[1] This application claims the benefit of priority under 35 U.S.C. § 119(e) from U.S. provisional patent application serial no. 61/576,727, filed December 16, 2011, and serial no. 61/576,922, filed December 16, 2011 ; the entire contents of each

provisional application being incorporated herein by reference.

BACKGROUND OF THE INVENTION

[2] The production of monoclonal antibodies and the development of recombinant antibody technology have made antibodies a clinically-important drug class. Many antibodies are on the market for human therapy, and many more are in clinical development, for a wide range of diseases, including autoimmunity and inflammation, cancer, and infectious disease. Examples of currently marketed antibodies include, among others, trastuzumab (HERCEPTIN®), adalimumab (HUMIRA®) and natalizumab (TYSABRI®) in the treatment of breast cancer, rheumatoid arthritis, and multiple sclerosis, respectively.

[3] IgG is the antibody class most widely adopted for therapeutic antibodies. Currently most (if not all) commercial therapeutic antibodies are of the IgG class.

Production and purification procedures for human IgG antibodies are well established and reliably performed. See, e.g., Beyer et al., Journal of Immunology 346; 26-37 (2009).

[4] In recent years, IgA antibodies have attracted increased attention as potential therapeutics for infectious and malignant diseases.. Despite this, IgA antibodies have not been commercially advanced. One of the main problems hampering work with IgA antibodies is the lack of established methods for production and purification. Clinical grade IgG antibodies are usually produced in mammalian cell expression systems to achieve post-translational modification and glycosylation patterns similar to humans. Vector systems are commercially available and optimized for expression in mammalian cells, e.g., in CHO cells - one of the most commonly used mammalian expression systems. For the production of IgA antibodies, however, expression vectors and cell lines have not been used consistently. And despite several existing approaches to the purification of IgA antibodies, no particular purification method has achieved broad acceptance. See, e.g., 2009 Beyer.

[5] Several different mammalian expression systems have been used (at least on a research scale) to make IgA antibodies, including (i) murine cell lines (e.g.,

Chintalacharuvu et al. Proc. Natl. Acad. Sci. USA 94; 6364-6368 (June 1997), describing anti-hapten mouse-human chimeric secretory IgA antibodies made by expressing monomeric IgA in murine transfectomas and then transfecting with human SC gene) and (ii) baby hamster kidney (BHK) cell lines (e.g., Dechant et al. Blood 100(13); 4574-4580 (December 2002), describing anti-cancer chimeric monomeric IgA antibodies made in BHK cells). Nevertheless, the literature seems to focus on making IgA antibodies using CHO cell lines.

[6] For example, one article describes making monomeric, dimeric, and secretory anti-virus IgA antibodies in CHO cells. Berdoz et al. Proc. Natl. Acad. Sci. USA 96; 3029-3034 (March 1999). In particular, this article describes generating single clones to express each of the anti-virus monomeric IgA (mlgA clone 22), dimeric IgA (dlgA clone F), and secretory IgA (slgA clone 6) antibodies, and that clones expressed up to 20ug of antibody per 1 x 10 ⁶ CHO cells in 24 hours. The article proposes that the described CHO approach makes possible the production of large quantities of secretory IgA for clinical applications, i.e., milligram quantities. Nonetheless, the same group later resorted to a method of combining, in vitro, a dimeric IgA (expressed from CHO) and a secretory chain to make fully assembled secretory IgAs. Favre et al. (2003), Journal of Chromatography B, 786, pp 143-151.

[7] Another article describes making anti-cancer monomeric IgA antibodies in CHO cells. Dechant et al. Journal of Immunology 179; 2936-2943 (2007). The article describes that the disclosed CHO system expressed 3 (IgAl) or 5 (IgA2) pg/cell/day. This article further describes challenges in using the CHO system to express IgAs, including a lack of homology between human and mouse receptor systems, difficulties with large scale production and purification, and obstacles in obtaining desired glycosylation. Still another article describes the serum-free production and purification of chimeric anticancer monomeric IgA antibodies in CHO cells. 2009 Beyer. This article describes that the disclosed CHO system expressed 2.2pg/cell/day, and that the disclosed CHO system expressed 2mg per week per roller bottle unit over 6 months (based on data that is not shown). This article further describes that difficulties in obtaining enough recombinant IgA material with desired specificity for in vivo studies remained in 2009 because of (among other reasons) a lack of established methods for production and purification. [8] It has been proposed that plants expression systems have potential advantages over traditional mammalian expression systems, including lower production costs, the absence of animal pathogens, rapid scalability, and the ability to fold complex proteins accurately. See, e.g., Ma et al. Nat Rev Genet 4; 794-805 (October 2003). Plant expression systems have been used for a variety of recombinant proteins, but there is relatively little literature on the expression of IgA antibodies in plant systems. Although one article describes an anti-virus monomeric IgA made in maize (Karnoup et al.

Glycobiology 15(10); 965-981 (May 2005)), most literature on this subject seems to be linked to one group, Planet Biotechology Inc. This group describes in several articles an anti-bacterial IgA-G hybrid antibody made in transgenic tobacco. See, e.g., Ma et al. Science 268(5211), 716-719 (May 1995); Ma et al. Nature Medicine 4(5); 601-606 (May 1998); Wycoff Current Pharmaceutical Design 10(00); 1-9 (2004).

[9] Specifically, the 1995 and 1998 Ma articles describe the expression of an anti- Streptococcus mutans chimeric secretory IgA-G hybrid antibody in tobacco. The hybrid antibody is described as containing murine monoclonal kappa light chain, hybrid Ig A-G heavy chain, murine J-chain, and rabbit secretory component. The antibody was made by successive sexual crossing between four transgenic N. tabacum plants and filial recombinants to form plant cells that expressed all four protein chains simultaneously. The parent is identified as the IgG antibody Guy's 13.

[10] The 2004 Wycoff article also describes expressing in transgenic tobacco the Guy's 13-based chimeric secretory IgA/G hybrid antibody (called CaroRx in this article) and cites to the 1995 and 1998 Ma articles (among others). This article briefly mentions that Planet Biotechnology Inc. had at some point transformed plants with all four chains (heavy chain, light chain, J-chain, and secretory component chain) simultaneously to produce a slgA that was fully human except for murine variable regions. But this is based on unpublished results and is not further discussed. Apparently, the CaroRx antibody was difficult to purify; the affinity of Protein A for the murine Ig domain was too low and Protein G was necessary for sufficient affinity chromatography. Furthermore, the article states that several other chromatographic media had shown little potential as purification steps for the hybrid slgA-G from tobacco leaf extracts. The article also indicates that the authors were unable to control for human-like glycosylation in tobacco, but that such was unlikely to be a problem because people are exposed to plant glycans every day in food without ill effect.

[11] WO9949024, which lists Wycoff as an inventor, Planet

Biotechnology Inc. as the applicant, describes the use of the variable regions of Guy's 13 to make a secretory antibody from tobacco. The application contains only two examples - the first a working example and the second a prophetic example. Working Example 1 describes the transient production of an anti-S. mutans (variable regions from Guy's 13) in tobacco. The tobacco plant was transiently transformed using particle bombardment of tobacco leaf disks. Transgenic plants were then screened by Western blot "to identify individual transformants expressing assembled human slgA" (p. 25). Prophetic Example 2 states that in a transformation system for Lemna gibba (a monocot), bombardment of surface-sterilized leaf tissue with DNA-coated particles "is much the same as with" tobacco (a dicot). The prophetic example also stops at screening by immunoblot analysis for antibody chains and assembled slgA, and states that the inventors "expect to find fully assembled slgA."

[12] Despite the increased attention to plant expression systems for antibodies and the use of IgA antibodies as therapeutic agents, secretory IgA antibodies have not been commercially advanced. Accordingly, it is desirable to have an expression system for the efficient production and purification of therapeutic secretory IgA antibodies.

SUMMARY OF THE INVENTION

[13] The present invention relates to the discovery that duckweed can serve as a useful system for the expression and production of fully-formed secretory IgA antibodies. Accordingly, a first aspect of the invention relates to a method of producing a secretory IgA antibody, which comprises expressing in a stably-transformed duckweed a fully- formed secretory IgA antibody. The duckweed that is expressing the secretory IgA antibody can be a mature plant. Once the stably -transformed duckweed is obtained, expression of the secretory IgA antibody generally involves culturing the duckweed, including growing mature plants. Subsequently the secretory IgA antibody can be recovered, typically by extracting the antibody from the tissue of the duckweed, and optionally purifying the resulting antibody composition.

BRIEF DESCRIPTION OF THE SEQUENCES [14] SEQ ID NO:l provides the amino acid sequence of a human κ light chain constant region (CL) (UniProtKB/Swiss-Prot database entry P01834

(IGKC_HUM AN)) .

[15] SEQ ID NO:2 provides the amino acid sequence of a human λΐ light chain constant region (CL) (UniProtKB/Swiss-Prot database entry P0CG04

(LAC1_HUMAN)).

[16] SEQ ID NO:3 provides the amino acid sequence of a human λ2 light chain constant region (CL) (UniProtKB/Swiss-Prot database entry P0CG05

(LAC2_HUM AN)) .

[17] SEQ ID NO:4 provides the amino acid sequence of a human λ3 light chain constant region (CL) (UniProtKB/Swiss-Prot database entry P0CG06

(LAC3_HUM AN)) .

[18] SEQ ID NO:5 provides the amino acid sequence of a human λ6 light chain constant region (CL) (UniProtKB/Swiss-Prot database entry P0CF74

(LAC6_HUM AN)) .

[19] SEQ ID NO:6 provides the amino acid sequence of a human λ7 light chain constant region (CL) (UniProtKB/Swiss-Prot database entry A0M8Q6

(LAC7_HUM AN)) .

[20] SEQ ID NO:7 provides the amino acid sequence of a human IgA a-1 heavy chain constant region (Cal-Ca2-Ca3) (UniProtKB/Swiss-Prot database entry P01876 (IGH A 1 _HUM AN) ) . [21] SEQ ID NO: 8 provides the amino acid sequence of a human IgA a-2 m(l)-allotype heavy chain constant region (Cal-Ca2-Ca3) (UniProtKB/Swiss-Prot database entry P01877 (IGHA2_HUM AN)) .

[22] SEQ ID NO: 9 provides the amino acid sequence of a human IgA a-2 m(2)-allotype heavy chain constant region (Cal-Ca2-Ca3). (UniProtKB/Swiss-Prot database entry P01877 (IGHA2_HUMAN) with indicated modifications for allotype 2 variant).

[23] SEQ ID NO: 10 provides the amino acid sequence of a human IgA a-

2 (n)-allotype.

[24] SEQ ID NO: 11 provides the amino acid sequence of a human J- chain (amino acids 23-159 UniProtKB/Swiss-Prot database entry P01591).

[25] SEQ ID NO: 12 provides the amino acid sequence of a human secretory component (amino acids 19-603 of UniProtKB/Swiss-Prot database entry P01833 (PIGR_HUM AN) , RCSB Protein Data Bank structure 2OCW).

[26] SEQ ID NO: 13 provides the amino acid sequence for a signal peptide (heavy chain secretion signal).

[27] SEQ ID NO: 14 provides the amino acid sequence of the heavy chain variable region of infliximab (cA2).

[28] SEQ ID NO: 15 provides the DNA sequence for the infliximab heavy chain IgA2m(n) optimized for maize.

[29] SEQ ID NO: 16 provides the amino acid sequence for a signal peptide (light chain secretion signal). [30] SEQ ID NO: 17 provides the amino acid sequence of the light chain variable region of infliximab (cA2).

[31] SEQ ID NO: 18 provides the DNA sequence for the infliximab light chain optimized for maize.

[32] SEQ ID NO: 19 provides the amino acid sequence of a natural signal peptide (secretion signal) for a human J-chain (amino acids 1-22 UniProtKB/Swiss-Prot database entry P01591).

