William Parker, Department of Surgery, Duke University Medical Center, Box 2605, Durham, NC 27710, USA. Tel.: 919 681 3886; fax: 919 681 7263; e-mail: email@example.com
A recently improved understanding of gut immunity has merged with current thinking in biological and medical science, pointing to an apparent function of the mammalian cecal appendix as a safe-house for symbiotic gut microbes, preserving the flora during times of gastrointestinal infection in societies without modern medicine. This function is potentially a selective force for the evolution and maintenance of the appendix, and provides an impetus for reassessment of the evolution of the appendix. A comparative anatomical approach reveals three apparent morphotypes of the cecal appendix, as well as appendix-like structures in some species that lack a true cecal appendix. Cladistic analyses indicate that the appendix has evolved independently at least twice (at least once in diprotodont marsupials and at least once in Euarchontoglires), shows a highly significant (P < 0.0001) phylogenetic signal in its distribution, and has been maintained in mammalian evolution for 80 million years or longer.
The appendix in humans is a narrow extension from the terminal end of the cecum, with dimensions of about 10 cm by 7–8 mm, and has an internal (luminal) diameter of 1–3 mm. The appendix has been considered, variably, an evolutionary vestige and a synapomorphy uniting all hominoids (apes and humans). The lack of a known function of the appendix has long been associated with the thought that the human appendix is an evolutionary remnant of a cecum that is utilized to ferment plant material. Charles Darwin, for example, noted the apparent lack of a function of the appendix in humans, and concluded that it must be an evolutionary remnant from a primate ancestor that ate leaves (Darwin, 1871). In fact, numerous researchers have argued that the appendix lacks a particular function in humans (Huntington, 1903; Johnston, 1919; Royster, 1927). On the other hand, data from phylogenetic and immunological studies have long suggested that the appendix may have some specific yet unknown function for which it is adapted (e.g. Berry, 1900; Keith, 1912; Neiburger et al., 1976; Gorgollon, 1978; Scott, 1980; Spencer et al., 1985; Zahid, 2004).
Because the appendix in humans is associated with substantial amounts of lymphoid tissue, termed gut-associated lymphoid tissue (GALT), it was proposed more than a century ago that the appendix has some sort of immune function (Berry, 1900). Further, studies have consistently shown that, in animals lacking an appendix, the terminal part of the cecum is rich in lymphoid tissue (Berry, 1900; Malla, 2003). This observation is true across a wide range of species, including birds (Berry, 1900) such as Columba livia (pigeon), Bernicla jubata (maned goose), Polioaetus plumheus (plumbeous fish eagle), and Leptoptilos javanicus (lesser adjutant), and carnivores such as Felis domesticus (cat; Berry, 1900; Malla, 2003) and Canis lupus familiaris (dog; Malla, 2003). Thus, it is thought that the immune function carried out by the appendix is analogous to the immune function carried out by the terminal portion of the cecum in animals lacking an appendix. However, an apparent function of the human appendix in the immune-mediated maintenance of the commensal bacterial flora of the gut was only recently identified (Bollinger et al., 2007). Such a delay was understandable: It was not until 2003 that it was realized that the immune system apparently supports growth of beneficial (symbiotic) bacteria in the mammalian gut in the form of microbial communities called biofilms (Bollinger et al., 2003; Everett et al., 2004; Sonnenburg et al., 2004). This conclusion was based on several lines of evidence (Everett et al., 2004; Sonnenburg et al., 2004), including the relatively recent observations that immunoglobulin A (IgA) and mucin, two of the most abundantly produced molecules by the immune system, both support growth of microbial biofilms in laboratory experiments and are associated with microbial biofilms in the mammalian gut (Bollinger et al., 2003; Sonnenburg et al., 2004). Finally, it was found that microbial biofilms are more abundant in the human appendix compared with other areas of the human colon (Bollinger et al., 2007). The implications of these new findings were then considered in light of an array of previously known observations, including the following: (a) the architecture of the human gut, including the relative isolation of the appendix from the main flow of the digestive tract and the narrow lumen (internal opening) of the appendix, (b) the association of the appendix with immune tissue, (c) the association of appendicitis with modern medicine and hygiene, (d) the enormous biological impact of diarrhoeal illness in the absence of modern medicine, sewer treatment systems and clean drinking water, (e) the observation that microbial biofilms are constantly in a state of growth and shedding, (f) the importance of biofilms to bacterial colonization in virtually all environments and (g) the vital role of symbiotic gut bacteria in mammalian biology. In light of these and other considerations, it was concluded (Bollinger et al., 2007) that an apparently important function of the human appendix is service as a well-adapted centre or ‘safe-house’ for maintenance of the symbiotic gut bacteria. Thus, in the event that the intestine becomes infected with a pathogenic species and the faecal material is flushed rapidly from the colon as a defence mechanism (diarrhoeal response), the appendix, which contains normal gut bacteria that form biofilms in a constant state of growth and shedding, serves as a source of bacteria to re-inoculate the gut with its normal gut flora, thus facilitating reestablishment of the normal intestinal flora following infection of the gut (Bollinger et al., 2007). Such rapid reconstitution of the gut flora is likely critical for survival in an environment where diarrhoeal illness is, in concert with malnutrition or perhaps lack of water, extremely life-threatening, particularly to the very young. Diarrhoeal disease, for example, has been at times the single greatest cause of disability-adjusted life years lost in the most populated countries in South-East Asia (WHO, 2001). This assessment provides an explanation for the observation that, regardless of the importance of the function of the appendix throughout evolutionary history, the structure is not apparently important in the face of medical care and hygienic practices associated with developed countries.
