Skip to main content

Remote homology searches identify bacterial homologues of eukaryotic lipid transfer proteins, including Chorein-N domains in TamB and AsmA and Mdm31p



All cells rely on lipids for key functions. Lipid transfer proteins allow lipids to exit the hydrophobic environment of bilayers, and cross aqueous spaces. One lipid transfer domain fold present in almost all eukaryotes is the TUbular LIPid binding (TULIP) domain. Three TULIP families have been identified in bacteria (P47, OrfX2 and YceB), but their homology to eukaryotic proteins is too low to specify a common origin. Another recently described eukaryotic lipid transfer domain in VPS13 and ATG2 is Chorein-N, which has no known bacterial homologues. There has been no systematic search for bacterial TULIPs or Chorein-N domains.


Remote homology predictions for bacterial TULIP domains using HHsearch identified four new TULIP domains in three bacterial families. DUF4403 is a full length pseudo-dimeric TULIP with a 6 strand β-meander dimer interface like eukaryotic TULIPs. A similar sheet is also present in YceB, suggesting it homo-dimerizes. TULIP domains were also found in DUF2140 and in the C-terminus DUF2993. Remote homology predictions for bacterial Chorein-N domains identified strong hits in the N-termini of AsmA and TamB in diderm bacteria, which are related to Mdm31p in eukaryotic mitochondria. The N-terminus of DUF2993 has a Chorein-N domain adjacent to its TULIP domain.


TULIP lipid transfer domains are widespread in bacteria. Chorein-N domains are also found in bacteria, at the N-terminus of multiple proteins in the intermembrane space of diderms (AsmA, TamB and their relatives) and in Mdm31p, a protein that is likely to have evolved from an AsmA/TamB-like protein in the endosymbiotic mitochondrial ancestor. This indicates that both TULIP and Chorein-N lipid transfer domains may have originated in bacteria.


Both eukaryotic and bacterial cells require lipids to be distributed away from the site of their synthesis by lipid transfer proteins [1, 2]. There are 24 different protein superfamilies, each with a different fold that forms a tube, bridge or shuttle that can transfer bilayer lipids [3, 4]. Each fold creates a hydrophobic cavity, which lowers the activation energy for extraction of lipid from a bilayer to facilitate lipid traffic. To date ~ 50% of the lipid transfer protein families are known to have bacterial members, which can indicate ancestral modes of action of the eukaryotic proteins [3].

TUbular LIPid binding (TULIP) lipid transfer domains were discovered 20 years ago in human serum proteins such as bactericidal/permeability increasing protein (BPI) [5]. Two homologous conical TULIP domains 6 nm long, ≤2.5 nm in diameter dimerize head-to-head to create an elongated cylinder with two hydrophobic pockets (Fig. 1a), which are sometimes continuous (Fig. 1b) [5, 7,8,9,10]. SMP domains (standing for synaptotagmin, mitochondrial and lipid binding proteins) are TULIP homologues initially predicted by structural bioinformatics [11], since confirmed crystallographically [7, 12,13,14,15,16]. Three families of TULIP-like proteins have been identified in bacteria. The E. coli protein YceB and its homologues have the same fold as eukaryotic TULIPs [9, 17,18,19], though its core is shorter than any other solved TULIP structure (Fig. 2a). This domain in YceB is described as DUF1439, where “DUF” stands for domain of unknown function. It follows an N-terminal lipid-anchor predicted to be attached to the extra-cellular side the intermembrane space [19, 20]. TULIP-like proteins have also been found among the OrfX cluster associated with botulinum neurotoxin serotypes-E/F [9, 10]. OrfX2 and P47 (plus closely related OrfX3) have 2 TULIP-like domains, though dimerization is side-by-side, not head-to-head. According to PFAM, P47 has homologues across divergent bacterial phyla and in some fungi.

Fig. 1

TULIP domains form elongated cylindrical dimers with lateral openings leading to hydrophobic pockets. a Cartoon of head-to-head TULIP dimers. The two slightly conical TULIP domains contain a pocket that accommodates 1 or 2 lipids, with surface openings through which hydrophilic headgroups project. b Structure of Cholesteryl Ester Transfer Protein (CETP, 4EWS_A), showing transparent protein (N- and C- terminal TULIPs coloured blue and red), and space-filling spheres of four bound lipids [6]. c Side and end views of the ribbon structure of SMP domain in Extended Synaptotagmin-2 (4P42), showing the central super-helical β-meander (yellow) that is flanked by partially helical elements (blue preceding and red following) [7]

Fig. 2

YceB may dimerize similarly to eukaryotic TULIPs. a YceB dimer in 3L6I crystals, with solid surfaces of chains A (light blue) and B (pink), with histidines used to tag and purify the domain shown as spheres (2 in chain A = blue, 3 in chain B = red) [17]. b Two views of the Cα backbone of 3L6I_A with the TULIP core in light blue and accessory elements preceding and following the core in dark blue, showing side chains only for the histidines introduced for tagging. c β-meander for the dimerization interface between the two domains of BPI (bright and dull yellows) aligned with two copies of the three β-strands in YceB (one blue, one red). d Two views of a model YceB dimer, made by omitting the short C-terminal helix, positioning two monomers into the BPI dimerizing interface shown in (c)

Chorein-N is a domain of ~ 110 residues found at the N-terminus of two related, long, universal eukaryotic proteins: VPS13 (3000–7000 residues) and ATG2 (1000–2000 residues) [4, 21, 22]. Like SMP TULIP domains, Chorein-N domains appear from combined structural and biochemical studies to transfer lipids between closely apposed organelles [4, 21]. The name Chorein-N comes from the domain first being described in human VPS13A, which has the alternate name “Chorein” because VPS13A mutations cause a choreiform syndrome [23].

Here bacterial TULIP-like proteins and eukaryotic Chorein-N domains have been used in remote homology searches for bacterial homologues. Among three new TULIP families, one has precisely the same dimerization interface as found in BPI. Newly described bacterial TULIPs were analysed for likely targeting signals to suggest the parts of bacterial cells where they bind and transfer lipids. Chorein-N domains were found in bacteria at the N-terminus of the closely related intermembrane space proteins AsmA and TamB, and also in Mdm31p in fungal mitochondria, which is presumed to have evolved from a common ancestor with AsmA/TamB. An additional Chorein-N domain is at the N-terminus of one of the bacterial TULIPs. All of the bacterial TULIPs and Chorein-N proteins are therefore implicated to be lipid transfer proteins. Finding both domains in bacteria and their derivatives indicates that they originated in bacteria before eukaryote evolution.