[33] SEQ ID NO:20 provides the DNA sequence for a J-chain optimized for maize.

[34] SEQ ID NO:21 provides the amino acid sequence for a signal peptide (SC-chain secretion signal).

[35] SEQ ID NO:22 provides the DNA sequence for an SC-chain optimized for maize.

[36] SEQ ID NO:23 provides the amino acid sequence for a rice a- amylase signal peptide (secretion signal).

[37] SEQ ID NO: 24 provides the amino acid sequence of the heavy chain variable region of adalimumab (D2E7).

[38] SEQ ID NO: 25 provides the amino acid sequence of the light chain variable region of adalimumab (D2E7).

[39] SEQ ID NO:26 provides the DNA sequence for a J-chain optimized for Lemna. [40] SEQ ID NO: 27 provides the DNA sequence for an SC-chain optimized for Lemna.

[41] SEQ ID NO:28 provides the DNA sequence for the adalimumab heavy chain IgAl optimized for Lemna.

[42] SEQ ID NO: 29 provides the DNA sequence for the adalimumab light chain κ optimized for Lemna.

[43] SEQ ID NO:30 provides the amino acid sequence of the heavy chain variable region of ustekinumab (CTNO-1275).

[44] SEQ ID NO:31 provides the amino acid sequence of the light chain variable region of ustekinumab (CTNO-1275).

[45] SEQ ID NO:32 provides a complete Lemna-optimized UKB-SA1 heavy chain IgAl DNA (including DNA encoding signal peptide SEQ ID NO:23).

[46] SEQ ID NO:33 provides a complete Lemna-optimized UKB-SA1 light chain DNA (including DNA encoding signal peptide SEQ ID NO:23).

BRIEF DESCRIPTION OF THE DRAWINGS

[47] Figures 1 A and IB show the amino acid sequences of various human antibody light chain subtypes and allotypes. Figure 1 A shows the amino acid sequence of a human κ light chain constant region (C _L ) (UniProtKB/Swiss-Prot P01834) (SEQ ID NO:l); Figure IB shows the amino acid sequences of a human λ light chain constant region (C _L ) allotypes (UniProtKB/Swiss-Prot P0CG04, P0CG05, P0CG06, P0CG74, and A0M8Q6; SEQ ID NOS:2 to 6). [48] Figure 2 shows the amino acid sequences of the constant regions of human IgAl (SEQ ID NO:7), IgA2 m(l) (SEQ ID NO:8), IgA2 m(2) (SEQ ID NO:9), and IgA2(n) (SEQ ID NO: 10) antibody a heavy chains;

[49] Figure 3 shows the amino acid sequence of a human J-chain (a.a. 23-

159 of UniProtKB/Swiss-Prot entry P01591, SEQ ID NO:l l).

[50] Figure 4 shows the amino acid sequence of a human secretory component (a.a. 19-603 of UniProtKB/Swiss-Prot database entry P01833

[PIGR_HUM AN] , SEQ ID NO:12).

[51] Figure 5 shows the structure of the SIgA vector constructs of

Reference Example 1.

[52] Figure 6 shows reducing and non reducing gels of an anti-TNFa

SIgA of Example 2. The label "A" shows non-reducing SDS-PAGE analysis

demonstrating expression of complete SIgA while the label "B" shows reducing SDS- PAGE analysis.

[53] Figures 7A through 7C show the structure of vector constructs

SynAOl (Fig 7A), SynA02 (Fig 7B) and SynA03 (Fig 7C) used for expression of an anti- p40 (anti-IL 12/23) SIgA in Lemna in Example 4.

[54] Figure 8 shows reducing and non reducing gels of an anti-p40 (anti-

IL-12/23) SIgA in Example 5. The A gel shows non-reducing SDS-PAGE analysis demonstrating expression of complete SIgA. The B gel shows reducing SDS-PAGE analysis. [55] Figure 9 shows the degradation of an anti-p40 (anti-IL- 12/23) SIgA in Example 7 compared to ustekinumab (IgGl) and to colostral secretory IgA in simulated intestinal fluid (SIF). Gel A compares the SynAOl-WT antibody of the invention (UKB-SA1) and gel B compares the SynAOl-GO antibody of the invention (UKB-SAlgO).

DETAILED DESCRIPTION OF THE INVENTION

[56] The present invention is based on the discovery that secretory IgA antibodies can be advantageously expressed by a stably-transformed duckweed. Though secretory IgA ("SIgA") antibodies are large and complex, duckweed can nonetheless be stably-transformed to express the fully-formed SIgA antibody. Further, the duckweed expression system lends itself to the recovery of the fully-formed SIgA antibody.

Because good expression rates can be obtained and the antibody can be subsequently isolated/recovered, the use of stably-transformed duckweed permits the production of SIgA antibodies in bench-scale, clinical-scale, and/or commercial-scale amounts. In contrast, other plant expression systems, namely tobacco and lettuce, gave very poor expression of SIgA and essentially no recoverable yield. Surprisingly, transient CHO expression systems also gave very low titers of SIgA expression. The discovery that a duckweed expression system can effectively express a fully-formed SIGA antibody is thus a surprising and beneficial advance.

[57] The antibodies that can be expressed by duckweed according to the present invention are essentially any secretory IgA antibody. An antibody, as is well known, has a basic structural unit that consists of two heavy chain proteins (heavy chains) and two light chain proteins (light chains), which are bound together by non- covalent and covalent (e.g., disulfide bonds) interactions into a single unit. The heavy and light chains have N-terminal variable regions and C-terminal constant regions. The variable regions of the light and heavy chains together form an "antigen binding region." Because the antibody has two heavy and light chains, the antibody has two antigen binding regions.

[58] Antibodies are classified based on the heavy chain constant region, e.g., classified as IgG, IgA, IgM, IgE, IgD, etc. In nature, the heavy chain constant regions of the various classes are produced by different genes: the IgA class heavy chains are uniquely encoded-for by a genes, the IgG class heavy chains by γ genes, and so forth.

The heavy chain constant regions also impart the various classes with differences in their physio-chemical properties, their isotypic antigentic determinants, and/or in their biological function. Lefranc et al., The Immunoglobulin FactsBook, Academic Press

2001, Chapter 2, (ISBN 0-12-441351-X). The constant region of the IgA heavy chains

(Ca) typically have three domains that are referred to as Cal, Ca2, and Ca3. Unlike the heavy chain constant region, the light chain constant region is not used for classification.

In humans, for example, all classes use one of two types of light chain constant regions, namely the CK (kappa) or Ck (lambda) type. The amino acid sequences of human kappa

(SEQ ID NO: l) and several lambda light chain constant regions (SEQ ID NOS:2-6) are provided in Figs. 1A and IB, respectively. The definition and structure of antibodies are well known to workers skilled in this art, such as described in, e.g., Alberts, B. et al, Molecular Biology of the Cell 3 ^rd Edition, Chapter 23, Garland Publishing Inc., New York, NY, 1994, and Nezlin, R., The Immunoglobulins. Structure and Function (1998) Academic Press (ISBN 0-12-517970-7).

[59] The C-terminal section of two IgA antibodies, i.e., C-terminal ends of the Ca3 region (also referred to as the "tail piece"), can be joined together via a J- chain to form a dimer. Dimeric IgA has four antigen binding regions; two from each IgA monomer. Typically the four antigen binding regions (and their CDRs) are identical for reasons such as ease of manufacture. But the antigen binding regions can, in certain circumstances, be different, e.g., different CDRs binding different epitopes on the same antigen or event different antigens (such as in the case of bispecific antibodies). Typically the CDRs of the four antigen binding regions are identical. A secretory chain, sometimes called a secretory component or SC-chain, can be attached to the dimeric IgA antibody. The SC-chain provides increased resistance to proteolysis especially in the intestinal tract. The SC-bound dimeric IgA is referred to herein as "secretory IgA" or "SIgA." Interestingly, in mammals such as humans, SIgA formation requires the cooperation of two cell types: the plasma cells to produce dimeric IgA and epithelial cells to add the SC- chain onto the dimeric IgA post-secretion from the plasma cells. See Pabst (2012) Nat Rev Immunol. 2012 12:821-832 for a recent review on this subject. But surprisingly, in the present invention, a single transformed duckweed plant cell can produce and assemble a fully-formed SIgA.

[60] Heavy chain constant regions that qualify as an IgA-class antibody are well known in the art. Generally the amino acid sequence of the heavy chain constant regions of an IgA, regardless of how it is produced (e.g., naturally or recombinantly), corresponds to an amino acid sequence encoded for by an a-gene. In addition, IgA antibodies have characteristic antigenic determinants unique to IgA-class antibodies and different from the antigenic determinants of other classes of antibodies, such as IgG-class antibodies (see, e.g., Nezlin, R., The Immunoglobulins. Structure and Function (1998) Academic Press (ISBN 0-12-517970-7); Lefranc et al., The Immunoglobulin FactsBook, Academic Press 2001, Chapter 2, (ISBN 0-12-441351-X)). Furthermore, IgA antibodies are the only isotype that is known to specifically bind to the FcaR (see, e.g., Alberts, B. et al, Molecular Biology of the Cell 3 ^rd Edition, Chapter 23, Garland Publishing Inc., New York, NY, 1994; Lefranc et al., The Immunoglobulin FactsBook, Academic Press 2001 , Chapter 2, (ISBN 0-12-441351-X)).

[61] Accordingly, for purposes of the present invention, the terms "IgA antibody," "monomeric IgA," "dimeric IgA" and "SIgA" each refers to antibodies that contain the heavy chain constant regions of an IgA class of immunoglobulin, e.g., which corresponds to an amino acid sequence that can be encoded for by a genes and which react with an antibody specific for the IgA-class heavy chain. The amino acid sequence "corresponds" in that it is identical to, or contains only minor variations

(insertions/deletions/substitutions) from, an amino acid sequence produced by any a gene, an individual human's IgA heavy chain sequence, or a human IgA heavy chain consensus sequence. Indeed, variations can and do exist in the amino acid sequence of the IgA heavy chain constant region without moving such antibodies outside of the IgA class. For clarity, because the heavy chain sequence is determinative of the Ig class, a recombinant antibody containing the IgA heavy chain constant regions and further containing the antigen binding regions encoded for by DNA sequences obtained from a known IgG antibody is still an "IgA antibody." On the other hand, a secretory IgA antibody modified to replace the Ca2 heavy chain constant domain (encoded for by the IgA-specific a-gene) with a Ογ2 heavy chain constant domain (encoded for by the IgG- specific γ-gene) is not an IgA antibody, and is instead a hybrid IgA/IgG antibody. Such a hybrid is not within the scope of the terms "monomeric IgA," "dimeric IgA," and "SIgA" antibodies, and thus is not a secretory IgA antibody according to the invention.

[62] Minor variations of the heavy chain constant regions are permitted only to the extent that the overall antibody class, framework, and functionality of SIgA is maintained; e.g., J-chain binds to monomers and SC-chain binds to the dimeric structure and provides proteolysis resistance. Such variations include conservative substitutions. Exemplary conservative substitutions are shown in Table 1. The amino acids in the same block in the second column and preferably in the same line in the third column may, for example, be substituted for each other.

Table 1

[63] Typical minor variations of the constant regions from the normal or naturally-occurring sequence involve only conservative changes to the amino acid sequence using the recognized substitutions, insertions and/or deletions.. Generally, the variations (substitutions, insertions, and/or deletions) of a constant domain of the heavy or light chain means no more than 10 and usually no more than 5 amino acid additions, deletions, and/or substitutions (either naturally-occurring or genetically-engineered), in any Cal, Ca2, or Ca3 domain in comparison to a normal IgA constant domain. The sum of these minor variations in the constant domains of the SIgA antibody of the invention is usually less than 20 amino acids (acid/deletions/substitutions) and often less than 10 or less than 5.