Bacterial biofilms and high concentrations of GALT are in fact associated with the terminal portion of the cecum in rats (Palestrant et al., 2004) and in mice (Swidsinski et al., 2005), and with the human appendix (Bollinger et al., 2007) more so than with other portions of the large bowel of these animals. Microbial biofilms have also been found in the ceca of nonhuman primates (Palestrant et al., 2004) and the koala (Phascolarctos cinereus; McKenzie, 1978). However, it should be noted that the distribution of bacterial biofilms and immune tissue in the gut of many species, including some species possessing an appendix, has not yet been described, and more work in this area is warranted.
The evidence that the appendix may be well suited to serve an important role in humans challenges the belief that this anatomical structure lacks a function, and brings to the forefront the question as to when in mammalian evolutionary history the structure and its current function evolved. In light of the apparent function of the appendix, and of the possible adaptive advantage the appendicular morphology provides for the maintenance of the gut flora, the comparative anatomy and evolution of appendicular morphology are examined and discussed.
Materials and methods
The methods applied in this study are threefold. First, a comparative anatomical approach was adopted to describe instances of appendicular morphology among mammals. Second, a phylogenetic analysis was conducted in which cecal and appendicular characters were mapped onto a mammalian molecular consensus phylogeny. Considered together, these approaches contribute to an understanding of the evolution and function of the cecal appendix across mammals. Finally, an assessment of microbial biofilm formation was conducted on the gastrointestinal tract of an outgroup species (amphibian) as a test case for whether the biofilms found in the mammalian appendices and ceca might also characterize the proximal bowel of an outgroup for the appendix and cecum. This final approach provides insight into the potentially long evolutionary history of the immunological function of the proximal large bowel.
Assessment of the cecal morphology in Mus musculus (common laboratory mouse) was conducted in the laboratory. All procedures involving laboratory mice were approved by the Duke University Animal Care and Use Committee. Mice were fed ad libitum prior to euthanization. Following anesthetization using 5% isofluorane in oxygen, whole blood was withdrawn from the inferior vena cava, and the heart and lungs were removed en bloc as prescribed under a protocol not related to this project. The colon, cecum and small bowel were then dissected free and photographed.
The data collected during the comparative anatomical analysis were used to generate categorical variables capturing the morphological variation present in the appendix of the major mammalian lineages. Taxonomic groups at the family level were included as terminal operational taxonomic units whenever possible given the availability of data. In the event that the cecal and appendicular morphology had been described in a limited number of species within a taxon, those species were interpreted as representative of the morphology of the entire taxon. In the event that the presence of an appendix in a taxon was variable, some analyses were conducted with those taxa defined as having an appendix, and additional analyses were conducted with those taxa described as variable. The data matrix of appendicular characters is presented in Table 1.
Table 1. Data matrix of character states mapped onto the consensus phylogeny.
Analysis 1 employed the strict definition of an appendix as a cecal appendage with a distinct junction. Analysis two treats true appendices the same as appendix-like structures without an evident cecum or without an evident junction between appendix and cecum. Some of the cladistic analyses described in the text were conducted using three character states, with absent = 0, variable = 1 and present = 2, whereas other analyses used only two states, with absent = 0, variable = 1 and present = 1.
We defined the appendix strictly based on morphology as a relatively narrow and extended, close-ended structure at the apex of the cecum that is clearly distinguished from the cecum by a relatively abrupt change in the diameter of the bowel between the cecum and the appendix. Although the cecal appendix has been characterized in some cases, especially in rabbits and in humans, as having a thickened wall and a high concentration of lymphoid tissue, these properties are generally associated with the apex of ceca with no appendix (Berry, 1900), and these properties have not been evaluated in many species that have an appendix based strictly on a morphological assessment. Thus, at the present time, the presence of lymphoid tissue or the thickness of the tissue are less useful parameters than the shape of the bowel as defining characteristics of a true cecal appendix.
Some taxa were identified which exhibit small appendix-like structures in the apparent absence of a cecum, or which exhibit very long, tapering ceca that terminate in long, narrow appendix-like structures, but which have no apparent abrupt change in the diameter of the bowel that would mark a junction between the cecum and an appendix. Whether these structures may simply be unusual ceca or whether they may share a common evolutionary history with true appendices is not known; thus, two analyses were conducted. Analysis 1 was conducted using only true appendices as defined above by morphological features, while Analysis 2 treated true appendices and appendix-like structures as the same (Table 1). For both Analysis 1 and Analysis 2, taxa exhibiting variable characters were treated using two approaches in the analyses: as a character state the same as taxa with an appendix (i.e. absent = 0, variable = 1, present = 1) or as exhibiting an intermediate state (i.e. absent = 0, variable = 1, present = 2). For the optimization, the ACCTRAN procedure in paup 4.0 was used.
For all analyses, we report the most parsimonious result or results with the least number of independent evolutionary gains of an appendix, as gains of an appendix during the evolutionary process would presumably be less likely than the loss of an appendix. To systematically find the most parsimonious result with the least number of gains, a gain was weighted as higher than a loss. Using a model where taxa with an appendix were treated the same as taxa having variable expression of an appendix, a gain (change from state 0 to state 1) was weighted as 1.1 vs. a loss of an appendix (change of state 1 to state 0) which was weighted as 1.0. This approach ignores all transitions between the variable state and the appendix-present state, essentially weighting those transitions as zero in the calculations.