YceB may dimerize like eukaryotic TULIPs

The unit cell of the YceB crystal deposited at PDB (3L6I) contains a head-to-head dimer of TULIP-like domains. However, the dimeric interface includes a polyhistidine tag, which creates a sharp bend in the dimer (114°) (Fig. 2a), compared to BPI and homologues which are far straighter (155°) (Fig. 2b) [5]. We wondered if the dimeric interface is affected by the tag. 3L6I is extended by three strands and a short helix (Fig. 2a). Omitting the helix, we found that the strands make a dimerization β-meander similar to that of BPI (Fig. 2c), potentially allowing dimerization similar to eukaryotic TULIPs (Fig. 2d). Thus, it is possible that head-to-head dimerization as in BPI is also found in YceB, but that the His-tag added to YceB for purification altered the dimerization interface.

HHsearch predicts TULIP domains in bacterial proteins DUF4403, DUF2140 and DUF2993

We seeded HHsearch with the known bacterial TULIPs: P47, OrfX2 and YceB, searching for homologous families. HHsearch predicts homology by comparing multiple sequence alignments (MSAs) of the query with libraries of MSAs covering PDB, Pfam and proteomes of model organisms. P47 and OrfX2 produced no significant hits, except to each other (ppSS = 100% over 389 residues, e-value from sequence alone = 1e− 50). Searches with YceB produced significant hits to four regions in three bacterial proteins families (Fig. 3a, column 1).

Fig. 3

Homologies between YceB and 4 other bacterial TULIP domains. a All vs. all pairwise alignments of bacterial TULIP YceB (#1) with four newly proposed bacterial TULIPs (#2–5): DUF4403 N and C-terminus; DUF2140; and Rv0817c, a DUF2993 protein in M. tuberculosis. The arthropod protein Takeout (#6) is also included. The strength of alignments is colour coded according to the predicted probability of shared structure (ppSS, large numbers), with the number of aligned columns shown in brackets. b Arrangement of secondary structural elements predicted in DUF4403. Each TULIP domain core is a superhelix (coloured in rainbow order blue to red), and three additional β-strands: one before the core 1 (“-1” - white) and two after the core (“+ 1” & “+ 2” - black). A loop links the two domains (gray). c I-TASSER model of DUF4403 (residues 17–423), each core domain rainbow coloured (N = blue <− > C = red), and interface strands coloured as in B (“-1” white, “+ 1” “+ 2” black). d DUF4403-N and -C structures predicted by contact co-evolution in GREMLIN [24, 25]. Colouring as in c. This method predicts helices immediately following the core of both domains. e & f DUF2140 and DUF2993C predicted and coloured as in d. In both cases, the initial α-helix and β-strand are short. DUF2993C includes the final 15 residues of Rv0817c, beyond the end of DUF2993, predicted as an extra short strand and helix

The DUF4403 family has two predicted TULIPs in what is classed as a single domain of ~ 460 aa. DUF4403 is a full length homologue of BPI, with the same combined pattern of α/β elements (Fig. 3b). Both DUF4403 TULIP domains were predicted to have three additional β-strands per domain to form a β-meander interface that precisely replicates BPI or CETP [5]. I-TASSER modelled the DUF4403 domain as a full length homologue of pseudodimeric BPI (Fig. 3c) [26]. Ab initio folding by GREMLIN of each half of DUF4403 (see Methods) confirmed the prediction, and added a helix to the dimer interface of both domains (Fig. 3d).

DUF2140 family domains, which are found exclusively in monoderm organisms such as Bacilli (one per species), aligned strongly with DUF4403 (Fig. 3a). GREMLIN made a TULIP-like model, complete with dimerization interface for DUF2140, although the domain starts off with a shorter helix and shorter first β-strand than eukaryotic TULIPs (Fig. 3e).

Proteins with DUF2993 occur in actinobacteria, cyanobacteria and some firmicutes (2 or 3 per species). DUF2993 domains are also reported by PFAM in some mosses and algal picoplankton, presumably in plastids derived from an ancestral cyanobacterium. Among the three family members in M. tuberculosis, Rv0817c and Rv0479 are essential for growth, but Rv3243c is not [27, 28]. The proteins consist of ~ 270 aa, with a predicted signal sequence followed by DUF2993 (~ 230 aa) then 15 aa at the extreme C-terminus. The C-terminal of Rv0817c (residues 121–255) showed strong TULIP-like homology, aligning with both DUF2140, and YceB (Fig. 3a, row #5). The core of the TULIP in DUF2993 is shorter than other TULIPs, even shorter than DUF2140 (Fig. 3f and Additional file 1: Figure S1).

Overall, these results suggest that TULIP domains are widespread in bacteria, both monoderms and diderms, most being able to dimerize by the same mechanism as BPI, and one having the same pseudodimeric form as BPI. The degree of homology detectable by HHsearch is low between known eukaryotic TULIPs and the previously known bacterial TULIP domains P47/OrfX2 and YceB (Fig. 4, columns/rows 5/6). In contrast, applying this tool to link eukaryotic TULIPs and the newly described bacterial TULIPs produced stronger homologies (Fig. 4, columns/rows 7–10). DUF4403-C and Juvenile Hormone Binding Protein (JHBP, the family of arthropod TULIPs that includes Takeout) aligned strongly (Fig. 4, red outlines). The single strongest eukaryotic–bacterial homology was with DUF4403-N as query and BPI-N as target (ppSS 63% over 167 residues, Fig. 4, row 1). Note that the tool is asymmetric, and this hit is cited rather than the weaker one carried out in reverse (Fig. 4, column 1) [11, 29]. Overall, the strength of alignments between BPI-N/JHBP and DUF4403-N were stronger than between BPI-N/JHBP and either BPI-C or SMP (Fig. 4). This shows the significance of the alignment between DUF4403 and eukaryotic TULIPs.