[64] Accordingly, at a minimum, SIgA includes any recognized amino acid sequence that is generally accepted as being within the IgA class. For example, information on the structure and function of IgA can be found in Snoeck et al., Vet. Res. 37; 455-467 (2006) and "Mucosal immune defense: Immunoglobulin A", C.S. Kaetzel ed., Springer, New York (2007) ISBN 978-0-387-72231-3. Electronic databases, such as RCSB Protein Data Bank, can also establish a known IgA sequence or portion/domain thereof. The constant domains contained in the SIgA antibodies of the invention can be human, non-human, or a combination of these. Preferred are mammalian constant regions. Most preferred are human constant regions.

[65] In humans there are two recognized IgA subclasses: IgAl and IgA2 which differ in the hinge section between the Cal and Ca2 domains of the heavy chain. In IgAl this hinge section is relatively long and in nature typically O-glycosylated. In IgA2 the hinge section is relatively short and in nature lacks glycosylation. Both IgAl and IgA2 SIgA antibodies are usually present in mucosal secretions. In humans, the IgA2 subclass has three known allotypes: IgA2m(l), IgA2m(2) and IgA2m(n). Unlike the subclasses, only one specific allotype will be found in a normal healthy individual. The m(l) allotype is strongly prevalent in the Caucasian population (98%) and varies between 23% and 96% for other populations. The m(2) allotype has a high prevalence in populations of African and Asian descent (50-70%). The m(n) allotype - which is considered to be a hybrid of the m(l) and m(2) allotypes - has been reported to be genetically possible, but has not been actually observed in any population. See

Chintalacharuvu et al, Journal of Immunology 152, 5299-5304 (1994). Accordingly, SIgA antibodies of the present invention preferably contain human IgA heavy chain constant regions of the IgAl or IgA2 sub-types, including IgA2m(l), IgA2m(2) and IgA2m(n) allotypes, and combinations thereof (e.g., one constant domain from an IgAl and another constant domain from an IgA2).

[66] Typically, the SIgA antibodies of the invention comprise the Cal domain, the hinge section, and the -Ca2-Ca3 domains of of an IgA antibody (with or without minor variations), including a human IgAl and/or IgA2 antibody. In this embodiment, the Cal domain, the hinge section, and the -Ca2-Ca3 domains can be of an IgAl, an IgA2m(l) allotype, an IgA2m(2) allotype, or a combination thereof. Amino acid sequences of a human IgAl heavy chain constant region (SEQ ID NO:7), a human IgA2m(l) heavy chain constant region (SEQ ID NO:8), a human IgA2m(2) heavy chain constant region (SEQ ID NO:9), and a human IgA2(n) heavy chain constant region (corresponding to the Cal and Ca2 regions of an IgA2m(2) and the Ca3 region of an IgA2m(l)) (SEQ ID NO: 10) are respectively shown in Figure 2.

[67] A modified, shortened, or removed linker/hinge section between

Cal and Ca2 in IgAl has been reported to increase resistance against proteases (for example see Senior et al, J. Immunol. 2005; 174: 7792-7799). Such can be incorporated into the SIgA antibodies of the present invention.

[68] The J-chain is a protein that attaches to the tail piece of a monomeric

IgA to join two monomeric IgAs to form a dimer. The J-chain is normally of mammalian origin, such as human, murine, rat, rabbit, sheep, cow, or goat origin, but is preferably of human origin. An example of the amino acid sequence of a human J-chain is set forth in Figure 3 (SEQ ID NO:l 1). Usually, the sequence of the mammalian J-chain is the same as the naturally-occurring sequence, but it can be subject to minor variations as described above for constant regions generally, e.g., up to 10 amino acid insertions, substitutions, or deletions. The minor variations do not significantly alter the function of the J-chain, and in particular the ability to join two monomeric IgA antibodies to form a dimer and to enable attachment of the SC-chain.

[69] The secretory component, also referred to as "SC" or "SC-chain," is a protein that binds to the dimeric IgA framework and imparts increased resistance against proteolysis upon the antibody to which it is bound. Generally, the secretory component is of mammalian origin, such as human, murine, rat, rabbit, sheep, cow, or goat origin, but is preferably of human origin. The SC is the result of cleavage of the Polymeric IgA-receptor (PIGR) which usually occurs at a specific position. Some variation can occur in the position of the cleavage resulting in variant forms of SC.

Usually, the sequence of the mammalian-derived SC-chain is the same as the naturally- occurring sequence, but it can be subject to minor variations as described above for constant regions generally, e.g., up to 10 amino acid insertions, substitutions, or deletions. The minor variations do not significantly alter the function of the secretory component, e.g., the ability to stabilize the SIgA against proteolysis. An example of the amino acid sequence of a human secretory component is set forth in Figure 4 (SEQ ID NO: 12).

[70] The antigen binding region binds to and neutralizes a target antigen.

Typically, the antigen binding region comprises a heavy and light chain variable region pair, each containing hypervariable regions (CDRs, which directly interact with the antigen) and the supporting framework regions. The CDRs in each heavy and light chain variable region are separated from each other and from the constant domain by the framework regions, which serve to maintain the CDRs in the correct binding

conformation. In general each variable part of an immunoglobulin heavy or light chain contains 3 different CDRs and four framework regions. For a more detailed description of antibody antigen binding regions, see for example C.A. Janeway et al, "Immunobiology" 6 ^th Edition, Chapter 3, pp 110-115; Garland Science Publishing, New York, 2005 (ISBN 0815341016). Regarding framework regions in particular, see for example WO92/22653 (discussing that although framework regions do not directly interact with antigen, framework regions can influence binding of the CDRs with antigen, such as binding strength and/or downstream events). [71] In embodiments, the antigen binding region can target bacteria, viruses, and other pathogens; pollen and other allergens; cancer targets (e.g., tumor- specific antigens; tumor-associated antigens); and proinflammatory mediators (among other targets). In some embodiments, the antigen binding regions of the SIgAs of the invention bind to and neutralize a proinflammatory cytokine or a receptor thereof. These cytokines and receptors include TNF-alpha (preferably soluble TNF-alpha), IFN-gamma, IFN-alpha, GM-CSF, CXCL10/H 0, IL-l-beta, IL-l-alpha, IL-4, IL-5, IL-6, IL-12, IL- 13, IL-17A, IL-18, IL-20, IL-22, IL-23, IL-1 receptor, IL-2 receptor, IL-4 receptor, IL-5 receptor, IL-6 receptor, and IL-17 receptor.

[72] Secretory IgA antibodies specifically and preferentially bind with high affinity to a corresponding antigen, such as a human proinflammatory cytokine. A variety of protocols for performing binding, competitive binding, or immuno -radiometric assays to determine the specific binding capability of an antibody are well known in the art (see for example Maddox et al, J. Exp. Med. 158, 1211-1226, 1993). Such

immunoassays typically involve the formation of complexes between the specific protein and its antibody and the measurement of complex formation (e.g., binding to or unbinding from the specific protein). Generally, the affinity of secretory IgA antibodies of the invention is at least two-fold, at least 10-fold, at least 50-fold, at least 100-fold, or at least 1000-fold or greater than the affinity of the antibodies for a non-specific protein such as, for instance, BSA or casein. Typically the secretory IgA antibodies of the present invention exhibit a binding affinity constant (K _D ) with respect to the antigen of 10 ^"7 M or lower, preferably 10 ^"8 M, 10 ^"9 M, 10 ^"10 M' 10 ^"U M, or 10 ^"12 M or lower. [73] Generally the secretory IgA antibodies of the present invention neutralize the antigen to which it is bound. For the present invention, the term

"neutralizes" means inhibits/reduces the effect of the antigen to some degree, such as by at least 30%, at least 35%, at least 40%, and at least 45%. Typically the inhibition / reduction in the effect of the antigen at least 50%. For example, in some embodiments, the secretory IgA antibodies of the present invention inhibit/reduce the effect of an antigen to which it is bound by at least 55%, at least 60%, at least 65%, at least 70%, at least 75%, at least 80%, at least 85%, at least 90%, at least 95%, or at least 98%.

[74] Secretory IgA antibodies of the invention can be non-human antibodies, chimeric antibodies, humanized antibodies, human antibodies, or other combinations of human and non-human sequences/regions. A chimeric antibody is an antibody having an antigen binding region (CDRs and framework) originating from a first species (typically a mouse) and heavy chain constant regions originating from a second species (typically a human). Exemplary chimeric antibodies are described in Knight, D.M, et al. (1993) Mol. Immunol. 30: 1443-1453; WO 92/16553). A humanized antibody is a human antibody onto which non-human (typically murine) CDRs have been grafted. In the humanized antibody, certain human supporting framework amino acid residues can be replaced with their counterparts from the non-human parent antibody. Such an antibody containing certain non-human framework residues is still a humanized antibody. See, e.g.,. Carter et al. (1992) Proc Natl Acad Sci U SA. 1992 89(10):4285-9. In the humanized antibodies, the sequence of the supporting framework into which the non-human CDRs are grafted can be obtained from any human isotype/class, preferentially from IgG or IgA, and may be modified to improve the properties thereof (e.g., antigen binding and/or downstream effects). A human antibody is fully-human, containing only human constant and variable regions, i.e., having only human heavy and light chains (derivable from human genomic sequences by naturally-occurring recombination and mutation processes, consensus sequences, phage display selection of specificities with the highest binding affinities for a given target, etc.). Likewise, a non- human antibody contains only non-human constant and variable regions, i.e., having only non-human heavy and light chains. Exemplary fully human antibodies are described in pub no WO97/29131 and references cited therein.

[75] The SIgA antibodies of the present invention are monoclonal antibodies. A "monoclonal antibody" refers to a population or collection of antibodies that are substantially identical because they were all produced by clones of a single cell. For the present invention, a monoclonal SIgA is a SIgA containing monoclonal monomeric IgA antibodies, a J-Chain, and an SC-chain that were all produced by clones of a single cell.

[76] The method of the present invention expresses a SIgA antibody as described above in a stably-transformed duckweed. Generally, a genetically modified duckweed is a known expression system for producing various proteins (see U.S. Patent No. 6,040,498), including for the production of monoclonal antibodies (see U.S. Patent No. 7,632,983). Duckweed is the common name for the members of the

monocotyledonous family Lemnaceae. The five known genera and 38 species of Lemnaceae are all small, free-floating, fresh-water plants whose geographical range spans the entire globe: genus Lemna (L. aequinoctialis, L. disperma, L. ecuadoriensis, L. gibba, L. japonica, L. minor, L. minuscula, L. obscura, L. perpusilla, L. tenera, L.

trisulca, L. turionifera, L. valdiviana); genus Spirodela (S. intermedia, S. polyrrhiza); genus Wolffia (Wa. angusta, Wa. arrhiza, Wa. australina, Wa. borealis, Wa. brasiliensis, Wa. columbiana, Wa. elongata, Wa. globosa, Wa. microscopica, Wa. neglecta) genus Wolfiella Wl. caudata, Wl. denticulata, Wl. gladiata, Wl. hyalina, Wl. lingulata, Wl. repunda, Wl. rotunda, and Wl. neotropica), and genus Landoltia (L. punctata). For clarity, the term "duckweed' as used in the present invention includes the foregoing species, genetically modified variants thereof (e.g., modified to control glycosylation, secretion, etc.), and any other genera or species of Lemnaceae, if they exist, optionally in a genetically modified form. Typically the genus Lemna is preferred, especially the species L. minor and L. gibba in natural or genetically modified forms. Also, the use of the term "duckweed," or any genus or species thereof, is meant to include individual plant cell(s), nodules, as well as whole plants including mature plants having root and fronds, unless otherwise indicated by context or express statement.

[77] The stably-transformed duckweed of the present invention means that the nucleic acid sequences and/or genes needed to produce the desired SIgA have beenintroduced into the genome of a duckweed by transgenesis. This modification is considered a stable transformation because the nucleic acid introduced to the duckweed have integrated into the genome and are capable of being inherited by the progeny thereof, more particularly, by the progeny of multiple successive generations.