The second approach to the cladistic analyses for both Analysis 1 and Analysis 2 was to treat taxa with variable expression of an appendix as an intermediate between taxa possessing an appendix and taxa lacking an appendix. For this analysis, a gain (0–1 or 0–2) was weighted as 1.1, a loss (2–0 or 1–0) was weighted as 1.0, and a transition between variable and present states (1–2 or 2–1) was weighted as 0.9. This approach, in contrast to ignoring transitions between the appendix-present state and the appendix-variable state, considers these transitions on an almost equal footing to a variety of other possible evolutionary events, including the de novo gain of an appendix. Thus, for example, this approach would favour five independent gains of the appendix over one gain plus five changes from the appendix-present state to the appendix-variable state. Although some appropriate weighting scheme might possibly yield an accurate result, the selection of any such weighting scheme would be arbitrary, and indeed a truly accurate weight is likely to be variable depending on the taxa. Considering this, we present results in which transitions between appendix-present and appendix-variable states are essentially weighted as zero (ignored by grouping the appendix-variable and appendix-present states together), and another set of results in which these transitions are only weighted slightly less than an independent gain of an appendix. Although neither approach is optimal, these two approaches taken together yield a range of results that potentially provide a better perspective on the evolutionary history of the cecal appendix than any one analysis alone.
The most parsimonious evolutionary scenario was taken to be that with the lowest sum of total weighted evolutionary events, counting each event as the weighted value of that event. These weighting schemes resulted in the identification of only one maximally parsimonious tree for each analysis with a minimum of evolutionary gains of an appendix.
To determine if character optimization would be reliable, the degree of phylogenetic signal in the distribution of the cecal appendix was assessed using a random tree analysis, as suggested by Laurin (2004, 2005). For this approach, 10 000 random trees were generated using equiprobable trees in Mesquite 2.6 (Maddison & Maddison, 2008), and the most parsimonious occurrence of the appendix was assessed in those trees as in the actual tree, incorporating the above mentioned weighting schemes.
The weighted number of evolutionary events found in the optimization analysis of the actual tree was then compared with the numbers found on the random trees. If the number of events of the appendix across the consensus phylogeny is less than that found in 95% of the random trees, then the null hypothesis (that the character is randomly distributed in phylogeny) could be rejected at the 0.05 confidence level. The assessment of random trees in this manner thus allowed the determination of whether the distribution of the appendix has a phylogenetic pattern more so than would be expected by chance. In addition to the analysis of the number of appearances of the appendix, the number of independent losses of an appendix in the actual tree was compared with the number of independent losses on the random trees.
Because North American bullfrogs (Rana catesbeiana) lack an appendix and cecum, an analysis of the intestinal biofilms in this species (a representative of an outgroup in the cladistic analyses) will potentially provide insight into the immunological function of the gastrointestinal tract prior to the evolution of the cecum or appendix. North American bullfrogs were obtained from Rana Ranch Bullfrog Farm (Twin Falls, ID, USA).
All procedures involving R. catesbeiana were approved by the Duke University Animal Care and Use Committee. Frogs were fed ad libitum prior to euthanization. The frogs were euthanized by immersion in a solution containing 1% tricaine methanesulfonate (MS222) until no corneal response was noted, followed by double pithing. Samples were evaluated for the presence of biofilms adjacent to the gut epithelium using the method previously described (Palestrant et al., 2004) as follows: Bulk stool was removed from sections of intestine, and the sections were placed in an embedding medium for frozen tissue specimens (Tissue-Tek O.C.T. Compound; Sakura Finetechnical Co., Ltd, Tokyo, Japan) so that the luminal side of the sample, with remaining faecal material, was face up and covered with the O.C.T. compound. Samples were then flash frozen in liquid nitrogen and stored at −80 °C until use. Four-micrometer cryosections were prepared from frozen biopsies, mounted on microscope slides, and stored at −80 °C until needed. Slides were allowed to warm to room temperature and fixed in 95% ethanol, gradually hydrated to distilled water, rinsed with 1% acetic acid for six seconds, and then washed twice in phosphate buffered saline, pH 7.4. Slides were incubated with 0.1% acridine orange (Sigma, Saint Louis, MO, USA) in 67 mm phosphate buffer, pH 6.0 for 3 min, washed with phosphate buffer for 1 min and differentiated in 100 mm CaCl2 for 30 s. Slides were coverslipped with Vectashield mounting medium, and evaluated using a Zeiss 410 confocal light microscope (Carl Zeiss, Inc., Thornwood, NY, USA). Photos were taken of the areas at the border between the epithelium and the lumen, and a false colour image was constructed from the black and white images to highlight the presence of biofilms in the images.
The appendix of the rabbit is easily viewed as an extension of its already enlarged cecum (Archer et al., 1963), with a readily discernable boundary between the cecum and the appendix marked by a sharp change in the diameter of the bowel (Fig. 1e; Snipes, 1978; Kurohmaru et al., 1984; Berry, 1900). Both the diameter of the lumen and the length of the rabbit’s appendix are 3–5 times larger than the corresponding measurements of the typical human appendix. The appendix of the prehensile-tailed porcupine (Coendou prehensilis) is said to be long and resembling that of a rabbit (Stevens & Hume, 1995): beyond that, its morphology has not been thoroughly described. The kangaroo rat (species of the genus Dipodomys) is also said to possess a cecal appendix (Midgley, 1938), but the features of the appendix have not been described. However, the cecal appendices of several rodent species have been described in at least some detail. For example, the lemming (Lemmus lemmus) is depicted by Behmann as having a cecal appendix substantially smaller than the primate appendix, with dimensions of approximately 10 mm long by 2–3 mm wide (Behmann, 1973). Other examples from rodents include the cecal appendices of the scaly-tailed flying squirrel (Anomalurus derbianus; Chivers & Hladik, 1980) and the meadow vole (Microtus pennsylvanicus; Golley, 1960). Both of these rodents have cecal appendices substantially smaller than the primate appendix, each being 7–8 mm in length and about 3 mm in diameter (Fig. 1g,l). The fact that the Cape dune mole-rat possesses an appendix was only recently discovered (Kotze et al., 2006). It consists of a constricted cecal apex, similar to those found in the rabbit and prehensile-tailed porcupine, and is positioned lateral to the cecum against the left abdominal wall (Fig. 1i; Kotze et al., 2006). The appendix of the Cape dune mole-rat is somewhat larger than the appendices in scaly-tailed flying squirrels and in meadow voles, being approximately 15 mm in length and 4 mm in diameter. A recent evaluation of six species of mole-rat found that only two of the six species evaluated possessed a cecal appendix (Kotze et al., 2008). Interestingly, no statistically significant variation in the cecum size of the same six mole-rat species was observed based on assessment of the fraction of surface area or the fraction of length contributed by the cecum to the overall digestive tract (Kotze et al., 2009). Thus, the presence or absence of a cecal appendix in mole-rats does not correspond to cecum size.
