Imaging cellulose synthase motility during primary cell wall synthesis in the grass Brachypodium distachyon

The mechanism of cellulose synthesis has been studied by characterizing the motility of cellulose synthase complexes tagged with a fluorescent protein; however, this approach has been used exclusively on the hypocotyl of Arabidopsis thaliana. Here we characterize cellulose synthase motility in the model grass, Brachypodium distachyon. We generated lines in which mEGFP is fused N-terminal to BdCESA3 or BdCESA6 and which grew indistinguishably from the wild type (Bd21-3) and had dense fluorescent puncta at or near the plasma membrane. Measured with a particle tracking algorithm, the average speed of GFP-BdCESA3 particles in the mesocotyl was 164 ± 78 nm min−1 (error gives standard deviation [SD], n = 1451 particles). Mean speed in the root appeared similar. For comparison, average speed in the A. thaliana hypocotyl expressing GFP-AtCESA6 was 184 ± 86 nm min−1 (n = 2755). For B. distachyon, we quantified root diameter and elongation rate in response to inhibitors of cellulose (dichlorobenylnitrile; DCB), microtubules (oryzalin), or actin (latrunculin B). Neither oryzalin nor latrunculin affected the speed of CESA complexes; whereas, DCB reduced average speed by about 50% in B. distachyon and by about 35% in A. thaliana. Evidently, between these species, CESA motility is well conserved.

observed. Because the movement is tied closely to glucan extension, the speed of movement provides a proxy for the reaction rate and thereby can reveal details of the reaction mechanism.
This approach has been used to characterize cellulose synthesis in the hypocotyl of A. thaliana seedlings 8,9 . There, the speed of movement is approximately 300 nm min −1 , a rate that corresponds to each glucan in the complex being elongated by about ten glucose units per second 7 . Imaging tagged CESA complexes has helped show how cellulose synthesis is influenced by the cytoskeleton as well as by specific, regulatory proteins [1][2][3] .
While the information thus mined has been revelatory, so far the movement of the cellulose synthase complex has to our knowledge been reported for A. thaliana only, and in that species almost exclusively for the hypocotyl. It is not known to what extent the results from the A. thaliana hypocotyl can be generalized even to other organs, let alone to other species.
To characterize the motility of cellulose synthase in another species, we chose the grass Brachypodium distachyon. In view of its small stature and genome, as well as its phylogenetic affinity to wheat and related temperate grasses, B. distachyon is emerging as a popular model grass 10,11 . Grasses evolved a cell wall that differs in composition and structure from most other taxa 11,12 , and it is reasonable to allow that the mechanistic details of cellulose synthesis might likewise differ.
Cellulose synthase catalytic subunits are encoded by the CESA gene family. In any cell, active cellulose synthase is believed to involve CESA subunits from three distinct clades. For primary cell walls in A. thaliana, the clades are named CESA1, CESA3, and CESA6-like. In B. distachyon, the CESA family structure has been characterized 13 . It resembles that of other land plants and the putative orthologs in B. distachyon have been named to follow as much as possible the nomenclature established for A. thaliana. Both species have a small family of CESA6-like genes (three in B. distachyon; four in A. thaliana) and a single CESA1 gene. In contrast, A. thaliana has one CESA3 gene while B. distachyon has two (BdCESA3 and BdCESA2); however, mRNA from BdCESA2 is weakly expressed in stems, leaves, and roots 13 .
Here, we generated plants in which B. distachyon CESA3 and CESA6 are tagged with a green fluorescent protein (GFP), and used them to characterize CESA motility in both root and mesocotyl and as affected by selected inhibitors.

Results
Transgenic lines expressing GFP-tagged CESA proteins. To generate plants for imaging CESAs, we made constructs in which mEGFP 14 is fused to N-terminal B. distachyon CESA coding sequences, obtained from wild-type cDNA, and is driven by a Zea mays ubiquitin promoter. We worked with BdCESA3 (putative ortholog of AtCESA3 13 ) and BdCESA6 (one of three CESA6-like family members in B. distachyon 13 ), in view of their strong expression in seedlings 13 and the functionality of the A. thaliana orthologs when similarly tagged 7,15 . By the second generation after transformation, some lines expressing either construct appeared indistinguishable from the wild type (Fig. 1). Although expression of the transgene varied between the lines, only those with substantially higher expression were associated with aberrant plant development. For imaging, we selected lines with intermediate expression: two lines (T1 and T7) for GFP-BdCESA3 and one line (T2) for GFP-BdCESA6. Qualitatively, these three lines were similar and here we present results for GFP-BdCESA3.
