Fine Spatiotemporal Control of Nitric Oxide Release by Infrared Pulse

Aug 26, 2013 - Kyoto Prefectural University of Medicine, 13, Taishogun, ... Chemistry, School of Science and Technology, Kwansei Gakuin University, Sa...
0 downloads 0 Views 3MB Size
Download PDF
Articles pubs.acs.org/acschemicalbiology

Fine Spatiotemporal Control of Nitric Oxide Release by Infrared Pulse-Laser Irradiation of a Photolabile Donor Hidehiko Nakagawa,*,†,∥ Kazuhiro Hishikawa,† Kei Eto,‡ Naoya Ieda,† Tomotaka Namikawa,⊥,# Kenji Kamada,⊥,# Takayoshi Suzuki,†,§ Naoki Miyata,† and Jun-ichi Nabekura‡ †

Graduate School of Pharmaceutical Sciences, Nagoya City University, 3-1, Tanabe-dori, Mizuho-ku, Nagoya, Aichi 467-8603, Japan National Institute of Physiological Sciences, 38, Nishigonaka, Myodaiji, Okazaki, Aichi 444-8585, Japan § Kyoto Prefectural University of Medicine, 13, Taishogun, Nishitakatsukasa-cho, Kita-ku, Kyoto 403-8334, Japan ∥ Japan Science and Technology Agency, PRESTO, 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan ⊥ Research Institute for Ubiquitous Energy Devices, National Institute of Advanced Industrial Science and Technology (AIST), Ikeda, Osaka 563-8577, Japan # Department of Chemistry, School of Science and Technology, Kwansei Gakuin University, Sanda, Hyogo 669-1337, Japan ‡

S Supporting Information *

ABSTRACT: Two-photon-excitation release of nitric oxide (NO) from our recently synthesized photolabile NO donor, Flu-DNB, was confirmed to allow fine spatial and temporal control of NO release at the subcellular level in vitro. We then evaluated in vivo applications. Femtosecond near-infrared pulse laser irradiation of predefined regions of interest in living mouse brain treated with Flu-DNB induced NO-releasedependent, transient vasodilation specifically at the irradiated site. Photoirradiation in the absence of Flu-DNB had no effect. Further, NO release from Flu-DNB by pulse laser irradiation was shown to cause chemoattraction of microglial processes to the irradiated area in living mouse brain. To our knowledge, this is the first demonstration of induction of biological responses in vitro and in vivo by means of precisely controlled, twophoton-mediated release of NO.

N

We have previously reported on photoinduced NO release from 4-substituted 2,6-dimethylnitrobenzenes.7 In this type of compounds, the steric effect of the two methyl groups at the ortho-positions causes the nitro group to take a twisted conformation with respect to the benzene ring, and thereby facilitates isomerization of the nitro group to nitrite, followed by release of NO. These dimethylnitrobenzene derivatives are unique as NO donors in their mechanism of NO release. However, their maximum absorption band for NO release is in the UVA range, which is harmful to living cells. To overcome this limitation, it would be highly advantageous to develop compounds working at longer wavelength, ideally in the nearinfrared (NIR, 700−900 nm) range, where photoirradiation is less biologically harmful and penetrates deeper into tissues. To achieve this, we adopted the two-photon excitation (TPE) technique, which is already used for fluorescence bioimaging, etc. Among our previously synthesized 4-substituted 2,6dimethylnitrobenzene derivatives, we found that Flu-DNB (1,

itric oxide (NO) has pleiotropic functions in biological systems, including roles in blood pressure regulation, neuromodulation, and biodefense.1−4 Since NO is unstable under ambient conditions, it is difficult to apply NO directly to biological systems, and therefore, various NO donors have been developed and employed for biological studies.5 Indeed, NO donors have been indispensable for biomedical research on NO, but many of them are spontaneous NO donors that release NO via spontaneous decomposition. Once such donors are added to media, NO release begins immediately at a rate that depends on the nature of the donor and the solution conditions, so that it is practically impossible to control the amount, rate, or location of NO treatment. However, endogenous NO formation is considered to be precisely regulated by nitric oxide synthases and signal transduction machinery. Thus, controllable NO donors seem to be essential for detailed investigation of the physiological roles of NO and for developing potential therapeutic agents. Photocontrolled release of NO is expected to be useful for this purpose,6 allowing noninvasive or minimally invasive injection of NO with a photosyringe in a highly spatially and temporally controlled way. © 2013 American Chemical Society

