FUNCTIONAL ANALYSIS OF THE CELLULOSE SYNTHASE CLASS SPECIFIC REGION IN PHYSCOMITRELLA PATENS

Cellulose synthases are found in a wide range of organisms, from bacteria to land plants. However, the cellulose synthases found in land plants (CESAs) form large, multimeric, rosette-shaped cellulose synthase complexes (CSCs) and have three unique regions not found in other cellulose synthases; the N-terminal zinc-binding domain, the Plant Conserved Region (P-CR) and the Class Specific Region (CSR). The CSR, a portion of the large cytoplasmic region that contains the catalytic domain, has been predicted to be involved in CSC formation through in silico modeling. This project tested the hypothesis that the CSR is necessary for clade-specific CESA function. We have developed a complementation assay in the moss Physcomitrella patens based on the ppcesa5KO-2B line, which does not produce gametophores. In P. patens, CESAs in the A Clade (CESA3, 5 and 8) have similar CSR structures that are distinct from those in the B Clade (CESA4, 6, 7, 10). The ppcesa5KO-2B line is complemented by overexpression of PpCESA3 and PpCESA8, but not by the overexpression of CESAs in Clade B. An overexpression vector containing a Clade A CESA with a Clade B CSR was unable to rescue ppcesa5KO phenotype, indicating that the CSR is necessary for clade-specific CESA function. However, an overexpression vector containing a Clade B CESA with a Clade A CSR was also unable to rescue the ppcesa5KO phenotype. This signifies that the CSR is not the sole determinant of clade-specific CESA function. A vector containing a Clade A CESA and the C-terminus of a Clade B CESA was able to rescue the ppcesa5KO phenotype. This suggests that the functionality of the C-terminal region, containing six transmembrane helices, is not a determinant of clade-specific function. A vector containing a Clade A CESA with the N-terminus of a Clade B CESA was unable to rescue the ppcesa5KO phenotype. This suggests that CSR may be interacting with an N-terminal domain, possibly the P-CR or zinc-binding domain to confer clade-specific function.

, and the University of Rhode Island.

Cellulose Applications
Cellulose is the most abundant biopolymer on earth and is important in a range of applications such as textiles, food and biofuels. Lignocellulosic biofuels are produced in a variety of ways, from thermal conversion to microbial conversion and chemical conversion (Carroll and Somerville 2009). The most cost efficient methods are microbial and chemical conversion. These methods require size reduction of biomass on the submillimeter level, pretreatment to increase accessibility of cell wall polysaccharides, hydrolysis of polysaccharides to simple sugars and conversion of those sugars into fuel, usually ethanol (Carroll and Somerville 2009). One of the main obstacles in efficient cellulosic biofuel production is the recalcitrance of cellulose. As cellulose exists as long microfibrils, cellulases may be prevented sterically from cleaving glycosidic linkages.
The cellulose microfibril may additionally prevent hydrolysis due to extensive hydrogen bonding of the glucan chains to one another (Carroll and Somerville 2009). The mechanism of cellulose biosynthesis is still not completely understood, making improvement on cellulose recalcitrance difficult. Thus, it is necessary to understand cellulose structure and synthesis in order to engineer crops better suited for biofuels.

Cellulose Structure
Cellulose occurs as microfibrils composed of hydrogen-bonded β (1, 4)-glucan chains. Cellulose has been characterized as having six different allomorphs, with the first two, cellulose Iα and Iβ, being the native celluloses and the other four, II, III, V, and VI, being chemically and physically altered derivatives (Sturcova et al. 2004;Thomas et al. 2013;Wada et al. 2004). In algae, the Iα and Iβ allomorphs are found with a high degree of crystallinity, in part due to the large, ribbon-like microfibrillar bundles. In higher plants, these allomorphs predominate, but are more disordered than their algal counterparts, forming thin, ropelike microfibrillar bundles (Sturcova et al. 2004;Thomas et al. 2013;Wada et al. 2004). The exact number of chains in a land plant cellulose microfibril is unknown. Early studies using NMR estimated thirty-six glucan chains in a microfibril (Delmer 1999;Ha et al. 1998;Herth 1983). However, land plant cellulose microfibrils are long, but very thin, making accurate measurements from NMR very difficult. More recent studies, using a combination of NMR with WAXS, SAXS and SANS for more accurate measurement of microfibril diameter, estimate 18-24 glucan chains (Newman et al. 2013;Thomas et al. 2013).
Cellulose synthases are present in many organisms, from bacteria to land plants. In contrast to bacterial cellulose synthases (BCSs), land plant cellulose synthases (CESAs) have additional domains, including the N-terminal zinc-binding domain, the plantconserved region (P-CR) and the class-specific region (CSR) (Figure 1) (Arioli 1998;Ihara et al. 1997;Pear et al. 1996). The CSR is notable in that among CESAs of the same class, there is sequence similarity, but between CESAs of different classes, there is little similarity, thus the former designation of hypervariable region (Vergara and Carpita 2001). A recent crystal structure of the bacterial cellulose synthase RsBcsA of Rhodobacter sphaeroides confirmed that the conserved D, D, D, QXXRW motif is responsible for catalysis of β (1, 4)-glucan synthesis using UDP-glucose as a substrate (Morgan et al. 2013).