Fig. 4

All–vs.–all HHsearches for TULIPs. MSAs were made from 3 rounds of PSI-BLAST of each of four eukaryotic TULIPs (BPI-N and BPI-C, Juvenile Hormone Binding Protein (JHBP) from arthropods and the SMP domain of Nvj2p) and six bacterial proteins (P47/OrfX2, YceB/DUF1439, DUF4403-N and -C, DUF2140 and DUF2993). Hydrophobic signals were removed from BPI, DUF4403, YceB, and DUF2140, and the whole N-terminus of DUF2993 was removed. BPI and DUF4403 MSAs were divided into N- and C-terminal domains. For each of the 10 MSAs, the HHpred server returned 1000 hits from PFAM (v32.0). The diagram shows the ppSS (and number of aligned columns) of the highest target in each of the 10 PFAM families for the ten domains. Hits are colour-coded according to ppSS as indicated. “0” indicates ppSS < 0.5%, or no match in the top 1000 hits. The search with P47 matched both halves of the DUF4403 pseudodimer. The results between two domains are in many cases asymmetric depending on which is query and which template, which is caused by HHsearch making query and target templates differently, as reported previously [11, 29]. The strongest homology between eukaryotic and bacterial proteins occurred with (i) the combination of JHBP < > DUF4403-C (either as query; outlined in red) and (ii) DUF4403-N as query with BPI-N as target, outlined in blue

The N-terminus of DUF2993, AsmA and TamB is homologous to the N-terminus of VPS13

Next, we looked in detail at the N-terminus of DUF2993, using Rv0817c as an example. Between a predicted N-terminal transmembrane domain embedded in the inner membrane (1–25) and the TULIP domain, there is a α/β region of 100 residues (Fig. 5a). HHsearch indicated that the most homologous protein families for this region are the unstudied DUF2125 and the “Chorein-N” domain (Fig. 5b), recently crystallised for the first time and shown to form a hydrophobic scoop that forms a cavity that can transfer lipids [4, 21]. Searches for bacterial homologues of Chorein-N identified homology not only to the N-terminus of DUF2993, but far more significant homology to the N-termini of two E. coli proteins on the intermembrane space side of the inner membrane: AsmA (assembly suppressor mutation A) and TamB (translocation and assembly module-B) (Fig. 5c). With a ppSS > 90% over the majority of residues in the domain, these hits strongly indicate homology between Chorein-N and AsmA/TamB [29]. The region of fungal Mdm31p previously identified as homologous to AsmA/TamB [30] is also moderately homologous to Chorein-N, although it shows no homology with the N-terminus of DUF2993 (Fig. 5c).

Fig. 5

Homologies between N-termini of DUF2993 AsmA, TamB and Chorein-N. a Secondary structural elements in residues 1–124 of the M. tuberculosis DUF2993 protein Rv0817c, also showing the confidence of the structural prediction by PSI-PRED (range 0–9, transmembrane domain excluded). b HHpred hits in PFAM for the N-terminus of Rv0817c. After identifying itself, the next hit is DUF2125 (ppSS = 56%, 57 columns), and the third hit is Chorein-N (ppSS = 45%, 37 columns). c All vs. all pairwise alignments of the N-terminus of Rv0817c, Chorein-N, Mdm31p (residues 131–312) and the N-termini of AsmaA and TamB (130, 115, 182, 180 and 150 residues respectively). A second alignment for TamB has a profile made after just one round of BLAST (bottom line) to exclude any AsmA sequences, which mix with TamB sequences with multiply iterated PSI-BLAST searches. Colour coding as Fig. 3a. d Secondary structural elements in residues 1–95 of the Chorein-N domain (from crystal structure 6CBC – 100% confidence). e and f Secondary structural elements in the N-termini of AsmA (1–120 aa) and TamB (1–125 aa), with confidence of the prediction as in a

The structural elements in Chorein-N and the N-termini of AsmA and TamB (Fig. 5d/e/f) match those in Rv0817c (Fig. 5a) and Mdm31p (data not shown). Homology between AsmA TamB and Mdm31p was previously demonstrated using HMMER, which is less sensitive than HHsearch [31,32,33]. This homology is so close that the 3rd iteration of PSI-BLAST seeded with TamB includes proteins annotated as AsmA (data not shown). AsmA/TamB have multiple other homologues in the intermembrane space, for example four in E. coli: YicH, YhjG, YdbH and YhdP [33]. Mdm31p has the analogous localization, in the intermembrane space of mitochondria, being found mostly in fungi [34], and some homologues in amoebae (data not shown).

The most significant finding here is that the Chorein-N lipid transfer domain is present in the N-termini of the bacterial proteins AsmA, TamB and related proteins in the intermembrane space, as well as in the N-terminus of Rv0817c.

Location of predicted bacterial TULIPs and Chorein-N domains

After predicting the existence of new bacterial TULIP domains and hydrophobic scoops similar to Chorein-N, a major question is whether their cellular locations will allow them to transfer lipids. DUF2140 is the only new domain in monoderms (Gram +ve), some of which also express OrfX proteins. DUF2140 is predicted to be N-terminally anchored to the outside of the cell, with an unstructured ~ 20 aa linker before the TULIP. The function of these domains cannot be to transfer lipids between distinct membranes. Instead, they may capture or export specific lipids (Fig. 6a).

Fig. 6

Predicted locations of potential bacterial lipid transfer proteins. Locations predicted from analysis of hydrophobic N-termini and acylation signals for: a DUF2140, TULIP in Gram +ve bacteria; b TULIPs in Gram –ve bacteria: DUF4403, YceB/DUF1439 and Rv0817c/DUF2993 (c) TamB in Gram –ve bacteria such as E. coli, with N-terminus homologous to Chorein-N and C-terminus forming a narrower hydrophobic tubular “β-taco”. All dimensions, including the inter-membrane space 20 nm across as in E. coli, lengths of TULIP domains, location and length of linkers, and membrane anchors are drawn to approximate scale, except the hydrophobic seams/openings, which are enlarged and drawn straight, compared to narrower helical openings found in nature. Shading: cytoplasm blue, inner membrane pink, intermembrane space yellow, outer membrane green. The three different types of domain with hydrophobic cavities are shown in a key at the bottom

TULIPs in diderms might be involved in intracellular lipid traffic between the inner and outer membranes. DUF4403 domains are preceded by a hydrophobic region and are predicted to be secreted into the intermembrane space. 66% are predicted to have this region cleaved, leaving the bulk of the protein anchored to the membrane by acylation [35]. Of these, the + 2 residue following the site of lipidation is only rarely aspartate (2%), suggesting that the large majority of DUF4403 domains are anchored in the outer membrane [36]. The linker sequence between this site and the folded domain is 20–25 aa, so the domain is unlikely to be able to stretch across the intermembrane space (Fig. 6b). YceB/DUF1439 is predicted to be localized similarly to DUF4403, 55% having a cysteine for acylation after processing of the hydrophobic region, and 95% of the + 2 residues indicating outer membrane residence. Since these proteins have much shorter linkers than DUF4403 (6–8 residues), they are predicted to localize even more closely to the outer membrane (Fig. 6b).