Received: May 22, 2013 Accepted: August 26, 2013 Published: August 26, 2013 2493

dx.doi.org/10.1021/cb400361m | ACS Chem. Biol. 2013, 8, 2493−2500

ACS Chemical Biology

Articles

Chart 1. Chemical Structure of Flu-DNB and Proposed Mechanism of NO Release upon Photoirradiation

due to DAR-4 M T (Figure 1a,b). This result indicated that UVA irradiation induced NO release in Flu-DNB-loaded cells,

Chart 1a) released NO in response to femtosecond nearinfrared pulse laser irradiation, as well as conventional UVA range photoirradiation (Chart 1b).8 Here, we show that very fine spatiotemporal control of NO release can be achieved with Flu-DNB. We first established that control of NO release at the subcellular level in vitro can be achieved with this system. We then confirmed its utility in vivo, by observing NO-dependent vasodilation, as well as chemoattraction of microglial processes, in living mouse brain in response to femtosecond near-infrared pulse laser irradiation through a cranial window after Flu-DNB administration. Though NO release utilizing the TPE technique has been reported,9−11 the NO donors that were used employed a nitrosyl-chelated metal ion moiety for NO release and are unsuitable for biological applications. The present report is the first to demonstrate precisely controlled NO release in cells and in vivo using a TPE-type NO donor.

■

Figure 1. Confocal fluorescence images of cells treated with DAR-4 M AM in the presence or absence of Flu-DNB with photoirradiation in the UVA range. HCT116 cells were treated with or without Flu-DNB for 24 h and then with DAR-4 M AM. The treated cells were photoirradiated with UVA light (325−385 nm). Representative images are shown in panels a−d: (a) Confocal images of cells treated with FluDNB but not photoirradiated; (b) treated with Flu-DNB and photoirradiated; (c) not treated with either Flu-DNB or photoirradiation; (d) not treated with Flu-DNB but photoirradiated. All cells were treated with DAR-4 M AM.

RESULTS AND DISCUSSION UVA-Induced NO Release in Flu-DNB-Treated Cultured Cells. HCT116 human colon cancer cells were treated with 25 μM Flu-DNB for 24 h, and then 10 μM DAR-4 M AM, a prodrug form of a fluorogenic NO probe, was loaded into the cells for NO detection. DAR-4 M AM is intracellularly hydrolyzed to the active form (DAR-4M), which reacts with NO in the presence of molecular oxygen to afford a red fluorescent product, DAR-4 M T.12 The cells were also treated with L-NNA, a nitric oxide synthase (NOS) inhibitor, to suppress possible endogenous NO formation from NOS in HCT116 cells. The treated cells were irradiated with UVA (325−385 nm) and observed by confocal fluorescence microscopy (see Supplementary Methods for experimental procedures). Staining of the cells with Flu-DNB was confirmed by the observation of the green fluorescence signal of Flu-DNB (Supplementary Figure S1), in which the fluorescein moiety is advantageous for checking cellular distribution of the compound. Further, while a very low fluorescence signal of DAR-4 M T was observed in nonirradiated cells, the cells irradiated with UVA showed significant red fluorescence signals

demonstrating that Flu-DNB can release NO in a cellular system. UVA irradiation in the absence of Flu-DNB did not cause any increase of DAR-4 M T fluorescence (Figure 1c,d). Photoirradiation at the wavelength and intensity used in this experiment did not have any apparent effect on cell viability, and Flu-DNB itself did not show severe cytotoxicity at the concentration used (IC50 of Flu-DNB >100 μM by MTT assay) in the absence of photoirradiation. Near-Infrared (NIR) Pulse Laser-Induced NO Release in Flu-DNB-Treated Cultured Cells. Next we examined NO release in cells treated with Flu-DNB and exposed to NIR pulse laser irradiation. We had previously confirmed that Flu-DNB 2494

