevant solute proton pools that resonate at frequencies chemically shifted by parts per million (p.p.m.) from bulk water. (B) Irradiation of amide (top) or amine (bottom) protons using a frequency-selective saturation pulse. Chemical exchange transfers amide (top) or amine (bottom) magnetization to bulk water, resulting in indirect saturation of the bulk water ( $M_{SS}^{W}$ ). (C) CEST spectra (vertically offset for clarity) measured at different pH (6.0, 6.5, and 7.0). The red and green portions of the curve illustrate the pH-dependent response of the amide and amine protons, respectively.

M_{SS}^{W} (ω) ∕ M_{0}^{W} = 1 ∕ (1 + k f T_{1 W})

(1)

f = [solute protons] ∕ [water protons]

(2)

CEST (ω) = 1 - M_{SS}^{W} (ω) ∕ M_{0}^{W}

(3)

The chemical exchange rate of protons from the solute pool to bulk water is both temperature and pH dependent. Within the narrow physiologic temperature range (35°C to 39°C), the chemical exchange rate is dominated by modulations in pH.^{5, 18} The pH dependence of the solute proton exchange rate is described by equation (4), where k_b is the base-catalyzed exchange rate and pK_W is the ionization constant of water (pK_W=15.4 at 37°C).^{19, 20}

k = k_{b} 10^{pH} - p K_{W}

(4)

Amide and amine protons associated with peptides have been shown to have different k_b and therefore have a differential response to pH changes.^{5, 13, 14} As pH increases, amine (k_amine=10^pH−4.2) and amide (k_amide=5.57 × 10^pH−6.4) base-catalyzed exchange rates increase. As expected from equations (1) and (4), saturation of the amide proton pool (i.e., ω=3.50 p.p.m.) causes $M_{SS}^{W} (ω) ∕ M_{0}^{W}$ to decrease as pH increases (Figure 1C) producing an increase in the amide CEST effect (equation (3)). Paradoxically, the amine proton CEST effects appear to decrease as pH increases (Figure 1C), apparently defying equations (1) and (3). However, the apparent decrease in CEST effect occurs because the amine proton exchange rate increases into the fast-exchange regime (k >Δω) when pH increases above ∼6.5 pH units. Since the amide and amine CEST effects change in opposite directions with increasing pH, the ratio of these two effects is expected to be sensitive to pH.

Amine/Amide Concentration-Independent Detection pH Measurement

Amine/amide concentration-independent detection is based on the CEST MRI experiment. Direct measurement of pH using the CEST effect amplitude of a single solute proton pool is difficult due to the concomitant effects of solute concentration, saturation pulse amplitude and duration, and T₁ relaxation. Ward and Balaban¹² first proposed a ratiometric approach to measure pH using the CEST effects from two different exchange sites on one CEST agent. Briefly, a single solute may have two different proton exchange sites (i.e., site 1 and site 2). Rearranging equation (1) combined with equation (2) for both sites results in equation (5).

\frac{(M_{0} - M_{S})^{Site 1, 2}}{M_{S}^{Site 1, 2}} = k^{Site 1, 2} T_{1 W} {(\frac{[solute protons]}{[water protons]})}^{Site 1, 2}

(5)

Taking the ratio of equation (5) for two different sites produces equation (6),¹² where T_1W effects and solute concentration effects cancel leaving only a ratio of pH-dependent exchange rates.

\frac{M_{S}^{Site 2} (M_{0} - M_{S})^{Site 1}}{M_{S}^{Site 1} (M_{0} - M_{S})^{Site 2}} = \frac{k^{Site 1} T_{1 W} {(\frac{[solute protons]}{[water protons]})}^{Site 1}}{k^{Site 2} T_{1 W} {(\frac{[solute protons]}{[water protons]})}^{Site 2}}

(6)

In the present study, amine protons resonating at 2.75 p.p.m. and amide protons resonating at 3.50 p.p.m. on endogenous proteins were considered to be site 1 and site 2, respectively. The bulk water magnetization after saturation at 6.0 p.p.m. was used as a common reference point for both amine and amide CEST measurements (equation (7)). This approach minimizes endogenous macromolecule MT-related signal loss while avoiding contamination from NOEs, which occur between −2.0 and −4.0 p.p.m. (Supplementary Information).^{6, 7, 21} The common reference point also reduces the number of CEST images required for pH measurement.

AACID = \frac{M_{Z} (3.50 p . p . m .) \times (M_{Z} (6.0 p . p . m .) - M_{Z} (2.75 p . p . m .))}{M_{Z} (2.75 p . p . m .) - (M_{Z} (6.0 p . p . m .) - M_{Z} (3.50 p . p . m .))}

(7)

The reference point was chosen to be 6.0 p.p.m. because there are no observed CEST effects or NOEs at 6.0 p.p.m. in tissue or bovine serum albumin (BSA) phantoms and the 6.0 p.p.m. CEST image is unaffected by CEST effects at 3.5 and 2.75 p.p.m. The use of a single reference point for both CEST effects is justified when using a ratiometric method since any significant signal modulations at the reference frequency (i.e., due to T₂ relaxation or MT) will cancel.