Transformant lines are isolated by applying a selective pressure , resistance to which is encoded in the construct used for transgenesis Transformed duckweed lines carrying the exogenous DNA of interest do not necessarily express the SIgA of interest due to, for example, transgene silencing and epigenetic control of gene expression; however screening of different lines allows for the isolation of lines expressing the SIgA of interest at desired levels. This genetic modification causes the duckweed to express the desired SIgA antibody during its otherwise normal metabolic activity. The term "express" and its grammatical variants refers to the biosynthesis of the SIgA antibody, which includes the transcription, translation, and assembly of the antibody by the duckweed. Thus, expressing a SIgA antibody in a stably-transformed duckweed entails providing the stably -transformed duckweed with an environment that supports metabolic activity and/or growth for a sufficient period of time to produce the SIgA antibody. Generally this environment comprises providing light (natural and/or artificial) and a liquid medium typically based on water. Providing this environment is often referred to as "culturing" the duckweed. Methods of culturing duckweed including the media, supplements (if any), conditions, etc., are known in the art and have been disclosed in, e.g., US patents 6,040,498; 7,161,064; and 7,632983; and references cited therein, respectively.

[78] Culturing of transgenic duckweed of the invention can be performed in transparent vials, flask, culture bags, or any other container capable of supporting growth using defined media. In some embodiments of the invention large scale growth of duckweed, necessary to achieve industrial production levels, is carried out in bioreactor tailor-designed for growth of duckweed. In a preferred embodiment, duckweed bioreactors, which can be inoculated aseptically, support aseptic growth of duckweed. In even more preferred embodiments, a bioreactor can be directly connected to harvest bag to separate the media from the plant material, either of which can then be piped into downstream purification steps. Suitable bioreactors, methods/devices to inoculate them aseptically, and aseptic harvest bags are described in US patent 7,176,024 to Branson et al. or in US application 2010/209966 To Everett et al.

[79] In embodiments, the SIgA of the invention are expressed in auxotrophic duckweed requiring specific supplements for growth. This property allows for effective containment of the transgenic duckweed. Auxotrophic duckweed can be created by RNAi downregulation of expression of metabolic enzyme such a s threonine deaminase. Methods of making such auxotrophic duckweed, which require

supplementation of Isoleucine for growth are described in Ngyen et al. (2012) Transgenic res 21: 1071-83.

[80] The SIgA antibody expressed by the stably-transformed duckweed is fully-formed. This means that the antibody is intact, properly-folded, and functional (e.g., binds to its target antigen). It does not exclude, however, the formation of non- intact SIgA antibodies, such as monomeric IgA, dimeric IgA, or fragments thereof, or the formation of incorrectly folded SIgA antibodies. Rather, so long as at least some of the complete, intact, and properly folded SIgA antibody is expressed, then the transformed duckweed has produced fully-formed SIgA. Preferably, the stably -transformed duckweed expresses more SIgA antibodies than monomeric IgA antibodies, based on a mole basis. Likewise, the stably-transformed duckweed preferably expresses more SIgA antibodies than dimeric IgA antibodies (i.e., dimers devoid of SC-chain). Generally the stably-transformed duckweed expresses sufficient amounts of the fully-formed SIgA antibody that the SIgA antibody is at least 1 % of the total soluble protein (TSP), often at least 2%, and typically at least 5%. In practice, the expressed SIgA antibodies generally comprise 5 to 25% TSP. In terms of weight, the stably-transformed duckweed generally expresses at least 50 mg of SIgA antibody per kg of wet biomass, up to 500 mg per each kg of wet biomass. In general 12% TSP equates to about 1.0 gram of antibody.

[81] In a preferred embodiment, the culturing of the stably-transformed duckweed amounts to growing a clone population of stably-transformed duckweed plants, each expressing the same fully-formed SIgA antibody. The plants double their population in approximately every 30 to 60 hours. The plants grow on the surface of a relatively still (not circulating or turbulent) water-based media and are provided with light source and a CO ₂ source, such as air. In this way, a large biomass can be

accumulated which has expressed the fully-formed SIgA antibody, and thus large amounts of the antibody can be produced.

[82] Once the stably-transformed duckweed has expressed the fully- formed SIgA antibody, the duckweed preferably secretes the antibody. For purposes of the present invention, "secretion" means translocation of a polypeptide across the plasma membrane of the duckweed cell to region in between the cell membrane and the cell wall (known as the apoplast) and/or to the cell wall itself (where the secreted slgA remains associated with the cell wall). The SIgA secreted into the apoplast (and/or cell wall) generally does not cross the cell wall and diffuse into the liquid media (or culture media) in any detectable amounts. Typically, less than 1% of the fully-formed SIgA antibody crosses the cell wall and enters the liquid media/ external environment.

[83] Following expression of the fully-formed SIgA antibody, recovery of the antibody from the duckweed and/or the culture media is often desired. The first step, generally, is to separate the SIgA antibody from the duckweed. Typically, the fully- formed SIgA antibody is retained within the duckweed's apoplast and/or the duckweed's cell wall. Separation in this case generally requires extraction.

[84] Extraction of secreted SIgAs typically involves a homogenization step to disrupt the plant material and allow for release of the secreted SIgA from the apoplast and/or cell wall into the homogenization buffer, which is also referred to as extraction buffer or extraction media. Homogenization buffers and techniques are known in the art. Small scale homogenization can be performed manually, such as by using mortar-and-pestle crushing, and the like. Larger scale homogenization is preferably performed using a mechanical mixer, typically a high shear mixer such as a Silverson 275 UHS mixer, or similar apparatuses. The buffer is typically an aqueous phosphate buffer composition though such is not required. The buffer may contain additional functional ingredients as is known in the art. For example, to reduce proteolysis by metallated proteases, EDTA may be added to the extraction buffer, typically in amounts from 1 to 20 mM, including 5 to 10 mM. Also, one or more anti-oxidants, such as ascorbic acid, sodium metabisulfite, benzyl alcohol, benzoic acid, and the like, may be added during the homogenization process. Homogenization is generally followed by centrifugation and filtration to obtain a buffer solution that contains the SIgA antibodies and other soluble proteins.

[85] Homogenization is often followed by clarification; a step that seeks to remove certain naturally abundant impurities including host cell proteins such as RuBiSco, as well as non-proteinaceous impurities, such as tannins. This is usually achieved by acidic precipitation. For example, clarification can be performed by adjusting the pH of the filtrated homogenate to 4.5, followed by centrifugation (such as for 30 min at 12000), neutralization to pH 7.4, and an additional filtration step. In a preferred embodiment, pH adjustments are performed using 1 M citric acid pH 1.5, or 1M sodium acetate for acidification and 2M tris-base for neutralization, though other suitable pH adjusting agents can also be used instead of or in addition to such agents. Filtration is performed as known in the art, often by using a 0.22 μιη filter.

[86] The recovery of the SIgA antibodies from duckweed may end with the extraction buffer or the clarified material. However, for some uses, purification of the antibody is desired. Purification can be performed using known methods and techniques and generally comprises subjecting the clarified material to affinity chromatography (AC), size exclusion chromatography (SEC), and optional polishing steps. For

efficiency, AC usually precedes SEC, though such is not required.

[87] Methods of using affinity chromatography (AC) as a purification step to remove contaminant proteins and impurities are known in the art and are described in "Process Scale Purification of Antibodies" (2009), Edited by U. Gottschalk, J. Wiley and son, Hoboken, NJ, and references cited therein. Usually the SIgA antibody is bound to the affinity resin material while one or more impurities are not bound. The conditions are modified and the previously bound SIgA antibody is eluted from the column. The opposite can also be performed with the desired antibody passing though and the impurity or impurities being bound to the column. The light chain constant region can be the affinity target. Useful affinity columns include KappaSelect and Capto L from GE Healthcare Life Sciences (Piscataway, NJ, USA). When KappaSelect is used, the addition of MgCl ₂ is often advantageous. The use of Protein A as an AC column is usually avoided. Another kind of AC step is known as IMAC (immobilized copper affinity chromatography). IMAC can be used as the sole AC step or in combination with more traditional AC steps. When used, IMAC is often carried out first. If the crude antibody composition, such as the clarified material, contains EDTA, then it is

advantageous to use CuSO ₄ to neutralize the negative effects of EDTA, which interferes which the IMAC purification process. Often IMAC is used for small to medium scale purification of SIgA where the amounts are less than 10 g, typically less than 5 grams.

[88] Methods of using SEC for purification of monoclonal antibodies are known in the art. In general, SEC allows the separation of fully assembled SIgAs of interest from lower molecular forms (such a monomer of IgA, J-chain and SC-chain, or combinations thereof). Furthermore, SEC also permits a buffer change, such as, for example, the reformulation of the SIgA of interest into a new desired buffer. Suitable columns include, for example, a Sephacryl S300 HR column.

[89] Other purification steps can be employed as well. For example, ion exchange chromatography (IEX) can be useful for removing colored impurities associated with the plant material. Methods and techniques for performing IEX chromatographic purification of antibodies are known in the art and are described, e.g., in Graf et al. (1994) Bio separation, vol. 4, no. 1 pages 7-20, or in "Process scale

purification of antibodies (2009) Edited by U. Gottschalk, ed. J. Wiley and son, Hoboken, NJ, and references cited therein. Often IEX, such as anion exchange chromatography (AEX) or cation exchange chromatography (CEX), is performed before IMAC or other AC step is employed, but is not limited thereto and can be employed at other points of the purification and/or can be employed multiple times with the same or different exchange resin (e.g., AEX and subsequently CEX). In some embodiments an AEX column such as DOWEX 1X2 is employed, often before the AC column.

[90] Further polishing/purification steps can be added, as is known in the art. For example, after any and/or each purification step (e.g., chromatography step) an ultrafiltration (UF) step can be performed. Typically, a UF step is performed at or near the end of the polishing phase in order to increase purity and/or change the buffer or concentration of antibody in the buffer.

[91] The SIgA antibodies are often sufficiently recovered so as to be

"isolated" or in an isolated form. As used herein, the terms "isolate," "isolating" and "isolation" refer to separating the antibody from the host cells and native host cell proteins. The extent of separation is generally at least 50%, but is frequently at least 70%, 80%, 90%, 95%, 98%, 99%, 99.5%, or 99.9% (w/w). Isolation is thus related to purification and is generally achieved by completion of the recovery/extraction step, clarification, and/or capture steps described above. Preferably the antibody in isolated form has removed, or been separated from, at least 90%, more typically at least 99% (w/w) of the host cell proteins of the original composition. Typically, about 20% of the expressed slgA can be recovered, e.g., about 25-100mg per each kg of wet biomass. An isolated form of the SIgA antibody can have a concentration of 1 mg/ml or greater, often 2-10 mg/ml or greater.

[92] For some embodiments, especially where a pharmaceutical use is ultimately intended for the antibody, low amounts of incomplete SIgA antibodies are often desirable. This applies to isolated as well as non-isolated or crude forms of the SIgA antibodies. For instance, compositions where the amount of dimer IgA (no SC- chain) is less than 50%, more desirably less than 25%, and often less than 10%, based on the amount of SIgA antibodies in the composition, are often preferred. For clarity, where a composition contained 10 mg of SIgA, the amount of non-SC-chain dimeric IgA would preferably be less than 1 mg, i.e., less than 10% the amount of the SIgA antibody. The same is true for monomeric IgAs: the content of IgA monomers is desirably less than 50% the amount of SIgA's, more desirably less than 25% and often less than 10%. In some embodiments, the combined amount of dimer IgA (i.e., no SC-chain) and monomer IgA is less than 25% of the amount of SIgA in the composition, often less than 10%, and even less than 5%. The above amounts apply to both isolated and non-isolated forms of SIgA compositions. Accordingly, the low relative amounts of incomplete SIgA can be a result of the expression system (native low-production of incomplete SIgA), the result of separation and/or purification that removes incomplete SIgAs, or both. [93] Purified SIgA compositions are also useful. A purified composition contains SIgA antibodies in an amount of at least 85%, often at least 90%, more often at least 95%, and preferably at least 97%, 98%, or 99%, based on the total soluble protein content. The purified composition can be a solid, such as a lyophilized product, or a liquid. A typical liquid form contains no solids, e.g., no insoluble cell wall materials, and is often based on water as the main or sole solvent and optionally containing salts, pH adjusting agents, or buffers. A purified liquid composition generally contains the SIgA antibody in a concentration of 50 μg/ml or more, often at least 100 μg/ml, preferably at least 1 mg/ml.

