Marsupial Light Chains: Complexity and Conservation of λ in the Opossum Monodelphis domestica^1 2

Two identical heavy chains paired with two identical light chains is the generic structure of vertebrate Ig molecules. In mammals, the Ig heavy chains (IgH)⁴ are encoded at a single site in the genome, but the two types of Ig light chains (IgL), κ and λ, are encoded at separate, unlinked loci. The use of the two IgL types can vary between species, some having a bias or preferential use of one over the other (reviewed in 1 . Mice and rabbits, for example, use predominantly Igκ, whereas horses, sheep, and cattle use primarily Igλ (1, 2, 3, 4, 5, 6). The use of one IgL type over another correlates, in general, with the overall complexity of the loci in most species. Humans, for example, have a significant amount of Vλ and Vκ diversity and use both extensively, 60% Igκ:40% Igλ (2, 7, 8, 9). Mice, on the other hand, have only three functional Vλ segments but a large number of available Vκ and have a 95% Igκ:5% Igλ ratio (7, 9). The contributions that IgL make to Ab diversity can also vary greatly between species. Humans appear to have a significant amount of light chain diversity (7, 10). In contrast, the λ repertoire of cattle is restricted to a recurrent Vλ-Jλ rearrangement, even though they appear to have multiple functional Vλ and Jλ segments in their germline (5). Perhaps the most extreme case of limited contribution by light chains occurs in the camelids (camels and llamas), which produce a form of IgG lacking light chains entirely (11).

Our knowledge of the structure, diversity, and evolution of the mammalian IgL genes is based on studies of only one of the three major orders of mammals, the eutherians or “placental” mammals. To date there has been no reported IgL gene structure from either of the other two mammalian orders, the prototherians (egg laying monotremes, e.g., the platypus) or the metatherians (marsupials). The relationship of these three mammalian lineages has been a subject of continued debate over much of this century with most investigators placing the metatherians and the eutherians together as sister taxa, with the prototherians diverging earliest (12). However, more recent analysis of mitochondrial DNA supports the idea that prototherians and metatherians are sister taxa, with the eutherians splitting off first (12, 13). Possible times for the divergence of these groups range from less than 120 million years ago, during the Cretaceous Period, to possibly greater than 170 million years ago, during the Jurassic Period (14, 15). A more extensive analysis of metatherian and prototherian immunobiology provides a comparison between very distantly related mammalian species and should yield important knowledge into the evolution of mammalian immune systems.

In addition to their importance to mammalian evolution, marsupials also provide an opportunity to study mammals that are born comparatively less developed than mice or humans. Developmental immaturity combined with the lack of a placenta, which supports the transfer of maternal Ig, in most marsupial species creates unique immunological problems for metatherians (16). The opossum, Monodelphis domestica, has been established over the last decade as an important laboratory-bred marsupial for studies of many areas of comparative and biomedical research (17, 18). M. domestica are native to South America and are a member of the family Didelphidae, which contains the largest number of species within the marsupials, and Monodelphis is the most species-rich genus of the family (19). The Didelphidae are also thought to have diverged earliest from the rest of the metatheria and may contain some of the oldest extant mammalian species (20, 21).

We have begun characterizing the Ig genes of M. domestica, and we previously reported that the IgH repertoire was derived from two related group III type V_H families (22). To extend this analysis to opossum IgL, we have cloned and characterized Igλ-containing cDNAs and have found the presence of at least three highly divergent Vλ families, the absence of bias in the Vλ-Jλ combinations, and evidence that the duplicated Jλ-Cλ pair arrangements found in placental mammals is conserved in the opossum. It appears that the genetic complexity of the M. domestica Igλ locus is greater than that for the IgH locus, suggesting that λ light chains contribute significantly to the diversity of the Ig repertoire in this species.