All CEST contrast measurement methods are sensitive to inhomogeneities in the main magnetic field (B₀) because the CEST effects are measured at specific frequencies. Therefore, it is essential that B₀ maps are measured and a B₀ correction is applied (Supplementary Methods). The effects of the amplitude of the applied magnetic field (B₁) on CEST measurement are complex due to the B₁ dependence of saturation efficiency, direct saturation, and MT. Hence, it is also important to measure and correct for B₁ inhomogeneities using B₁ calibration curves (Supplementary Information).

Numerical Simulations

The Bloch-McConnell equations²² were used to numerically simulate CEST spectra based on a four-pool exchanging system (bulk water, macromolecular, amine, and amide protons) using specific saturation pulse parameters. First, to simulate the relationship between AACID values and pH, CEST spectra were generated for different pH using previously reported pH-dependent base-catalyzed exchange rates of the amide and amine protons.^{5, 13, 19} Ratiometric AACID values, calculated using equation (7) adopted from Ward and Balaban,¹² were measured directly from simulated spectra. In addition, macromolecular concentration was varied (9%, 10%, and 11%) to model potential changes in MT after ischemia.²³ Finally, normal tissue and IT were modeled by incorporating tissue-specific relaxation time constants and exchange rates for each pool (Table 1). Relationships of AACID–pH were simulated for both tissue types to study the cumulative effects of differences in water content, relaxation times, and consequent changes in direct saturation.

Table 1.

Parameters used to simulate CEST spectra for different tissue types

Simulation parameters
Proton pools	Bulk water	Macromolecule	Amine	Amide
Chemical shift (p.p.m.)	0	0	2.75	3.50
T₁ (seconds)	NT: 1.844^a IT: 1.992^a	0.1^b	1.0^c	1.0^c
T₂ (seconds)	NT: 0.045^a IT: 0.054^a	0.000015^b	0.05	0.05
Concentration	NT: 39.1 mol/L^d IT: 51.8 mol/L^d	0, 9, 10, 11% by weight	10.0 mmol/L	71.9 mmol/L
Exchange rate (1/s)	—	50^b	10^pH−4.2e	5.57 × 10^pH−6.4f

CEST, Chemical exchange saturation transfer; IT, ischemic tissue; NT, normal tissue; p.p.m., parts per million.

From ref. 25.

From ref. 22.

From ref. 1010.

From ref. 40.

Estimated from refs. 19, 13.

From ref. 5.

In Vitro Solutions

To study the AACID pH measurement, a series of solutions containing 10% (by weight) bovine serum albumin (BSA; Sigma-Aldrich, St Louis, MO, USA) dissolved in 600 μL of phosphate-buffered saline (PBS) were produced with pH ranging from 6 to 8 pH units. Solution pH was adjusted using 1 mol/L HCl and 1 mol/L NaOH and confirmed with a standard pH electrode (Cole-Parmer, Montreal, QC, Canada). To study the effect of protein concentration on AACID, a series of pH 7 solutions (600 μL) were produced with 6%, 8%, 10%, 12%, and 14% BSA. All phantoms were prepared in 5 mm nuclear magnetic resonance tubes. Physiologic total protein content has been previously shown to be ∼10% by weight in both rodent and human brains.²⁴ To measure potential contributions of CEST contrast from common brain metabolites, we prepared phantoms with each metabolite dissolved into pH 7.0 PBS solutions at physiologic concentrations (Sigma-Aldrich). We used the identical metabolites and concentrations as Cai et al¹⁰ to allow direct comparisons (Table 2).

Table 2.

Metabolite concentrations used for in vitro measurement of potential contaminant CEST effects

Metabolite concentrations
Metabolite	Concentration (mmol/L)
Creatine (Cr)	10
Myo-inositol (MI)	10
N-acetyl aspartate (NAA)	10
Glutamine (Gln)	2
Taurine (Tau)	2
Aspartate (Asp)	2
Glutamate (Glu)	10
γ-aminobutyric acid (GABA)	2

CEST, chemical exchange saturation transfer.

Magnetic Resonance Experiments

All phantom and animal MRI studies were performed in an Agilent 9.4-T small animal MRI scanner (Agilent, Palo Alto, CA, USA). The bore of the magnet is 31 cm and the maximum gradient strength is 400 mT/m. A long 4-second continuous wave RF saturation pulse with a B₁ amplitude of 1.5 μT was introduced before an RF spoiled two-dimensional fast spin echo pulse sequence for all phantom and animal CEST MRI studies. In vitro and in vivo¹H-MRI and ³¹P-MRS methods are described in detail in Supplementary Information.

General Animal Procedures

All animal protocols conform to the guidelines established by the Canadian Council on Animal Care and were approved by The University of Western Ontario Animal Use Subcommittee. For in vivo studies, C57BL/6 male mice (3 to 5 months old, 24 to 28 g) were used. Three animals were used for in vivo pH calibration and five different animals were used for MCAO demonstration. Two of the five animals used for MCAO demonstration were used for histology. Four different animals were used to measure the percent reduction in cerebral blood flow obtained by permanent MCAO using laser Doppler flowmetry. Animal and histologic preparation procedures are described in detail in Supplementary Information.