dx.doi.org/10.1021/cb400361m | ACS Chem. Biol. 2013, 8, 2493−2500

ACS Chemical Biology

Articles

Figure 2. Confocal fluorescence images of cells pretreated with DAR-4 M AM and irradiated with the 735 nm pulse laser in the presence or absence of Flu-DNB. Representative images in the presence of Flu-DNB are shown in panels a−d: (a,b) confocal images before (a) and after (b) pulse laser irradiation (735 nm) of the area indicated by the broken circle for uncaging NO; (c,d) images before (c) and after (d) pulse laser irradiation of the point indicated by the arrow for uncaging NO. Representative images in the absence of Flu-DNB are shown in panels e−h: (e,f) fluorescence image before (e) and after (f) pulse laser irradiation at the point indicated by the broken circle; (g,h) image before (g) and after (h) irradiation of cell indicated by the arrow. Changes of DAR-4 M T fluorescence after pulse laser irradiation to the points in vehicle- and Flu-DNB-treated cells are shown in panel i. Schematic illustration of the experimental settings of laser scanning microscopy was shown in panel j. Data represent means and s.e.m. (n = 3). **P < 0.01 versus fluorescein. Scale bars = 50 μm.

releases NO in vitro upon irradiation with a femtosecond pulse laser in the range of 720 to 800 nm.8 HCT116 cells were treated with Flu-DNB and DAR-4 M T in the same way as for the one-photon (UVA) irradiation experiments described above, and subjected to femtosecond NIR pulse laser irradiation at 735 nm. When the laser was scanned over a whole cell, a large increase of the fluorescence signal of DAR-4 M T was observed only in the irradiated area (Figure 2a,b). Irradiation at a single site within a cell resulted in an increase of DAR-4 M T fluorescence at the irradiated location (Figure 2c,d). These results indicate that Flu-DNB released NO in the cells under two-photon irradiation conditions and confirm that NO release can be spatiotemporally restricted by appropriately controlling the irradiation area. Vehicle treatment did not result in any fluorescence increase (Figure 2e−h), and no phototoxicity was apparent under the experimental conditions. NIR Pulse Laser-Induced NO Release in Flu-DNBLoaded Living Mouse Brain. Two-photon irradiation generated by a femtosecond pulse laser at NIR wavelength can illuminate spatiotemporally restricted areas of biological samples with minimal photodamage, suggesting it is suitable for in vivo application. Therefore, we next examined photoinduced NO release inside tissues by NIR pulse laser irradiation after Flu-DNB treatment. For this purpose, we chose mouse brain as

a model tissue because the surface of the brain tissue can be easily accessed through a burr hole in the skull, and we have previously shown that a shallow area of the brain can be observed by two-photon fluorescence microscopy.13 First, we investigated whether photoinduced NO release in mouse brain influences blood vessels. C57BL/6J mice were anesthetized and a cranial window was made. A small pore was carefully made on the dura without irritating the brain parenchyma and covered with a droplet of artificial cerebral fluid (ACSF) containing 100 mM Flu-DNB and 1% (v/v) DMSO as a cosolvent for 3 h. The cranial window was covered with a coverslip. Observation of Flu-DNB fluorescence showed that Flu-DNB was mainly distributed to vessel walls to a depth of at least 100 μm from the brain surface. For convenience in quantitative analysis of the movement of vessel walls, we selected arteries over around 20 μm in diameter on the surface of brain and examined whether photoinduced NO release induced a vasodilation response of these vessels in vivo. We found that NIR pulse laser irradiation focused on the blood vessel wall induced a transient, but significantly increased of the vessel diameter (Figure 3a,b). In contrast, photoirradiation of blood vessels in the presence of fluorescein (a partial structure of Flu-DNB lacking the NO-releasing moiety; used as a control) instead of Flu-DNB failed to induce any diametric 2495