Cellulose synthase complex movement: qualitative observations. In grasses, the first stem internode, made in the embryo, is called the mesocotyl. It is functionally analogous to the hypocotyl and likewise convenient for imaging. For imaging, we used a microscope set up for total internal reflection fluorescence (TIRF) microscopy but illuminated the sample at a sub-critical angle. This approach, sometimes called variable-angle epi-fluorescence microscopy 16 , allows deeper penetration of excitation light into the sample than does TIRF while still reducing background. In the mesocotyl of the tagged lines, bright puncta were present within a narrow focal plane, presumably corresponding to the plasma membrane and a thin layer of cortical cytoplasm (Fig. 2a). Puncta often appeared to be organized in lines and, among puncta, fluorescence intensity varied, even within a single cell, indicating that the endogenous and introduced CESA proteins might compete for space in the complex. In image sequences, particles appeared to move bi-directionally (Movie S1). As a way to assess motility visually, we made average projections, in which a moving particle appears as a line. Although some particles appeared stationary, a large majority of them moved (Fig. 2b).
To extend these results, we imaged tagged cellulose synthase complexes in the root. Helpfully for sub-critical angle illumination, the outer epidermal wall of the root is thinner than that of the mesocotyl but, unhelpfully, B. distachyon roots sprout a luxuriant mane of root hairs, making it difficult to bring the root close to the coverslip. Therefore, we exposed roots for several days to silver nitrate, at a concentration that inhibits root hair growth but has little or no effect on the growth of the root itself. In the tagged lines, puncta were visible in the root, and their appearance and motility appeared similar to the mesocotyl (Fig. 2c,d; Movie S2).
Because our intention is to compare CESA motility in B. distachyon to that of A. thaliana, we obtained a GFP-AtCESA6 line 15 , and imaged the tagged CESA in the hypocotyl. As expected, the GFP-AtCESA6 fluorescence formed puncta at the plasma membrane. In single frames and projections (Fig. 2e,f; Movie S3), the puncta appeared qualitatively similar to those in the B. distachyon mesocotyl.
Disruption of cellulose synthesis is indicated when the growth of a cylindrical organ becomes less anisotropic, that is, when the organ swells. Such swelling is caused by several classes of inhibitors, which have been used productively to probe cellulose synthesis. We selected an inhibitor of cortical microtubules, oryzalin; of cellulose synthesis, dichlorbenzyl-nitrile (DCB); and of actin filaments, latrunculin B. First, we ran dose-response curves to determine the sensitivity and responsiveness of B. distachyon (Fig. 3). The three compounds inhibited elongation and stimulated radial expansion. Latrunculin also inhibited growth of root hairs. Compared to A. thaliana, the grass B. distachyon was at least ten times more sensitive to oryzalin, about equally sensitive to DCB, and about five times less sensitive to latrunculin 17,18 . As for responsiveness, in A. thaliana, saturating doses of either oryzalin or DCB caused root diameter to triple 17 ; whereas in B. distachyon, while oryzalin caused almost the same degree of swelling, DCB did not quite double root diameter (Fig. 3). Saturating latrunculin increased root diameter by around 15%, a modest response level that is nevertheless similar in both species.
Cellulose synthase complexes move under the influence of cortical microtubules 7,19 but the nature of the relation between these elements is not fully understood. In A. thaliana, depolymerizing microtubules with oryzalin does not appreciably alter tagged CESA speed, at least for treatment times less than a day 7,20,21 . We treated seedlings for up to 2.5 h with 30 nM oryzalin and observed CESA motility in the B. distachyon mesocotyl. Qualitatively, the puncta moved as they did in untreated plants (Fig. 4a,b; Movie S4). After 1 day of treatment at this concentration, the mesocotyls were swollen, indicating activity of treatment.