Chemical Exchange Saturation Transfer Data Processing

All acquired CEST MR data were loaded into MATLAB (Mathworks, Natick, MA, USA) for analysis. Using custom MATLAB programs, B₀ and B₁ corrections were performed on a pixel-by-pixel basis and all CEST spectra were smoothed using the ‘smooth’ algorithm from the MATLAB curve fitting toolbox.

Statistical Analysis

Cerebral tissue pH measurements were made from pathologic and contralateral regions within each mouse brain and expressed as mean±s.d. Differences in mean pH between stroke and contralateral brain regions (n=5 mice) were tested using the Student's t-test within the MATLAB Statistics toolbox.

RESULTS

Numerical Simulations

Bloch-McConnell simulations showed that as pH increases within the physiologic pH range (i.e., pH=6.0 to 8.0 pH units), the CEST contrast associated with amine protons and amide protons changes in opposite directions (Figures 2A to 2C). These changes are most easily observed in Figure 2C, where amide and amine CEST effects, as defined by equation (3) are displayed with varying pH. Specifically, as pH increases from 6.0 to 8.0, the CEST effect increases for the amide protons at 3.50 p.p.m. Conversely, the CEST effect associated with amine protons at 2.75 p.p.m. decreases from pH 6.5 to ∼7.5. The resulting AACID values decrease linearly as pH increases (Figure 2D). Numerical simulations also show that pH-dependent exchange rate effects dominate AACID measurements in brain tissue when considering expected variations in macromolecule concentration, water content, and relaxation times in tissue after ischemia. First, varying macromolecular concentrations (9%, 10%, and 11%) modeled potential changes in MT after ischemia²³ (Figure 2E) and confirmed negligible effects on the simulated AACID–pH relationship. Second, the AACID–pH relationship was simulated for both normal and ischemic brain tissue to study the cumulative effects of differences in water content, relaxation, and consequent changes in direct saturation. Known changes in water content, and T₁ and T₂ relaxation²⁵ also produced negligible changes in AACID measurement (Figure 2F).

Figure 2.

Numerical simulations of amine/amide concentration-independent detection (AACID) dependence on pH, macromolecule concentration, and relaxation time constants in brain: (A) Simulated chemical exchange saturation transfer (CEST) spectra for varying pH in normal brain tissue. (B) Figure 1A magnified to highlight the amine and amide pH-dependent CEST effects. Note: M_Z and M₀ represent the region of interest magnetic resonance imaging (MRI) signal acquired with and without a saturation pulse, respectively. (C) Amine and amide CEST values measured directly from (A) using equation (3). (D) AACID values measured directly from (A) using equation (7). (E) AACID–pH relationships simulated for a range of macromolecular concentrations (% by weight) using a 4-second, 1.5-μT saturation pulse. (F) AACID–pH relationships simulated for normal and ischemic brain tissue using a 4-second, 1.5-μT saturation pulse.

In Vitro Studies

Solutions with varying pH were produced containing different concentrations of BSA dissolved in PBS. Although these solutions did not replicate all in vivo conditions (e.g., T₁ and T₂ relaxation time constants and macromolecule MT), they provided a physical model of an exchanging system containing both amide and amine protons similar to that found in vivo. These solutions showed the effects of changing pH, protein concentration, and temperature on CEST, and consequently AACID measurements. The CEST spectra of 10% BSA dissolved in PBS solutions with varying pH show three distinct CEST effects (Figures 3A and 3B). Amide proton CEST effects are centered at ∼3.50 p.p.m. and two different amine proton CEST effects are centered at ∼2.0 and ∼2.75 p.p.m. Incrementally increasing solution pH from 6.0 to 8.0 caused the amide CEST effect to increase and the amine CEST effect at ∼2.75 p.p.m. to decrease. The amide and amine CEST effects at ∼2.75 p.p.m. are plotted separately with varying solution pH in Figure 3C. Ratiometric AACID values decreased linearly when increasing pH from 6.5 to 8.5 (Figure 3D) as predicted by numerical simulations. Measurements of AACID were nearly identical for pH 7.0 phantoms containing 6%, 8%, 10%, 12%, and 14% BSA (Figure 3E). The AACID pH measurements obtained in a pH 7.0 10% BSA phantom were unaltered between 35°C and 39°C (Figure 3F).

Figure 3.

Amine and amide chemical exchange saturation transfer (CEST) dependence on pH, protein concentration, and temperature: (A) CEST spectra of 10% (by weight) bovine serum albumin (BSA) dissolved in phosphate-buffered saline (PBS) with varying pH at 37°C. (B) Figure 2A magnified. (C) Amine and amide CEST values measured directly from (A) using equation (3). (D) Amine/amide concentration-independent detection (AACID) values measured directly from (A) using equation (7). (E) AACID values measured in solutions containing a range of BSA concentrations at 37°C using a 4-second, 1.5-μT saturation pulse. (F) AACID values measured in a pH 7.0 10% BSA solution for a range of physiologically relevant temperatures using a 4-second, 1.5-μT saturation pulse.