Cellulose Synthase Complexes
In contrast to BCSs, the CESA subunits aggregate into six-particle rosette-shaped cellulose synthase complexes (CSCs) visible in freeze-fracture electron microscopy of plasma membranes of land plants (Doblin et al. 2002;Kimura et al. 1999;Mueller and Brown 1980) and their closest green algal relatives (Herth 1983). Vascular plants have multiple isoforms of CESAs, with the model plant Arabidopsis thaliana containing ten different isoforms (Doblin et al. 2002); three isoforms form CSCs that produce primary cell walls (Desprez et al. 2007;Persson et al. 2007), and three different isoforms form CSCs that produce secondary cell walls (Taylor et al. 2003). Although it is known which CESA isoforms compose the CSCs responsible for either primary or secondary cell wall synthesis in Arabidposis, the CESA stoichiometry of rosette CSCs is unknown. The exact number of CESAs in a CSC is also unknown, but it is generally assumed that the number of CESAs in a rosette is equal to the number of glucan chains in a cellulose microfibril.
Although previously speculated to contain about thirty six CESAs (Doblin et al. 2002), recent spectroscopic and diffraction data on cellulose microfibrils from celery collenchyma and spruce fibers suggests that CSCs are more likely to contain 18-24 CESAs (Thomas et al. 2013). Sequence analysis has shown that major differences between the six CESA classes, three from primary and three from secondary cell wall CSCs, occur in the CSR (Vergara and Carpita 2001). Taken together with the observation that the CSR is found in CESAs, but not BCSs, this suggests that the CSR plays a role in rosette CSC formation.

Figure 1:
A schematic of an Arabidopsis thaliana CESA protein in the plasma membrane. The three unique regions found only in land plant CESAs are marked; the Zinc-binding domain, the P-CR and the CSR. The large cytosolic loop contains the catalytic motif D, D, D, QxxRW, shown with red stars. Also marked is the known mutation in the P-CR of an AtCESA known as fra6. No mutations have been found in the CSR or zinc-binding domain.