A degenerate oligonucleotide (5′-CCNGGYTTYTGYTGRTACCA) complementary to the coding strand for the amino acid sequence WYQQKPG conserved in framework region 2 (FR2) of light chain V segments (also see 23 was used to amplify opossum V_L fragments by anchored PCR. The target for PCR was a commercially available M. domestica spleen cDNA library constructed using the λZAPII cloning vector (Stratagene, La Jolla, CA). The degenerate FR2 oligonucleotide was used in PCR as a reverse primer in combination with the T3 universal sequencing primer specific for a site flanking the cloning site in λZAPII. Successful amplification was achieved using 2 mM MgCl₂ and 55°C annealing temperature and Taq polymerase (Perkin-Elmer, Foster City, CA). For this study, all PCR products were cloned for sequencing or for use as probes using the pCR2.1 vector (Invitrogen, Carlsbad, CA) following the manufacturer’s recommended protocol. An oligonucleotide primer complementary to the 5′ region of M. domestica Cλ (5′-ACCATAGGCCATGACCATGG) was paired in PCR with the T3 primer to amplify Vλ region segments in an unbiased manner. The spleen cDNA library described above was used as target with the conditions of 3.0 mM MgCl₂ and 55°C annealing temperature.

In experiments to confirm the germline Jλ-Cλ pair arrangement, oligonucleotides for each known M. domestica Jλ segment (JλI, 5′-GTGTTCGGCAGTGGGACCAG; JλII, 5′-GTGTTCGGTGGTGGGACCAA; JλIII, 5′-GTGTTCGGTGCTGGGACCAA; JλIV, 5′-GTGTTCGGCCGTGGGACCAG; JλV, 5′-GTGTTTGGCGGTGGGACCAA; JλVI, 5′-GTGTTCGGCGGTGGGACCAG) were paired with the Cλ primer described above to amplify genomic fragments. Amplifications were performed using PCR with 2 mM MgCl₂ and a 60°C annealing temperature.

All genomic M. domestica DNA used were extracted from spleen tissue using standard protocols. For Southern blot analysis, genomic DNA were cut with various restriction endonucleases following the manufacturer’s recommended conditions (see figure legends). Digested DNA were electrophoresed through 1% agarose (FMC Bioproducts, Rockland, ME) and transferred to reinforced nitrocellulose for probing (Micron Separations, Westborough, MA). Phage plaque lifts for cDNA library screening were also made using reinforced nitrocellulose. All probes used in this study were prepared as DNA inserts excised from plasmids and labeled with [³²P]dCTP by the random primer method (Prime-it Kit, Stratagene). Hybridizations were done at 42°C in 50% formamide, 5× Denhardt’s solution, 5× SSC, 50 mM NaPO₄ (pH 6.5), 0.1% SDS, 5 mM EDTA, and 250 mg/ml sheared salmon sperm DNA. Final wash conditions were 65°C and 0.2× SSC.

DNA sequencing reactions were performed using the ThermoSequenase sequencing kit (Amersham, Arlington Heights, IL), and the reactions were analyzed using an automated DNA sequencer (Perkin-Elmer ABI Prism 377 DNA sequencer). All DNA sequences reported were derived by completely sequencing both strands of each clone. Sequences were analyzed using the Sequencher 3.0 program (Gene Codes, Ann Arbor, MI), and alignments were constructed using the CLUSTAL W program (24). All phylogenetic trees shown are reconstructed from nucleotide alignments. To align the nucleotide sequences, first the amino acid translations were aligned using CLUSTAL W with minor manual corrections, then nucleotide sequences were aligned and gapped manually based on the protein alignments to retain codon positions. Based on these nucleotide alignments, trees were reconstructed using the neighbor-joining method of Saitou and Nei (25).

To isolate clones containing opossum Igλ sequences, fragments of Vλ segments were first amplified from a spleen cDNA by anchored PCR using a FR2-specific degenerate primer and then, cloned and sequenced. Four unique clones were found to be homologous to the leader, FR1, and complementarity determining region 1 (CDR1) of known mammalian Vλ segments (not shown). The clones varied from 175 to 184 nucleotides in length and shared from 52% to 74% nucleotide similarity to mammalian Vλ, but less than 40% similarity to any mammalian Vκ sequences (not shown). Two of the PCR-generated clones were from different Vλ families, based on less than 50% similarity, and were used independently to screen a cDNA phage library constructed from M. domestica spleen RNA. Three clones were identified using each probe, and all six clones were found to contain full length λ light chain cDNAs containing variable and constant regions (Fig. 1).