Amine and amide CEST effects are also dependent on saturation pulse amplitude (B₁) and duration (t_sat) (Figures 4A and 4B). Similar to Cai et al¹⁰, CEST effects were measured in vitro for eight metabolite solutions (Table 2) at 37°C. The CEST contrast produced by these brain metabolites was measured at 3.50 and 2.75 p.p.m. at physiologic concentrations to determine the magnitude of contaminant contributions to AACID pH values (Figures 4C and 4D). Glutamate was the only metabolite to generate a significant contaminant CEST effect at 3.50 p.p.m. (Glutamate CEST≈2.3% compared with CEST from 10% BSA≈9.2%) and 2.75 p.p.m. (Glutamate CEST≈3.0% compared with CEST from 10% BSA≈19.4%). Note that the CEST effects from the amine protons associated with glutamate are optimal at ∼1.0 to 3.0 p.p.m. at physiologic pH.¹⁰ However, glutamate CEST contributions to AACID pH values are expected to be <5% due to ratiometric cancellation when using a 4-second, 1.5-μT saturation pulse at 9.4 T.

Figure 4.

Amine and amide chemical exchange saturation transfer (CEST) dependence on saturation pulse amplitude, duration, and metabolite contributions: amine and amide CEST effects from 10% (by weight) bovine serum albumin (BSA) dissolved in pH 7.0 phosphate-buffered saline (PBS) at 37°C measured following (A): a 4-second continuous wave RF saturation pulse with amplitudes ranging from 0 to 3 μT and (B) a 1.5-μT continuous wave RF saturation pulse with duration ranging from 0 to 6 seconds. CEST values measured at (C) 2.75 p.p.m. and (D) 3.50 p.p.m. for major brain metabolites at physiologic concentrations (37°C using a 4-second, 1.5-μT continuous wave RF saturation pulse) compared with 10% BSA.

In Vivo Amine/Amide Concentration-Independent Detection pH Calibration

Numerical simulations indicate that significant differences in macromolecular MT and proton exchange rates^{5, 10, 22} between BSA solutions and brain tissue necessitated a separate in vivo AACID pH calibration. The ³¹P-MRS (Figure 5A) and CEST (Figure 5B) spectra were acquired in vivo and postmortem in normal healthy mice (n=3). pH calibration was achieved by fitting a linear model to mean whole brain in vivo and postmortem AACID values plotted against the corresponding ³¹P-MRS pH measurements (Figure 5C). The calibration is valid assuming the P_i and PCr metabolites used to measure pH by ³¹P-MRS exist in the same physical environment as the amide and amine protons used to produce the AACID value⁵ and the pH dependence of the AACID measurement is linear in vivo, as it was in both numerical simulations and in vitro. The resulting linear relation obtained by ³¹P-MRS calibration described in equation (8) was used to produce quantitative pH maps in vivo.

Figure 5.

Amine/amide concentration-independent detection (AACID) pH calibration in vivo using ³¹P-MRS: pH measurements and chemical exchange saturation transfer (CEST) spectra acquired using the standard ³¹P-MRS and ¹H-MRI, respectively, in vivo and postmortem. (A) In vivo (bottom) and postmortem (top) whole brain ³¹P-MRS spectra for one mouse. (B) Corresponding CEST spectra. (C) AACID and pH measurements for three different mice. MRI, magnetic resonance imaging; ³¹P-MRS, phosphorus (³¹P) magnetic resonance spectroscopy.

pH = - 4.0 \times AACID + 12.8

(8)

Acute Acidosis after Middle Cerebral Artery Occlusion

After the pH calibration of AACID measurements in vivo, high-resolution pH maps were produced in mice after permanent MCAO. Several studies have reported local cellular acidosis after induction of ischemia.^{5, 26, 27} Therefore, an MCAO mouse model²⁸ was used to establish the feasibility of in vivo pH measurement using AACID across the range of pH values encountered in ischemic conditions.