Rv0817c is like the majority of DUF2993 proteins, being predicted to be secreted into the intermembrane space and anchored in the inner membrane by a N-terminal transmembrane domain. The Chorein-N domain and TULIP are continuous without breaks and may dimerise like 2 TULIP domains (Fig. 5a). The short linker at its extreme N-terminus implies that Rv0817c is closely anchored to the inner membrane (Fig. 6b). The same localization has been demonstrated for TamB, and predicted for AsmA [37]. The domain structure of these proteins is predicted to consist of tubular domains of different diameters and length up to 16 nm (Fig. 6c).


Prediction of new TULIP domains

Four new TULIP domains in three families spread across many bacterial phyla were detected by HHsearch. Previously, TULIP domains have been discovered through many different methods, both experimental and bioinformatic. The first solved TULIP structure was BPI. This was revealed to be a symmetrical pseudodimer with its N and C-terminal domains having similar structure and topology [5]. The widely conserved direct apposition of the two domains indicates they duplicated from a single progenitor domain that formed homodimers. However, the two domains of BPI show no detectable similarity by PSI-BLAST and only minor homology by HHsearch (ppSS = 35% over 105 residues), which indicates that even low levels of reported homology may be consistent with common origin. We note that several of the alignments detected between bacterial and eukaryotic TULIPs were stronger that this (Fig. 4).

One family of TULIPs, the SMP domains, was first discovered by prediction using HHsearch [11]. The homology that identified SMPs [11] is no stronger than that described here as linking eukaryotic TULIPs with the newly described DUF4403 family (Figs. 3a and 4). An alternative way to predict protein structure is contact folding, using a tool such as GREMLIN [24, 38]. Since contact folding has needed large numbers of sequences to sample enough co-evolution, bacterial proteins with homologues among metagenomes have been ideal. This method confirmed the presence of TULIP folds in all of DUF4403-N and C-termini, DUF2140 and DUF2993-C-terminus (Fig. 3c-f), acting as an independent verification of the structures predicted by HHsearch.

Origin of bacterial TULIPs

It is possible that the TULIP fold has evolved on more than one occasion by convergent evolution. This could explain the low level of homology between eukaryotic and bacterial proteins that has not been picked up previously, and it would place the eukaryotic and bacterial proteins in separate superfamilies, despite their similar overall 3D structure. Yet there is evidence from multiple sources that suggests divergent evolution. Firstly, there are multiple instances in the TULIP superfamily of low levels of sequence homology for proteins accepted as homologues, in particular the C-terminal domain of BPI, and SMP domains (Fig. 4). Secondly, features accompanying the core TULIP domain indicate that the overall unit evolved once from a common origin. The key observation here is that bacterial TULIP domains have three additional β-strands, one coded before the domain and two after, that form 6 stranded β-meanders for head-to-head dimerization. This is exactly the same arrangement as the dimerization interface of BPI and related proteins. This applies not only to DUF4403 (Fig. 3c), DUF2140 (Fig. 3e) and DUF2993 (Fig. 3f), but also possibly to YceB, after the possible artefact introduced by the polyhistidine tag in 3L6I is taken into account (Fig. 2). Another 6 stranded β-meander interface (with additional helices) is found between the TULIP domain of clostridial OrfX2 and a circularly permuted StART-like lipid transfer domain [10]. Importantly, such dimerization interfaces are not mandatory, as SMP domains have two other interfaces with either two or zero β-strands [3]. A parsimonious explanation is that the β-meanders evolved once alongside the TULIP, and spread into both bacteria and eukaryotes, with subsequent partial losses. To test such a prediction of divergence between structurally related families, the structure-sequence relationships must be studied in detail [39,40,41].

The TULIP domain is widespread in eukaryotes, so may have occurred first in early eukaryotes, in which case it then must have horizontally transferred at least 5 times into bacteria to explain all of YceB/DUF1439, DUF4403, DUF2140, DUF2993, P47/OrfX2. More parsimoniously, TULIP domains may have been present in bacteria prior to eukaryote evolution, with even more widespread losses since.

Bacterial TULIPs may need to cooperate with other lipid transfer proteins

The domain composition of the proteins containing TULIP domains may indicate their mode of action. DUF2140 and P47 are found in monoderms, where they may act to pick up lipids from the environment, as occurs for eukaryotic TULIPs such as lipopolysaccharide binding protein and Takeout [42, 43]. Alternatively they may detoxify cells by removing specific hydrophobic molecules [44]. In diderm bacteria such as E. coli and M. tuberculosis, the linkers between the membrane anchors and the TULIP domains of YceB (DUF1439) and DUF4403 are ≤25 aa, implying maximal extension ≤9 nm, and the domains themselves are not long enough to span the intermembrane space (≥20 nm) (Fig. 6b). So these TULIPs either do not carry out lipid transfer (acting similarly in diderms as in monoderms), or they cooperate with another lipid transfer protein in the intermembrane space. Such cooperation might involve lipid hand-off between two lipid transfer proteins anchored in the inner and outer membranes, similar to cholesterol hand-off between NPC2 and NPC1 in lysosomes [45].