In A. thaliana, the inhibitor of cellulose synthesis, DCB, has been observed to halt CESA motility 22 . Therefore, we treated B. distachyon with DCB and examined CESA motility in the mesocotyl. In seedlings treated with 5 µM DCB for 150 min, the density of puncta was not manifestly changed, but, as seen by the limited formation of tracks in the projection image, their motility was curtailed (Fig. 4c,d; Movie S5).
While microtubules have been the focus of efforts expended to understand the control of cellulose synthase motility, when actin is inhibited, roots and stems swell radially 17 , which could indicate a lowered rate of cellulose synthesis. Indeed, actin filaments have been implicated in trafficking CESA complexes 23,24 , and cell walls of seedlings treated with latrunculin have reduced amounts of cellulose 25 . Because short-term treatments with latrunculin did not appear to inhibit cytoplasmic streaming in B. distachyon, we exposed seedlings on plates to 250 nM Cellulose synthase complex movement: quantification. To quantify motility, we used a particle-tracking algorithm 26 . This routine finds particles meeting selection criteria and follows them through subsequent frames, without requiring manual intervention. In each sequence, particles that were tracked for longer than 30 seconds were used to determine average speed. In Figs 5 and 7, each bar shows the mean speed of For the mesocotyl, variability between image sequences (between different cells, often on different plants) was appreciable (Fig. 5). The data include results from both 3-and 4-day-old seedlings, but age did not appear to account for the variability. For the root, average speeds tended to be a little higher than for the mesocotyl but the trend was unlikely to be significant. In view of the need to inhibit root hairs, additional root sequences were not obtained. To allow direct comparisons, we also quantified the tagged CESA motility in the A. thaliana hypocotyl (Fig. 5). Again, the variability between cells was notable but comparable to that of B. distachyon. Likewise, the mean of the sequences for each species appeared similar.
To compare the species more accurately, we pooled the tracked particles from each sequence and assessed the complete distribution (Fig. 6). The distribution for B. distachyon appeared to be shifted to slightly lower values throughout its range. The mean CESA speed for the B. distachyon mesocotyl was 164 ± 78 nm min −1 (mean ± SD, n = 1451 particles) and for A. thaliana hypocotyl was 184 ± 86 nm min −1 (n = 2755 particles). Based on a t-test, there is less than one chance in a thousand that these means are the same. Nevertheless, the means differ by an amount that is numerically small while the variability between cells is large (Fig. 5), circumstances that hinder assessing what if anything the statistical difference means biologically.
Finally, we quantified CESA motility for the sequences obtained for the B. distachyon mesocotyl in seedlings exposed to the selected inhibitors (Fig. 7). Confirming visual inspection, the average speeds for oryzalin or latrunculin treatments were indistinguishable from those of the control. In contrast, 30 minutes of DCB was sufficient to decrease average speed by 50%, an effect that did not become larger by 150 minutes of treatment. In contrast, DCB was reported for the A. thaliana hypocotyl to bring CESA motility to a standstill 22 . Therefore, for comparison, we treated A. thaliana with DCB and observed that mean CESA speed was reduced, but to a lesser extent than in B. distachyon. Taken together our results demonstrate that CESA motility in growing organs is substantially conserved in these species.

Discussion
The B. distachyon mesocotyl is amenable for imaging CESA motility. In the etiolated state, it is thin enough to mount for microscopy with sub-critical angle illumination and there are abundant puncta available for tracking. Similar to the A. thaliana hypocotyl, the mesocotyl cytoplasm streams vigorously and contains labeled Golgi bodies and vesicles, which sometimes come close to the plasma membrane. Roots also could be imaged, provided that root hairs were eliminated. Possibly because this type of microscopy is sensitive to exact geometry, samples varied considerably in the apparent density of puncta and their overall brightness. This was true for both B. distachyon and A. thaliana. Therefore while we were able to quantify the speed of the puncta reliably, we quantified neither CESA density nor brightness.