After ischemia, anaerobic metabolism causes cellular acidosis.^{5, 26, 29} Quantitative pH maps produced at 2 (Figure 6D) and 5 hours (Figure 6I) after permanent MCAO (n=5) showed a localized pH decrease that corresponded anatomically to the subtle changes observed in T₁-, T₂- and diffusion-weighted MR images (Figures 6A to 6C and 6F to 6H). Regions of interest (ROIs) containing the ischemic core and contralateral brain are shown in Figures 6D and 6I. Figure 6E shows the CEST spectrum from the contralateral brain ROI (blue) superimposed onto the spectrum from the ROI containing the ischemic core (red). At 3.50 to 3.70 p.p.m., the CEST spectra do no overlap, indicating a distinct difference in amide CEST effects (red arrow). Whereas, at 2.75 p.p.m., the CEST spectra nearly overlap, indicating very little change in amine CEST effects (green arrow). This pattern of change was expected based on numerical and in vitro studies that also showed only small changes in amine CEST effect between pH 6.5 and 7.0 (Figures 2A, 2B, 3A, and 3B). Regions of significant acidosis in pH maps were consistently larger 5 hours after surgery compared with 2 hours. For example, in the mouse shown in Figure 5, the region of significant acidosis (pH6.6) within the imaging slice was ∼15 mm³ 2 hours after occlusion and increased to 29 mm³ at 5 hours. At 5 hours after occlusion, the pH values (mouse #: stroke mean±s.d., contralateral tissue mean±s.d.) calculated using the pH maps were (mouse 1: 6.76±0.19, 6.93±0.14), (mouse 2: 6.57±0.13, 6.90±0.13), (mouse 3: 6.67±0.22, 7.02±0.13), (mouse 4: 6.64±0.17, 7.04±0.15), and (mouse 5: 6.54±0.16, 6.94±0.16). Ischemic and contralateral tissue pH mean values were calculated using all pixels in the entire ischemic and contralateral sides, respectively. The average ischemic tissue pH (6.64±0.09) was significantly lower (P<0.01 by t-test, n=5 mice) than the average contralateral tissue pH (6.97±0.06). While pH maps were fairly consistent among different animals, specific quantitative measurements such as ischemic volumes and mean pH may vary with slice location. A series of molecular events initiated by cerebral ischemia lead to increased intracellular water content (i.e., edema) and consequently reduced water flow, leading to increased brain tissue water T₁ relaxation time constants.^{25, 30} However, the numerical simulations and in vitro results from the current study show that these physiologic changes have negligible effect on the ratiometric AACID pH measurement. After MRI, animals were immediately euthanized and histology was performed (n=2). Brain sections corresponding to the acquired pH maps were stained with 2,3,5-triphenyltetrazolium chloride and showed that regions of decreased pH corresponded to regions of low oxidative metabolism indicated by a lack of 2,3,5-triphenyltetrazolium chloride staining (white, Figure 6J). The mean (±s.d.) percent reduction in cerebral blood flow measured with laser Doppler flowmetry³¹ in C57BL/6 mice following MCAO was 87±7% (n=4).

Figure 6.

Cerebral ischemia: representative magnetic resonance (MR) images of a mouse brain 2 hours (A–E) and 5 hours (F–J) after permanent middle cerebral artery occlusion (MCAO) in radiologic orientation. (A, F) T₁-weighted image, (B, G) T₂-weighted image, (C, H) diffusion-weighted image, (D, I) pH map with ischemic (left) and contralateral (right) regions of interest (ROIs) drawn, and (E) chemical exchange saturation transfer (CEST) spectra measured from ischemic (red) and contralateral (blue) ROIs at 2 hours, (E) histology 2,3,5-triphenyltetrazolium chloride (TTC) stained for cell damage.

DISCUSSION

The current study shows a novel method to noninvasively quantify in vivo pH using CEST-MRI contrast. The ratiometric approach called AACID is insensitive to variations in tissue T₁, T₂, macromolecule concentration, and temperature over the relevant physiologic ranges after ischemia. Since the AACID measurement is a relative measure that compares two CEST effects, it is also insensitive to changes in the concentration of amine and amide protons relative to water protons that can be caused by changes in water content, protein content, and diffusion.^{12, 32} However, chemical shift corrections must be made to account for B₀ variation and CEST amplitude corrections must be made to account for B₁ variation on a pixel-by-pixel basis. Using AACID pH measurement, high-resolution maps of brain tissue pH were obtained in mice after permanent MCAO. Regions of significantly decreased pH at 2 and 5 hours after MCAO corresponded to regions of cellular damage.

Previous studies have shown that changes in pH can be measured by noninvasive MRI techniques.^{5, 8, 9} Although the AACID technique avoids several sources of error associated with previous CEST-based methods to measure pH, three assumptions were required to measure quantitative pH using AACID. These assumptions were that: (1) the ratio of exchangeable amide and amine proton populations remains constant throughout the brain and during ischemic conditions, (2) the relationship between AACID values and pH is linear in vivo, as it was found to be in numerical and in-vitro studies, and (3) direct water saturation has a negligible effect on the AACID ratio. The validity of each assumption is discussed below.

First, the AACID technique assumes that the ratio of exchangeable amide to amine proton populations is constant throughout the brain, even in regions with different protein concentrations. Consider that the measured CEST signals are from exchangeable amine and amide protons on mobile peptides and proteins, with the majority existing within the cell. There are 21 standard amino acids that constitute the structural units of peptides and proteins in eukaryotes,³³ along with many other less abundant amino acids that are formed by posttranslation modifications. Of the 21 amino acids, only lysine has exchangeable amine proton on its side arm. Although each amino acid contains an amine functional group, when peptide bonds form between amino acids, the amine is transformed to an amide group. Therefore, peptides and proteins may contain from one to hundreds of exchangeable amide protons, while only proteins with lysine-rich regions contain significant amounts of exchangeable amine groups. The average occurrence of lysine in proteins is ∼6%.³³ In the current study, pH maps did not show any significant contrast between gray and white matter in mouse brain (Figure 6D), suggesting that the amine-to-amide ratio is the same in both gray and white matter. However, we cannot exclude the possibility that the ratio of amine and amide proton populations changes in disease, which would result in inaccurate pH measurements.