[94] Recombinant production of slgAs in duckweed requires

transformation of duckweed by transgenesis. Stable transgenesis in duckweed can be obtained by different techniques as described in US patents 6,040,498 and 7,161,064 to Stomp et al. Briefly stable duckweed transgenesis can be achieved by DNA-coated particle bombardment, electroporation, and Agrobacterium spp. -mediated transformation. Preferably, transgenesis of duckweed is performed by using A. tumefaciens -mediated transformation. Briefly, Agrobacterium-mediated transformation is carried out by dedifferentiating fully grown duckweed plants or tissues, preferably tissues of

meristematic origin, into calli. Callus induction is carried out by growing duckweed in medium containing plant growth regulators and supplements. Calli can/will re- differentiate into organized nodules. Both nodules or calli can be infected with

Agrobacterium, according to the procedure described in US patents 6,040,498 and 7,161,064 to Stomp et al. Regeneration of plants from infected calli/nodules and concomitant selection for transformants by applying the desired selective pressure results in the isolation of transgenic duckweed lines carrying the exogenous DNA of interest.

[95] Construct for expression of SIgAs, to be used for transformation of duckweed, can be produced by using standard techniques for example, the techniques described in Sambrook & Russell, Molecular Cloning: A Laboratory Manual, 3 ^rd Edition, Cold Spring Harbor Laboratory, NY (2001) and Ausubel et al, Current Protocols in Molecular Biology (Greene Publishing Associates and Wiley Interscience, NY

(1989)).Vectors for transformation of duckweed have been described elsewhere, such as in US patents 6,040,498 and 7,161,064 to Stomp et al.. Preferably, an A. tumefaciens binary vector (generated, for example, by standard cloning in E. coli) is used to first transform A. tumefaciens; the transgenic line obtained can then be employed to transform duckweed. Preferably, such vectors contain multiple resistance genes, to allow for selection in bacteria and in duckweed. Genes for bacterial selection are known in the art. Suitable resistance genes for selection in plants have been described in US patents 6,040,498 and 7,161,064 to Stomp et al., and include gentamycin and kanamycin.

[96] For expression of SIgAs, multiple transformations can be performed with separate vectors including different cassettes coding for the J-chain, the SC-chain, the antibody H chain and L chain. In a preferred embodiment, a single expression vector is used for transformation that contains 4 cassettes each encoding for one of the structural subunit of the slgA (namely, H chain, L chain, SC-chain and J-chain). Construction of expression vectors containing multiple expression cassette for antibody expression have been described in US patent 7,632,983 to Dickey et al. [97] In embodiments, expression of the cassettes is driven by individual promoters, even when the cassettes are located in the same vector for single-step transformation. Examples of suitable promoter can be found in US patents US 4,771,002 to Stanton, US 5,428,147 to Barker et al, US 7,622,573 & US 8,034,916 to Dickey et al, disclosures of which are incorporated herein by references. Most preferably, a different promoter is used for each expression cassette (such as the chimeric A. tumefaciens octopine and mannopine synthase promoter, the L. minor polyubiquitin promoter

(LmUbq), Lemna aequinoctialis polyubiquitin promoter (LaUbq) and Spirodela polyrrhiza polyubiquitin promoter (SpUbq)). In a preferred embodiment, the expression vector includes cassettes coding for all 4 of the SIgA components, i.e. J-chain, SC-chain, H-chain and L-chain. In an even more preferred embodiment, each of the 4 cassettes is driven by a different promoter. In some embodiments, an inducible promoter(s) can be used. See, e.g., US patent 7,632,983 to Dickey et al Examples of such inducible promoters include heat shock gene promoters, cold-induced promoters, drought-inducible gene promoters, pathogen-inducible gene promoters, wound-in due ible gene promoters, and light/dark-inducible gene promoters, promoters from genes induced by abscissic acid, auxins, cytokinins, and gibberellic acid.

[98] In an advantageous embodiment, the vectors used for expression include, 5' of the coding sequence of the expression cassette, a signal peptide sequence placed in frame with the N-terminal portion of the protein of interest. Thus, each of these DNA sequences are "operably linked" to each other. The term "operably linked" means multiple nucleotide sequences that are placed in a functional relationship with each other. Generally, operably-linked DNA sequences are contiguous, and where necessary to join two protein coding regions, in reading frame. Such signal peptide sequence interacts with a receptor protein on the membrane of the endoplasmic reticulum (ER) to direct the translocation of the growing polypeptide chain across the membrane and into the endoplasmic reticulum for secretion. Presence of the signal peptide sequence ensures efficient secretion. This signal peptide is generally cleaved from the precursor

polypeptide to produce a mature polypeptide lacking the signal peptide. Suitable signal peptide include the Arabidopsis thaliana chitinase signal peptide, the Oryza sativa a- amylase signal peptide, or any other suitable duckweed signal peptide sequences, as described in US patent 7,632,983 to Dickey et al. In a most preferred embodiment, the sequence of the signal peptide used in the O. sativa a-amylase signal peptide. In some embodiments of the present invention, the secreted SIgAs are retained within the apoplast (the region between the plasma membrane and the cell wall) and/or associated with the cell wall.

[99] Other suitable nucleotide sequences including enhancers, 5' leader sequences, such as the leader sequence of L. gibba ribulose -bis -phosphate carboxylase small subunit 5B gene, 3' UTR sequences, introns, enhancers, "ACC" and "ACA" trinucleotides to be introduced directly upstream of the translation initiation codon of the nucleotide sequence of interest can be used to improve expression as disclosed in the art and in US patents 6,040,498 and 7,161,064 to Stomp et al as well as 7,622,573 & US 8,034,916 7,632,983 to Dickey et al, disclosures of which are all incorporated by reference herein. [100] The expression from the transgenic lines obtained can also be improved by optimizing the codon distribution of the encoded proteins for expression in duckweed. Duckweed-preferred codons, as used herein, refers to codons that have a frequency of codon usage in duckweed of greater than 17%. Likewise the codons can be optimized for expression in L. minor or L. gibba. In each case the codons have a frequency of codon usage of greater than 17%. Duckweed and Lemna ssp. codon optimization is known in the art and is carried out, e.g. as described in US patent 7,632,983 to Dickey et al.

[101] Another option is to modify the glycosylation profile of the duckweed. It is recognized that antibodies having more than one glycosylation site can have the same glycan species attached to each glycosylation site, or can have different glycan species attached to different glycosylation sites. In this manner, different patterns of glycan attachment yield different glycoforms of a glycoprotein. Monomeric IgAl antibodies have two conserved N-glycosylation sites (per chain): one on the CH2 region and one on the tailpiece. Monomeric IgA2 antibodies have an additional two or three N- glycosylation sites (per chain). Furthermore, the J-chain of dimeric IgA has one conserved N-glycosylation site, and the secretory component of secretory IgA has 7 conserved N-glycosylation sites. The stably -transformed duckweed can express secretory IgA antibodies having various glycan patterns, including for example, aglycosylated slgAs and slgAs having glycan patterns native to plant, mammalian (human), or insect cells. [102] The terms "N-glycan(s)" and "N-linked glycan(s)" are used interchangeably and refer to an N-linked oligosaccharide, e.g., one that is or was attached by an N-acetylglucosamine (GlcNAc) residue linked to the amide nitrogen of an asparagine residue in a protein. The predominant sugars found on glycoproteins are glucose (Glu), galactose (Gal), mannose (Man), fucose (Fuc), N-acetylgalactosamine (GalNAc), N-acetylglucosamine (GlcNAc), and sialic acid (e.g., N-acetyl-neuraminic acid (NeuAc)). The processing of the sugar groups occurs co-translationally in the lumen of the ER and continues in the Golgi apparatus for N-linked glycoproteins.

[103] For the purposes of the present invention, the term "G2 glycan," "G2 glycan species," and "G2 glycan structure" are used interchangeably and refer to an N- linked glycan having the GlcNAc2Man3GlcNAc2Gal2 structure, in which two terminal galactose (Gal) sugar residues are present. For the purposes of the present invention, the term "Gl glycan," "Gl glycan species," and "Gl glycan structure" are used

interchangeably and refer to an N-linked glycan having the GlcNAc2Man3GlcNAc2Gal structure, in which only one terminal galactose (Gal) sugar residue is present. For the purposes of the present invention, the term "GO glycan," "GO glycan species," and "GO glycan structure" are used interchangeably and refer to an N-linked glycan having the GlcNAc2Man3GlcNAc2 structure, in which no terminal galactose (Gal) sugar residues are present.

[104] For the purposes of the present invention, the term "high-mannose glycan," high-mannose glycan species," and "high-mannose glycan structure" are used interchangeably and refer to an N-linked glycan having the GlcNAc2ManX structure, wherein X is a whole number greater than three, such as 5, 6, 7, 8, or 9. For the purposes of the present invention, the term "Man5 glycan," Man5 glycan species," and "Man5 glycan structure" are used interchangeably and refer to an N-linked glycan having the GlcNAc2Man5 structure. The same is applicable for the terms Man6 glycan (species; glycan structure), Man7 glycan (species; glycan structure), Man8 glycan (species; glycan structure), Man 9 glycan (species; glycan structure), etc.

[105] In mammals, naturally-occurring N-glycans contain a fucose (Fuc) residue attached to the GlcNAc2Man3 core structure by an al,6 linkage. In plants, naturally-occurring N-glycans contain a fucose (Fuc) residue attached to the

GlcNAc2Man3 core structure by an al,3 linkage and further contain a xylose (Xyl) residue attached to the GlcNAc2Man3 core structure by a β 1,2 linkage. For the purposes of the present invention, a GO glycan containing the mammalian al,6-linked Fuc residue attached to the GlcNAc2Man3 core structure is referred to as a "G0F<6 > glycan." For the purposes of the present invention, a GO glycan containing the plant al,3-linked Fuc residue attached to the GlcNAc2Man3 core structure is referred to as a "G0F<3 > glycan," a GO glycan containing the plant i,2-linked Xyl residue attached to the GlcNAc2Man3 core structure is referred to herein as a "GOX glycan," and a GO glycan containing each of the plant al,3-linked Fuc residue and the plant i,2-linked Xyl residue attached to the GlcNAc2Man3 core structure is referred to herein as a "G0XF<3> glycan." In an embodiment, the invention relates to a secretory IgA antibody, or a population of secretory IgA antibodies, in which substantially all N-glycans lack Fuc and Xyl residues. [106] The stably-transformed duckweed can also contain a genetic modification that alters the glycan profiles. For example, the N-glycans of the SIgA can be expressed with reduced levels of fucose and xylose residue, preferably less than 10%, more preferably less than 1%. This modification from natural glycan profile can be achieved by several techniques, including knocking out endogenous al,3- fucosyltransferase (FucT) and i,2-xylosyltransferase (XylT), or otherwise inhibiting their transcription, translation into protein or enzymatic activity. In a preferred embodiment, the duckweed is transformed with at least one recombinant nucleotide construct that provides for the inhibition of expression of al,3-fucosyltransferase (FucT) and i,2-xylosyltransferase (XylT) in a plant. In a more preferred embodiment, these constructs triggers RNA interference targeting the mRNAs of al ,3-fucosyltransferase (FucT) and i ,2-xylosyltransferase (XylT). In an even more preferred embodiment, the construct is a RNA hairpin construct. These methods for altering the N-glycosylation pattern of proteins in duckweed are known in the art and are described in US patent 7,884,264 to Dickey et al. The use of the RNA hairpin construct can be advantageous for obtaining a glycan profile where at least 30% of the N-glycans are GO glycans lacking Fuc and Xyl residues and/or where the combination of GO glycans lacking Fuc and Xyl plus high-mannose glycans are at least 70% relative to the total amount of N-glycans in the plurality of secretory IgA antibodies.