Physcomitrella patens Cellulose Synthases and the Class-Specific Region
In the moss Physcomitrella patens, there are seven CESA isoforms, which is particularly interesting given that it lacks vascular tissue with secondary cell walls (Roberts and Bushoven 2007). P. patens has become established as a model system due to its ease of genetic manipulation. P. patens has a fully sequenced genome (Rensing et al. 2008) and has a high rate of homologous recombination, which enables targeted gene modification (Cove 2005). Additionally, a stream-lined process for genetic transformation has been developed and optimized (Roberts et al. 2011). P. patens also exists predominantly in the haploid phase of its life cycle, enabling rapid, efficient and facile genetic manipulation (Cove 2005).
Unlike in Arabidopsis, the specific functions of the CESA isoforms in P. patens are unknown (Roberts and Bushoven 2007). Whereas Arabidopsis CESAs are specific to primary and secondary cell wall synthesis (Desprez et al. 2007;Persson et al. 2007;Taylor et al. 2003), PpCESAs appear to function in the development of particular tissues (Goss et al. 2012). Additionally, seed plant CSCs contain three CESA isoforms (Desprez et al. 2007;Persson et al. 2007;Taylor et al. 2003), whereas CSCs in P. patens could be homo-or hetero-oligomeric (Goss et al. 2012). However, there are two phylogenetically distinct groups of CESAs in P. patens, referred to as Clade A and Clade B, which differ in the sequences of their CSRs (Appendix 1, 2) (Goss et al. 2012). Computational modeling indicates that CSRs of Clade B CESAs have a long, central α-helix, whereas in CSRs of Clade A CESAs, this α-helix is disrupted (Sethaphong, personal communication). Previous experiments in which a lesion was inserted into the catalytic region of PpCESA5 and was thus functionally knocked out indicate that PpCESA5, a member of Clade A, is expressed in the gametophore and has a striking mutant phenotype in which gametophores fail to develop (Goss et al. 2012). Interestingly, when ppcesa5KO-2B was complemented with PpCESA3 or PpCESA8, both in Clade A, the wild-type gametophore phenotype was fully rescued. However, when ppcesa5KO-2B was complemented with CESAs from Clade B (PpCESA4, 6, 7, 10), the wild-type phenotype was not rescued (A. Roberts, unpublished). As Clade A and Clade B differ in CSRs, this region may be implicated in the functional differences revealed by the complementation assays. More specifically, the central α-helix found in Clade B CESAs and absent in Clade A CESAs may contribute to CSR-specific function of the CESAs.
Further evidence of CSR function in CSC formation is demonstrated by the recent computational modeling of the Gossypium hirsutum CESA1 cytosolic domain. The catalytic residues of the GhCESA1 computational model align closely with that of the RsBcsA crystal structure, indicating that the catalytic mechanism is conserved across family 2 GTs and validating the computational model (Sethaphong et al. 2013). This model depicts the P-CR and CSR as facing away from the catalytic region into the cytoplasm, indicating that these regions may be involved in CSC formation. Models created using the program Rosetta Symmetry (Rohl et al. 2004) predict that the P-CR and CSR interact when the CESAs form dimers, trimers and hexamers (Sethaphong et al. 2013). A missense mutation in the P-CR of Arabidopsis thaliana leads to a decrease in cellulose deposition and wall thickness in fiber cell walls (Zhong et al. 2003), indicating that the P-CR is necessary for cellulose synthesis (Figure 1). However, no known mutations that affect cellulose biosynthesis are located in the CSR (Sethaphong et al. 2013). Coupled with evidence that Arabidopsis primary and secondary CSC components differ in CSR structure and that CSR structure is correlated with the ability of P. patens CESAs to complement the ppcesa5 mutant, the computational model of GhCESA1 suggests that the CSR plays a direct role in CSC formation.
In this study, I analyzed the role of the CSR in the functional differences between Clade A and Clade B CESAs in gametophore development using the ppcesa5KO-2B complementation system. By complementing ppcesa5KO-2B with a chimeric CESA5 containing the CSR from a Clade B CESA (CESA4), I determined that the ability of Clade A CESAs to complement the ppcesa5KO phenotype is dependent on the CSR. By complementing ppcesa5KO-2B with a chimeric CESA4 containing the CSR of a Clade A CESA (CESA5), I determined that the CSR is not the only region necessary for cladespecific CESA function. By complementing ppcesa5KO-2B with a chimeric CESA5 containing the C-terminus of CESA4, I determined that the C-terminus is not necessary for clade-specific function. By complementing ppcesa5KO-2B with a chimeric CESA5 containing the N-terminus of CESA4, I determined that the N-terminus, including the zinc-binding domain and P-CR, is necessary for clade-specific function.

General Strategy
Chimeric CESA expression vectors were constructed using PCR fusion and Invitrogen MultiSite Gateway cloning (Life Technologies, Grand Island, NY, USA).
PpCESA8, PpCESA4 and PpCESA5 cDNA clones were used as templates for PCR. PCR products were fused by a single overlap extension to produce chimeric CESAs (Atanassov et al. 2009), which were inserted with a triple hemagglutinin tag into the pTHAct1Gate (xt18) destination vector containing an actin1 promoter driving constitutive expression and sequences that target the expression vector to the P. patens 108 locus, which can be disrupted without producing a phenotype (Perroud and Quatrano 2006).

Primer Name
Primer Sequence Figure 2. A schematic detailing the primer design for each chimera and nomenclature of the chimeric CESAs.
described (Atanassov et al. 2009 (Atanassov et al. 2009). Using PpCESA cDNA as templates, F1, F2 and F3 fragments were amplified by PCR for the final overlap extension step to create chimeric Ppcesa PCR products.  for correct recombination by sequencing with primers P395 and P403 for PpCESA5 constructs and P404 and P405 for PpCESA4 constructs (Table 2, Figure 4).