Accurate quantitative pH measurement also requires a linear relationship between AACID and pH in vivo. This assumption is reasonable considering a linear relationship was observed in both numerical simulations and in vitro. Numerical simulations provided a convenient theoretical model to investigate complex exchange systems. Simulation of the AACID response to altered pH showed a linear relationship under both aqueous (i.e., no endogenous macromolecule MT) and tissue (i.e., with endogenous macromolecule MT) conditions. However, the simulations also showed that endogenous macromolecule MT significantly shifted the AACID response to pH, necessitating a separate in vivo pH calibration. Additional numerical simulations revealed that physiologically relevant regional changes in endogenous macromolecule MT effects (<10%) do not significantly affect the linear AACID–pH relationship. Measurements of AACID were also insensitive to changes in water content, and variations in T₁ and T₂ relaxation, suggesting that direct saturation has negligible effects at 9.4 T.

The CEST sensitivity of both the amide and amine protons depends on the chemical exchange rate. To detect CEST contrast, the exchange rate must be within the slow-to-intermediate exchange regime at 9.4 T. At physiologic pH and temperature, amide protons associated with endogenous protein have slow exchange rates of 10 to 300/s and APT contrast has been detected when using a low amplitude (∼1.2 μT) saturation pulse.⁵ Conversely, amine protons associated with free amino acids and metabolites have much faster exchange rates (500 to 10,000/s).^{10, 19} However due to the significant CEST effects observed at 2.75 p.p.m. from 10% BSA solutions (Figures 3A and 3B) using a 4-second, 1.5-μT continuous wave RF saturation pulse, the authors suspect that amine protons associated with peptides and proteins are in the slow-to-intermediate exchange regime (e.g., 100 to 1,500/s) at pH7.0.¹⁷In vitro results indicate that at 9.4 T, amine proton exchange rates pass through the slow, intermediate, and fast exchange conditions as pH increases from 6.0 to 8.0. Exchangeable amine protons associated with endogenous peptides and proteins resonating at ∼2.75 p.p.m. were also recently detected by spin-locking techniques using a 5-second saturation pulse with amplitudes as low as ∼2 μT.³⁴ Therefore, in the current experiment, a saturation pulse with a low amplitude (1.5 μT) and a long duration (4 seconds) was used to selectively saturate the relatively slowly exchanging amide and amine protons and avoid contaminant CEST contrast from fast exchanging amine protons associated with common brain metabolites such as glutamate and creatine.^{10, 11}

Since chemical exchange rate is sensitive to pH and temperature, the effects of temperature modulation of the system must also be considered. Zhou et al⁵ recently showed an increase in amide proton exchange rate of ∼1% per 1°C compared with an ∼76% increase in exchange rate after an increase in pH of only 0.4 pH units. These results suggest that base-catalyzed exchange rate effects mediated by pH dominate temperature (Arrhenius) mediated exchange rate effects within the physiologically relevant range. Theoretical modeling of temperature effects is difficult,⁵ therefore in vitro verification was performed in the current study (Figure 3E). Using in vitro protein solutions containing amine and amide protons, AACID values were measured at different temperatures (35°C, 36°C, 37°C, 38°C, and 39°C). As expected from Zhou et al, the amide and amine CEST effect is more sensitive to physiologically relevant changes in pH than to physiologically relevant changes in temperature and AACID values did not change significantly across the measured temperature range (Figure 3E). Within the range of temperature variation expected in cerebral ischemia,³⁵ amine and amide CEST effects were negligibly altered in comparison with the magnitude of the changes observed when pH was altered within the physiologic range.

The use of CEST-MRI contrast to measure pH has evolved over the last several years. The first approach to measure CEST effect was to divide the nuclear magnetic resonance signal intensity after a saturation pulse by the signal intensity without saturation. However, this measurement is contaminated by endogenous macromolecule MT effects, direct water saturation, and blood oxygen level-dependent effects.⁵ Therefore, Zhou et al⁵ introduced the novel MTR_asym curve to maximize CEST sensitivity, while eliminating the mentioned interferences. However when using MTR_asym, APT contrast is modulated by changes in tissue pH, as well as amide content, water content, T₁ relaxation time constants, and macromolecular NOEs; therefore, quantitative pH measurement using APT should be used cautiously.^{5, 6, 7} The AACID pH measurement avoids NOE contamination by normalizing the amide and amine signal intensities after saturation to the signal intensity at 6.0 p.p.m. after saturation. Normalization to 6.0 p.p.m. decreases CEST sensitivity at 2.75 and 3.50 p.p.m. compared with MTR_asym methods because the endogenous macromolecule MT effect is not completely removed. However, normalizing both 2.75 and 3.50 p.p.m. to 6.0 p.p.m. allows the measurement of two different CEST effects while acquiring only three images. The 6.0 p.p.m. point is not influenced by amine or amide CEST effects, or NOEs, and it provides a good indicator of local T₁-related signal loss, blood oxygen level-dependent effects, and macromolecule MT for normalization. Similar results are possible when using any reference at Δω>6.0 p.p.m.; however, CEST sensitivity decreases as Δω increases away from the water resonance because contributions from macromolecule MT will begin to dominate measured AACID values. The current normalization at 6.0 p.p.m provided a linear AACID–pH relation that ensured a straightforward in vivo calibration.

