Introduction

Sclerosing pneumocytoma is an uncommon, but unique, neoplasm of the lungs. It occurs more frequently in middle-aged adults, and is predominant in females [1]. Although sclerosing pneumocytoma is generally considered a benign neoplasm, rare cases with regional lymph node metastases and local recurrence have been reported [2,3,4]. Diagnosis of sclerosing pneumocytoma can be challenging, especially on fine-needle aspiration and intraoperative frozen sections, and it is frequently misdiagnosed as lung adenocarcinoma [5, 6].

Sclerosing pneumocytoma has characteristic histopathological features. It usually displays a mixed solid, papillary, hemorrhagic, and/or sclerotic pattern, and contains two distinct cellular components, namely the round cells and the surface cells. Studies using laser microdissection and clonal analysis demonstrated that the round cells and the surface cells exhibit a uniform pattern of monoclonality, indicating that both the cell types likely originate from a common precursor [7, 8].

The histogenesis and pathogenesis of sclerosing pneumocytoma remain largely unknown at present. The molecular alterations in sclerosing pneumocytoma have been investigated in many studies. It has been shown that sclerosing pneumocytoma frequently harbors nonspecific microsatellite alterations in p16 and Rb tumor suppressor genes, but lacks alterations involving EGFR, HER-2, and K-RAS [9]. Besides, nuclear expression of β-catenin proteins is frequently seen in sclerosing pneumocytoma, indicating that Wnt/β-catenin pathway is involved in the pathogenesis. However, mutation in the exon 3 of β-catenin gene is rarely found in this tumor [10].

Recently, Jung et al. conducted a large-scale study using whole-exome sequencing to investigate genomic alterations in sclerosing pneumocytoma. They identified a high frequency of AKT1 point mutations (overall 31 of 68, 46%) and recurrent β-catenin mutations (overall 3 of 68, 4%) [11]. This provides a milestone in understanding the molecular pathogenesis of sclerosing pneumocytoma. However, around half of the cases lacked identifiable cancer-related mutations in their study. This result has been particularly intriguing to us. Because sclerosing pneumocytoma is a disease with distinct and relatively uniform histopathological features, we expected that there should be a common genetic alteration in the majority of cases. We, therefore, speculated that sclerosing pneumocytoma might harbor certain kinds of genetic changes, which are difficult to detect by standard whole-exome sequencing analysis, such as structural variations, large insertions/deletions, and alterations in repetitive DNA sequences or non-coding genomic regions [12]. Internal tandem duplication, which may involve duplication of several tens to several hundreds of nucleotides within the coding exons, is also a probable genetic alteration. Internal tandem duplication is notoriously difficult to detect by standard bioinformatics pipelines, and specialized bioinformatics tools are usually required to identify this unique genetic change [13,14,15].

In the present study, we investigated the molecular alterations in sclerosing pneumocytoma, with a primary focus on internal tandem duplication. Using Genomon-ITDetector, a software package designed specifically for internal tandem duplication detection from cancer genome sequencing data [15], we discovered recurrent internal tandem duplications in AKT1 gene in half of the cases. Moreover, we found that AKT1 internal tandem duplications were mutually exclusive of other forms of AKT1 mutations, including point mutations and short indels. Taking all forms of AKT1 mutations together, we were able to detect AKT1 mutations in nearly all the sclerosing pneumocytoma cases in our study. These results suggest that AKT1 mutations are the major molecular alterations in sclerosing pneumocytoma.

Materials and methods

Analysis of PRJNA297066 sclerosing pneumocytoma cohort data

First, we analyzed the publicly available next-generation sequencing dataset for sclerosing pneumocytoma using bioinformatics tools for detecting internal tandem duplication to discover potentially unidentified molecular changes. Briefly, the raw sequence data from a total of 44 paired normal and sclerosing pneumocytoma tumor samples in the previous whole-exome sequencing study by Jung et al. (Project ID: PRJNA297066) [11] were download from the Sequence Read Archive database (https://www.ncbi.nlm.nih.gov/sra). The downloaded SRA files were converted to fastq files by fastq-dump in the NCBI SRA Toolkit. The sequencing reads were then aligned to the human reference genome (UCSC hg19) by Burrows–Wheeler Alignment tool to generate the aligned BAM files [16]. Genomon-ITDetector, a software package for detection of internal tandem duplications in cancer genome sequencing data, was then used to detect internal tandem duplications in the aligned BAM files [15]. Manual reviews and confirmation of the internal tandem duplications were done with Integrative Genomics Viewer [17]. The data of AKT1 and β-catenin point mutations were retrieved from the manuscript by Jung et al. [11].