TamB, AsmA, Mdm31p (and related proteins) are predicted to be lipid transfer proteins

HHsearch identified homology between Chorein-N domains and the N-termini of AsmA, TamB and Rv0817c, as well as Mdm31p (Fig. 5c). The solved structures for Chorein-N of both VPS13 and ATG2 have an α-helix closing off a half-tube with internal diameter 33 Å. The domain has a hydrophobic interior and can enclose the hydrophobic portions of multiple phospholipids [4, 21]. In both examples, the Chorein-N domain transfers phospholipids between membranes. The cavity formed in a model of the N-terminus of TamB based on Chorein-N from VPS13 (6CBC) has an entirely hydrophobic lining (data not shown). Thus, the homologies to Chorein-N predict that the N-termini of AsmA, TamB and Rv0817c (DUF2993) as well as Mdm31p are lipid transfer domains.

TamB, AsmA in bacteria and their relative Mdm31p in fungi have been studied to quite different extents to date. TamB is present in all diderms, [33]. It forms a long rod (16 nm long in E. coli) anchored by a single N-terminal transmembrane domain inserted in the cytoplasmic membrane (Fig. 6c) [46]. The C-terminus of TamB interacts with TamA and with the beta-barrel assembly machinery subunit A (BamA), both in the outer membrane [33, 47]. This has suggested that TamB functions in insertion of outer membrane proteins, e.g. autotransporters [48]. Contradicting this, TamB is found in species lacking autotransporters [33], and is not needed for cell-free reconstituted secretion of autotransporters [49]. AsmA is less studied, with the major finding being that mutation affects the folding of Omp85 β-barrels and membrane fluidity [37, 50, 51]. Mdm31p and its paralog Mdm32p in budding yeast are strongly linked to endoplasmic reticulum–mitochondrial lipid traffic [34, 52,53,54], but their function is unknown. Previously, an X-ray crystal structure has shown that part of the C-terminus of TamB forms a β-partly enclosed tube with a hydrophobic lining called a “β-taco” [30]. With an internal diameter of 24 Å, this is narrower than the Chorein-N domain. Given predictions that the β-taco repeats throughout most of TamB, the rod is mostly a long half-enclosed tube with a hydrophobic lining.

The identification of Chorein-N at the N-terminus of TamB suggests that it might be a carrier for lipids, rather other hydrophobic cargo such as hydrophobic polypeptide as previously suggested [30]. The same applies to AsmA, Mdm31p and related proteins. Since TamB and the other E. coli proteins are all too short to bridge across the intermembrane space [46], each one can only act as a lipid transfer protein if it hands-off lipids to other lipid transfer proteins in the outer membrane (Fig. 6).


Systematic searches have been carried out for bacterial homologues of two different lipid transfer domains previously only carried out in eukaryotes. For the TULIP superfamily, three widespread bacterial TULIP families were identified. Occurrence of the dimerization interface previously found in eukaryotic TULIPs suggests a unitary origin for the superfamily, most likely in bacteria. One bacterial TULIP family, DUF2993, are “Rosetta Proteins” [55]. Preceding a C-terminal TULIP, DUF2993 has at its N-terminus a domain related to Chorein-N, a eukaryotic lipid transfer domain recently identified in VPS13 and ATG2. A systematic search for bacterial proteins with Chorein-N identified homologues at the N-terminus of a large family of intermembrane space proteins in Gram –ve bacteria (AsmA, TamB) and their relative Mdm31p in fungal mitochondria. This implies that AsmA, TamB, Mdm31p and related proteins are involved in lipid traffic across the intermembrane space.



This was run online at the Tuebingen Toolkit using non-redundant target databases where clusters of similar sequences (sharing 70% identity or more) have been reduced to one sequence.

Pairwise HHsearch

Sequences used to seed searches are described in Table 1. HHsearch was run through the HHpred v3.0.0 online at the Tuebingen Toolkit [56]. Settings for MSAs were: 3 iterations with PSI-BLAST; e-value for inclusion 0.001; realign with Maximum Accuracy alignment algorithm – off; secondary structure scoring during alignment – on. Initial searches were of template libraries containing the most up-to-date PDB structures (October 5, 2018) and Pfam (versions 31 and 32, depending on when searches were carried out). PSI-BLAST (3 iterations) was chosen because of its bench-mark status in place of HHblits (3 iterations), which is the default choice in HHpred [57]. Use of MSAs made by HHblits tended to strengthen eventual alignments, but did not alter the categorization of alignments as strong or weak (data not shown). Representative MSAs were saved from these searches (numbers of sequences given in Table 1; all MSAs of bacterial proteins and Chorein-N, Takeout and Mdm31p are provided in Additional file 2). For Figs. 3a and 5c the representative MSAs were compared pairwise with the “Aligning two alignments” option of HHpred, where results are symmetrical, i.e. independent of which MSA is query and which is target. To split domains in DUF4403 and Rv0817c, and to remove N-terminal signal sequences, full-length MSAs were subdivided into sections using Jalview, deleting any entries with less than 20 residues. Details of all pairwise comparisons of MSAs are included in Additional file 3.

Table 1 Numbers of proteins in TULIP families under study

HHpred in PFAM

MSAs of ten proteins, 4 eukaryotic and 6 bacterial, were used to seed searches of PFAM. Creation of MSAs from all sequences was as above, with 3 PSI-BLAST iterations. The four eukaryotic domains included were: BPI (human sequence, divided into N-terminal domain with signal sequence removed) and C-terminal domain; arthropod TULIP (Circadian clock-controlled protein, A. echinatior F4WUM6); and SMP domain (from Nvj2p, S. cerevisiae). All ten MSAs were submitted to HHpred which reported the top 1000 hits in PFAM v32.0. Unlike the pairwise searches (above), these searches are not symmetrical [29], and where the results differ, we cite the stronger hit.

Other standard structural predictions

Secondary structural elements were predicted by PSI-PRED 3.0, which is called as a sub-routine within HHsearch. 3D structure was predicted by I-TASSER [26], and visualized in QtMG (CCP4MG). Intracellular targeting was determined by submitting sequence families to SignalP, PsortB, Secretome2.0 and by inspection for acylation sites. A structure of TamB-N (26–114) was made using the Modeller option after HHpred.