While CEST effects have been detected at 3.50 p.p.m. at clinical field strengths (1.5 to 3 T),³⁶ further work is required to assess the feasibility of measuring the 2.75 p.p.m. amine CEST effect at these lower magnetic field strengths. The AACID method is expected to have lower sensitivity to pH at lower magnetic fields because the amine proton exchange likely approaches the fast exchange rate condition and because direct saturation of water will be increased. However, the method would remain independent of NOE and would require less saturation power than at higher magnetic field strengths. The AACID technique is immediately applicable in humans using current 7 T and 9.4 T human MRI systems, although an initial pH calibration would be required. In addition, three-dimensional AACID pH maps can be generated with significantly higher spatial and temporal resolution using steady-state CEST acquisition methods.³⁷ Application of the AACID technique to measure pH in other diseases such as cancer is possible with proper in vivo pH calibration.³⁸

The quantitative resolution or the uncertainty in a determined value (e.g., pH value) is a useful parameter when comparing different measurement techniques. Using methods described by Pardue et al,³⁹ quantitative resolution is defined as the ratio of the uncertainty in measured response to the sensitivity of a measurement system. In the current study, the quantitative resolution is determined using the in vivo pH calibration data (Figure 5). Results in Figure 5C reveal a measured response uncertainty of 0.036 (i.e., standard deviation of AACID values from the linear fit in Figure 5C) and an AACID pH sensitivity of 0.253/pH. Therefore, the quantitative resolution of pH measurements is 0.14 pH units. Zhou et al⁵ report a slightly lower quantitative resolution, stating that pH changes of ∼0.2 pH units or greater can be detected using amide proton CEST effects.

CONCLUSION

AACID represents a novel technique to quantify in vivo pH using endogenous CEST contrast. Using AACID, CEST-MRI can provide noninvasive, high-resolution pH distribution maps within animal models and could be extended to human subjects. Quantitative pH measurement using AACID is independent of protein concentration, water content, temperature, and NOEs at 9.4 T. In vitro results indicate that glutamate may contribute ∼5% to the AACID measurement in vivo. After in vivo pH calibration, pH was measured by AACID in a mouse model of MCAO. Decreased pH was observed on the ischemic side of the brain at 2 and 5 hours after the start of cerebral ischemia. The tissue region showing decreased pH increased in size between 2 and 5 hours. Absolute in vivo pH measurement could provide an indicator of disease progression, severity, and treatment response.

Footnotes

The authors declare no conflict of interest.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the technical support of K Gilbert for the construction of the ³¹P radiofrequency coil used on the 9.4T MRI.

Notes

Supplementary Information accompanies the paper on the Journal of Cerebral Blood Flow & Metabolism website ()

References

Casey

Grinstein

Orlowski

. Sensors and regulators of intracellular pH. Nat Rev Mol Cell Biol 2010; 11: 50–61.

Kalimo

Rehncrona

Soderfeldt

Olsson

Siesjo

. Brain lactic-acidosis and ischemic cell-damage. 2. Histopathology. J Cereb Blood Flow Metab 1981; 1:313–327.

Raghunand

. Tissue pH measurement by magnetic resonance spectroscopy and imaging. Methods Mol Med 2006; 124: 347–364.

Moon

Richards

. Determination of intracellular pH by 31P magnetic resonance. J Biol Chem 1973; 248: 7276–7278.

Zhou

Payen

Wilson

Traystman

van Zijl

. Using the amide proton signals of intracellular proteins and peptides to detect pH effects in MRI. Nat Med 2003; 9: 1085–1090.

Liu

Zhou

Xue

Zuo

Wang

. Quantitative characterization of nuclear Overhauser enhancement and amide proton transfer effects in the human brain at 7 tesla. Magn Resort Med 2012; 70: 1070–1081.

Jin

Wang

Zong

Kim

. MR imaging of the amide-proton transfer effect and the pH-insensitive nuclear overhauser effect at 9.4T. Magn Reson Med 2013; 69: 760–770.

Sun

Cheung

Wang

. Association between pH-weighted endogenous amide proton chemical exchange saturation transfer MRI and tissue lactic acidosis during acute ischemic stroke. J Cereb Blood Flow Metab 2011; 31: 1743–1750.

Sun

Zhou

Sun

Huang

van Zijl

. Detection of the ischemic penumbra using pH-weighted MRI. J Cereb Blood Flow Metab 2007; 27: 1129–1136.

10.

Cai

Haris

Singh

Kogan

Greenberg

Hariharan

. Magnetic resonance imaging of glutamate. Nat Med 2012; 18: 302–306.

11.

Haris

Nanga

Singh

Cai

Kogan

Hariharan

. Exchange rates of creatine kinase metabolites: Feasibility of imaging creatine by chemical exchange saturation transfer MRI. NMR Biomed 2012; 25: 1305–1309.

12.

Ward

Balaban

. Determination of pH using water protons and chemical exchange dependent saturation transfer (CEST). Magn Reson Med 2000; 44: 799–802.

13.

Desmond

Stanisz

. pH mapping based on the ratiometric amide and amine relationship from endogenous CEST. Contrast Media Mol Imaging 2013; 8: 301–302.

14.

Zong

Wang

Kim

Jin

. Sensitivity and source of amine-proton exchange and amide-proton transfer magnetic resonance imaging in cerebral ischemia. Magn Reson Med 2013; 71: 118–132.

15.

