Abnormal Rab11-Rab8-vesicles cluster in enterocytes of patients with Microvillus Inclusion Disease

2.1. Clinical findings & Genetics

Patient 1 (male; MYO5B) had intermittent episodes of intractable diarrhea and poor growth from the first few months of life. MVID was diagnosed in his 5-year old sister during her first year of life, and the ultrastructural analyses of their diagnostic intestinal biopsies were suggestive of MVID. These affected siblings were born to consanguineous Asian parents. The affected sister could be weaned from total parenteral nutrition during her 4^th year of life. Sanger sequencing of the complete coding region (40 exons) and all flanking exon-intron boundaries of the MYO5B gene encoding Myosin 5b was performed in the index patient, i.e., the older affected sibling of patient 1, revealing a homozygous MYO5B mutation c.244G>A (p.Glu82Lys) in exon 3. Targeted sequencing of exon 3 in family members revealed the same mutation in homozygous state in patient 1 and in heterozygous state in both healthy parents. Results are based on NCBI mRNA reference sequence NM_001080467, in which the A of the ATG translation initiation codon was nucleotide 1. Primers are listed in Supplementary Table S1. The c.244G>A (p.Glu82Lys) mutation is very rare; it is not listed in public databases including the Exome aggregation consortium database, EXAC. The p.Glu82Lys mutation affects a highly conserved amino acid, and is a non-conservative amino acid substitution, which is likely to impact secondary protein structure, as these residues differ in polarity, charge, and size, all of which is supported by in-silico analyses (SIFT, PolyPhen2). The identified MYO5B missense mutation might still preserve residual Myo5b protein function.

Patient 2 (male; MYO5B) is the second child of healthy Asian patients who were not aware of any consanguinity between them, and he has a healthy sister. He started to have an intractable secretory diarrhea from birth, requiring total parenteral nutrition. An analysis of the whole coding region and all splice-sites of MYO5B revealed homozygosity for a c.414C>A (p.His138Gln) mutation in exon 4. The c.414C>A (p.His138Gln) mutation is very rare; it is not listed in public databases including the Exome aggregation consortium database, EXAC. The p.His138Gln mutation affects a highly conserved amino acid, and is a non-conservative amino acid substitution, which is likely to impact secondary protein structure, as these residues differ in polarity and size, all of which is supported by in silico analyses (SIFT, PolyPhen2).

Patient 3: MYO5B c.1323–2A>G was described previously (4, 17).

Patient 4: STX3 c.372_373dup (p.Arg125Leufs*7) was described previously (6).

2.2. TEM-morphology and PAS-cytochemistry

In the biopsies from all four MVID patients distinct accumulations of ~100 to 400nm-wide, pleomorphic endomembrane compartments occupied the subapical cytoplasm of a high percentage of enterocytes (Fig. 1A). Three-dimensional (3D) reconstruction using electron tomography confirmed mainly isolated vesicles, as well as partly branching, interconnected tubular compartments (Fig. 1B, C; Movie S1). These endomembrane aggregates reached frequently from the perinuclear area to the plasma membrane, or the terminal web, where still present. Neighboring, unaffected enterocytes and controls showed scattered vesicles and short tubules, likely early and recycling endosomes and/or exocytic vesicles. Similar patterns were seen in genome-edited and control CaCo2 cell models (see Fig. S1D in Ref. (15)).

In conventionally stained plastic- and cryo-sections the contents of the vesicles/tubules appeared homogeneously electron-dense or -lucent (Fig. 1A, B). Locally, dense compartments were continuous with lucent ones, including parts with mixed contents (Fig. 1A). Dense vesicles/tubules were found preferentially, though not exclusively, in the crypts (Fig. 1B). Lucent ones occurred all along the villi, where they frequently surrounded also microvillus inclusions, but also sporadically in the crypts. The relative abundance of dense and lucent vesicles/tubules varied throughout the samples and between the patients. Genome-edited CaCo2 cell models showed predominantly lucent vesicles/tubules (see Fig. 1G in Ref. (15)).

Consistent with previous studies (12), most of these vesicles/tubules proved PAS-positive (Fig. 1C, D, E); and so did late endosomes/multivesicular bodies, the huge, quite heterogeneous population of degradative compartments (lysosomes, autophagosomes; (Fig. 1D, Figure S1D)), the glycocalyx and substances contained within the lumina of microvillus inclusions (Figure S1B). PAS-staining intensity of the vesicles/tubules was not uniform, but PAS-reactive material was also detectable within the lucent vesicles/tubules thanks to special preparation and incubation protocols (Fig. 1E, F; for further details see the Materials and methods paragraph). In control biopsies, we merely observed a few dense, ~200 nm wide, PAS-positive vesicles (and occasionally short tubules) in the apical cytoplasm of immature crypt enterocytes (Fig. 1G).