Expression Clone Preparation
Plasmid DNA from sequence-verified expression clones was isolated using

Moss Subculture
Protonemal filaments from the P. patens cesa5KO-2 line (Goss et al. 2012) provided by Dr. Alison Roberts (University of Rhode Island, Kingston, RI, USA) were grown on BCDAT overlain with cellophane at 25°C with fluorescent lights at a photon flux density of 60 µM m -2 s -1 for 7 d. The tissue was then homogenized using Omni International hard tissue omni tip probes (USA Scientific, Ocala, FL, USA) in 4-6 ml of sterile water and plated on BCDAT overlain with cellophane and grown under the same conditions for 5-6 d for protoplast isolation according to Roberts et al (2011).

Protoplast Isolation & Transformation
Protoplasts were isolated according to Roberts et al (2011) using the P. patens cesa5KO-2 line. Briefly, protonemal tissue was digested using driselase, washed thrice in an isosmotic medium, mixed with the linearized expression vector and polyethylene glycol and heat shocked at 45°C for three minutes. Protoplasts were resuspended in top agar protoplast regeneration medium (PRMT) and plated on bottom agar protoplast regeneration medium (PRMB) overlain with cellophane. Transformed protoplasts were screened through two rounds of hygromycin selection to obtain stable transformants.

Complementation Assays
Stably transformed moss lines were arrayed on BCDAT plates and incubated at Statistics P values were assigned using the Two-Tailed Fisher's Exact Test of Independence (Sokal and Rohlf 1981). For each expression vector, data from each trial was pooled and compared against the corresponding opposite control data (i.e. a vector that did not rescue would be compared against the positive control) to determine the p value. Each expression vector was compared against the corresponding similar control to determine whether or not there was a significant difference.

Protein Isolation
Microsomal proteins were isolated from at least twelve lines from each nonrescuing transformation along with 3XHACESA8 lines (positive control) and xt18-GW lines (negative control) according to Hutton et al (Hutton et al. 1998) with modifications.
Tissue was ground using the Argos pellet mixer and pestle (Argos Technologies Inc., Elgin, IL, USA) in 100 µl of extraction buffer containing 50 mM HEPES, 0.5 M sucrose, 0.1 mM EDTA, and 4 mM M-ascorbic acid for five minutes on ice. The lysate was centrifuged at 10,000 x g for ten minutes at 4° C and the supernatant centrifuged again under the same conditions in order to remove plastids and nuclei. IgG peroxidase antibody for one hour at 25° C. Finally, the membrane was washed as previously described, blotted and exposed using X-ray film (Research Products International) after 20 minutes.

Experimental Design
The ppcesa5KO-2B complementation assay was developed to test the role of the CSR in clade-specific CESA function in P. patens. The ppcesa5KO mutants are characterized by an inability to form normal gametophores, most commonly developing abnormal buds and rarely producing small gametophores with irregular phyllotaxy (Goss et al. 2012). PpCESAs within the same clade (Clade A) can rescue the ppcesa5KO phenotype but PpCESAs from Clade B cannot (Roberts, unpublished Results were excluded if the positive and/or negative control was absent or produced too few lines (a result of either a low transformation rate or contamination). Results were also excluded if the positive and/or negative control behaved abnormally (i.e the positive control produced gametophores in less than 40% of stable, transformed lines).

Transformation with Positive and Negative Control Vectors
The ppcesa5KO phenotype is characterized by a complete absence of leafy gametophores, or occasionally stunted gametophores with abnormal phyllotaxy.
PpCESA5KO lines typically produce cellulose deficient buds that are unable to mature into normal leafy gametophores due to irregularities in cell expansion and cell division (Goss et al. 2012). The gametophores that are able to develop from these irregular buds are small, producing only one to three irregularly spaced, misshapen leaves.
Transformation with a 3XHAPpCESA5 expression vector fully and reproducibly rescued the ppcesa5KO phenotype. Between 40% and 100% of lines stably transformed with this vector produced gametophores, with an average of 74% of lines producing gametophores, and each line that produced gametophores also produced a full length protein when probed with an anti-hemagglutinin antibody (Table 3, Figures 5, 6 & 7). These lines produced several gametophores (usually more than four per colony) of normal size and phyllotaxy, with leaves occurring in a regular spiral pattern (Figure 8). Leaves were narrow and pointed with elongated cells occurring in distinct files (Figure 8).
Transformation with the xt18-GW empty vector consistently failed to rescue the ppcesa5 KO phenotype. Between 0% and 20% of lines produced gametophores, with an average of 5% of lines producing gametophores (Table 3, Figure 5 & 7). Gametophores produced by xt18-GW lines resembled those of the ppcesa5KO-2B line. Each line that produced gametophores typically only produced one or two per colony. The gametophores themselves were small, with only one to three misshapen, small leaves. The leaves were composed of irregularly shaped cells and lacked the conducting cells present in mature leaves and gametophores (Figure 8).