Zhou

Lai

Wilson

Laterra

van Zijl

. Amide proton transfer (APT) contrast for imaging of brain tumors. Magn Reson Med 2003; 50: 1120–1126.

16.

Hall

. Guyton and Hall Textbook of Medical Physiology: Enhanced E-book. In: Sciences ( Gruliow

, ed.). Saunders: Philadelphia, PA, 2010.

17.

Woods

Woessner

Sherry

. Paramagnetic lanthanide complexes as PARACEST agents for medical imaging. Chem Soc Rev 2006; 35: 500–511.

18.

McVicar

Suchy

Hudson

Menon

Bartha

. Simultaneous in vivo pH and temperature mapping using a PARACEST-MRI contrast agent. Magn Reson Med 2012.

19.

Liepinsh

Otting

. Proton exchange rates from amino acid side chains—implications for image contrast. Magn Reson Med 1996; 35: 30–42.

20.

Covington

Robinson

Bates

. Ionization constant of deuterium oxide from 5 to 50 degrees. J Phys Chem 1966; 70: 3820–3824.

21.

Jones

. NOE imaging in the human brain at 7T. Contrast Media Mol Imaging 2013; 8: 114–124.

22.

Hudson

RHE

Barrett

Jones

Pasternak

Bartha

. Four-pool modeling of proton exchange processes in biological systems in the presence of MRI-Paramagnetic Chemical Exchange Saturation Transfer (PARACEST) agents. Magn Reson Med 2008; 60: 1197–1206.

23.

Makela

Kettunen

Grohn

OHJ

Kauppinen

. Quantitative T-1 rho and magnetization transfer magnetic resonance Imaging of acute cerebral ischemia in the rat. J Cereb Blood Flow Metab 2002; 22: 547–558.

24.

Banay-Schwartz

Kenessey

Deguzman

Lajtha

Palkovits

. Protein-content of various regions of rat-brain and adult and aging human brain. Age 1992; 15: 51–54.

25.

Barber

Hoyte

Kirk

Foniok

Buchan

Tuor

. Early T1- and T2-weighted MRI signatures of transient and permanent middle cerebral artery occlusion in a murine stroke model studied at 9.4T. Neurosci Lett 2005; 388: 54–59.

26.

Combs

Dempsey

Maley

Donaldson

Smith

. Relationship between plasma glucose, brain lactate, and intracellular pH during cerebral ischemia in gerbils. Stroke 1990; 21: 936–942.

27.

Mabe

Blomqvist

Siesjo

. Intracellular pH in the brain following transient ischemia. J Cereb Blood Flow Metab 1983; 3: 109–114.

28.

Beraldo

Soares

Goncalves

Fan

Thomas

Santos

. Stress-inducible phosphoprotein 1 has unique cochaperone activity during development and regulates cellular response to ischemia via the prion protein. FASEB J 2013;27:3594–3607.

29.

Sun

Wang

Cheung

. Imaging acute ischemic tissue acidosis with pH-sensitive endogenous amide proton transfer (APT) MRI-correction of tissue relaxation and concomitant RF irradiation effects toward mapping quantitative cerebral tissue pH. Neuroimage 2012; 60: 1–6.

30.

Rosenberg

. Ischemic brain edema. Prog Cordiovasc Dis 1999; 42: 209–216.

31.

Goldlust

Paczynski

Hsu

Goldberg

. Automated measurement of infarct size with scanned images of triphenyltetrazolium chloride-stained rat brains. Stroke 1996; 27: 1657–1662.

32.

Ali

Liu

Shah

Flask

Pagel

. Using two chemical exchange saturation transfer magnetic resonance imaging contrast agents for molecular imaging studies. Acc Chem Res 2009; 42: 915–924.

33.

Fasman

. Prediction of protein structure and the principles of protein conformation. Plenum Press: New York, NY, 1989.

34.

Jin

Wang

Zong

Kim

. Magnetic resonance imaging of the Amine-Proton Exchange (APEX) dependent contrast. Neuroimage 2012; 59: 1218–1227.

35.

Schwab

Spranger

Aschoff

Steiner

Hacke

. Brain temperature monitoring and modulation in patients with severe MCA infarction. Neurology 1997; 48: 762–767.

36.

Jones

Schlosser

van Zijl

Pomper

Golay

Zhou

. Amide proton transfer imaging of human brain tumors at 3T. Magn Reson Med 2006; 56: 585–592.

37.

Jones

Polders

Hua

Zhu

Hoogduin

Zhou

. In vivo three-dimensional whole-brain pulsed steady-state chemical exchange saturation transfer at 7T. Magn Reson Med 2012; 67: 1579–1589.

38.

Desmond

Moosvi

Stanisz

. Mapping of amide, amine, and aliphatic peaks in the CEST spectra of murine xenografts at 7T. Magn Reson Med 2013.

39.

Pardue

. The inseparable triad: Analytical sensitivity, measurement uncertainty, and quantitative resolution. Clin Chem 1997; 43: 1831–1837.

40.

Venkatesan

Lin

Gurleyik

Paczynski

Powers

. Absolute measurements of water content using magnetic resonance imaging: Preliminary findings in an in vivo focal ischemic rat model. Magn Reson Med 2000; 43: 146–150.