Folding through co-evolution of contacts (“GREMLIN”)

The overall process for making models from co-evolution intra-chain contacts, has been described previously [58]. The following overall pipeline is referred to as GREMLIN. In brief, the MSAs were generated using an iterative procedure including HHblits [57]. In cases where there were not enough sequences from UniProt, metagenomic sequences were used to enrich the MSA. The MSA was then fed to GREMLIN to detect co-evolving residue pairs [25]. Contact map alignment was then used to search against the PDB to find structural elements with similar contact patterns. Top hits were hybridized to create a pool of models. These models were then iteratively refined using two additional rounds of hybridization.

Availability of data and materials

HHsearch datasets as used by HHpred (online server) in this study can also be downloaded and analysed off-line, being available in the “hh-suite” repository, at



Assembly suppressor mutation A


Beta-barrel assembly machinery


Bactericidal/permeability increasing protein


Cholesteryl ester transfer protein


Domain of unknown function


Juvenile hormone binding protein


Multiple sequence alignment

ppSS :

Predicted probability of shared structure


Translocation and assembly module


Tubular lipid binding domain


  1. 1.

    Chiapparino A, Maeda K, Turei D, Saez-Rodriguez J, Gavin AC. The orchestra of lipid-transfer proteins at the crossroads between metabolism and signaling. Prog Lipid Res. 2016;61:30–9.

    Article  CAS  Google Scholar 

  2. 2.

    May KL, Silhavy TJ. Making a membrane on the other side of the wall. Biochim Biophys Acta. 2017;1862(11):1386–93.

    Article  CAS  Google Scholar 

  3. 3.

    Wong LH, Gatta AT, Levine TP. Lipid transfer proteins: the lipid commute by shuttles, bridges and tubes. Nat Rev Mol Cell Biol. 2018;20:85–101.

  4. 4.

    Kumar N, Leonzino M, Hancock-Cerutti W, Horenkamp FA, Li P, Lees JA, Wheeler H, Reinisch KM, De Camilli P. VPS13A and VPS13C are lipid transport proteins differentially localized at ER contact sites. J Cell Biol. 2018;217:3625–39.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. 5.

    Beamer LJ, Carroll SF, Eisenberg D. Crystal structure of human BPI and two bound phospholipids at 2.4 angstrom resolution. Science. 1997;276(5320):1861–4.

    Article  CAS  Google Scholar 

  6. 6.

    Liu S, Mistry A, Reynolds JM, Lloyd DB, Griffor MC, Perry DA, Ruggeri RB, Clark RW, Qiu X. Crystal structures of cholesteryl ester transfer protein in complex with inhibitors. J Biol Chem. 2012;287(44):37321–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. 7.

    Schauder CM, Wu X, Saheki Y, Narayanaswamy P, Torta F, Wenk MR, De Camilli P, Reinisch KM. Structure of a lipid-bound extended synaptotagmin indicates a role in lipid transfer. Nature. 2014;510(7506):552–5.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. 8.

    Kolodziejczyk R, Bujacz G, Jakob M, Ozyhar A, Jaskolski M, Kochman M. Insect juvenile hormone binding protein shows ancestral fold present in human lipid-binding proteins. J Mol Biol. 2008;377(3):870–81.

    Article  CAS  Google Scholar 

  9. 9.

    Gustafsson R, Berntsson RP, Martinez-Carranza M, El Tekle G, Odegrip R, Johnson EA, Stenmark P. Crystal structures of OrfX2 and P47 from a botulinum neurotoxin OrfX-type gene cluster. FEBS Lett. 2017;591(22):3781–92.

    Article  CAS  Google Scholar 

  10. 10.

    Lam KH, Qi R, Liu S, Kroh A, Yao G, Perry K, Rummel A, Jin R. The hypothetical protein P47 of Clostridium botulinum E1 strain Beluga has a structural topology similar to bactericidal/permeability-increasing protein. Toxicon. 2018;147:19–26.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Kopec KO, Alva V, Lupas AN. Homology of SMP domains to the TULIP superfamily of lipid-binding proteins provides a structural basis for lipid exchange between ER and mitochondria. Bioinformatics. 2010;26(16):1927–31.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. 12.

    AhYoung AP, Jiang J, Zhang J, Khoi Dang X, Loo JA, Zhou ZH, Egea PF. Conserved SMP domains of the ERMES complex bind phospholipids and mediate tether assembly. Proc Natl Acad Sci U S A. 2015;112(25):E3179–88.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  13. 13.

    Jeong H, Park J, Lee C. Crystal structure of Mdm12 reveals the architecture and dynamic organization of the ERMES complex. EMBO Rep. 2016;17(12):1857–71.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. 14.

    Jeong H, Park J, Jun Y, Lee C. Crystal structures of Mmm1 and Mdm12-Mmm1 reveal mechanistic insight into phospholipid trafficking at ER-mitochondria contact sites. Proc Natl Acad Sci U S A. 2017;114(45):E9502–11.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. 15.

    Lees JA, Messa M, Sun EW, Wheeler H, Torta F, Wenk MR, De Camilli P, Reinisch KM. Lipid transport by TMEM24 at ER-plasma membrane contacts regulates pulsatile insulin secretion. Science. 2017;355:eaah6171.

  16. 16.

    Kawano S, Tamura Y, Kojima R, Bala S, Asai E, Michel AH, Kornmann B, Riezman I, Riezman H, Sakae Y, et al. Structure-function insights into direct lipid transfer between membranes by Mmm1-Mdm12 of ERMES. J Cell Biol. 2018;217(3):959–74.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Kuzin AP, Neely H, Seetharaman J, Chen CX, Janjua H, Cunningham K, Ma L-C, Xiao R, Liu J, Baran MC et al: Crystal structure of the uncharacterized lipoprotein yceb from e. coli at the resolution 2.0a. northeast structural genomics consortium target er542. released Jan 26 2010:

  18. 18.

    Vance SJ, McDonald RE, Cooper A, Smith BO, Kennedy MW. The structure of latherin, a surfactant allergen protein from horse sweat and saliva. J R Soc Interface. 2013;10(85):20130453.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. 19.

    Wong LH, Levine TP. Tubular lipid binding proteins (TULIPs) growing everywhere. BBA Mol Cell Res. 2017;1864(9):1439–49.

    CAS  Google Scholar 

  20. 20.