The Class-Specific Region in Clade-Specific Function
Transformation with a vector containing CESA5 with the CSR of another Clade A CESA (3XHAppcesa5CSR8) was able to rescue the ppcesa5KO phenotype, producing normal gametophores in the same numbers as 3XHAPpCESA5 transformed moss lines (p < 0.0001, Figure 5 A, Table 3). The ability of 3XHAppcesa5CSR8 to rescue the ppcesa5KO phenotype indicates that CSRs within the same clade can perform cladespecific functions in Clade A CESAs.
Transformation with a vector containing CESA5 with the CSR of a Clade B CESA (3XHAppcesa5CSR4) failed to rescue the ppcesa5KO phenotype, producing no gametophores in both trials (p < 0.0001, Figure 5 B, Table 3). This was consistent with the negative control, xt18-GW transformations, which rarely produced gametophores that tended to be stunted with an abnormal phyllotaxy ( Figure    To test whether failure to rescue the ppcesa5KO phenotype could be explained by poor expression or misfolding and subsequent destruction of proteins coded by the transgenes, transformants that did not rescue the ppcesa5KO phenotype were analyzed for accumulation of the transgenic protein through Western blots probed with an antihemagglutinin antibody and an anti-tubulin antibody to evaluate protein loading. Initially, Western blots were performed on microsomal protein extracts from 3XHAPpCESA5 transformed moss lines that both produced and did not produce gametophores ( Figure 6).
All lines that produced gametophores showed the predicted band at 125 kD, indicating that the full-length 3XHAPpCESA5 protein was expressed. Six lines that did not produce gametophores produced the predicted band and the other four lines did not. One interpretation of this result is that a minimum dose of PpCESA5 proteins is needed to successfully rescue the ppcesa5 KO phenotype. It is also possible that some of the stable transformants do not contain the intact vector, as evidenced by the lack of a 125 kD band in most of the gametophore-less lines, instead containing an truncated vector with the functional resistance cassette.
Tubulin bands across the Western blots were often faint and of variable intensity, indicating that protein loading was uneven. Despite using the Pierce BCA Protein Assay, protein concentrations were too low to measure accurately. Additionally, a linear trend line was used initially to calculate protein concentration against BSA standards instead of the recommended exponential trend line. Both of these factors contributed to inconsistent protein loading, as indicated by variable tubulin levels across the blots ( Figure 6). Using larger quantities of moss tissue to increase protein concentration of the samples will make calculating protein concentration more reliable.
All 3XHAppcesa4CSR5 lines produced the predicted band at 125 kD, indicating that the full-length protein was expressed ( Figure 6). This expression validates the inability of 3XHAppcesa4CSR5 to rescue the ppcesa5 KO phenotype. In contrast, only seven out of twelve 3XHAppcesa5CSR4 transformed lines produced the predicted band at 125 kD ( Figure 6). However, this still indicates that full-length 3XHAppcesa5CSR4 is unable to rescue the ppcesa5 KO phenotype. The lower apparent expression rate may be due to uneven protein loading, as in this particular Western blot, a fainter anti-HA band correlated to a fainter anti-tubulin band ( Figure 6). Regardless, the Western blots indicated that the chimeric proteins were intact and not truncated, indicating that failure to rescue cannot be solely attributed to lack of protein expression.