For these experiments, we expressed tagged CESA sequences from B. distachyon in a wild-type background; thus, we cannot guarantee that the imaged constructs are functional. Nevertheless, the tagged proteins are probably functional. First, the constructs made here for B. distachyon were designed to mimic constructs of A. thaliana CESA genes that are functional 7,15 . Second, in some of our transgenic lines, aberrant phenotypes occurred and were correlated with high expression of the transgene; in contrast, the wild-type appearance of the lines used for imaging suggests that at least gain-of-function phenotypes are absent. Finally, in A. thaliana, tagged CESA genes expressed in a wild-type background have a pattern of motility that is indistinguishable from that observed when the tagged CESA gene is expressed in the cognate mutant background 27,28 . Until mutants in B. distachyon CESA genes for the primary cell wall are isolated, studying CESA motility in a wild-type background will be a necessary expedient. To the best of our knowledge, there are no such mutants available in any grass species.  Compared to the A. thaliana lines expressing the transgene in a mutant background, the more or less similar fluorescence in our B. distachyon lines presumably reflects a more or less similar ratio of tagged to un-tagged subunits in motile complexes. Be that as it may, tagged subunits appear to be incorporated to a substantial extent and if this had caused an appreciable decrement in the rate of cellulose synthesis, then we would expect to have seen short and swollen seedlings.
We find that CESA particles in B. distachyon had a mean speed and a frequency distribution closely resembling that of the A. thaliana hypocotyl, albeit shifted to slightly lower values. Nevertheless, the mean speed we measured for CESA movement in the hypocotyl (184 nm min −1 ) is lower than usually reported for this material. From a survey of 15 papers that report CESA speeds, the average speed is 295 nm min −1 and the lowest value is 209 nm min −1 (Table 1). Our value is thus lower than previously reported.
For the data in Table 1, CESA speed was quantified by using kymographs. To determine whether our low rate reflects biology (e.g., growth conditions) or an analytical difference between kymographs and particle tracking,  we analyzed a subset of our image sequences by using kymographs. For a total of 513 particles, the kymograph rate was 276 ± 6 nm min −1 (mean ± SE), a value well within the range of those in the literature but higher than the rate we found by particle tracking on the same sequences (209 ± 2 nm min −1 , 1340 particles). Therefore, our rate appears to be low because of methodology rather than biology. Particle tracking and kymographs sample CESA particles differently. Tracking applies an automated methodology that requires particles conform to a fixed size and minimum brightness; particles that are adjacent, cross paths, or blink substantially are likely to be excluded. On the other hand, the kymograph has subjective steps. In particular, the kymograph begins with the user drawing a line on the image superimposed on a line in the projection image representing coherent movement. In projection, the faster and longer-lived the particle, the more prominent the line, and thus users might be drawn to over-sample rapidly moving, longer-lived complexes. Even though the basis for the numerical discrepancy remains to be elucidated, we believe that this discrepancy should not detract from our conclusion that CESA particles in B. distachyon and A. thaliana move at essentially the same speeds.
As reported for A. thaliana, we found for B. distachyon that CESA speed was affected by treatment with neither a microtubule inhibitor (oryzalin) for a few hours nor with an actin inhibitor (latrunculin) for a day. Interfering with actin causes root swelling in B. distachyon, as it does in A. thaliana, although the swelling is substantially less than caused by inhibiting microtubules. Inhibiting actin has been associated with altered delivery of cellulose synthase complexes to the plasma membrane apparently through interfering with the movement of Golgi bodies [23][24][25]29 . Although we did not analyze the density or patchiness of CESA puncta in the presence or absence of latrunculin, qualitatively the two conditions appeared similar.
The cellulose synthesis inhibitor, DCB, inhibited CESA motility by about 50%. In A. thaliana, the compound was reported to stop CESA movement completely and to increase the density of CESA puncta at the plasma membrane 22 . Nevertheless, we found with particle tracking that DCB caused a reduction of CESA speed in both species, rather than a cessation. Insofar as the effects of DCB on CESA motility in A. thaliana have been reported apparently in only one publication, it is difficult to assess why we report a less severe effect. While methodology might again be at play, tracking apparently reports a higher speed than kymographs for DCB treatment, which is the opposite of what happens for untreated material.