In addition to the traditional morphological pattern in which the appendix is manifested as a projection from the distal end of the cecum, at least one mammal possesses an appendix despite very little to no obvious cecum. Hume depicts the appendix in wombats as being about 2 cm long by 0.75 cm in diameter, and describes a digestive tract with a cecum that is virtually absent (Fig. 1d; Hume, 1989, 1982). In fact, the cecum is sufficiently small that early researchers debated whether or not the appendix of the wombat was located at the apex of the cecum, as in primates (Owen, 1868; Flower, 1872; Lönnberg, 1902; Mitchell, 1916). The confusion stemmed from dispute as to the definition of the cecum itself in wombats, and later anatomical studies confirmed that the appendix is indeed an extension at the apex of the true cecum in Vombatidae (Lönnberg, 1902; Mitchell, 1916).
Interestingly, Stevens and Hume describe structures that appear superficially identical to appendices in the proximal large bowel of monotremes (Stevens & Hume, 1995). These structures, present in the absence of an obvious cecum, are described in both of the extant monotremes, the platypus (Ornithorhynchus anatinus) and in the echidna (Tachyglossus aculeatus). The structure in the platypus, for example, is 2.5 cm long and about 3 mm in diameter (Fig. 1n; Krause, 1975), about one-third the size but approximately equivalent in shape to the human appendix. The structure in the echidna is smaller, being slightly over one cm in length and 3–4 mm in diameter (Fig. 1o; Stevens & Hume, 1995), and is similar in dimensions to the appendix of the Cape dune mole-rat. It is unknown whether these structures in the monotremes are true appendices of a small cecum, or whether these structures actually correspond to the cecum in other animals. If these appendix-like structures do represent the cecum in this earliest extant branch of the mammalian line, it suggests the possibility that, at least in some cases, the cecum might have originated as a nondigestive organ first, and developed later into a larger structure that can function as a chamber for bacterial fermentation.
Several other mammalian species are characterized by unique cecal morphology. The koala and greater glider, both marsupials, possess long tapering ceca (Fig. 1m; Hume, 1989), which appear superficially similar to the cecum and appendix found in the rabbit. Although the apex of the koala’s cecum has been described as an appendix (Keith, 1912), the cecum of the koala lacks any apparent stricture which might mark the junction between cecum and appendix, and thus the koala cannot be considered to have a true cecal appendix based on the definition we use. The greater glider, on the other hand, has an abrupt stricture in the cecum, albeit less pronounced than that in the rabbit, which may mark the transition from cecum to appendix. However, while the appendices of the rabbit and the human are characterized by a small lumen and dense immune tissue, it is unclear whether the distal ceca of the koala and the greater glider share these traits. Nevertheless, it seems reasonable to hypothesize that a long, tapering cecum may serve digestive and immunologic functions in much the same way as the cecum plus the appendix does in the rabbit. In support of this idea is the observation by electron microscopy that the cecum of the koala is replete with bacterial biofilms (McKenzie, 1978) formed by bacteria which are vital to the ability of the koala to digest eucalypt leaves (Osawa et al., 1993a,b). As previously described (Bollinger et al., 2007), the presence of bacterial biofilms is an indicator of the apparent immunologic function of the apex of the cecum or, when present, the appendix. In addition to the koala and the greater glider, the ground cuscus (Phalanger gymnotis), a cecal fermenting marsupial native to New Guinea, is also depicted as having what superficially appears to be an appendix at the apex of its cecum (Fig. 1f; Hume, 1999). However, the 1/4 cm appendix or appendix-like structure in the ground cuscus has not been characterized in detail, and it remains unknown whether the structure is associated with substantial immune tissue.
Among nonhuman primates, varying expressions of cecal appendages exist and there has been debate over the specific characters used to define an appendix (Fisher, 2000). While hominoids have been found to consistently possess a narrow distinct apex (Fig. 1a,b), thick apex walls, and a concentration of lymphoid tissue, other primate taxa are known to exhibit various combinations of these traits (Fisher, 2000). Unfortunately, the data are still quite limited for most primate taxa (Fisher, 2000). However, the data which are available demonstrate a wide distribution of appendices on the primate phylogenetic tree, suggesting an ancient history of cecal appendages among primates.
The aforementioned mammalian taxa with a well-documented appendix are taxonomically and phylogenetically distant (lagomorphs, rodents, marsupials and primates), although they all fall into either the Euarchontoglire or marsupial clades. In addition, their diets are variable, although the nonprimate species known to possess an appendix are characterized by an herbivorous diet, and many demonstrate a preference for grasses or fibrous leaves (e.g. Owen, 1868; Borradaile, 1955). However, many herbivorous taxa survive without appendices. Unlike the Cape dune mole-rat (Kotze et al., 2006) and the wombat (Hume, 1989), which are both colon fermenters, the rabbit, pika and porcupine are cecal fermenters, with large ceca that function as chambers for bacterial fermentation (e.g. Dalquest et al., 1952; Stevens & Hume, 1995; Kotze et al., 2006). In those species, the appendix may serve as an extension of this fermentation chamber. The otherwise indigestible cellulose in their primary food sources is broken down into useable nutrients by the resident microflora in the cecum. In the rabbit, cellulose is broken down into carbohydrates in the appendix (Borradaile, 1955). On the other hand, the appendices of the Cape dune mole-rat and the wombat are of such small size that they cannot significantly contribute to fermentation beyond that carried out by the remainder of the rather large colon of those species. There are few data available, however, on the degree of bacterial fermentation in the appendices of other species.