The N and C-termini in Clade-Specific Function
In order to determine other regions necessary in clade-specific CESA function in P. patens, N and C-terminal swaps were created between PpCESA5 and PpCESA4. The  The 3XHAppcesa5/5/4, 3XHAPpCESA5 and xt18-GW transformed ppcesa5KO-2B lines were examined with bright field light microscopy to assess differences in gametophore development. Whole gametophores from 3XHAppcesa5/5/4 lines showed no difference in size or phyllotaxy from gametophores from 3XHAPpCESA5 lines ( Figure 8). Closer examination of individual leaves revealed no difference in leaf morphology, nor any difference in cell size, shape and pattern within the leaves ( Figure   8).
Expression levels of both transgenes were analyzed through Western blots using an anti-HA antibody and an anti-tubulin antibody. All microsomal protein extracts from twelve lines of 3XHAppcesa4/5/5 produced the expected band at 125 kD, indicating that the full-length protein was expressed ( Figure 6). In the first blot ( Figure 6), all protein extracts from gametophore producing 3XHAppcesa5/5/4 lines produced the expected band at 125 kD. Several protein extracts from lines that did not produce gametophores also produced a fainter band at 125 kD. However, it is difficult to determine whether or not protein expression levels are different between gametophore-producing and gametophore-less lines, as anti-tubulin staining shows that protein loading was not properly normalized. In the second blot, all microsomal protein extracts from gametophore producing lines of 3XHAppcesa5/5/4 transformed moss, with the exception of the last lane, in which the sample had not been loaded into the well, produced the expected band at 125 kD. Protein from lines that did not produce gametophores did not produce the expected band ( Figure 6). However, the presence of the tubulin band (55 kD) is varied in intensity across the samples, indicating that protein levels were not properly normalized. This is most obvious in the positive control, in which neither the HA nor tubulin band is visible ( Figure 6). This opens up the possibility that the transgene is being expressed, but is undetected by the blot.

The Experimental Assay
A complementation assay should be able test hypotheses about the function of a specific gene and its encoded product in a reliable and reproducible manner. In order to achieve this, a complementation assay must be based on a mutant with a measurable phenotype that is easy to score. In the ppcesa5KO mutants, the phenotype is clear: an inability to form leafy gametophores, only rarely forming stunted, abnormal gametophores (Goss et al. 2012). This phenotype is rescued by the constitutive expression of the disrupted PpCESA5 gene, demonstrating that this phenotype is the result of a PpCESA5 deficiency. Additionally, this phenotype is rescued by other PpCESAs of the same clade, but not by PpCESAs of Clade B (Roberts, unpublished), providing an ideal system in which to test clade-specific functional differences between the CESAs. Strengthening this system is the use of positive and negative controls, which provide standards of comparison for statistical analysis. Since constitutive expression of We expect that complemented lines resulted from integration of the vector into the 108 locus with no deleterious effect on the phenotype. However, failure to complement in some lines could result from integration of the vector at other loci, which could have an effect on gametophore development independent of the ppcesa5KO lesion. It has been reported that PEG-mediated transformation generates a small percentage of polyploidy cells, which are slow-growing (Schaefer and Zryd 1997). Colonies that failed to produce gametophores were often small and it is possible that these lines were polyploid and failed to produce gametophores due to developmental delay. Some stable noncomplemented lines may have resulted from integration of a truncated vector that included the resistance marker, but lacked either the actin1 promoter or the CESA coding sequence. This could be tested by variability in protein expression levels across the lines.
In summary, the advantages of the ppcesa5KO complementation assay include a) a clear phenotype consisting of abnormal buds and the absence of leafy gametophores, b) the ability to score complementation as either full (normal gametophores), partial (abundant abnormal gametophores/fewer normal gametophores), or none (no gametophores or very few abnormal gametophores) rescue, c) functional discrimination between the Clade A and Clade B CESAs, d) positive and negative controls for validation of results and comparison of rescues, e) creation of chimeras through PCR fusion without compromising the structural integrity of these closely related proteins, f) the ability to test the expression and integrity of the transgenic proteins through Western blotting with anti-HA and g) the ability to exclude genotyping the mutants.

The CSR is Necessary for Clade-Specific CESA Function
Transformation with an expression vector containing a Clade A CESA with the CSR of another Clade A CESA (3XHAppcesa5CSR8) rescued the ppcesa5KO phenotype ( Figure 5 A, Table 3), indicating that CSRs can perform clade-specific functions when inserted into a different CESA from the same clade. Additionally, as the chimeric CESA was not misfolded, this may indicate that the other CSR swaps are unlikely to misfold as well.

