Introduction
Acidosis is a common microenvironmental feature of most solid tumors [
1]. Acidification of the extra cellular space is a consequence of high rates of glucose metabolism combined with inefficient tumor perfusion [
2,
3]. Low extracellular pH can induce tumor cell migration, invasion and metastasis by poorly defined mechanisms [
4,
5]
. Acute exposure of cells to acidic pH has been shown to cause up-regulation of several secreted proteases, such as cathepsins D, L, and/or B, [
6‐
8] and increase expression of the matrix metalloproteinases (MMP); MMP-2 (gelatinase A) and MMP-9 (gelatinase B) in vitro [
9,
10]. Additionally, pretreatment of melanoma cells with acidic pH before tail vein injection leads to increased experimental metastases in vivo [
11,
12]. We have shown previously that neutralizing the acid pH of tumors with oral NaHCO
3 reduced spontaneous metastases without affecting systemic pH [
13]. Mathematical reaction-diffusion type models were used to quantitatively describe the effect, thus providing a theoretical framework within which to interpret the data. The p
K
a of NaHCO
3 in water is 6.14 [
14]; and models predicted that the most efficacious buffers would have a p
K
a of ~7.0. Additionally, the models assumed that bicarbonate was working as a buffer per se, and did not assume a specific interaction with carbonic anhydrases, which are used in vivo to dehydrate HCO
3
−. The current study was undertaken to test the efficacy of a non-bicarbonate/non-volatile buffer with a higher p
K
a on tumor growth and metastases. For this, we have used a non-volatile buffer with a p
K
a of ~6.9, 2-imidazole-1-yl-3-ethoxycarbonyl-propionic acid (IEPA).
IEPA is a synthetic imidazole buffer that has been used to map the extracellular pH, pHe, in different animal models using magnetic resonance spectroscopic imaging, MRSI [
15‐
17]. In this paper, we studied the effect of IEPA on spontaneous and experimental metastases. We first showed that oral IEPA was tolerable at doses up to 200 mM, as evidenced by animal weight gain. We demonstrated that IEPA was absorbed into the systemic circulation using MRS. Measurement of the pH in the primary subcutaneous tumor showed that IEPA slightly increased the tumor pH. IEPA also had a moderate but statistically insignificant effect on the growth rate of the primary tumor, which was consistent with previous results using bicarbonate. We then observed that IEPA had a significant inhibitory effect on experimental metastases in the PC3M prostate model. Although the mechanism by which inhibiting metastases has not yet been determined, these data combined with the previous bicarbonate studies [
14] demonstrate that non-volatile buffers are effective in reducing tumor metastases, and thus the effects appear to be due to buffering per se.
Materials and methods
Chemicals
IEPA (2-imidazole-1-yl-3-ethoxycarbonyl-propionic acid) was obtained from Dr. Paloma Ballesteros (Laboratory of Organic Synthesis and Molecular Imaging by Magnetic Resonance (Ref. Lab. 250) Faculty of Sciences, UNED Pº Senda del Rey, 9 28040-Madrid, Spain), and was dissolved in tap water at a concentration of 200 mM. The pH of the IEPA solution was adjusted to 7.4 (using 1 M HCl) and then given to the mice to drink in place of their regular tap water. Cell culture media and supplies were obtained from Invitrogen, Carlsbad, CA. Luciferin was obtained from Gold Biotechnology, St. Louis, MO.
Animals housing and diet
All animals were maintained in accordance with IACUC standards of care in pathogen free rooms, in the Moffitt Cancer Research Center (Tampa, FL) Vivarium. All imaging and measurements were performed within the facility. Three days prior to inoculation with tumor cells, 4–6 week old male beige SCID mice (Harlan, Madison, WI) were placed in two cohorts per experiment, which were allowed to drink either tap water or 200 mM IEPA. The weights of the water bottles were recorded before and after providing them to the animals, thereby tracking the amount of liquid consumed per cage. Animal weights were measured and recorded twice weekly, and the overall health of each animal was noted to ensure timely endpoints within the experiment.