VGH-TPE sclerosing pneumocytoma cohort samples

We also included an independent cohort of patients with sclerosing pneumocytoma who underwent surgical resection in Taipei Veterans General Hospital between 2007 and 2017 (VGH-TPE cohort). The samples from this cohort consisted of 40 sclerosing pneumocytoma tumor samples, 25 of which had paired normal lung samples. All the tissues had been fixed in formalin before being embedded in paraffin blocks. The hematoxylin and eosin-stained slides were reviewed by a pathologist (Y.C.Y). The percentages of histological growth patterns, including hemorrhagic, solid, papillary, and sclerotic, were semi-quantitatively recorded with 5% increments for each tumor. The predominant histological pattern was then defined as the pattern with the highest percentage.

Targeted next-generation sequencing in VGH-TPE sclerosing pneumocytoma cohort

We performed targeted next-generation sequencing to sequence all exons of AKT1 and β-catenin genes in the tumor and paired normal samples in the VGH-TPE sclerosing pneumocytoma cohort. Briefly, sections cut from formalin-fixed paraffin-embedded specimens were used for DNA extraction. The collected DNA was purified with GeneRead DNA FFPE Kit (Qiagen, Inc.), according to the manufacturer’s instructions. The quality and quantity of the DNA were verified using NanoDrop 2000 (Thermo Fisher Scientific, Inc.) and Fragment Analyzer (Agilent formerly Advanced Analytical (AATI), Inc.). A customized DNA panel was designed using the online tool, QIAseq Targeted DNA Custom Panel Builder, covering all coding regions of AKT1 and β-catenin genes. Genomic DNA samples (100 ng) were used for library preparation with multiplex PCR target enrichment panel, according to the manufacturer’s instructions. The purified libraries were pooled and then sequenced with MiSeq instrument (Illumina, Inc.).

For single nucleotide variations and short indels, variant detection was performed by smCounter2 workflow, a software package for detecting somatic variants in QIAseq Targeted DNA Panel [18]. Variant allele frequency was obtained from the variant unique molecular identifier allele fraction data in the reports generated by the smCounter2. Low-frequency variants (with variant allele frequency <5%) were filtered out because they are likely to represent subclonal mutations. Synonymous mutations were also filtered out.

For detection of internal tandem duplications, the process was similar to that used for PRJNA297066 sclerosing pneumocytoma cohort data. The reads were aligned with the hg19 human reference genome using the Burrows–Wheeler Alignment tool. Detection of internal tandem duplications was performed by Genomon-ITDetector. Manual reviews of the detected internal tandem duplications were conducted using the Integrative Genomics Viewer.

Targeted PCR, gel electrophoresis, and Sanger sequencing

Hematoxylin–eosin stained tissue slides were reviewed by the pathologist to select the tumor regions. The selected tumor region was dissected manually from the corresponding areas in consecutive tissue sections and, after deparaffinization, subjected to genomic DNA extraction. Briefly, the dissected tumor tissue was transferred to an eppendorf tube containing proteinase K solution. The tube was then incubated at 56 °C for 16 h, followed by an inactivation step at 95 °C for 10 min. Exons 3 and 4 of the AKT1 gene (NCBI reference sequence: NM_001014431.1) were amplified by PCR in a total volume of 25 μL containing 100 ng of template DNA, 1 X Taq Master Mix Red (Ampliqon III), and 0.5 μM of sense and antisense primers. The primer sequences are listed in Supplementary Table 1. The sense and antisense primers used for amplification of AKT1 exon 3 were AKT1-E3S1 and AKT1-E3A1, and those for exon 4 were AKT-E4S1 and AKT1-E4A1, respectively. The PCR was carried out for 35 cycles of 95 °C for 40 s, 56 °C for 40 s, and 72 °C for 40 s, followed by 5 min extension at 72 °C. The amplicons were electrophoresed on 4% agarose gels. For AKT1 exon 3 E17K mutation analysis, the PCR products were subjected to Sanger sequencing with AKT1-E3A2 primers. For AKT1 exon 4 internal tandem duplication analysis, the PCR products, with two major bands identified on agarose gels, were extracted, purified, and then subjected to TA-cloning using T&ATM Cloning Vector Kit (Yeastern Biotech Co. Ltd., Taiwan), according to manufacturer’s instructions. After transformation, single colonies were picked and subjected to Sanger sequencing with M13 forward primers.

Statistical analysis

The association of clinical and pathological characteristics with molecular alterations was analyzed by Student’s t-test (for continuous variables), or chi-square test/Fisher’s exact test (for categorical variables), when appropriate. Analyses were performed using SPSS software (IBM, Armonk, NY, USA). A P-value <0.05 was considered to be statistically significant.