A final consideration regarding the comparative anatomy of the cecal appendix is the observed variation of anatomy between individuals within a given species. Scott, for example, noted ‘considerable variation’ in the shape of the cecum in a variety of primate species (Scott, 1980). Fisher (2000) noted four primate species (Callicebus moloch, Callicebus personatus, Macaca nigra and Colobus polykomos) in which the presence of the appendix varies from individual to individual within that species, and indicated that other species probably have variable expression of an appendix, but have yet to be identified because of a lack of study. In fact, expression of the cecal appendix is not entirely uniform even in Homo sapiens, with dozens of cases of congenital absence of the vermiform appendix having been reported (Pester, 1965), and at least one case of two vermiform cecal appendices in one individual being reported (Travis et al., 2008). Nevertheless, as the congenital absence of an appendix in humans is relatively rare, we will consider the appendix to be ‘present’ rather than ‘variable’ when conducting the cladistic analyses described below.
A careful examination of the intestinal morphology of Mus musculus (the common laboratory mouse) reveals that intra-species variability in the expression of the cecal appendix is not limited to primates. Mus musculus has been depicted by Komarek as having a distinct apex to its cecum (Komarek, 2004), although it is unclear whether the junction between the apex and the corpus of the cecum is sufficiently well defined to qualify the cecal apex as a true cecal appendix. However, examination of the cecal morphology of laboratory specimens reveals (Fig. 2) that some specimens do indeed have a distinct cecal appendix, while others completely lack any stricture whatsoever distinguishing the apex of the cecum from the corpus of the cecum, and yet others possess an intermediate morphology similar to that described by Komarek (2004). This variability is observed even in inbred strains of Mus musculus raised under relatively uniform and unvarying laboratory conditions (Fig. 2), suggesting that the variability of appendix morphology within this species is not because of genetic variability between individuals or to any obvious dietary factors.
As described in the Materials and methods, Analysis 1 employs a strict definition of an appendix, while Analysis 2 considers appendix-like structures the same as true appendices. Results are presented below from both Analysis 1 and Analysis 2 using two approaches. First, any taxa with a variable character state were treated the same as taxa with an appendix (i.e. absent = 0, variable = 1, present = 1). This essentially weights any transitions between the appendix-variable state and the appendix-present state as zero for the analysis. Secondly, any taxa with a variable character state were treated as exhibiting an intermediate state (i.e. absent = 0, variable = 1, present = 2). This approach assigns a relatively high weight (0.9) to any transitions between the appendix-variable state and the appendix-present state. The distribution of appendices present across mammals is shown in Fig. 3, with the results from Analysis 1 indicated by solid branches of the phylogenetic tree, and the results from Analysis 2 shown by the combination of solid (true appendix) and checkered (appendix-like structures) branches.
Analysis 1: employing a strict definition of an appendix
As predicted, the result of the character change analysis indicated that the primitive condition for mammals is the absence of an appendix (Fig. 3). In the analysis characterizing all taxa as either present or absent for the appendix (taxa with variable expression of the appendix were characterized as having an appendix; left side of Fig. 3), the results indicate that the appendix has evolved independently five times: at the basal primate node, at the base of the glires (rodents and lagomorphs), in the vombatidae (wombats), in the phalangeridae (cuscus and brush-tailed possum) and in the pseudocheiridae (ring-tailed possums and gliders). The appendix was lost in three primate clades (indriids, tarsiids and the cheirogaleids), and in two rodent clades (sciurids and the caviidae/chinchillidae clade).
Analysis 1 was also conducted in which a ‘variable state’ was assigned to all taxa having some species with an appendix and some without an appendix (Fig. 3, right side of diagram). A substantial number of taxa having an appendix are characterized as having a variable state (11 of 23, 48% for Analysis 1; see Table 1). However, the results from this analysis were similar to the results obtained treating the appendix-variable state the same as the appendix-present state, with the exception of the results in the primate clade (Fig. 3, compare left and right sides of diagram). Treating the variable appearance of the appendix as an intermediate state resulted in five independent evolutionary occurrences of the appendix in the primate clade. This nonintuitive result probably reflects an inadequacy of the computational approach to deal with the effects of frequent loss of the appendix in phylogeny rather than actual multiple independent gains of the appendix in closely related species.
Analysis 2: treating appendix-like structures and appendices together
The distribution of appendices present across mammals when treating appendix-like structures as appendices is indicated on the left side of Fig. 3, with taxa having appendix-like structures indicated by the checkered branches of the cladogram. In the analysis characterizing all taxa as either present or absent for the appendix (taxa with variable expression of the appendix were characterized as having an appendix), seven appearances of the appendix were identified. There were differences to the hypothesized evolution of the appendix within marsupials and monotremes compared with those found in Analysis 1. In addition to the appearances in primates and glires described for Analysis 1 above, the appendix also appeared in Caenolestidae and at the base of the monotreme clade (platypus and echidna). In addition, the appendix appeared three times in the diprotodont marsupials, including at the base of the Vombatidae/Phascolarctidae clade (left hand side of Fig. 3).