Cell culture and inoculation
PC3M cells (-Luc6 clone) were obtained from Caliper (Hopkinton, MA). The cells were cultured using MEM/EBSS media, supplemented with 10% fetal bovine serum, 1% penicillin streptomycin, 1% nonessential amino acids, 1% sodium pyruvate and 1% MEM vitamins. In preparation for inoculation into mice, the cells were trypsinized and rinsed once with sterile phosphate buffered saline (PBS) prior to resuspension at a concentration of 5 × 106 cells in 200 μl PBS. For primary tumor injection, animals were prepared by removing the hair from the injection site, and 200 μl containing 5 × 106 cells in PBS were injected subcutaneously into the right flank of each mouse. For experimental metastases, 200 μl containing 5 × 106 cells in PBS were injected directly and slowly (over the course of 1 min) into the tail vein of each mouse. In both preparations, cell distributions were verified by bioluminescent imaging immediately following injection.
Bioluminescent imaging
Animals were anesthetized with isoflurane and injected intraperitoneally with 10 μl per g body weight of sterile d-luciferin substrate prepared in PBS at 15 mg/ml (resulting dose 150 μg/g body weight). After 5 min, mice were transferred to the thermo-regulated, light-tight chamber of the In Vivo Imaging System, IVIS-200 (Caliper; Hopkinton, MA). Photographic images were acquired first, and the bioluminescent images were overlaid on top of these images. Bioluminescent images were acquired by measuring photons emitted from luciferase-expressing cells and transmitted through the tissue. The exposure time for the bioluminescent image acquisition ranged from 0.5 s (whole tumor images) up to 2 min (lung metastases) to ensure non-saturation, and differences in exposure time were corrected by expressing data as total flux in photons/sec, rather than photon counts. Images were analyzed using the LivingImage software (Caliper; Hopkinton, MA).
Necrosis counting
The center section (~5 mm) of the subcutaneous tumor of each animal was fixed in paraffin blocks prior to staining one 4 μm thick cross-sectional sample per animal with hematoxylin and eosin for histology. Histology slides were scanned using the Aperio™ (Vista, CA) ScanScope XT with a 20×/0.8NA objective lens (200×) at a rate of 2 min per slide via Basler tri-linear-array. Image analysis was performed using an Aperio Genie® v1 customized algorithm in conjunction with Positive Pixel Count v9 with the following optimized thresholds [Hue value = 0.2; Hue width = 0.6; color saturation threshold = 0.05; IWP(High) = 210; Iwp(Low) = Ip(High) = 160; Ip(low) = Isp(High) = 80; Isp(Low) = 0]. The algorithm was applied to the entire slide’s digital image to determine the percentage of necrosis by detecting the number of pixels that satisfy the color and intensity specification defined above (necrotic), divided by the number of pixels in non-necrotic tissue. The training algorithm developed above was quality controlled by a practicing pathologist.
Magnetic resonance imaging and spectroscopy
MR images and spectra were obtained on a Varian MR imaging spectrometer ASR310 (Agilent Life Sciences Technologies, Santa Clara, CA) with a 30 cm horizontal clear bore operating at a field strength of 7 T. For reference, a high resolution spectrum of IEPA in D2O was obtained on a Varian Nuclear Magnetic Resonance spectrometer with a 54 mm vertical bore opening at a field strength of 9.4 T. For in vivo spectroscopic imaging, naïve mice were allowed to drink IEPA for 3 days prior to imaging. The animals were sedated using isoflurane, placed in the animal cradle for insertion into the bore of the 7 T Varian MRI and maintained warm using a continuous warm air blower (Small Animal Instruments, Inc., Stonybrook, NY). Temperatures were measured using a fiber optic endorectal thermometer in conjunction with the MR compatible animal monitoring system (Model 1025, Small Animal Instruments, Inc. Stony Brook, NY.). SCOUT images were taken to verify location, and T2 weighted images for anatomical identification were obtained using an fast spin echo (FSEMS) pulse sequence, with FOV = 40 × 80 (mm), 15 coronal slices, 1 mm thick, no gap, TR = 2450 s, effective TE = 72 ms, with fat suppression on. Spectra were obtained using a stimulated echo (STEAM) localization sequence on a 2 × 2 × 2.5 mm3 voxel in the bladder with 256 averages (flip angle 90 deg, TE 9.44, TM 8.01, and TR 2000), for an 8.5 min acquisition. Images and spectra were processed using the Varian Vnmrj software or using MATLAB (MathWorks, Inc, Natick, MA).