The six full length cDNA sequences shown in Fig. 1 are grouped by nucleotide similarity in the V region. The presence of at least two Vλ families, which have been designated opossum Vλ1 (clones 2c, 3c, and 4c) and Vλ2 (clones 7c, 10c, and 12c), is apparent in the opossum Igλ repertoire. The separation of these sequences into two Vλ families is based on a typically >87% similarity among sequences in the same family and <56% similarity between the families.

To rapidly screen for the presence of additional Vλ families, an oligonucleotide primer complementary to coding sequence near the 5′ end of the Cλ region was paired with a primer specific for the T3 promoter sequence flanking the cloning site in the phage vector used to construct the cDNA library. This approach amplifies V domain sequences, using the spleen cDNA library as target, without bias for Vλ or Jλ sequences. The sequence of the Cλ primer was complementary to nucleotides 422–441 in the Cλ region shown in Fig. 1, which is a sequence common to all 6 Cλ regions found so far. A total of 40 unique Vλ-Jλ rearrangements were amplified from the cDNA library and then, cloned and sequenced. Of these new sequences, the majority (36 total) clearly grouped with the Vλ1 family, while 2 grouped with the known Vλ2 family (sequences 46p and 62p in Fig. 2). The remaining 2 clones (sequences 18p and 25p in Fig. 2) shared 97% nucleotide similarity to each other, but <65% similarity to any Vλ1 or Vλ2 sequences, and defined a third Vλ family, opossum Vλ3. One clone (51p in Fig. 2) was grouped as a Vλ1 member but clearly contains a FR3 from the Vλ2 family. Whether this clone contains a bona fide germline V segment that may have undergone gene conversion or recombination, or is an artifact of template jumping during PCR, remains to be determined.

An unusual feature that distinguishes the three families is the presence of consistently shorter CDR1 and CDR2 regions in the Vλ2 and Vλ3 families when compared with the Vλ1 members. The Vλ2 and Vλ3 segments are one codon shorter than Vλ1 in CDR1 and four and three codons shorter in CDR2, respectively. In addition, the CDR3 regions created by the Vλ-Jλ junction are also consistently shorter in those clones that contain rearrangements involving Vλ2 and Vλ3 family members. The length of the CDR3 does not appear to correlate with a bias in Vλ-Jλ combinations. Based on FR4 sequences, we estimate there to be at least six functional Jλ segments in the M. domestica Igλ locus (indicated by the Roman numerals next to the FR4 sequences in Fig. 2). All six Jλ segments can be found in rearrangements that contain a Vλ1 and long CDR3 regions, while four of six Jλ segments can be found in rearrangements that contain a Vλ2 or Vλ3 with comparatively shorter CDR3 regions. In summary, there appears to be no relationship between the combination of particular Jλ with specific Vλ segments, and the length of the CDR3 region does not associate with particular Jλ segments.

To estimate the number of Vλ gene segments present in the M. domestica genome, Southern blot hybridizations were performed using representative clones from each of the three families as probes (Fig. 3). A Vλ1 probe hybridized to an average of 20 restriction fragments in the M. domestica genome (Fig. 3,A). This same blot was stripped and rehybridized with probes specific for the Vλ2 (Fig. 3,B) and Vλ3 (Fig. 3 C) families, which revealed 8 and 4 genomic fragments, respectively.

An alignment of the nucleotide (Fig. 2) or amino acid sequence encoded (Fig. 4) by the six full-length cDNAs revealed three distinct pairs of sequences based on FR4 and C regions. Unlike the order of the clones presented in Fig. 2, the six sequences in Fig. 4 are grouped based on similar FR4 regions (amino acid positions 105–116) to illustrate the paired relationships. The FR4 regions of cDNA clones 2c and 7c are nearly identical, and there is significant similarity in the FR4 regions of clones 3c and 12c, as well as 4c and 10c. Comparison of the six Cλ sequences reveals identical paired patterns of similarity; in other words, cDNA clones with similar FR4 sequences share similar Cλ sequences. The most likely explanation for this pattern is the presence of multiple functional Jλ segments, each with its own Cλ downstream.