Analysis 2 was also conducted in which a ‘variable state’ was assigned to all taxa having some species with an appendix and some without an appendix (Fig. 3, right side of diagram). A substantial number of taxa having an appendix or appendix-like structure are characterized as having a variable state (12 of 26, 46% for Analysis 2; see Table 1). As in Analysis 1, treating the appendix-variable state as an intermediate state resulted in five independent evolutionary occurrences of the appendix in the primate clade, but had no notable effect on the results in other clades.
Detection of a phylogenetic signal
In both Analysis 1 and Analysis 2, regardless of whether taxa with variable expression of the appendix were treated the same as taxa with an appendix or as an intermediate state, none of the 10 000 randomly generated trees scored as low as the actual tree (P <0.0001) in their respective analyses. The sum of all weighted events in the real tree was at least 2 U less than the lowest scoring randomly occurring tree in all analyses conducted. For example, in Analysis 1, considering taxa with a variable expression of the appendix the same as taxa with consistent expression of an appendix, the sum of all weighted events occurring in the actual tree was 10.5, whereas the lowest sum obtained in any of the random trees was > 14.8. As another example, in Analysis 2, treating taxa with a variable expression of the appendix as an intermediate state, the sum of all weighted events occurring in the actual tree was 16.8, whereas the lowest sum obtained in any of the random trees was > 19.6. Thus, the distribution of the appendix shows a highly significant phylogenetic signal based on these analyses.
Alternatives to the most parsimonious results
Although the above analyses each yielded one most parsimonious result with a minimum number of gains of the appendix, other results are worth consideration. Considering only true cecal appendices (Analysis 1) and ignoring transitions between appendix-variable and appendix-present taxa, the evolution of the appendix at the base of the Euarchontoglires would require only one additional evolutionary step compared with the most parsimonious analysis. In the same analysis, placing the appearance of the appendix at the base of the diprotodont marsupials requires an additional two steps.
Considering true cecal appendices and appendix-like structures (Analysis 2) and ignoring transitions between appendix-variable and appendix-present taxa, the evolution of the appendix at the base of the Euarchontoglires would require only one additional evolutionary step compared with the most parsimonious analysis, as in Analysis 1 above. However, in Analysis 2, again ignoring transitions between appendix-variable and appendix-present taxa, the number of steps required to position the appendix at the base of the diprotodont marsupials is one extra step, rather than two as in Analysis 1.
Biofilm distribution in an outgroup
In mammalian species evaluated to date, microbial biofilms tend to be most prominent in the appendix and in the apex of the cecum in animals without an appendix. In contrast, biofilms are generally absent in the distal large bowel and in the small intestine. Analysis of the biofilm distribution in the gut of the frog, which lacks a cecum or appendix, revealed a distribution of microbial biofilms strikingly similar to that seen in mammals (Fig. 4). Despite the fact that the entire length of the large bowel of the frogs examined was only about 3 cm in length, microbial biofilms were extensively associated with the proximal large bowel, and decreased in size and abundance in the mid and especially the distal large bowel (Fig. 4). Biofilms were not observed in the small bowel of the animals (Fig. 4). These observations suggest that immune-mediated biofilm formation in the proximal large bowel is independent of the architecture of the bowel, and point toward an ancient origin for the support of microbial biofilms in the proximal large bowel of vertebrates.
It is now evident that animals with a gut evolve as ‘superorganisms’, which include not only the animal species, but also the microorganisms that live within the animal’s digestive tract (Backhed et al., 2005; Moran, 2006; Ley et al., 2008). This co-evolution is perhaps not surprising, given the vital importance of the symbiotic relationship to both the host species and the intestinal microbes living in the gut of that species. This view, in conjunction with the finding that the appendix can be described as a ‘safe-house’ for biofilms containing commensal bacteria (Bollinger et al., 2007), provides a new perspective in which to evaluate the evolution of the appendix.
The word appendix, which derives from the same Latin root as ‘appendage’, indicates in anatomy ‘a supplementary, accessory or dependant part attached to a main structure’ (Dorland & Newman, 2003). Implicit in this definition is an identifiable boundary between the appendix and the ‘main structure’. Thus, in case of the cecal appendix, the identification of a boundary between the cecum and the appendix is a necessary criterion for the presence of a true cecal appendix. Morphologically, the boundary is evident as an abrupt change in the diameter of the bowel between the cecum and the appendix. With this view, we have defined the true appendix based on morphology as a relatively narrow and extended, close-ended structure at the apex of the cecum that is clearly distinguished from the cecum by a relatively abrupt change in the diameter of the bowel between the cecum and the appendix. However, this strict definition might eliminate structures that are evolutionarily related to a true appendix but lacking an observable cecum or junction separating cecum from appendix. Thus, for the purpose of the cladistic analyses, we defined an ‘appendix-like structure’ as a relatively narrow and extended, close-ended structure resembling a true appendix, but either lacking a distinct junction separating it from the cecum (as in the koala), or present in the absence of an identified cecum (as in monotremes). This approach provides some degree of accommodation in the event that the strict definition of an appendix as currently applied is unnecessarily narrow. For example, further study in monotremes may reveal a small cecum, and further study in the koala may reveal a distinct junction between cecum and appendix marked by such factors as increased wall thickness and high concentrations of lymphoid tissue as in the appendix of the human and rabbit.
The adjective ‘vermiform’, meaning ‘wormlike’, is often used to describe the cecal appendix of primates, but is generally not used to describe the cecal appendix of other animals. In our description, we did not breach this historical usage of the term vermiform. However, it seems highly likely that the primate appendix shares a common function and perhaps even common evolutionary origins with the appendices of other mammals. Thus, although historical precedent limits the use of the term vermiform almost exclusively to the primate’s cecal appendix, the structure and biological function of the primate’s appendix is probably best understood within the context of the cecal appendices of a wide range of mammals, including primates and nonprimates alike.