Electrode measurement of pH
Animals were sedated using isoflurane, and placed on a warming surface to maintain appropriate body temperature for the duration of the experiment. Both the needle microelectrode and the reference electrode were obtained from Microelectrodes, Inc., (Bedford, NH). A shallow small (<5 mm) incision was made in an alternate (non tumor) site and the 1 mm reference electrode was placed subcutaneously therein. A needle micro electrode (OD 0.8 mm with a beveled end) was inserted up to 1.3 cm into the center of the tumor, and was held in place for up to 1 min, until pH readings stabilized. The needle was rotated once in each location, to allow the pH electrode to re-read at the same depth in order to make two independent measurements per location. The pH was measured at three locations, one near the center/core of the tumor, one in a mid region of the tumor, and one at the rim of the tumor; these values were averaged to report a mean for each animal. After the pH was measured in the primary tumor, the animals were euthanized (29 days after subcutaneous injection of the primary tumor cells). Before and after the pH was measured in each animal, the pH electrodes were used to measure a standard pH 7 buffer solution (Thermo Fisher Scientific, Inc., Waltham, MA).
Statistics
Non-paired data were analyzed for differences in their mean by two-tailed Student’s
t-test. Chauvenet’s criterion was used for outlier analysis of independent data sets. A mouse was rejected from its set based on an expected deviance value of below one-half. The expected deviance values for all other physical measurements were above one-half showing two otherwise valid sets [
18].
Discussion
Previous work has shown that a volatile buffer, i.e. bicarbonate, was effective in inhibiting spontaneous and experimental metastases [
13]. The aim of the current study was to investigate the effects of non-volatile buffers, specifically IEPA, to test the hypothesis that metastases inhibition was due to buffering and was not specific to bicarbonate. Metastasis is the leading cause of death from cancer [
23] and can be promoted by low pH and acidity; which is a common feature of solid tumors. The extracellular pH (pHe) of solid tumor is acidic, e.g. pHe of 6.5–7.1, compared to normal tissues with a pHe of 7.3–7.4. Experimental observations have shown that acid-mediated invasion can occur via destruction of the extracellular matrix, which is promoted by proteases and glycosidases and may convert in situ to locally invasive cancers [
29]. An ‘acid-mediated invasion hypothesis’ was developed using mathematical models supported by empirical observations [
14], and states that: (1) tumors export H
+ into the adjacent stroma; (2) this results in damage to supporting stroma accompanied by cell death and extracellular matrix (ECM) break down; which (3) facilitates the invasion of the tumor cells and consequently colonization at distant sites. As a test of the model we observed that chronic oral administration of sodium bicarbonate (NaHCO
3), a physiologic buffer, was able to increase tumor pHe and reduce lung metastasis in spontaneous and experimental breast and prostate animal models. Neutralization of tumor acidity was measured with fluorescence microscopy of a dorsal window chamber and
31P magnetic resonance spectroscopy. Even though NaHCO
3 was effective and inexpensive, its p
K
a may be lower than optimal, and it was unclear whether it was working as a buffer, or if the effect was specific for bicarbonate. Therefore, in this study we examined a non-volatile buffer with a higher p
K
a value for its effects on spontaneous metastasis.
We used IEPA with pK
a ~ 7, and the PC3M prostate cancer cell line was chosen because, as shown in our previous work, it was responsive to bicarbonate. IEPA was shown to be absorbed into the blood stream, and had no negative effects on the animal’s growth or behavior. In a subcutaneous tumor model, we found that IEPA only slightly decreased the growth rate of the primary tumor. However this decrease in size was accompanied with a small but significant increase in the primary tumor pH and a substantial change in tumor anatomy; with a lower percentage necrosis when compared to control animals.
Although bone marrow is the most common site for prostate cancer metastasis, it is also known to metastasize to the lungs [
24]. Spontaneous metastases, as measured by in vivo upper thoracic bioluminescence, were decreased in the IEPA-treated animals. Experimental metastases following tail vein injection were measured by total and thoracic in vivo bioluminescence, and these were also substantially decreased in the IEPA-treated animals. Hence, this non-volatile buffer was effective in reducing spontaneous and experimental metastases, compared to bicarbonate. This suggests that the effect is due to pH buffering and is not due to a specific bicarbonate effect per se. The fact that both spontaneous and experimental metastases were inhibited also suggests that the effect of the buffer therapy is on extravasation or/and colonization and not intravasation.