dx.doi.org/10.1021/cb400361m | ACS Chem. Biol. 2013, 8, 2493−2500

ACS Chemical Biology

Articles

Figure 3. Two-photon fluorescence images of vessels in mouse brain treated with Flu-DNB. Mouse brain treated with Flu-DNB or fluorescein was irradiated with the pulse laser (735 nm) at the ROI indicated by red lines. Two-photon images were obtained simultaneously by scanning with the acquisition pulse laser (950 nm). The observed fluorescence was from Flu-DNB or fluorescein. Representative images are shown in panels a−d: (a,b) two-photon fluorescence images of a vessel in mouse brain treated with Flu-DNB immediately before (a) and during (b) the uncaging pulse irradiation; (c,d) images of mouse brain without Flu-DNB immediately before (c) and during (d) the uncaging pulse irradiation. Changes of vessel diameter after pulse laser irradiation in fluorescein (control) and Flu-DNB-treated mouse brains are shown in panel g. Data represent means and s.e.m. (n = 21 for fluorescein and n = 28 for Flu-DNB). The vessel diameter before irradiation is shown by a yellow scale in each pair of images. Close-up views of panels a and b are shown as panels e and f, respectively. Enlargement of vessel diameter in the Flu-DNB treated brain is indicated by a pair of yellow arrows in panels a, b, e, and f, whereas it was not apparent in the control (d). A schematic illustration of the experimental settings of laser scanning microscopy is shown in panel h. **P < 0.01 versus DMSO. Scale bars = 30 μm.

expressing GFP in microglia were anesthetized, and a cranial window for imaging was made to evaluate the motility of microglia.17 The brain was treated with 100 mM Flu-DNB for 1 h in the same way as described for the vasodilation experiments, and then subjected to NIR pulse laser irradiation and twophoton fluorescence microscopy. Microglial cells expressing GFP were intensely but sparsely observed in the mouse brain. Resting microglia have extensive processes that extend in all directions. Small circular areas adjacent to the processes of a microglial cell were selected as regions of interest (ROI) at a depth of several tens to a hundred micrometers. Each ROI was irradiated with a femtosecond pulse laser at 735 nm, and the GFP image was simultaneously obtained by two-photon fluorescence microscopy at 950 nm for GFP. As shown in Figure 5a,b and Supplementary Figure S5, the processes of microglial cells were attracted toward the site of stimulating NIR irradiation. In the absence of Flu-DNB, pulse laser irradiation had no attracting effect on glial processes (Figure 5c,d and Supplementary Figure S6). This result suggested that

change (Figure 3c,d). Thus, photoirradiation of Flu-DNB at blood vessels induced vasodilation in living mouse brain. To confirm that NO released by photoirradiation to FluDNB caused this vasodilation, we examined the effect of the guanylyl cyclase inhibitor 1H-1,2,4-oxadiazolo[4,3-a]quinozalin-1-one (ODQ) because it is well established that the vasodilation effect of NO is mediated by NO-dependent guanylyl cyclase activation and cGMP upregulation.14,15 The photoinduced increase of vessel diameter in the Flu-DNB treated brain was attenuated in the presence of ODQ (Figure 4). These results indicate that the vasodilation induced by NIRpulse irradiation of Flu-DNB occurred via the NO-cGMP pathway. It is well-known that microglial cells respond to various chemical and physical stimuli by extending processes toward the stimulus, and NO could be one of the chemoattractants for microglia.16 Thus, we examined whether NO released from FluDNB in the brain of a living mouse showed chemoattractive activity for microglial processes. Transgenic Iba1 mice 2496

dx.doi.org/10.1021/cb400361m | ACS Chem. Biol. 2013, 8, 2493−2500

ACS Chemical Biology

Articles

Figure 4. Two-photon fluorescence images of vessels in mouse brain treated with Flu-DNB and ODQ. Mouse brain treated with Flu-DNB was irradiated with the uncaging pulse laser (735 nm) at the ROI indicated by the red lines. Two-photon images were obtained simultaneously by scanning with the acquisition pulse laser (950 nm). Then, the mouse brain was treated with ODQ for 30 min, and two-photon images of the same field were obtained. The observed fluorescence was due to Flu-DNB. Representative images are shown in panels a−d: (a,b) two-photon fluorescence images of a vessel in mouse brain treated with Flu-DNB immediately before (a) and during (b) the uncaging pulse irradiation; (c,d) images of the same field as panel a after additional treatment with ODQ immediately before (c) and during (d) the uncaging pulse irradiation. Changes of vessel diameter after pulse laser irradiation in the presence of Flu-DNB before ODQ treatment and after ODQ treatment are shown in panel e. The vessel diameter before irradiation is shown by a yellow scale in each pair of images. Enlargement of a vessel diameter without ODQ, indicated by a pair of yellow arrows in panel b, was attenuated after ODQ treatment (d). Data represent means and s.e.m. (n = 14). **P < 0.01 versus without ODQ. Scale bars = 30 μm.