The present analysis strongly supports the idea that the appendix is frequently lost during evolution. More than 45% of taxa containing species with an appendix are polymorphic, also containing species without an appendix. Although this observation could hypothetically be accounted for by multiple, independent gains in evolution, the frequency of finding the appendix in certain clades seems unreasonably high for this explanation to be valid: Of the taxa examined, 10 of 13 primate taxa (77%), 10 of 13 glire taxa (77%) and three of nine (33%) diprotodont marsupial taxa contain species with an appendix or appendix-like structure. Further, as the random tree analysis indicates, the fact that the appendix is found only in euarchontoglires and diprotodont marsupials strongly supports the presence of at least two gains of the appendix. In addition, the observation that the appendix is variably expressed in a variety of euarchontoglires and diprotodonts (shown as polymorphic in Fig. 2) is probably best explained by a frequent loss of the appendix during evolution. This conclusion suggests that the appendix is a biologically expensive structure and/or the conditions for which it is necessary are not present in all ecological niches.
Cecal morphologies that may reflect the recent evolutionary loss of an appendix are of interest. In case of Rattus norvegicus, for example, the apical region of the cecum is rich in lymphoid tissue but is not clearly separated from the basal region by a change in the diameter of the bowel. This lymphoid rich region of the cecum ‘is thought to be analogous with the vermiform appendix of man’ (Hofstetter et al., 2006). Based on the cladistic analyses reported here, the apical region of the rat’s appendix might indeed be evolutionarily derived from the cecal appendix of an ancestor. The kangaroo (Macropus giganteus), as another example, has a small cecum with two distinct segments, one basal and one apical. The apical segment of the kangaroo’s cecum is not defined as a cecal appendix because there is no profound change in diameter marking the junction between the two portions of the cecum. However, the apical segment of the cecum in this diprotodont marsupial, as in the rat, may reflect a structure analogous to the cecal appendix that was present in an extinct ancestor. This idea is particularly attractive since several other diprotodont marsupials possess an appendix or an appendix-like structure. With this in mind, it seems possible that a common ancestor of the diprotodont marsupials, a clade which includes the kangaroo, may have possessed an appendix or appendix-like structure.
The idea that the loss of an appendix may be a far more likely evolutionary event than the gain of an appendix brings to light several alternatives for the evolution of the appendix in addition to the ones indicated by the parsimony analyses. The most glaring possibility is that the appendix evolved only once during the evolution of the placental mammals, occurring at the base of the Euarchontoglire clade. This scenario involves two additional losses of the appendix (in the Cynocephalidae and the Scandentia) but eliminates the gain of the appendix at the base of the primates. Similarly, if the appendix evolved only once in diprotodont marsupials, then at least four additional losses of the appendix occurred (one in the kangaroo), but two gains are eliminated. Unfortunately, any devised weighting scheme applied to the cladistic analyses to compensate for the relative propensity for appendix loss vs. appendix gain would be arbitrary, and thus conclusions regarding the exact number of appearances of the cecal appendix in phylogeny are likely to be elusive.
Whether the appendix evolved independently in glires and in primates and whether the appendix evolved more than once in diprotodont marsupials remain interesting questions. Regardless of the answer to these questions, the appendix has apparently evolved independently and convergently at least once in Euarchontoglires and at least once in diprotodont marsupials. This finding adds to an extensive list of morphological features that have evolved independently and convergently in Australian marsupials and again in placental mammals in the rest of the world (Strickberger, 2000; Coyne, 2009).
The appendix has evolved independently at least twice and has been extensively maintained, albeit not uniformly, in at least three and perhaps four groups of mammals; glires, primates, Diprotodont marsupials and perhaps monotremes. The possibility that the appendix occurred at the base of the primates suggests that the appendix may have been preserved in that clade since before the two primate suborders, Strepsirrhini and Haplorrhini, diverged an estimated 60–63 million years ago (Gingerich, 1986; Shoshani et al., 1996; Pouydebat et al., 2008). Similarly, the conclusion that the appendix evolved at the base of the glires indicates that the appendix has been maintained since before the K–T extinction, when rodents and lagomorphs diverged approximately 80 million years ago (Bininda-Emonds et al., 2007). However, if the appendix was only acquired once in placental mammals, at the base of the Euarchontoglires, then the acquisition of the appendix is pushed back further as far as to 92 million years ago, after the first Euarchontoglires appeared (Bininda-Emonds et al., 2007), but before Euarchonta and Glires diverged. In contrast, the appearance of the appendix in the diprotodont marsupials was substantially more recent, occurring sometime after 54 million years ago when that clade emerged (Bininda-Emonds et al., 2007). Nevertheless, the ancient origin or origins of the cecal appendix in the Euarchontoglires and the parallel evolution of the structure in the marsupials and placentals strongly support the idea that the function of the structure is biologically important.
Although microbial biofilms in the proximal large bowel are apparently a hallmark of immune support for the microbial flora in a wide range of mammalian species, the biofilm distribution in the gut of an outgroup for mammals had not been evaluated previously. Thus, the observation that biofilms are distributed in frogs in a manner similar to mammals, with a preference for the proximal large bowel (Fig. 4), strongly suggests an ancient origin for a pro-microbial immune function in the proximal large bowel. Specifically, this observation suggests that the adaptations supporting biofilm growth by commensal bacteria are more ancient than blind sacs of the gut, such as the cecum, which are involved in fermentation. In support of this idea, microbial biofilms are not only strengthened by secretory Immunoglobulin A (SIgA) produced by the adaptive immune system, but can also be supported by mucin (Orndorff et al., 2004; Bollinger et al., 2005), a major biomolecule produced by the more ancient innate immune system. This finding points toward increased immune support of the gut microbes as one of the potential driving forces for the evolution of blind sacs in the proximal large bowel.