To investigate if the increase in tumor pH could be responsible for the remarkable decrease in metastases, we used the subcutaneous (spontaneous) model to measure the pH of the primary tumor. We observed a slight (statistically significant) increase in the primary tumor pH with IEPA treatment, accompanied by a substantial decrease in necrotic tumor anatomy. Although statistically significant, the change in pHe of the primary tumor was very slight (0.06 pH units) and the physiological relevance of this change is unclear. In work with other tumor models, we have observed that the effect of bicarbonate on tumor pH was reduced in large and/or rapidly metabolizing tumors (data not shown). The slight effects seen in this system are also consistent with reaction-diffusion models, which also predict that a larger and more substantial effect may take place at the site of colonization, when tumors are small (ca. 1 mm) and poorly perfused [
14]. Unfortunately, no methods are available that can measure the pHe of these small nascent colonies. Thus, we propose that the major effect of buffering occurs at metastatic sites, where acid load is lower due to smaller foci and where reducing acidity can inhibit local invasion [
25‐
27].
Several mechanisms could potentially contribute to the effect of buffers on tumor metastasis. Angiogenic factors, such as vascular endothelial growth factor (VEGF-A) and interleukin 8 (IL-8) are upregulated by acidic pHe [
28]. Neovascularization influences the dissemination of cancer cells throughout the body ultimately leading to metastasis formation. IEPA may possibly function through increasing pHe thereby affecting angiogenesis and consequently metastasis. The acid-mediated invasion can occur via destruction of the extracellular matrix, which is promoted by proteases and glycosidases and may convert in situ to locally invasive cancers [
29]. Matrix metalloproteinases (MMP-2 and MMP-9) are believed to be critical for invasion. Even tough the MMPs have alkaline pH optima, low pH will increase their activation and release [
11,
30]. Lysosomal proteases (cathepsins) are more active at acid pH, are released in invasive cancers, and function to proteolytically activate MMPs [
31]. Thus, it is possible that the buffers, by neutralizing tumor acidosis, are decreasing the proteolytic enzyme activation and prevent the ECM degradation which leads to decrease metastasis.
In summary significant reduction of metastasis with IEPA buffer support the hypothesis that non-volatile buffers with p
K
a ≥ 7 should be more effective in buffering extra cellular acidity [
14]. This study specifically used prostate cancer model PC3M because in our previous work we have shown that this model was highly responsive to bicarbonate treatment. Previous work has shown that bicarbonate was ineffective in inhibiting metastases from B16 melanoma. These cells are more metabolically active and produce H
+ at a higher rate, compared to PC3M (data not shown) and hence, we hypothesize that they may be inhibitable with more buffering power, such as that of non-volatile buffers.
These findings suggest that non-volatile buffers, such as imidazoles may have the potential for clinical benefit. Sodium bicarbonate is currently in two clinical trials as a buffer; however compliance is low because of the taste. Thus, there is a need to find a palatable and effective alternative. The current work shows evidence that other buffers may be considered as an alternative to the NaHCO
3 in clinical trials with the possibility of better tolerance in addition to greater efficacy. This study indicated no toxicity in mouse models, in agreement with previous findings [
32,
33]. Additionally, the measurements performed in the study by Garcia Martin et al. [
15] indicate that the IEPA was able to extravasate only in the areas of the brain affected by the glioma and not in normal brain. Thus IEPA is not expected to cross the normal blood brain barrier within detectable limits.
In conclusion, this study demonstrates that a non-volatile buffer can reduce the incidence and growth of experimental metastases in a model of prostate cancer. Previous work using buffer therapy has exclusively used bicarbonate, which may act through alternative pathways, e.g. through activation of carbonic anhydrases. The current work demonstrates that alternative, non-volatile buffers can be equally efficacious as bicarbonate. Thus the mechanism of metastasis inhibition appears to be limited to pH buffering. It also introduces alternative buffers that may be more applicable in a clinical setting for acute or chronic use.