of which requires direct access to the brain for each application, the combination of Flu-DNB and NIR pulse irradiation could attract microglial cells to specified areas repeatedly with very high spatiotemporal accuracy in the same mouse brain tissue. Thus, this technique should be advantageous for manipulating NO-related biological events in the body of living animals. We first confirmed that Flu-DNB released NO in cells in response to UVA irradiation, as shown in Figure 1b. This result indicates that Flu-DNB works as an NO donor in cellular systems and is available to treat cells with NO in a wellcontrolled manner. The cellular localization of Flu-DNB appeared to be the membrane compartment in cytoplasm (Supplementary Figure S1). Although we examined colocalization of the fluorescence of Flu-DNB with a mitochondrial fluorescent dye, the two were not completely colocalized. FluDNB seemed to be distributed to hydrophobic regions nonspecifically due to its physicochemical properties. In this experiment, we also used DAR-4 M AM as a fluorogenic NO probe to detect NO released in the cells. In the fluorescence images, DAR-4 M T, the NO-reacted product of the NO probe, seemed to be localized in the cells in a particulate manner. It has been suggested that DAR-4 M T is localized to mitochondria,12 so at least a part of the fluorescence of DAR4 M T seen in our study should be distributed to mitochondria. In this experiment, our focus was on controlled intracellular NO release from our compound, Flu-DNB, so we did not further evaluate the localization of DAR-4 M T. Flu-DNB was also found to release NO upon NIR pulse laser irradiation at 735 nm. When the stimulating laser irradiation was restricted to designated areas, the increase of DAR-4 M T fluorescence was observed only within those areas. This means that NO release from Flu-DNB is able to be controlled by NIR pulse laser irradiation. When a single cell was irradiated, DAR-4 M T fluorescence was increased within the irradiated cell and was localized to particulate areas in the cell, which could be

Figure 5. Microglial dendrite attraction by NO released from FluDNB. Mouse brain expressing GFP in microglia, treated with FluDNB, was irradiated with the uncaging pulse laser (735 nm) in the ROI (broken circles). The two-photon images were obtained simultaneously by scanning with the acquisition pulse laser (950 nm). Representative images are shown in panels a−d: (a) two-photon fluorescence images of microglia expressing GFP in mouse brain treated with Flu-DNB at the beginning of uncaging pulse irradiation; (b) image of the same field as panel a, but after uncaging pulse irradiation; (c) image of the brain without Flu-DNB treatment at the beginning of uncaging pulse irradiation; (d) image of the same field as panel c, but after uncaging pulse irradiation. Average fluorescence changes of the ROI before and after pulse laser irradiation in the presence or absence of Flu-DNB are shown in panel e. Areas inside broken circles were irradiated, and the same sites are indicated by the arrows. Data represent means and s.e.m. (n = 4). Scale bars = 30 μm.

Flu-DNB released NO upon pulse laser irradiation and the NO chemoattracted microglial processes toward the irradiated area. In contrast to spontaneous NO donors such as NOC5, the use 2497

dx.doi.org/10.1021/cb400361m | ACS Chem. Biol. 2013, 8, 2493−2500

ACS Chemical Biology

Articles

In conclusion, we have developed the first system that allows fine control of NO release in cells and in vivo by using a femtosecond NIR pulse laser to induce NO release from FluDNB. We show that this system can be used to manipulate blood vessels and microglia in vivo. The results of our cellular and in vivo experiments indicate that NO expresses its biological activity adjacent to the site where it is produced and is rapidly scavenged before it can permeate far. In other words, NO activity as a second messenger appears to be much more local than had previously been thought.