[107] Accordingly, the present invention also relates to a method of expressing a plurality of secretory IgA antibodies containing multiple N-glycans, such as two or more different N-glycans, using any of the above-mentioned techniques. Generally, the plurality of secretory IgA antibodies contains at least about 30% GO glycans (preferably GO glycans lacking Fuc and Xyl residues) relative to the total amount of N-glycans in the population. Alternatively, or in addition thereto, the plurality of secretory IgA antibodies can contain at least about 25% high-mannose glycans (e.g., Man5, Man6, Man7, Man8, and/or Man9 glycans) relative to the total amount of N- glycans in the population. Often the GO glycans (preferably GO glycans lacking Fuc and Xyl residues) and high-mannose glycans (e.g., Man5, Man6, Man7, Man8, and/or Man9 glycans) together are the majority of glycans present in the plurality of secretory IgA antibodies, such as at least 70% relative to the total amount of N-glycans in the plurality of secretory IgA antibodies.

[108] The glycoform of an antibody, and the nature of glycan species, can be determined by measuring the glycosylation profile thereof. The term "glycosylation profile" means the characteristic fingerprint of the representative N-glycan species that have been released from an antibody, either enzymatically or chemically, and then analyzed for their carbohydrate structure, for example, using LC-HPLC, or MALDI-TOF MS, and the like. See, for example, the review in Current Analytical Chemistry, Vol. 1, No. 1 (2005), pp. 28-57." For more information on glycosylation of therapeutic antibodies, see, e.g., Fernandes et al., Eur. Biopharm. Rev., Summer 2005, pp. 106-110; Jefferis, Nature Reviews / Drug Discovery, vol. 8, Mar. 2009, pp. 226-234.

[109] The duckweed lines, which produce the SIgAs, can be cryopreserved for indefinite periods of time to ensure maintenance of lines having specific desired characteristic and to protect them from genetic modifications that may occur over time. Methods of cryopreservation of duckweed are known in the art and are described in US patent application US20120190004 to Parson et al.

Reference Example 1: Transient expression of an anti-TNF-a secretory I A antibody in plants.

a) Preparation of cDNA constructs

[110] The amino acid sequences of a suitable leader sequence (e.g., secretion signal, SEQ ID NO: 13), the heavy chain variable region of infliximab (SEQ ID NO: 14), and the human a2(n) IgA heavy chain constant region (Chintalacharuvu et al, Journal of Immunology, 1994, 152, 5299-5304; SEQ ID NO:10) were joined together. Cleavage of the signal sequence (a.a. 1-19) corresponded to the predicted cleavage site using the SignalP program (available on the world wide web at

.cbs.dtu.dk/services/SignalP).

[Ill] The resulting amino acid sequence was back-translated into a cDNA sequence optimized for expression in maize (Zea mays) (SEQ ID NO: 15) (see Liangjiang Wang and Marilyn J. Roossinck., "Comparative analysis of expressed sequences reveals a conserved pattern of optimal codon usage in plants." Plant Mol Biol (2006) 61 :699- 710).

[112] Similarly the cDNA sequence for the light chain of the construct was obtained by joining the sequences of a suitable secretion signal (SEQ ID NO: 16), the light chain variable region of infliximab (SEQ ID NO: 17), and the human κ Ig light chain constant region (SEQ ID NO: l), and back-translating the obtained amino acid sequence into a cDNA sequence optimized for expression in maize (Zea mays) (SEQ ID NO: 18).

[113] The cDNA sequence for the human J-chain was obtained by joining the amino acid sequences of secretion signal (SEQ ID NO: 19) and J-chain sequence (SEQ ID NO: l 1), both obtained from UniprotKB/Swiss-Prot database entry P01591, and back- translating the obtained amino acid sequence into a cDNA sequence optimized for expression in maize (Zea mays) (SEQ ID NO:20).

[114] The cDNA sequence for the SC-chain was obtained by joining the amino acid sequences of a suitable secretion signal (SEQ ID NO:21) and SC-chain sequence (SEQ ID NO:12, amino acids 19-603 of UniprotKB/Swiss-Prot database entry P01833), and back-translating the obtained amino acid sequence into a cDNA sequence optimized for expression in maize (Zea mays) (SEQ ID NO:22). The used secretion signal sequence was derived from the natural SC secretion signal by the addition of codons for two extra amino acids in order to obtain more favorable splicing sites for the construction of plasmid vectors.

[115] The cDNA's of the four constructs (the heavy chain, the light chain, the J-Chain and the SC-Chain) were obtained from a commercial source.

b) Vector construction and cloning strategy

[116] The cDNAs for the heavy chain of the construct (HC) and of the J- chain (JC) were ligated into the pGA15 and pGA14 plasmid vectors, respectively, using Pacl and Ascl restriction sites. After transfer to E. coli K12 XL10 gold and expansion, the constructs were each transferred to separate pRAP plasmids using Ncol and Kpnl restriction sites, resulting in the expression cassettes 35S:HC:Tnos and 35S:JC:Tnos. The pRAP plasmids containing the expression cassettes were transferred to and expanded in E. co/i K12 DH10B.

[117] The cDNAs for the light chain (LC) and of the SC-chain (SC) were ligated into the pGA14 and pGA15 plasmid vectors, respectively, using Pacl and Ascl restriction sites. After transfer to E. coli K12 XL10 gold and expansion, the constructs were each transferred into separate pTR2 plasmids using Ncol and Xbal restriction sites, resulting in the expression cassettes TR1'TR2':LC:T35S and TR1'TR2':SC:T35S. The pTR2 plasmids containing the expression cassettes were transferred to and expanded in E. co/i K12 DH10B.

[118] The light chain expression cassette TRl 'TR2' :LC:T35S was transferred to the pRAP vector containing the heavy chain expression cassette

35S:HC:Tnos using Hindlll restriction sites, and transferred to and expanded in E. coli K12 DH10B. Finally, the combined cassettes containing HC and LC (35S:HC:Tnos : TR1'TR2' :LC:T35S ) were transferred to a pBIN+ expression vector using Ascl and Pad restriction sites. This pBIN+ vector containing the combined HC and LC cassettes was transferred to Agrobacterium tumefaciens strain MOG101 using electroporation.

Similarly the cassettes containing JC and SC (35S:JC:Tnos : TR1'TR2' :SC:T35S) were combined in a pBIN+ expression vector. This pBIN+ vector containing the combined JC and SC cassettes was transferred to Agrobacterium tumefaciens strain MOG101 using electroporation. [119] The two transfected A. tumefaciens strains were used in combination for transient plant transformation and expression of the full SIgA construct. For vector information see also van Engelen et al., Plant Molecular Biology 1994, 26: 1701-1710. Information on A1MV leader sequence: van der Vossen et al,. Nucleic Acids Research 1993, 21: 1361-1367. pBIN+ is described in: van Engelen et al., Transgenic Research 1995, 4: 288-290. (In the referenced literature sources: pRAP35 = pCP031). By way of example, Fig. 5 shows a schematic representation of the building of the pBIN+ vector containing the combined HC and LC cassettes.

c) Transient expression in tobacco plants

[120] The youngest fully expanded leaves of six week old tobacco plants were infiltrated with a mixture of the two A. tumefaciens strains by placing a 2-ml syringe (without needle) containing the bacterial cell suspension at the lower side of a leaf and gently pressing the suspension into the leaf. The infiltrated area was usually clearly visible. Expression took 4-6 days with optimum levels at days 5-6.

[121] The leaves were frozen at -80°C, crushed and 2 ml extraction buffer per gram of leaf material was added. The extraction buffer was PBS pH=7.4; 0.02% Tween-20 (v/v); 2% polyclar AT (v/v); and 1% inhibitor cocktail (v/v). The suspension was homogenized with an Ultra Turrax (Janke & Kunkel). Solid material was removed by centrifugation (15 minutes, lO.OOOg). The solid material was extracted twice more with extraction buffer, the same as above except the inhibitor cocktail was replaced with 10 mM PMSF. The combined extracts were stored at -20°C. Using SDS-page and immunoblotting, formation of a complete SIgA could be shown, but the predominant antibody expressed was the monomeric IgA. The expression level for the IgA antibodies (including all molecular forms) was less than 0.15 mg/ kg of tobacco leaves. The majority of the expressed IgA was monomeric IgA, and slgA was effectively not recoverable. d) Transient expression in lettuce

[122] The SlgA construct was transiently expressed in lettuce by vacuum infiltration using the same vectors as for expression in tobacco (see:Negrouk et al., Highly efficient transient expression of functional recombinant antibodies in lettuce., Plant Science 2005, 169, 433-438). Full grown crops of lettuce Lactuca sativa L. (Oak Leaf lettuce) and vacuum infiltrated with a mixture of the two A. tumefaciens strains and harvested 3-5 days after infiltration. Formation of complete SlgA was not certain, and in any event was expressed at insubstantial or trace levels; i.e., <0.01 mg/ kg of lettuce and recovery was not possible.

Example 2: Stable expression of anti-TNF-a secretory IgA antibody (ADB-SAlg) in Lemna

[123] Anti-TNF-a SlgA was produced in Lemna by stable insertion of

DNA coding for the following proteins: The amino acid sequence of the heavy chain consisted of the rice amylase secretion signal (SEQ ID NO:23) attached to the N-terminal amino acid of the heavy chain variable region of adalimumab (anti-TNF-α IgGl , Humira ®, CAS number 331731-18-1) (SEQ ID NO:24) attached to the N-terminal amino acid of a human IgAl heavy chain constant region (SEQ ID NO:7). The amino acid sequence of the light chain of the Anti-TNF-α SlgA combines the rice amylase secretion signal (SEQ ID NO: 23) with the TNF-a binding light chain variable region of adalimumab (SEQ ID NO:25) and the human κ-light chain constant region (SEQ ID NO:l). The amino acid sequence of the J-chain consisted of the rice amylase secretion signal (SEQ ID NO:23) attached to the N-terminal amino acid of the human J-chain (SEQ ID NO:l 1). The amino acid sequence of the SC-chain consisted of the rice amylase secretion signal (SEQ ID NO:23) attached to the N-terminal amino acid of the human SC-chain (SEQ ID NO: 12). The produced SIgA, having IgAl heavy chain constant region, human κ-light chain constant region, and adalimumab heavy and light chain variable regions, combined with human J-chain and SC-chain is referred to as ADB-SAl . ADB-SAlg specifically refers to material produced in Lemna with modified N-glycosylation (i.e. GO glycosylation lacking fucose and xylose)

[124] Genes were designed for each of the four components with Lemna minor preferred codon usage (63-67% GC content). Tables with suitable preferred codon use in Lemnaceae can be found in PCT application WO2005/035768 and in relevant references contained therein. The synthetic genes also contained the rice a-amylase signal sequence (GenBank M24286; SEQ ID NO:23) fused to the 5' end of their coding sequences. Restriction endonuclease sites were added to allow cloning into A.

tumefaciens binary vectors. Design of DNA sequences and vector construction was performed by Biolex Therapeutics, Inc., Pittsboro, NC, USA. DNA sequences were produced by DNA2.0 (Menlo Park, CA, USA).