Taken together, the vast majority of the subapically clustering vesicles and tubules in the here presented MVID patients and genome-edited human CaCo2 cell lines were identical to structures historically termed “secretory granules”, as originally described from enterocytes of MVID patients (1, 2).

The other ultrastructural hallmarks of MVID are shortened and/or lacking brush border microvilli, as well as microvillus inclusions and/or basolateral microvilli. Notably, these features were rare or possibly even absent in the newly described patients 1 and 2 with mutated MYO5B (based on data available so far; Figure S1A–C), but prominent in MYO5B patient 3 (see Fig. 4A in ref. (17)). Patient 4 with a STX3 mutation showed few microvillus inclusions, but consistently basolateral microvilli (see Fig. 1D in Ref. (6). Denuded enterocytes occurred regularly in this patient, not only along the villi, but prominently also in the crypts (Figure S1D); such features have so far not been reported in any case of MVID, but could be also observed in two other, yet undescribed cases with STX3 mutations (our unpublished observations). Table 1 summarizes the varying subcellular phenotypes in enterocytes from the MVID patients and in the cell models presented here. In addition, drawings (Figure 2) illustrate in a generalized manner the predominant ultrastructural features of MVID enterocytes as compared to healthy controls.

2.3. Immunolabeling

To characterize the enigmatic vesicles/tubules and attribute them to a well-defined class of organelles we performed immunogold labeling on thawed cryo-sections (complemented with IF and pre-embedding IEM for patient 2). Biopsies and CaCo2 lines were probed with antibodies against membrane proteins that control proper trafficking of apical cargo in polarized epithelia, but show unorthodox, subapical IF-signals in MVID tissues and models. Amongst them, the small GTPase Rab11 appeared as a highly promising candidate (17–19). Indeed, in MVID samples and all genome-edited cell models presented here Rab11 was consistently seen at the membranes of subapically accumulating vesicles/tubules (Fig. 3A–C). To a minor extent Rab11 could also be detected in the cytoplasm, at the Trans Golgi network (TGN), brush border remnants and lateral plasma membrane domains (Tab. 2). Three different anti-Rab11/Rab11a antibodies gave almost identical results. Similarly, another key player in targeted transport of exocytic vesicles, Rab8a, also occurred regularly at the membranes of the subapical vesicles/tubules (Fig. 3D, E), in addition to its canonical distribution (Tab. 2). In the controls these markers were confined to their typical locations as reported from IEM (20–23) and IF-studies.

As a member of the Rab11-Rab8 cascade the t-SNARE Stx3 mediates fusion of exocytic vesicles with the apical plasma membrane (7, 24–26). In samples lacking functional Myo5b, Stx3 was consistently found associated with subapical vesicles/tubules (Fig. 4A, B) and throughout the subapical cytoplasm, together with only moderate labeling of the apical plasma membrane (Fig. 4B). In the controls, Stx3 was located predominantly at the brush border, whereas the STX3-mutant biopsy and the respective genome-edited CaCo2 cell models showed no labeling at all (Tab. 2).

Furthermore, of particular interest was the distribution of apical transporters, since mislocalization of NHE3 and CFTR had been reported from intestinal epithelia of MVID-patients (19, 27). In biopsies of duodenum from our MVID patients, anti-NHE3 strongly labeled lucent and dense vesicles/tubules (Fig. 4C, D), the endoplasmic reticulum, and to a minor extent, the brush border (Tab. 2). Similar results were obtained with HA StrepTactin II (HS) tagged NHE3 and CFTR expressed in the genome-edited CaCo2 models (e.g., Figure S3A, B). Controls showed normal distribution of NHE3 (and HS-CFTR). Double labeling experiments complemented these observations (Fig. 3E, Figure S2A, B), though the respective labeling densities were considerably lower than with one marker only. Occasionally, immunogold label of cargo (endogenous or tagged), Stx3 or Rab8a on the one hand, and Rab11a on the other hand, was apparently located at the very same, distinct structural entity (i.e., one and the same vesicle/tubule; e.g., Figure S2A).

Markers for late endosomes and (autophago)lysosomes specifically labeled their respective compartments (e.g., Lamp1 (Lysosomal acid membrane protein1): Fig. 4E; LBPA (Lysobisphosphatidic acid), or p62/SQSTM1 (sequestosome-1): data not shown). Subapical vesicles/tubules, however, remained generally unstained (Fig. 4E), except for very few cases where dense tubules were evidently connected with Lamp1-positive organelles.

In addition, we performed immunofluorescence microscopy for Rab11, Stx3 and NHE3 (Figure 5) as well as pre-embedding IEM for Rab11 (not shown) with biopsies from patient 2 because of limited material. The data generally supported our IEM findings and were consistent with the literature.