The mutant ppcesa5KO transformed by an expression vector containing a Clade
A CESA with the CSR of a Clade B CESA (3XHAppcesa5CSR4) did not produce gametophores, retaining the knock out phenotype (Figure 5 B, Table 3). This result supports the hypothesis that the CSR is necessary for clade-specific function, as the change from Clade A CSR to Clade B CSR leads to loss of function of PpCESA5. This may be due to the inability of the chimeric CESA to join a cellulose synthase complex, as supported by the in silico modeling of Gossypium hirsutum CESA1 catalytic domains as oligomers (Sethaphong et al. 2013). However, it may also be due to the inability of the chimeric CESA to interact with other non-CESA proteins, as the proteins necessary for CSC formation and function are still unknown (Delmer 1999;Doblin et al. 2002;Guerriero et al. 2010). Alternatively, the CSR may play a previously unidentified role in CESA function unrelated to protein-protein interactions. Although misfolding of the transgenic protein cannot be ruled out, Western blotting results indicate that the transgene is transcribed and translated to produce a protein of the expected mass that is not rapidly degraded. In order to determine whether or not the CSR is involved in CSC formation, it would be useful to perform co-immunoprecipitation experiments on wild type P. patens and 3XHAppcesa5CSR4 lines. To further test the hypothesis that the CSR is responsible for clade specific function, it would be useful to transform ppcesa5KO-2B lines with PpCESA5 with the CSR of the third Clade A gene (PpCESA3) in addition to PpCESA5 with the CSRs of the untested Clade B genes (PpCESA7 and PpCESA10). This would demonstrate whether the clade-specific function of the CSR is consistent across all P.

The CSR is Not Sufficient for Clade-Specific CESA Function
Transformation with an expression vector containing a Clade B CESA with the CSR of a Clade A CESA (3XHAppcesa4CSR5) did not rescue the ppcesa5KO phenotype, predominantly producing no gametophores and only occasionally producing malformed gametophores consistent with the knock out phenotype (Figure 5 C, Table 3).
Transformed lines all produced the expected band at 125 kD ( Figure 6) in Western blots.
The inability of the PpCESA5 CSR to confer Clade A-specific function on a Clade B CESA suggests that whereas the CSR is important for clade-specific CESA function in P.
patens, it is not sufficient. It follows that other CESA features are also required for cladespecific function. To further test this hypothesis, it would be useful to transform ppcesa5KO-2B lines with PpCESA4 with the CSRs of PpCESA 8 and PpCESA3. Until these tests are performed, it remains possible that one of the Clade CSRs is sufficient to confer Clade A-specific function. If the CSR is not sufficient, then this complementation system could be used to determine the region or regions that are also necessary for cladespecific functions.

The N-terminus is Required for Clade-Specific CESA Function
Transformation with an expression vector containing the N-terminus of a Clade B CESA and the CSR and C-terminus of a Clade A CESA (3XHAppcesa4/5/5) did not rescue the ppcesa5KO phenotype (Table 3 However, fewer colonies from these transformations produced gametophores, suggesting that the chimera may not be fully functional. Every line of the 3XHAppcesa5/5/4 transformants that produced gametophores also produced the expected 125 kD band, with the exception of one in which the protein sample was not loaded into the well. However, three lines that did not produce gametophores produced a weaker 125 kD band ( Figure   6). The incomplete rescue of this chimera could be attributed to a fully functional protein that was expressed at a lower level in ppcesa5KO than the 3XHAPpCESA5 and 3XHAppcesa5CSR8 chimeras. This is supported by the lower level of transgene expression relative to tubulin expression ( Figure 6), although it is difficult to tell due to uneven protein loading. The presence of gametophores could be due to a dosage effect, in which a certain amount of the transgene is required to rescue the ppcesa5KO phenotype, as mentioned above. Alternatively, the protein may lack functional integrity, such as a higher propensity for misfolding. In order to test for a dosage effect and functional integrity, Western blots with higher protein concentrations that are properly normalized could reveal a lesser intensity in lines that do not produce gametophores, indicating a dosage effect or truncated proteins, which would indicate a higher rate of protein misfolding and thus suggest a lack of functional integrity.

APPENDIX 5: Individual Trials of N-and C-Terminal Swaps
Individual trials of ppcesa5KO-2B phenotypic rescue by (