[125] The ADB-SAlg antibody was expressed in Lemna minor by transfection via an A. tumefaciens binary vector containing DNA sequences encoding all four of the SIgA components: J-chain (SEQ ID NO:26), SC-chain (SEQ ID NO:27), H- chain (SEQ ID NO:28) and L-chain (SEQ ID NO:29). To prepare this vector, independent expression cassettes were created containing a promoter and also DNA sequences encoding the protein and terminator for the J-chain, SC-chain, H-chain and L- chain. The H chain was fused to the modified chimeric octopine and mannopine synthase promoter with Lemna gibba 5' leader from ribulose bis-phosphatecarboxylase small subunit-1. The L-chain, SC-chain and J-chain genes were fused to high expression Lemnaceae Ubiquitin promoters L. minor polyubiquitin promoter (LmUbq), Lemna aequinoctialis polyubiquitin promoter (LaUbq) and Spirodela polyrrhiza polyubiquitin promoter (SpUbq), respectively. Sequences of these promoters have been disclosed in PCT application WO2007/124186. These expression cassettes were then cloned into a single A. tumefaciens transformation vector EC2.2 (a modification of the A. tumefaciens binary vector pBMSP3, which is a derivative of pBINPLUS. See Ni, M., Cui, D., Einstein, J., Narasimhulu, S., Vergara, C.E., and Gelvin, S.B., Plant J. 7, 661-676, (1995), van Engelen, Transgenic Res. 4:288-290 (1995), and Gasdaska et al.,

Bioprocessing J., 3:50-56 (2003)), with the appropriate restriction sites to create the final transformation vector SynB02. This vector also contained the gentamicin

acetyltransferase-3-I gene (aacCl) which confers resistance to gentamicin and allows for selection of transgenic L. minor lines.

[126] SynB02 was used to create an additional transformation vector to generate a glycan optimized version of anti-TNF-a SIgA, having GO glycosylation lacking fucose and lacking xylose. A chimeric hairpin RNA was used to silence endogenous L. minor mRNAs encoding a-l ,3-fucosyltransferase (Fuctl, GenBank DQ789145) and p-l,2-xylosyltransferase (Xyltl, GenBank DQ789146). A DNA sequence for this chimeric RNAi molecule was fused to the high expression SpUbq promoter and subsequently moved into the SynB02 vector creating the final

transformation vector SynB03. Further details on production of glycan optimized proteins in Lemnaceae can be found in PCT applications WO2007/084672,

WO2007/084922, WO2007/084926 and in Cox, K.M., Nature Biotechnology 2006, 12: 1591-1597.

[127] Lemna minor strain 8627 was transfected with vector SynB03, and glycosylation modified Lemna minor strain XF04 was transfected with vector SynB02. Once transformed plants were regenerated (approximately three months), single plants were harvested from the antibiotic selection plates and propagated separately in liquid growth media, without selection antibiotic, for further screening and characterization. Thus several hundred individual transgenic plant lines from each construct were generated. Independent transgenic lines were harvested and clonally propagated in individual harvest jars. For screening of transgenic lines, clonal lines were

preconditioned for 1 week at light levels of 150 to 200 μηιο1/ηι ² ·8 in vented plant growth vessels containing SH medium (Schenk R.U. et al., Can. J. Biol. 1972, 50: 199-204) without sucrose. Fifteen to twenty preconditioned fronds were then placed into vented containers containing fresh SH medium, and allowed to grow for two weeks. Tissue samples from each line were collected and frozen for analysis.

[128] To determine SIgA expression, frozen tissue samples were homogenized and centrifuged, and the supernatant was removed and screened by an ELIS A method using sheep anti -human IgA secretory chain (AbD Serotec catalog #5111- 4804 - 1:1000 dilution) coated plates to capture the SIgA antibody. The samples were then detected using a goat anti-human kappa light chain HRP conjugated antibody (Sigma catalog #A7164 - 1:2000 dilution). The highest-expressing lines from this primary screening were then grown again for two weeks in small research vessels under the optimal growth conditions. The resulting tissue was harvested and the ELIS A was performed to determine the percent of the total soluble protein that is the expressed SIgA antibody (ADB-SAlg). The results are summarized in Table 2.

Table 2

[129] Results of non-reducing and reducing SDS-PAGE analyses of purified material obtained from Lemna transfected with construct SynB02 are shown in Fig. 6. ADB-SAlg can be isolated from lemna culture and purified as described in Example 3.

Example 3: Isolation and purification of anti-TNF-a secretory IgA antibody ADB- SAlg from Lemna.

[130] Biomass from transgenic Lemna expressing anti TNF-a SIgA, having variable regions that are the amino acid sequence of the variable regions (antigen binding regions) of adalimumab, was homogenized in 50mM Sodium phosphate, 0.3M

Sodium chloride, buffer pH 7.4, at a buffer to tissue ratio of 4:1. An acid precipitation step was performed on the crude extract to remove rubisco and other plant proteins by adjusting the extract to pH 4.5 using 1M Sodium acetate, pH 2.5. The precipitate was removed by centrifugation of the material at 14,000 x g for 30 minutes at 4°C. The supernatant was adjusted to pH 7.4 and loaded on DOWEX (Dowex 1X2 anion exchange resin) to remove colored impurities. The flow-through fraction containing anti TNF-a SIgA was 0.22 μηι was filtered prior to chromatography using affinity chromatography and Size Exclusion Chromatography.

[131] Affinity purification: A KappaSelect (GE Healthcare prod. Nr. 17-

5458-03) column was prepared according to manufacturer instructions (28-9448-22 AA). The column was equilibrated with 5 cv of TBS buffer (50mM Tris, 0.15M Sodium chloride, pH 7.4). The supernatant was loaded on the column. Approximately 5 mg slgA/ml resin was loaded on columns of up to 1L KappaSelect. Non-binding material was washed from the column with 5 cv TBS buffer. The product was eluted from the column using 10 cv of 25 mM Sodium Acetate, pH 6.6 buffer containing 3.5 M MgCl ₂ . The fractions containing the secretory proteins were pooled and immediately diluted 4- fold using TBS buffer (50mM Tris, 0.15M Sodium chloride, pH 7.4). MgCl ₂ was replaced by buffer exchange with at least 10 volumes of TBS buffer using a Pall

Centramate cassette ultrafiltration system and a Pellicon 2 Mini Filter (Millipore prod.Nr. P2C005C01 - PLCCC 5K, regenerated cellulose) with 5kDa cut-off.

[132] SEC purification: Material obtained by KappaSelect

chromatography was further purified on a Sephacryl S300 HR column. The column was equilibrated with 2 cv of PBS buffer at a flow rate of 1.0 ml/min. The KappaSelect eluate was concentrated 3-4x using ultra filtration using a Pellicon 2 Mini Filter (Millipore prod.Nr. P2C005C01 - PLCCC 5K, regenerated cellulose) with 5kDa cut-off. [133] Typically, a concentration of 3-5 mg/ml and feed volume of 50ml was used for a 1L column. The feed was applied using an AKTA purifier at 1.0 ml/min. Elution was performed with at least 2 column volumes of PBS buffer at room

temperature and a flow rate of 1.0 ml/min. Fractions were collected and sufficiently pure fractions were pooled.

Example 4: Stable expression of anti-IL-12/23 secretory IgA Antibody based on ustekinumab (UKB-SA1) in Lemna

a) Construction of vectors

[134] Synthetic genes were designed for each of the 4 different protein chains of an anti-IL-12/23 secretory IgA. The amino acid sequence of the heavy chain consisted of the rice α-amylase secretion signal (SEQ ID NO:23) joined to the N-terminal amino acid of the variable part of the heavy chain of anti-IL12/23 IgGl antibody

Ustekinumab (Stelara ®, CAS number 815610-63-0, SEQ ID NO:30) which in turn is joined to the N-terminal amino acid of the constant part of a human IgAl heavy chain (SEQ ID NO:7). The amino acid sequence of the light chain consisted of the rice a- amylase secretion signal (SEQ ID NO:23) joined to the N-terminal amino acid of the light chain sequence of Ustekinumab (CAS number 815610-63-0), which combines an anti-IL-12/23 binding variable part (SEQ ID NO:31) with a human κ-light chain constant part (SEQ ID NO:l). The SC-chain consisted of the rice α-amylase secretion signal (SEQ ID NO:23) joined to the N-terminal amino acid of the amino acid sequence of amino acids 19 to 603 of the human polymeric immunoglobulin receptor disclosed in

UniProtKB/Swiss-Prot database entry P01833 (SEQ ID NO:12). The J-chain sequence consisted of the rice a-amylase secretion signal (SEQ ID NO:23) joined to the N-terminal amino acid of the amino acid sequence of amino acids 23 to 159 of the human sequence disclosed in UniProtKB/Swiss-Prot database entry P01591 (SEQ ID NO:l 1).

[135] Genes were designed for each of the four components with Lemna minor preferred codon usage (63-67% GC content). Tables with suitable preferred codon use in Lemnaceae can be found in PCT application WO2005/035768 and in relevant references contained therein. Restriction endonuclease sites were added to allow cloning into A. tumefaciens binary vectors. Design of DNA sequences and vector construction was performed by Biolex Therapeutics, Inc., Pittsboro, NC, USA. DNA sequences were produced by DNA2.0 (Menlo Park, CA, USA).

[136] The anti-IL-12/23 SIgA (UKB-SA1) was expressed in Lemna minor by transfection via an Agrobacterium tumefaciens binary vector containing DNA sequences encoding all four of the SIgA components: J-chain, SC-chain, H-chain and L- chain. To prepare this vector independent expression cassettes were created containing promoter, DNA sequences encoding the protein and terminator for the J-chain (SEQ ID NO:26), SC-chain (SEQ ID NO:27), H-chain (SEQ ID NO:32) and L-chain (SEQ ID NO:33). The H chain was fused to the modified chimeric octopine and mannopine synthase promoter with Lemna gibba 5' leader from ribulose bis-phosphatecarboxylase small subunit-1. The L-chain, SC-chain and J-chain genes were fused to high expression Lemnaceae Ubiquitin promoters L. minor polyubiquitin promoter (LmUbq), Lemna aequinoctialis polyubiquitin promoter (LaUbq) and Spirodela polyrhiza polyubiquitin promoter (SpUbq), respectively. Sequences of these promoters have been disclosed in PCT application WO2007/124186. These expression cassettes were then cloned into a single A. tumefaciens transformation vector EC2.2 (a modification of the A. tumefaciens binary vector pBMSP3, which is a derivative of pBINPLUS. See Ni, M., Cui, D.,

Einstein, J., Narasimhulu, S., Vergara, C.E., and Gelvin, S.B.. Plant J. 7, 661-676, (1995), van Engelen Transgenic Res. 4:288-290 (1995), and Gasdaska et al., Bioprocessing J., 3:50-56 (2003)), with the appropriate restriction sites to create the final transformation vector SynAOl (Fig. 7A). This vector also contained the gentamicin acetyltransferase-3-I gene (aacCl) which confers resistance to gentamicin and allows for selection of transgenic L. minor lines, and was used to produce UKB-SA1 with wild-type

(unmodified) N- and O-glycosylation.