A variety of species may be described as having a cecum that is greatly diminished in size compared with the ceca of cecal fermenters. These structures typically contribute a negligible volume to the digestive tract as a whole, indicating that they do not significantly contribute to fermentation or digestion at least in terms of adding volume to the digestive tract. These structures are typically 1–10 cm in length depending on the species, with a diameter typically < 2 cm and often < 1 cm. Such structures are found even in nonmammalian vertebrates, indicating a remarkably frequent convergent evolution of appendices or appendix-like structures, and suggesting the putative importance of those structures. A variety of birds, for example, are depicted as having paired ‘small ceca’ or appendix-like structures located at the proximal end of the large bowel. Avian species with paired appendix-like structures (Stevens & Hume, 1995) include the red-tailed hawk (Buteo jamaicensis), the hoatzin (Opisthocomus hoazin) and the emu (Dromaius novaehollandiae). Furthermore, one or two small ceca or appendix-like structures are also described in the mid-to-hindgut of a variety of teleosts (Stevens & Hume, 1995), including the sea chub (Kyphosus sydneyanus), the catfish (Bagarius bagarius), the knifefish (Notopterus notopterus) and cod (Raniceps raniceps).
As an extension of this thinking, a re-evaluation of the function of ceca in a variety of species may be warranted: it seems plausible that the morphology associated with ceca that comprise a relatively small percentage of the overall digestive tract may be specifically adapted for an immunologic function without substantial contribution to fermentation. For example, kangaroos (e.g. Macropus giganteus), hare-wallabies (e.g. Lagorchestes hirsutus), llamas (Lama glama), cattle (Bos primigenius taurus) and colobus monkeys (Colobus abyssinicus) have relatively small ceca compared with the rest of their voluminous digestive tracts (Dellow, 1979; Stevens & Hume, 1995; Hume, 1999; Reece, 2004). These species are all foregut fermenters, presumably rendering the digestive functions of a cecum of secondary importance. Here again the cecum in these animals may perform an immunologic function in the absence of a substantial digestive (fermentation) function, much like the appendix in animals with a substantial cecum. Given the cladistic analyses reported here, it seems probable that the small ceca in at least some species are evolutionarily derived from ceca with an appendix that had appeared early and independently in the evolution of diprotodont marsupials (e.g. kangaroos and wallabies) and primates (e.g. colobus monkeys). Thus, the evolution of small ceca in some species that are foregut fermenters seems likely to have involved loss of the digestive function of a cecum with an appendix, with maintenance of the immunologic function of the cecal appendix.
These observations, in addition to the demonstration that an amphibian species also harbours biofilms containing beneficial bacteria in the proximal section of its large bowel in a manner similar to mammals, raises the question as to whether the evolution of the cecal appendix is directed toward re-establishment of the immunologic function that may have been sacrificed during enlargement of the cecum for digestive purposes. In this scenario, it is plausible that a long, narrow appendix-like structure could have evolved as an immune structure in some cases prior to the evolution of a larger cecum with digestive function, and could have in fact been a precursor for that larger structure in some cases. The presence in extant monotremes of a very small, appendix-like structure without any evident digestive function in the absence of any apparent cecum is certainly consistent with this idea.
New developments in the field of immunology and gut biology have indicated that the function of the mammalian appendix involves immune-mediated maintenance of the normal gut flora, especially during bouts of intestinal distress. Such a function is likely enhanced by the narrow lumen of the appendix, which is expected to provide some degree of protection from any potential pathologic contaminants in the faecal stream. This function could be a selective force for the evolution and maintenance of the appendix in a variety of species, suggesting that re-evaluation of the evolutionary origins of the appendix is warranted.
Three morphotypes of cecal-appendices can be described among mammals based primarily on the shape of the cecum: a distinct appendix branching from a rounded or sac-like cecum (as in many primate species), an appendix located at the apex of a long and voluminous cecum (as in the rabbit, greater glider and Cape dune mole-rat), and an appendix in the absence of a pronounced cecum (as in the wombat). In addition, long, narrow appendix-like structures are found in mammals that either lack an apparent cecum (as in monotremes) or lack a distinct junction between the cecum and appendix-like structure (as in the koala). A cecal appendix has evolved independently at least twice, and apparently represents yet another example of convergence in morphology between Australian marsupials and placentals in the rest of the world. Although the appendix has apparently been lost by numerous species, it has also been maintained for more than 80 million years in at least one clade, supporting the idea that the structure has a vital biological function. The presence of biofilms in the proximal large bowel of the bullfrog, an amphibian which lacks an appendix and cecum, suggests an early history of immune-maintenance of commensal bacteria in the proximal large bowel of vertebrates, and points toward maintenance of the normal bacterial flora in the gut as a potential driving force for the evolution of blind sacs in the proximal large bowel.
This work was supported in part by the Fannie E. Rippel Foundation, the Duke Cancer Center, and by grants R01 HL60232-03 and P30 DK34987 from the National Institutes of Health. The authors thank Roxanne Wilson, Julie Fuller, Anna L. Keyte, Carolyn M. Berger, Diana L. Diesen and Heather M. Derby for technical assistance. The authors also wish to thank Jeffrey C. Oliver for assistance with Mesquite, Zoie Holzknecht for help with preparation of the manuscript and Michel Laurin for his generous guidance in performing the cladistic analyses. In addition, the authors thank Kathleen K. Smith, Diana L. Diesen, Sanet H. Kotze, and Carolyn M. Berger for contribution of animal tissues critical to the study.