To confirm the presence of multiple Jλ-Cλ pairs in the opossum genome, primers were designed to be unique for each of the six known FR4 regions (Jλ) and paired with the Cλ primer for PCR using genomic DNA as a target. PCR amplification with each FR4 primer paired with the Cλ primer yielded products ∼1.8 kb long, which were cloned and partially sequenced. Sequences internal to the primers confirmed that the amplified fragments contained an intron with predicted splice sites flanked by Jλ and Cλ segments, and each clone had a unique restriction map (Fig. 5). These results confirm the presence of multiple functional Jλ-Cλ pairs in the opossum genome. A Southern blot of M. domestica genomic DNA probed for Cλ revealed typically six to eight fragments (Fig. 6) consistent with the estimate of at least six unique Jλ segments based on FR4 sequences and the presence of Jλ-Cλ pairs.

Pairwise comparisons of the opossum sequences with Vλ sequences from placental mammals revealed greater similarity between the opossum Vλ families and Vλ sequences from other species than that found between opossum Vλ families. To illustrate these relationships, a phylogenetic tree was constructed using opossum Vλ aligned to representative Vλ sequences from other mammals. Fig. 7 A shows a tree based on nucleotide alignment of the FR regions of the 3 opossum Vλ families and the 10 human Vλ families. Also included in the alignment were sequences from mice, rabbits, and two artiodactyl species, cattle and sheep. There was no difference in the tree topology when CDR sequences were included in the alignments (not shown). The overall topology of the tree, or relationship among the mammalian Vλ sequences, is in general agreement with that reported in more extensive analysis of vertebrate V_L sequences (26, 27). Mouse Vλ2 was excluded from the alignment because it is highly similar to mouse Vλ1. The rabbit and artiodactyl sequences cluster on their own branches, whereas the 2 mouse sequences and 10 human Vλ families are more dispersed around the tree. The opossum Vλ families also intersperse with the sequences from mice and humans. This result suggests that the gene duplication events that produced these families predate the evolutionary separation of mammals.

Phylogenetic trees constructed from nucleotide alignments of all six Cλ sequences from the cDNA clones in Fig. 1 with Cλ regions from other species revealed a strikingly different pattern of evolution at the C end of the opossum Igλ compared with the V end (Fig. 7 B). In the case of Cλ, the duplication events appear to have occurred after speciation. The Cλ regions of the opossum all cluster at the end of a long branch and, likewise, the duplicated Cλ regions of mouse, human, and rabbit all cluster on their own branches. Two avian Cλ sequences were included for comparison. Mouse Cλ4 was excluded because it is a pseudogene and highly similar to Cλ1. Several relationships support the validity of this tree, including the intraspecies clustering of Cλ sequences previously noted (see 26 , and the common branch that the mouse surrogate light chain Cλ5 shares with Cλ1; Cλ5 is thought to be derived from Cλ1 (28).

The gray, short-tailed opossum, M. domestica, has been an important model for studies of marsupial immunobiology (30), but much work remains to develop the reagents needed to study marsupial immune systems at the level of sophistication achieved for mice and humans. Toward improving our knowledge of immunogenetics in this species, and to gain insight into the evolution of mammalian Ag receptors, we have been characterizing the opossum homologues of immunologically relevant genes including IgH, the recombination activating gene-1, and terminal deoxynucleotidyl transferase (22, 31, 32). We present here the first molecular characterization of a marsupial IgL. The λ light chain repertoire of the opossum is derived from at least 3 ancient V families, which total ∼30 gene segments. These V segments appear to randomly recombine with available J segments, giving a potential combinatorial diversity for opossum λ comparable to that described in humans. The first important conclusion from our results is that λ has been retained in the metatherian lineage. This is not unexpected given that both λ- and κ-like sequences have been described in all vertebrate groups, including sharks (23, 33, 34, 35, 36, 37, 38, 39).