    Tokuda H, Matsuyama S. Sorting of lipoproteins to the outer membrane in E. coli. Biochim Biophys Acta. 2004;1693(1):5–13.

    Article  CAS  Google Scholar 

  21. 21.

    Valverde DP, Yu S, Boggavarapu V, Kumar N, Lees JA, Walz T, Reinisch KM, Melia TJ. ATG2 transports lipids to promote autophagosome biogenesis. J Cell Biol. 2019;218(6):1787–98.

    Article  CAS  Google Scholar 

  22. 22.

    Chowdhury S, Otomo C, Leitner A, Ohashi K, Aebersold R, Lander GC, Otomo T. Insights into autophagosome biogenesis from structural and biochemical analyses of the ATG2A-WIPI4 complex. Proc Natl Acad Sci U S A. 2018;115(42):E9792–801.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. 23.

    Ueno S, Maruki Y, Nakamura M, Tomemori Y, Kamae K, Tanabe H, Yamashita Y, Matsuda S, Kaneko S, Sano A. The gene encoding a newly discovered protein, chorein, is mutated in chorea-acanthocytosis. Nat Genet. 2001;28(2):121–2.

    Article  CAS  Google Scholar 

  24. 24.

    Ovchinnikov S, Park H, Varghese N, Huang PS, Pavlopoulos GA, Kim DE, Kamisetty H, Kyrpides NC, Baker D. Protein structure determination using metagenome sequence data. Science. 2017;355(6322):294–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. 25.

    Kamisetty H, Ovchinnikov S, Baker D. Assessing the utility of coevolution-based residue-residue contact predictions in a sequence- and structure-rich era. Proc Natl Acad Sci U S A. 2013;110(39):15674–9.

    Article  PubMed  PubMed Central  Google Scholar 

  26. 26.

    Roy A, Kucukural A, Zhang Y. I-TASSER: a unified platform for automated protein structure and function prediction. Nat Protoc. 2010;5(4):725–38.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. 27.

    Griffin JE, Gawronski JD, Dejesus MA, Ioerger TR, Akerley BJ, Sassetti CM. High-resolution phenotypic profiling defines genes essential for mycobacterial growth and cholesterol catabolism. PLoS Pathog. 2011;7(9):e1002251.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. 28.

    Sassetti CM, Boyd DH, Rubin EJ. Genes required for mycobacterial growth defined by high density mutagenesis. Mol Microbiol. 2003;48(1):77–84.

    Article  CAS  Google Scholar 

  29. 29.

    Fidler DR, Murphy SE, Courtis K, Antonoudiou P, El-Tohamy R, Ient J, Levine TP. Using HHsearch to tackle proteins of unknown function: a pilot study with PH domains. Traffic. 2016;17:1214–26.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. 30.

    Josts I, Stubenrauch CJ, Vadlamani G, Mosbahi K, Walker D, Lithgow T, Grinter R. The structure of a conserved domain of TamB reveals a hydrophobic beta taco fold. Structure. 2017;25(12):1898–906 e1895.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. 31.

    Finn RD, Clements J, Eddy SR. HMMER web server: interactive sequence similarity searching. Nucleic Acids Res. 2011;39(Web Server issue:W29–37.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. 32.

    Soding J. Protein homology detection by HMM-HMM comparison. Bioinformatics. 2005;21(7):951–60.

    Article  Google Scholar 

  33. 33.

    Heinz E, Selkrig J, Belousoff MJ, Lithgow T. Evolution of the translocation and assembly module (TAM). Genome Biol Evol. 2015;7(6):1628–43.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. 34.

    Dimmer KS, Jakobs S, Vogel F, Altmann K, Westermann B. Mdm31 and Mdm32 are inner membrane proteins required for maintenance of mitochondrial shape and stability of mitochondrial DNA nucleoids in yeast. J Cell Biol. 2005;168(1):103–15.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. 35.

    Juncker AS, Willenbrock H, Von Heijne G, Brunak S, Nielsen H, Krogh A. Prediction of lipoprotein signal peptides in gram-negative bacteria. Protein Sci. 2003;12(8):1652–62.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. 36.

    Zuckert WR. Secretion of bacterial lipoproteins: through the cytoplasmic membrane, the periplasm and beyond. Biochim Biophys Acta. 2014;1843(8):1509–16.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. 37.

    Deng M, Misra R. Examination of AsmA and its effect on the assembly of Escherichia coli outer membrane proteins. Mol Microbiol. 1996;21(3):605–12.

    Article  CAS  Google Scholar 

  38. 38.

    Monastyrskyy B, D'Andrea D, Fidelis K, Tramontano A, Kryshtafovych A. New encouraging developments in contact prediction: assessment of the CASP11 results. Proteins. 2016;84(Suppl 1):131–44.

    Article  CAS  Google Scholar 

  39. 39.

    Murzin AG. OB (oligonucleotide/oligosaccharide binding)-fold: common structural and functional solution for non-homologous sequences. EMBO J. 1993;12(3):861–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. 40.

    Theobald DL, Wuttke DS. Divergent evolution within protein superfolds inferred from profile-based phylogenetics. J Mol Biol. 2005;354(3):722–37.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. 41.

    Remmert M, Biegert A, Linke D, Lupas AN, Soding J. Evolution of outer membrane beta-barrels from an ancestral beta beta hairpin. Mol Biol Evol. 2010;27(6):1348–58.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. 42.

    Wright SD, Ramos RA, Tobias PS, Ulevitch RJ, Mathison JC. CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science. 1990;249(4975):1431–3.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Hamiaux C, Stanley D, Greenwood DR, Baker EN, Newcomb RD. Crystal structure of Epiphyas postvittana takeout 1 with bound ubiquinone supports a role as ligand carriers for takeout proteins in insects. J Biol Chem. 2009;284(6):3496–503.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. 44.

    Choudhary V, Schneiter R. Pathogen-related yeast (PRY) proteins and members of the CAP superfamily are secreted sterol-binding proteins. Proc Natl Acad Sci U S A. 2012;109(42):16882–7.

    Article  PubMed  PubMed Central  Google Scholar 

  45. 45.

    Wang ML, Motamed M, Infante RE, Abi-Mosleh L, Kwon HJ, Brown MS, Goldstein JL. Identification of surface residues on Niemann-pick C2 essential for hydrophobic handoff of cholesterol to NPC1 in lysosomes. Cell Metab. 2010;12(2):166–73.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  46. 46.