Compared to DCB, the cellulose synthesis inhibitor isoxaben has been used more frequently to perturb CESA motility. However, isoxaben was developed as a herbicide for grass crops and consequently these species are relatively insensitive. As for DCB, while this compound affects A. thaliana and B. distachyon over a similar range of concentrations (Fig. 3), DCB might exert a weaker effect on cellulose synthesis in grasses compared to eudicots. In A. thaliana, roots exposed to a saturating dose of DCB swell to three times their initial diameter 17 ; whereas, those of grasses swell to a more limited extent ( Fig. 3; ref. 30 ). Furthermore, while DCB is categorized as a cellulose synthesis inhibitor 31 , investigations that established this mechanism were done on eudicot species mainly. In one example for a grass (maize roots), DCB inhibited cellulose synthesis rather modestly and in addition stimulated synthesis of hemicellulose 32 . While DCB inhibits elongation similarly in B. distachyon and A. thaliana, compounds closely related to DCB are powerful inhibitors of electron transport 31 that also inhibit growth. Even in eudicots, the molecular mechanism of DCB action against cellulose synthesis remains to be elucidated and mechanisms in grass species might be partially distinct.
Compared to each other, B. distachyon and A. thaliana have cellulose synthase complex motility that is almost indistinguishable, including the average speed of all particles measured, the variation in average speed among cells, the shape of the relative frequency distribution, and the response to DCB. The overall uniformity of cellulose synthase motility between these two species suggests that the operation of the complex is tightly constrained.

Materials and Methods
Seedling growth and handling. Brachypodium distachyon (L.) line Bd 21-3 was used, except that the cDNA used for cloning was made from Bd 21, which is the reference genome. Seeds were stripped of glumes and surface-sterilized in 70% ethanol for 2 min, in 20% household bleach and 0.1% Triton X-100 for 5 min, and then rinsed at least three times in water. Seeds were plated on half-strength Murashige-Skoog (MS) medium (pH 5.7), solidified with 0.8% Bactoagar. In some cases, plates were supplemented with 10 µM AgNO 3 , added to medium after autoclaving. After stratification at 4 °C for at least 5 days, plates for tagged CESA imaging were wrapped with aluminum foil, and placed vertically for 3 to 4 days in a growth chamber at 25 °C. For dose-response experiments, plates were exposed to continuous light (approximately 100 µmol m −2 s −1 ). Seeds from an Arabidopsis thaliana (L.) line (Columbia background) that expresses GFP-CESA6 15 were sterilized, stratified, and grown on modified Hoagland's medium, as described by Rahman et al. 18 except that plates were placed in the dark at 22 °C.
Oryzalin, DCB, and latrunculin B were dissolved in dimethyl-sulfoxide (DMSO) to create stock solutions and stored at −20 °C. For dose-response curves, agar-nutrient plates were made by adding an appropriate volume of stock solution to molten agar, with controls receiving the same volume of DMSO. Five-day-old wild-type seedlings were transferred onto the plates, the position of the root tip scored on the back of the plate and the plates returned to the growth chamber. After the indicated interval, root tip positions were scored again, the plates scanned, and elongation was measured by using ImageJ 33 on the scanned images as the distance along the root between the two score marks. Immediately after scanning the plates, root diameter was measured through a stereo dissecting microscope either from captured images or directly by eye with an ocular micrometer. In swollen roots, diameter was measured at its apparent maximum; for others, diameter was measured at the position where root hairs become full grown.
Construction of mEGFP-BdCESA transgenic lines. Total RNA was extracted from B. distachyon tissue (7-day-old whole seedlings) and reverse transcribed into cDNA with Invitrogen SuperScript III Reverse Transcriptase. Specific primers (with added attB sites) were used to amplify CESA3 (Bradi1g54250.1) and CESA6 Scientific RepoRTS | 7: 15111 | DOI:10.1038/s41598-017-14988-4 (Bradi1g53207) coding sequences into pENTRp5p2 ( Table 2). For tagging, we used mEGFP 14 in pENTRp1p5r. A multi-site Gateway reaction united the mEGFP and the CESA coding sequences in the plant expression vector pIPKb002 34 , in which expression of the tagged CESA genes is driven by a ubiquitin promoter from Zea mays. All constructs were confirmed by digestion with SfiI and sequencing.
Plant lines were made following an established protocol 35 . Briefly, plasmids containing mEGFP-BdCESA3 and mEGFP-BdCESA6 were transformed into agrobacterium line AGL1 by electroporation, and positive clones were tested with colony PCR. Transformed AGL1 lines were co-cultured with 3-month-old callus derived from embryos, and after three rounds of hygromycin selection, surviving callus was transferred onto regeneration medium. Regenerated plants were transferred to soil and propagated in the greenhouse.