mitochondria or other organelles, as discussed above. It is important to note that, while the increase of DAR-4 M T fluorescence was clearly observed in the irradiated cell, almost no increase could be seen in neighboring cells. This suggested that NO released in the cell did not substantially diffuse to adjacent cells under our conditions. This seems reasonable, even though NO is a hydrophobic gas that is considered to be readily permeable through the cell membrane, because cells have many NO-reactive components and DAR-4 M also traps NO in the cell. Although it is difficult to evaluate the concentration of NO in the irradiated cell, it should be less than 70 nM because it has been reported that 100 μM Flu-DNB produced 8 μM NO when irradiated with a 30 mW femtosecond pulse laser (720 nm) for 10 min, whereas in our experiment, a 3 mW NIR pulse laser (735 nm) was scanned over the whole cell for a total of 45 s (ca. 15 μs/pixel). This concentration range of NO might be small enough for NO to be completely captured before it can pass through the cell membrane. Further, when the NIR pulse laser (735 nm) was directed at a small region of interest within a cell, we observed intense fluorescence at the irradiated site, with only a slight increase of fluorescence within the whole cell, and almost no increase in neighboring cells. This observation implies that NO produced at the subcellular irradiation site is predominantly captured in the immediate vicinity of the irradiation site. In the vasodilation experiment, photoinduced vasodilation was observed during the irradiation in the presence of FluDNB, but not in the presence of fluorescein, a partial structure of Flu-DNB lacking the NO-releasing moiety. ODQ treatment attenuated the photoirradiation response. Thus, this response was caused by NO release from the compound. NO activates soluble guanylyl cyclase to increase intracellular cyclic GMP as a second messenger for vasodilation,14,15 suggesting that the inhibitory effect of ODQ on the photoresponse of the vessels is exerted through the physiological NO-cGMP pathway. Vasodilation of blood vessels was observed only during the photoirradiation. This is consistent with the report that FluDNB releases NO during photoirradiation.8 Although it was difficult to identify the cell types in which Flu-DNB was localized, the fluorescence was most intense around the vessel wall area. It was considered that the FluDNB-induced vasodilation response was likely owing to NO molecules released in the intensely Flu-DNB-stained area. In cellular experiments, we observed that DAR-4 M T fluorescence was markedly increased when sites intensely stained with Flu-DNB were irradiated. The results of the cell experiments suggested that the released NO diffused only in the vicinity of the production site under our conditions. In the vascular experiments, Flu-DNB was localized to the vessel wall area, and irradiation induced only a local vasodilation response. Considering these results together, it seems likely that FluDNB is localized to vascular smooth muscle cells under our conditions. Thus, the use of Flu-DNB in combination with NIR-pulse laser irradiation allowed us to manipulate the vasodilation response of blood vessels in living mouse brain in a spatiotemporally well-controlled manner. In addition, we also demonstrated the effectiveness of spatiotemporal control of NO release from Flu-DNB for studying NO-induced chemoattraction of microglial processes in vivo. Compared with microinjection of conventional spontaneous NO donors in living tissues, our method allows us to repeatedly manipulate NO release in a specified region of interest in vivo.

■

METHODS

Chemistry. The photocontrollable NO donor Flu-DNB was synthesized according to our previous report.8 Its structure and purity were confirmed by 1H NMR, 13C NMR, MS, EA, and melting point measurements. Flu-DNB was stored at −30 °C as a powder until use. DAR-4 M AM and ODQ were purchased from Sekisui Medical Co. Ltd., and Wako Pure Chemical Ind. Ltd., respectively. Cell Culture and Treatment for in Vitro Experiments. HCT116 human colon cancer cells were from American Type Culture Collection and were maintained in McCoy’s-5A culture medium (Sigma-Aldrich) containing 10% (v/v) fetal bovine serum supplemented with penicillin and streptomycin at 37 °C in a humidified incubator under 5% (v/v) CO2 in air, according to the ATCC instructions. For cellular experiments, cells were seeded into glassbottomed 3 cm plastic dishes at 5 × 105 mL−1 in a volume of 2 mL and cultured for 1 day before Flu-DNB treatment, then further cultured for 1 day. On the day of experiment, the medium was replaced with culture medium without bovine serum, and the cells were treated with 10 μM DAR-4 M AM (DMSO 0.2% (v/v) as a cosolvent) for 30 min in a CO2 incubator. Then the medium was replaced with 1 mL of the culture medium with serum, and the cells were incubated for 60 min in a CO2 incubator, washed with Dulbecco’s phosphate-buffered saline (D-PBS; Sigma-Aldrich), and photoirradiated in D-PBS. Pulse Laser Irradiation and Confocal Imaging of Cultured Cells. HCT116 cells were treated with 10 μM Flu-DNB (1% (v/v) DMSO as a cosolvent) and incubated for 24 h in the culture medium. The treated cells were washed with D-PBS twice and subjected to photoirradiation. Pulse laser irradiation was performed with a modelocked Ti-sapphire laser (pulse width of