[137] SynAOl was used to create additional transformation vectors to generate a glycan optimized version of UKB-SA1, further identified as UKB-SAlgO. UKB-SAlgO has GO glycosylation and lacks fucose and xylose. The term "GO

glycosylation" (also referred to as "GO") is a term of art indicating a particular type of human N-glycosylation defined by the lack of any galactose residues. See, e.g.,

Fernandes et al., "Demonstrating Comparability of Antibody Glycosylation during Biomanufacturing," Eur. Biopharm. Rev., Summer 2005, pp. 106-110; Jefferis,

"Glycosylation as a strategy to improve antibody-based therapeutics," Nature Reviews / Drug Discovery, vol. 8, Mar. 2009, pp. 226-234. A chimeric hairpin RNA was used to silence endogenous L. minor mRN As encoding a-l,3-fucosyltransferase (Fuctl,

GenBank DQ789145) and p-l ,2-xylosyltransferase (Xyltl, GenBank DQ789146). A DNA sequence for this chimeric RNAi molecule was fused to the high expression SpUbq promoter and subsequently moved into the SynAOl vector creating the final transformation vector SynA02 (Fig. 7B). Also the neomycin phosphotransferase II gene (ΝΡΤΠ) was moved into SynAOl replacing aacCl to produce transformation vector SynA03 (Fig. 7C). This exchange allows for kanamycin selection instead of gentamicin selection of transgenic glycan optimized L. minor lines. Further details on procedures for production of glycan optimized proteins in Lemnaceae can be found in PCT applications WO2007/084672, WO2007/084922, WO2007/084926 and in Cox, K.M., Nature Biotechnology 2006, 12: 1591-1597.

b) Transformation of and Expression using Lemna

[138] Lemna transformation vectors SynAOl, SynA02 and SynA03 were transvected into Agrobacterium tumefaciens strain C58Z707 (Hepburn et al., J. Gen. Microbiol. 1985, 131: 2961-2969) by electroporation. Agrobacterium colonies were selected using gentamycin (SynAOl and SynA02) or kanamycin (SynA03) and analyzed for the presence of the appropriate binary vector using a PCR based assay. A single colony was selected for each transformation vector and taken forward into L. minor transformation process (as follows).

[139] Partially dedifferentiated Lemna tissue (Lemna minor strain 8627) was incubated with Agrobacterium harboring the expression cassette plasmid by briefly dipping the tissue into the solution. The tissue was then placed on co-cultivation plates for two days in continuous light at 25°C. Following co-cultivation, the tissue was transferred to antibiotic selection plates and returned to continuous light at 25°C. The tissue was transferred weekly to fresh antibiotic selection plates. Cefotaxime was included in the antibiotic selection plates to eradicate the Agrobacterium. Gentamicin was included in these plates to select for transgenic tissue obtained from vectors SynAOl and SynA02 where there is a selectable marker gene included in the transferred genetic cassette which confers gentamycin resistance. Kanamycin was included in plates to select for transgenic tissue from vector SynA03.

[140] UKB-SA1 with unmodified wild-type (WT) glycosylation was obtained by transfection of Lemna minor strain 8627 (Biolex Therapeutics Inc.) with vector SynAOl. UKB-SAlgO, having GO glycosylation lacking fucose and lacking xylose, was obtained by transfecting Lemna minor strain 8627 with vector SynA02, or by transfection of the N-glycosylation modified Lemna minor strain XF04 (Biolex Inc.) with vector SynA03. Once transformed plants were regenerated (approximately three months) single plants were harvested from the antibiotic selection plates and propagated separately in liquid growth media, without selection antibiotic, for further screening and characterization. Thus several hundred individual transgenic plant lines from each construct were generated. Independent transgenic lines were harvested and clonally propagated in individual harvest jars. For screening of transgenic lines, clonal lines were preconditioned for 1 week at light levels of 150 to 200 μιηο1/ιη2- s in vented plant growth vessels containing SH medium (Schenk R.U. et al., Can. J. Biol. 1972, 50: 199-204) without sucrose. Fifteen to twenty preconditioned fronds were then placed into vented containers containing fresh SH medium, and allowed to grow for two weeks. Tissue samples from each line were collected and frozen for analysis. [141] To determine SIgA expression, frozen tissue samples were homogenized, centrifuged and the supernatant was removed and screened by an ELISA method using sheep an ti -human IgA secretory chain (AbD Serotec catalog #5111-4804 - 1:1000 dilution) coated plates to capture the SIgA antibody. The samples were then detected using a goat anti-human kappa light chain HRP conjugated antibody (Sigma catalog #A7164 - 1 :2000 dilution). The highest lines from this primary screening were then grown again for two weeks in small research vessels under the optimal growth conditions, the resulting tissue was harvested and the ELISA was performed to determine the percent of the total soluble protein that the SIgA antibody is expressed. The results are summarized in Table 3.

Table 3

Example 5: Isolation and purification of UKB-SA1 and UKB-SAlgO secretory IgA antibody from Lemna.

[142] Biomass from transgenic Lemna expressing UKB-SA1 or UKB-

SAlgO, having variable regions that are the amino acid sequence of the variable regions (antigen binding regions) of Ustekinumab, was homogenized in 50mM Sodium phosphate, 0.3M Sodium chloride, buffer pH 7.4, at a buffer to tissue ratio of 4:1. An acid precipitation step was performed on the crude extract to remove the enzyme RuBisCo and other plant proteins by adjusting the extract to pH 4.5 using 1M Sodium acetate, pH 2.5. The precipitate was removed by centrifugation of the material at 14,000 x g for 30 minutes at 4°C. The supernatant was adjusted to pH 7.4 and filtered to 0.22μηι prior to IMAC chromatography.

[143] IMAC purification: A Chelating Sepharose FF (GE Healthcare prod.

Nr. 17-0575-01) column was prepared according to manufacturer instructions (28-4047- 39 AC). The column was charged with 3-5 column volumes (cv) of 0.1M Copper sulfate. Excess copper was washed with 3-5 cv of double distilled water. The column was equilibrated with 3-5 cv of PBS buffer (50mM Sodium phosphate, 0.15M Sodium chloride, pH 7.4); 3-5 cv of 0.1M Sodium acetate, pH 4.0 buffer; 3-5 cv PBS buffer with 0.5M Imidazole; and 3-5 cv PBS buffer.

[144] The supernatant was loaded on the column. Approximately 3.8 mg

SIgA/ml resin was loaded on columns of up to 350 ml Chelating Sepharose. Non-binding material was washed from the column with 10 cv PBS buffer, 10 cv of 0.1M Sodium acetate, pH 4.0 buffer, and 10 cv PBS buffer. The product was eluted from the column using 10 cv of PBS buffer containing 0.075 M Imidazole (a gradient of 0-0.075 M imidazole was also used but did not lead to improved results). The fractions containing the secretory IgA antibodies were pooled.

[145] The column was regenerated by removing copper using a 0.2M

EDTA, 0.3M sodium chloride solution, followed by treatment with 0.1N NaOH.

[146] SEC purification: Material obtained by IMAC chromatography was further purified on a Sephacryl S300 HR column. The column was equilibrated with 2 column volumes of PBS buffer at a flow rate of 0.5 ml/min. The IMAC eluate was concentrated 3x using ultra filtration (e.g., using a 5-or 30 kDa regenerated cellulose (hydrosart) membrane) by using, for example, a spin filter (Vivaspin 15R) or a UF cassette (Sartorius 305 144 59 01 E). Typically, a concentration of 5-7 mg/ml and feed volume of 15-20 ml was used for a 360 ml column. The feed was applied using an AKTA purifier at 0.5 ml/min. Elution was performed with at least 2 column volumes of PBS buffer at room temperature and a flow rate of 0.5 ml/min. Fractions were collected and sufficiently pure fractions were pooled.

[147] Results of non-reducing and reducing SDS-PAGE analyses of purified material obtained from Lemna transfected with construct SynAOl are shown in Fig. 8 as gels A and B, respectively.

[148] Some batches of material produced by the purification method described above were further purified using affinity chromatography using CaptureSelect human IgA (BAC).

[149] Using an Akta Explorer 10 system a column loaded with 30.9 ml

Capture Select IgA (BAC) was equilibrated using 3 column volumes (CV) 20mM Tris pH 7.0 buffer at a flow rate of 5 ml/min. A solution containing 75.9 mg UKB-SAl in Tris buffer pH 7.0 was loaded on the column at a flow rate of 2.5 ml/min. After a wash step with 5 CV of 20 mM Tris buffer pH 7.0 at 5 ml/min the product was eluted with 5 CV of 20 mM Tris buffer pH 7.0 containing 3.5 M MgCl ₂ at 5 ml min.

[150] The eluate was first dialyzed twice against 5 L 20 mM Tris buffer pH 7.0 using snakeskin lOkDa dialysis tubing for at least 2 hours, followed by dialysis against 5 L PBS puffer pH 7.4 for at least 2 hours. The solution was concentrated using a stirred cell (Amicon 8200, Millipore, overhead pressure 30 Psi) with 30 kDa regenerated cellulose membrane filter (Millipore) to a final concentration of approximately 1 mg/ml. The obtained product was further purified using SEC purification as described previously. Glycosylation modified product UKB-SAlgO was purified using the same method.

[151] No substantial differences were observed in purification of the two glycosylation forms, wild-type UKB-SA1 and GO N-glycosylated UKB-SAlgO.

Example 6: Binding of UKB-SA1 and UKB-SAlgO to IL12.

[152] The binding of purified anti-IL- 12/23 SIgA, with variable regions taken from Ustekinumab and produced in Lemna as in Examples 11 and 12, to IL-12 was determined. The binding of both wild type glycosylation (UKB-SA1) and G0- glycosylation (UKB-SAlgO) products were determined in comparison to Ustekinumab (Stelara®) and colostral SIgA. Plates were coated with IL-12 (Abeam, AB52086) 1 μg/ml. Detection of bound UKB-SA1 / UKB-SAlgO (secretory IgA) and Ustekinumab (IgGl) antibodies was performed using a 1:1500 fold dilution of anti human kappa chain antibody (Abbiotec, cat. no.250987), ΙΟΟμΙ per well, for one hour at RT, washing 3 times, 30 seconds with 200 μΐ PBS/0.05% Tween with shaking, followed by incubation with a 1:1500 fold dilution of donkey anti mouse HRP conjugated (Emelca biosciences, MS3001), ΙΟΟμΙ per well, for one hour at RT.

[153] The UKB-SA1, UKB-SAlgO and Ustekinumab antibodies all bound with high affinity to IL-12 under the conditions of this assay. For UKB-SA1 and UKB- SAlgO antibodies, binding occurred independent of the type of glycosylation. Colostral

SIgA had minimal to no binding to IL-12.

Example 7: Proteolytic stability of UKB-SA1 and UKB-SAlgO

[154] The stability of UKB-S Al and UKB-SAlgOwith antigen binding regions having the amino acid sequence of Ustekinumab antigen binding regions (i.e., the variable heavy and light chains), produced in Lemna, was determined in simulated intestinal fluid (SIF, 0.05M phosphate buffer pH 6.8 containing 10 mg/ml pancreatin). Both the form with wild-type Lemna glycosylation obtained with vector SynAOl, and the GO glycosylation variant obtained with vector SynA02 were analysed. Stability was compared to Ustekinumab (IgGl) and to aspecific human colostral SIgA which was purified to contain only kappa light chains and a-1 heavy chains.

[155] 90 μΐ of a 1 mg/ml solution of the material to be tested was added to

810 μΐ of SIF at 37°C. 50 μΐ samples were drawn at T= 0, 5, 15, 30, 60 and 120 min and immediately frozen in liquid nitrogen. Samples were analyzed by non-reducing SDS- Page gel electrophoresis. Briefly; 17 μΐ of a 0.15M solution of iodoacetamide in LDS sample buffer was added to each of the frozen samples. Samples were thawed and applied to Criterion Tris-HCl gel (12.5%, 18 well, 30 μΐ comb (Biorad, 345-0015). After electrophoresis gels were treated with Krypton protein stain and analyzed. Stability of the samples was qualitatively assessed visually. Results are shown in Fig. 9 where gel A is the WT glycosylation (UKB-SA1, labeled SynAOl -WT in Fig. 9) and gel B is GO glycosylation (UKB-SAlgO, labeled SynA02-G0 in Fig.9). [156] Both glycosylation forms (WT and GO) of Lemna produced anti-

IL12/23 SIgA exhibited a stability that was comparable to natural human colostral SIgA. The IgGl antibody Ustekinumab degraded under these conditions at such a high rate that detection at T=0 was not possible.

[157] Each of the patents, patent applications, and journal articles mentioned above are incorporated herein by reference in their entirety. The invention having been described it will be obvious that the same may be varied in many ways and all such modifications are contemplated as being within the scope of the invention as defined by the claims appended hereto.