For checking expression of the transgenes, we employed semi-quantitative RT-PCR, with primers shown in Table 2. The cDNA was synthesized (as above) with RNA isolated from leaves of 3-week-old plants. As a loading control, we used ubiquitin conjugating enzyme 18 (UBI18) based on its stable expression 36 . Imaging. At the time of imaging, B. distachyon seedlings were 3 to 4 days old and A. thaliana were 4 to 5 days old. Whole seedlings were mounted in water under a coverslip, previously cleaned for 1 min in a plasma, and sealed with a home-made mixture of Vaseline, lanolin, and paraffin ("VALAP"). For inhibitor treatment, chemicals were freshly diluted in liquid growth medium immediately before treatment. Controls received DMSO only. Three-day-old seedlings were treated with inhibitor in liquid growth medium for the indicated times. For longer treatments, two-day-old seedlings were transplanted onto plates that contained the inhibitor and incubated vertically in darkness at 25 °C. Plates containing inhibitors were made as described above for the dose-response experiments.
Imaging was performed on an Olympus inverted microscope (IX81) set up for total internal reflection microscopy. Fluorescence was collected through a 60X, 1.45 NA oil immersion objective. The mEGFP was excited at 488 nm by an argon ion laser beam passing through the exciting filter Chroma ZET488 and reflected off of a dichroic mirror (Semrock Di01-R405/488/ 561/635). Fluorescent emission was collected through the band pass emission filter (Chroma ET 525/50bp-OD8). Excitation energy at the samples was adjusted to 1.0 mW power. Images were acquired with an Andor iXon DU897E camera in EMCCD gain mode, driven by Metamorph software. Vertical drift was corrected by autofocus (Olympus IX2-ZDC2). Typical exposures were 100 or 500 msec. Frame intervals for B. distachyon mesocotyl were 0.6 or 1.6 sec; for the root, frames were averaged and the interval between averaged frames was 0.6, 1.6, or 1.9 sec; and for the A. thaliana hypocotyl were 1, 2, or 3 sec. Because the presence of the cell wall precluded the use of total internal reflection, the angle of illumination was adjusted individually for each set of images to minimize background while permitting observation of the cell membrane.
When present, image drift in the focal plane, primarily due to growth of the plant, was corrected by using the program distribution field tracking, implemented for MATLAB 37 . This program finds the best match of a specified image region (the region to be stabilized) in the next frame. Unlike programs that compare image regions pixel-by-pixel, this one represents each image location with a normalized histogram (or probability distribution) representing the image values in the vicinity of each pixel 37 . This allows for good matches of one image patch to another, even when there are small perturbations in the positions of image elements, since it is the distribution of nearby values that must match, rather than the exact pixel values. The distribution field tracking algorithm is able to find stable matches of an image region even when there is significant variability in its exact appearance from frame to frame. CESA particles were tracked by using another MATLAB program 26 . Particles were identified by using a mask that selects features with a radius approximately equal to 3 pixels (the "feature size"). Particles were rejected if the calculated radius of gyration squared was greater than 6 pixels squared, or if the maximum intensity was less than several standard deviations above the background intensity for the sequence. All particles had an eccentricity of approximately 0.5, most likely limited by the TIRF optics, so eccentricity was not used as a criterion. To link particles between frames and form trajectories, the maximum displacement between successive frames was set to 2 pixels. Because GFP blinks, we permitted a particle to skip 1 frame; a trajectory was ended if a particle went dark for more than one successive frame. Additional post-processing was used to further reject trajectories with marginal particle identification due to blinking or a low signal-to-noise ratio. Specifically, a region outside of the  mask was used to define a background intensity, then the integrated background was subtracted from the integrated intensity of each feature. If the result was negative, or if the localization error 38 associated with the feature is larger than one pixel, then the feature is dropped from the trajectory. If this resulted in a particle being absent for more than one successive frame, the trajectory was omitted. All particle trajectories that lasted more than 30 seconds were used.
For each particle trajectory, speed was calculated as the vector sum of the regression coefficients of the x coordinate versus time and the y coordinate versus time.
For kymograph analysis, we used the 'Multi-Kymograph' plug in (http://imagej.net/Multi_ Kymograph) for Image J, as described by those authors. Data availability. The datasets generated for the current study are available from the corresponding author on reasonable request.