    Shen HH, Leyton DL, Shiota T, Belousoff MJ, Noinaj N, Lu J, Holt SA, Tan K, Selkrig J, Webb CT, et al. Reconstitution of a nanomachine driving the assembly of proteins into bacterial outer membranes. Nat Commun. 2014;5:5078.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. 47.

    Iqbal H, Kenedy MR, Lybecker M, Akins DR. The TamB ortholog of Borrelia burgdorferi interacts with the beta-barrel assembly machine (BAM) complex protein BamA. Mol Microbiol. 2016;102(5):757–74.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. 48.

    Selkrig J, Mosbahi K, Webb CT, Belousoff MJ, Perry AJ, Wells TJ, Morris F, Leyton DL, Totsika M, Phan MD, et al. Discovery of an archetypal protein transport system in bacterial outer membranes. Nat Struct Mol Biol. 2012;19(5):506–10 S501.

    Article  CAS  Google Scholar 

  49. 49.

    Norell D, Heuck A, Tran-Thi TA, Gotzke H, Jacob-Dubuisson F, Clausen T, Daley DO, Braun V, Muller M, Fan E. Versatile in vitro system to study translocation and functional integration of bacterial outer membrane proteins. Nat Commun. 2014;5:5396.

    Article  CAS  Google Scholar 

  50. 50.

    Misra R, Miao Y. Molecular analysis of asmA, a locus identified as the suppressor of OmpF assembly mutants of Escherichia coli K-12. Mol Microbiol. 1995;16(4):779–88.

    Article  CAS  Google Scholar 

  51. 51.

    Xiong X, Deeter JN, Misra R. Assembly-defective OmpC mutants of Escherichia coli K-12. J Bacteriol. 1996;178(4):1213–5.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. 52.

    Osman C, Haag M, Potting C, Rodenfels J, Dip PV, Wieland FT, Brugger B, Westermann B, Langer T. The genetic interactome of prohibitins: coordinated control of cardiolipin and phosphatidylethanolamine by conserved regulators in mitochondria. J Cell Biol. 2009;184(4):583–96.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. 53.

    Miyata N, Goda N, Matsuo K, Hoketsu T, Kuge O. Cooperative function of Fmp30, Mdm31, and Mdm32 in Ups1-independent cardiolipin accumulation in the yeast Saccharomyces cerevisiae. Sci Rep. 2017;7(1):16447.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. 54.

    Miyata N, Fujii S, Kuge O. Porin proteins have critical functions in mitochondrial phospholipid metabolism in yeast. J Biol Chem. 2018;293:17593–605.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  55. 55.

    Marcotte EM, Pellegrini M, Ng HL, Rice DW, Yeates TO, Eisenberg D. Detecting protein function and protein-protein interactions from genome sequences. Science. 1999;285(5428):751–3.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. 56.

    Zimmermann L, Stephens A, Nam SZ, Rau D, Kubler J, Lozajic M, Gabler F, Soding J, Lupas AN, Alva V. A completely Reimplemented MPI bioinformatics toolkit with a new HHpred server at its Core. J Mol Biol. 2018;430(15):2237–43.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  57. 57.

    Remmert M, Biegert A, Hauser A, Soding J. HHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment. Nat Methods. 2012;9(2):173–5.

    Article  CAS  Google Scholar 

  58. 58.

    Ovchinnikov S, Park H, Kim DE, DiMaio F, Baker D. Protein structure prediction using Rosetta in CASP12. Proteins. 2018;86(Suppl 1):113–21.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references


Thanks to Dr. Sergey Ovchinnikov, Harvard, for discussions and for folding proteins through co-evolution of contacts (“GREMLIN”) to make the models used in Fig. 3. Thanks also to Dr. Louise Wong, UCL, for critical comments on the manuscript.


TPL was supported by grant # BB/M011801/1 from the Bioinformatics and Biological Resources Fund of the BBSRC (UK).

Author information




TPL carried out all analyses, except for producing the GREMLIN models in Fig. 3 (by Dr. Sergey Ovchinnikov – see Acknowledgment). TPL wrote the manuscript. The author read and approved the final manuscript.

Corresponding author

Correspondence to Timothy P. Levine.

Ethics declarations

Ethics approval and consent to participate

Not applicable

Consent for publication

Not applicable

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This manuscript was initially submitted to BMC Structural Biology on 14 November 2018 and was peer reviewed at that journal

Supplementary information

Additional file 1: Figure S1. Predicted structural elements in short bacterial TULIP domains. Comparison of conserved structural elements in DUF2993 (16 examples) with DUF1439 (2 examples, top) and DUF2140. (2 examples, bottom). Residues are coloured according to the ClustalX scheme. Insertions are shown with the number of residues (e.g. “+ 14”) . Deletions are indicated by a black bar across the missing residues. Most DUF2993s have either a very short second β-strand or it is absent. There are also deletions in the final loop. 109 aa of Rv0817c align with the 127 aa core of YceB, both of which are shorter than DUF4403-C (180 aa) and DUF4403-N (220 aa). In Rv0817c, a further 28 aa align with 38 aa of YceB that forms the β-dimerization interface. The reduction in residues in DUF2993’s TULIP derives from (i) deletions before and in the second β-strand, as in DUF2140; (ii) short loops, especially the final loop; (iii) reduced residues in the final helix (14 fewer), covering the same distance by forming an extended polypeptide.

Additional file 2. MSAs for pairwise comparisons in HHpred (see Table 1).

Additional file 3. Pairwise comparisons in HHpred, corresponding to results in Figs. 3a and 5c. Also, included for comparison: BPI_N vs BPI_C.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Levine, T.P. Remote homology searches identify bacterial homologues of eukaryotic lipid transfer proteins, including Chorein-N domains in TamB and AsmA and Mdm31p. BMC Mol and Cell Biol 20, 43 (2019).

Download citation


  • AsmA
  • Chorein-N domain
  • HHpred
  • Lipid transfer protein
  • Mdm31p
  • Translocation and assembly module-B (TamB)
  • Tubular lipid binding domain (TULIP)
  • VPS13