Phylogenetic analysis of Vλ and Vκ sequences from several vertebrates revealed the presence of multiple Vλ groups, but only a single Vκ cluster, hence Vλ has been referred to as being “polyphyletic” compared with the Vκ (26, 27). In addition, phylogenetic analysis of Vλ sequences by Hayzer (26), Sitnikova and Su (27), and Zezza et al. (36) all generally agree that human and mouse Vλ families intersperse with genes from other vertebrates, while sequences from other species generally remain clustered with their phylogenetic origin (i.e., all rabbit Vλ clustered within a single group, all avian Vλ clustered within a single group, etc.). Reconstruction of a phylogenetic tree that includes the opossum Vλ sequences, shown here, reveals the interspersion of marsupial and placental mammal sequences. Although convergent evolution of metatherian and eutherian Vλ gene segments could account for this interspersion, the most likely explanation is the separation of the three Vλ lineages before the divergence of metatherians and eutherians, which probably occurred more than 100 million years ago and may have been as long ago as 175 million years (14, 15).

Mammalian V_H sequences do not show a similar evolutionary interspersion between marsupials and placental mammals. There are two V_H families in M. domestica, and both cluster on the same branch within the mammalian group III lineage (22). V_H sequences from two other marsupial species, one a complete sequence from the North American opossum Didelphis virginiana, the other a partial sequence from the Australian brushtail possum Trichosurus vulpecula, also cluster with the M. domestica V_H sequences on a common marsupial branch (Ref. 22 and our unpublished observations). Opossum Vλ gene segments, in contrast, retained a wider germline diversity, perhaps to compensate for less diversity in the heavy chain.

The gene duplication event that separated the mouse JCλ1-JCλ3 pair from the JCλ2-JCλ4 pair was reported to be very old, on the order of 240 million years ago, based on nonsynonomous substitution rates (40). Our analysis of mammalian Cλ also supports gene duplications in mice that are more ancient than those found in most mammals, as indicated by the long branch lengths for mouse Cλ in Fig. 7 B. However, these duplications, like those in all mammals, not only occurred after the separation of metatherians and eutherians, probably much less than 200 million years ago, but occurred after the separation of the species themselves. The mammalian Igλ and Igκ loci have followed distinct patterns of evolution in their gene organization. The Igλ loci, in general, contain duplicated J-C units, whereas the Igκ loci have a single C segment downstream from duplicated J segments (7, 8, 9). This pattern of multiple tandem J-C duplications in the λ locus in placental mammals is clearly conserved in the opossum and, therefore, conserved across mammalian orders that may be separated by as many as 175 million years. In contrast, the avian Igλ, represented by chickens and ducks, contains only a single Jλ and Cλ region (37, 38, 39), and the λ-like genes in cartilaginous and boney fishes are organized in duplicated units of [V_L-J_L-C_L] (33, 34, 35). A light chain related to mammalian λ has been identified in an amphibian and found to contain more than one of each J_L and C_L segment, although the organization of these genes has not been reported (23). It is interesting that while the tendency to undergo J-C duplications in λ is conserved across mammalian orders, the duplications themselves appear species specific and not conserved. In other words, the mammalian λ locus appears to consistently evolve by duplicating the J and C segments as a unit, although the duplications present in modern mammals likely occurred after the separation of the species. The presence of paralogous Jλ-Cλ pairs within a species without orthologous relationships between species was reported by Hayzer (26) in a more extensive analyis of eutherian Cλ sequences. It is curious as to why the λ locus in mammals would continue to independently evolve as (V)_n-(J-C)_n, while parallel evolution in the κ locus proceeded as (V)_n-(J)_n-C. We have recently identified variable and constant region sequences from the opossum that are clearly the homologues of Igκ (G.H.R. and R.D.M., unpublished observations), but the complexity and organization of κ in the opossum remains to be determined. It will be interesting in the future to compare how the κ locus has evolved in metatherians as well.