Finally, biopsy specimens from patient 4 with mutated STX3 provided the unique opportunity to document the IEM localization of endogenous Myo5b in the vicinity of vesicular compartments. Whereas control biopsies showed a rather inconspicuous distribution of Myo5b throughout the (sub)apical cytoplasm and at the apical plasma membrane, Myo5b was regularly found at membranes of the vesicles and tubules in enterocytes of patient 4 (Fig. 6A, B). Enterocytes from MVID patients with MYO5B mutations were devoid of label, except for some negligible background that could result from aberrant expression of Myo5b mutants (Figure 6C). Similar data were obtained from a Myo5b-rescue cell line (Myo5b KI HS-Myo5b: (15)). HS-Myo5b was found close to (Rab11-positive) vesicular/tubular compartments in the cell periphery (Figure S3C, D), at the apical plasma membrane in the vicinity of Stx3, and throughout the (sub)apical cytoplasm.

4.1. Human biopsy samples and Genetics

Written informed consent from the patients’ parents was obtained and snap biopsies from duodenal, jejunal, and colonic sites were obtained. Written informed consent for testing of MVID genes or genome-wide testing was obtained from the patients’ parents following genetic counseling. The study was approved by the Ethics committee of the Medical University of Innsbruck (study no. UN3987) according to the principles outlined by the Helsinki declaration.

Genomic DNA from leucocytes was extracted using the MagAttract DNA Mini M48 Kit (Qiagen, Hilden, Germany). PCR and sequencing was performed as reported (5).

4.2. Antibodies

The following primary antibodies were used: rabbit anti-Rab8a (56), rabbit anti-Rab11 (71-5300, Zymed), rabbit anti-Rab11a (VU57: (69)), mouse anti-Rab11a (8H10: (70)), rabbit anti-Syntaxin3 (ab133750, Abcam), rabbit anti-NHE3/SLC9A3 (HPA036669, Sigma), mouse anti-LAMP1 (H4A3, Developmental Studies Hybridoma Bank), mouse anti-LBPA (courtesy by Jean Gruenberg), guinea pig anti-p62 (PROGEN), chicken anti-Myosin 5b (71), mouse anti-HA.11 (clone 16B12, Covance MMS-101R-500).

4.3. Immunofluorescence microscopy (IF)

Sample preparation of natively frozen tissue and microscopy have been described previously (15, 17).

4.4. Immunogold-EM of thawed cryo-sections (IEM)

We generally processed biopsy and CaCo2-cell culture samples as previously described (72), taking advantage of the improved ultrastructure preservation achieved with a slight modification of the method published by Pujol and colleagues (73). In particular, tissues/cells were pre-fixed with 4% (w/v) formaldehyde dissolved in 0.1M phosphate buffer (pH 7.2; 24 hrs up to several days (including sample shipment); RT), followed by post-fixation with 4% (w/v) formaldehyde in 0.1M sodium borate buffer (pH 11, 1hr; RT). Subsequent sample processing and indirect single and double labelling was performed according to standard protocols (74, 75). Bound primary antibodies were visualized by relevant secondary antibodies (IgG molecules) conjugated to colloidal gold (Fig. 6) and/or by secondary antibodies (Fab’ fragments) conjugated to Nanogold (Nanoprobes; Figs. 3, 4 S2, S3). For double labelling we applied primary antibodies simultaneously, followed by detection with Nanogold-conjugates and 10nm-colloidal gold conjugates (in that order; Figs. S2, S3D). Nanogold conjugates, and optionally colloidal gold conjugates were subjected to silver enhancement (HQ-Silver, Nanoprobes).

4.5. Pre-embedding IEM

Frozen tissue (patient #2) was freeze-substituted and rehydrated (76) followed by permeabilization with Triton X (0.05%, 30min at RT), indirect immunogold labelling (using Nanogold-conjugates), silver enhancement, en-bloc staining with OsO4 and UA and, finally, epoxy resin embedding (77).

4.6. PAS cytochemistry at EM-level

Standard incubations were performed according to Thiéry (16) as described (36). In case of weak staining (e.g., Fig. 1E) incubation with periodic acid (PA) and thiocarbohydrazide (TCH) was optionally extended up to 15hrs each (16), and/or silver proteinate treatment (SP) was carried out at +50°C (78). Alternatively, formaldehyde-fixed biopsies were subjected to high-pressure freezing, freeze-substitution and resin embedding as described for native cell samples (15), followed by standard incubations (60 min PA, 120 min TCH, 30 min SP at RT: (16)) – an approach, that yielded considerably intensified PAS-reaction product (Fig. 1F).

4.7. Electron microscopy and electron tomography

EM and tomography were described previously (15, 72).