The preferential use of one light chain isotype over another, as seen in many mammals, appears to correlate with the overall complexity, or number, of available V_L segments, although sheep and horses may indicate that there are exceptions (3, 4). Humans have similar numbers of available Vλ and Vκ segments and use both light chain types nearly equally (60:40, κ:λ). Mice have a strong bias for Igκ and nearly 50-fold more V segments in their Igκ locus than Igλ (7, 9). Conversely, sheep have 10-fold more V segments in their Igλ locus than Igκ, and a 20:1 bias for Igλ expression (1, 6). The opossum, M. domestica, has ∼30 Vλ segments that are divided among 3 evolutionarily diverse families. We would expect, although we have not yet shown, that λ should contribute significantly to the expressed Ig diversity in this marsupial. When the expressed Vλ repertoire was sampled, Vλ segments from the Vλ1 family far outnumbered the other 2 families in V-J rearrangements cloned. Although we cannot rule out the possibility that this may reflect some bias in V-J recombination or selection for B cells expressing Vλ1, it is consistent with and easily explained by the number of Vλ segments in each family. Based on Southern blot analysis, Vλ1 appears to have twice as many segments as Vλ2 and five times as many as Vλ3. While it remains to be determined what percentage of the germline V segments in each family are functional, the frequency at which a Vλ segment is expressed likely reflects its representation in the genome, rather than a bias or preferential use.

A curious aspect of the structure of the opossum Vλ domains is the coincident length variation of all three CDRs. Members of the Vλ2 and Vλ3 families encode shorter CDR1 and CDR2 regions, or conversely, Vλ1 members encode longer CDR1 and CDR2. The V-J rearrangements generated using Vλ2 and Vλ3 segments contain shorter CDR3 regions as well. Vλ family-specific CDR length is also apparent in the alignment of human Vλ (29). Three of the human Vλ families (Vλ4, -5, and -9) have significantly longer CDR2 regions. In the opossum, the longer CDR3 found in rearrangements using Vλ1 does not correlate with FR4 sequence. Furthermore, in the seven rearrangements that contain a Vλ2 or Vλ3 isolated so far, four of the six putative J segments are present. These results support a lack of bias in the Vλ-Jλ recombinations and suggest that it is not the choice of J segment that creates the length variation of the CDR3 depending on whether Vλ1 vs Vλ2 or Vλ3 are being rearranged. The alternative explanation is that the length of the regions in the germline Vλ2 or Vλ3, which contribute to the CDR3, are shorter. We are presently cloning the germline Vλ segments to see if Vλ2 and Vλ3 members contain a shorter CDR3. It is also possible that during V-J recombinations involving a Vλ2 or Vλ3 there is additional nucleotide trimming at the junction to create shorter CDR3s or, conversely, more N region additions made by terminal deoxynucleotidyl transferase when a Vλ1 member is recombined. Shorter CDRs translate into shorter Ag binding loops in those Abs that contain a Vλ2 or Vλ3. Lack of N region additions and shorter CDR3 regions have been shown to increase Ag receptor cross-reactivity or Ag promiscuity in Ig and TCR (41, 42). It is possible that in the opossum Ig repertoire the Abs that contain a Vλ2 or Vλ3 have a broader specificity, although this remains to be experimentally determined.

PCR amplification with primers specific for opossum Jλ and Cλ segments from genomic DNA produced six unique J-C introns. Attempts to produce completely inbred lines of M. domestica have been unsuccessful to date, and we cannot presently determine whether or not the opossums we are using are homozygous or heterozygous at the Igλ locus (43). Therefore, given that we were able to clone at least six unique J-C introns, there may be as few as three functional Igλ(J-C) pairs in the opossum genome. Nonetheless, this provides the opossum with multiple functional Jλ segments to use in V-J recombination. Interestingly, in cattle, although multiple J-C pairs exist, the λ light chain repertoire appears to be dominated by a single V-J recombination (5). As pointed out earlier, there was no apparent bias in the Vλ-Jλ recombinations in the opossum spleen cDNA library.

In summary, the λ repertoire of the opossum is more heterogeneous than that of many placental mammals, such as the artiodactyls, and contains more available germline segments than rodents. The kinds of gene duplications that have occurred at the λ locus in placental mammals have also occurred independently in the marsupial lineage. Given the apparent combination of any V with any J, we would predict that the overall organization of the Igλ locus is probably similar to that found in humans with an array of V segments upstream of an array of J-C pairs.