Results

Discovery of AKT1-internal tandem duplication in PRJNA297066 sclerosing pneumocytoma cohort

We analyzed the whole-exome sequencing raw sequence data from PRJNA297066 sclerosing pneumocytoma cohort using the Genomon-ITDetector software, and then performed manual review with Integrative Genomics Viewer. Surprisingly, a high frequency of recurrent in-frame internal tandem duplications in the AKT1 gene was discovered in 22 of the 44 tumor samples (50%) (Table 1, Fig. 1 and Supplementary Fig. 1). The internal tandem duplications were clustered in exon 4 of AKT1, which encodes the Pleckstrin homology domain of the AKT1 protein. The internal tandem duplications were predicted to result in duplications of 7 to 15 amino acids in this region (Fig. 2a). In a subset of the cases (SH54T, SH67T, SH57T, SH41T, SH58T, SH09T, and SH03T), a novel junctional amino acid was introduced by the internal tandem duplications. In SH03T, there was an additional point mutation in the middle of the duplicated segment, which resulted in a threonine (T) to serine (S) substitution. The AKT1-internal tandem duplications were only detected in the tumor samples, and were absent in all the normal samples. Moreover, we found that AKT1-internal tandem duplications and AKT1 point mutations were mutually exclusive, and were found in 41 of the 44 tumor samples (93%).

Table 1 AKT1 internal tandem duplication and other mutations in AKT1 and β-catenin genes in the PRJNA297066 sclerosing pneumocytoma cohort
Full size table
Fig. 1
figure 1

Representative screenshot showing alignments around the AKT1 exon 4 in a tumor sample (SH58T) positive for AKT1 internal tandem duplication, using the Integrative Genomics Viewer. The regions between the two dotted lines are the parental sequences of duplicated nucleotides. The duplicated nucleotides (arrow) are aligned either to the right or left of the parental sequences, and they are marked as soft-clipped sequences by the alignment software. This is a characteristic feature of internal tandem duplications

Full size image
Fig. 2
figure 2

a Schematic of the AKT1 protein structure and predicted AKT1 protein sequences from the tumor samples positive for internal tandem duplication in the PRJNA297066 cohort. The wild-type AKT1 protein sequence (amino acids 60–85) is shown on the top, with the predicted protein sequence of each internal tandem duplication-positive case shown below. The first segments of the duplicated sequences are shown in green, and the second segments are shown in blue. Novel junctional amino acids (red) were introduced by the internal tandem duplications in cases SH54T, SH67T, SH57T, SH41T, SH58T, SH09T, and SH03T. In SH03T, there was an additional point mutation in the middle of the duplicated segment, which resulted in a threonine (T) to serine (S) substitution. b Schematic of the AKT1 protein structure and predicted AKT1 protein sequences from the tumor samples positive for internal tandem duplications in the VGH-TPE cohort. The wild-type AKT1 protein sequence (amino acids 60–85) is shown on the top, with the predicted protein sequence of each internal tandem duplication-positive case shown below. The first segments of the duplicated sequences are shown in green, and the second segments are shown in blue. Novel junctional amino acids (red) were introduced by the internal tandem duplications in cases V18T, V20T, V28T, V16T, V35T, V37T, V33T, and V04T. In V23T, there was an additional point mutation in the middle of the duplicated segment, which resulted in a proline (P) to alanine (A) substitution. PH Pleckstrin homology domain, Kinase Kinase domain N N-terminal, C C-terminal

Full size image

Validation of AKT1-internal tandem duplications in the VGH-TPE sclerosing pneumocytoma cohort

To validate our findings in the PRJNA297066 sclerosing pneumocytoma cohort, we performed targeted next-generation sequencing to sequence all the exons of AKT1 and β-catenin genes in an independent cohort (VGH-TPE cohort). This cohort consisted of 40 patients with sclerosing pneumocytoma who underwent surgery in our hospital. Similar to the findings in the PRJNA297066 cohort, we detected recurrent in-frame internal tandem duplications in the AKT1 gene in 20 of the 40 tumor samples in the VGH-TPE cohort (50%) (Table 2 and Supplementary Fig. 2). The internal tandem duplications clustered in exactly the same region of the AKT1 protein Pleckstrin homology domain as in the PRJNA297066 cohort and were predicted to result in duplications of 7 to 16 amino acids in this region (Fig. 2b). In a subset of the cases (V18T, V20T, V28T, V16T, V35T, V37T, V33T, and V04T), a novel junctional amino acid was introduced by the internal tandem duplications, which was also very similar to that observed in the PRJNA297066 cohort. In addition, in the case V23T, there was a point mutation in the middle of the duplicated segment, which resulted in a proline (P) to alanine (A) substitution. This finding was similar to that obtained for the case SH03T in the PRJNA297066 cohort.

Table 2 AKT1 internal tandem duplications and other mutations in AKT1 and β-catenin genes in the VGH-TPE sclerosing pneumocytoma cohort
Full size table

The analysis of AKT1 exon 4 using genomic PCR and electrophoresis on a 4% agarose gel revealed that cases carrying AKT1-internal tandem duplication yielded a higher molecular weight product (Fig. 3a). Sanger sequencing of the distinct PCR products showed the presence of in-frame tandem duplications and the duplicated sequences were identical to those derived from targeted next-generation sequencing (Fig. 3b).

Fig. 3
figure 3

a Targeted PCR and gel electrophoresis of AKT1 exon 4 from five representative samples (V01T, V39T, V40T, V33T, and V23T) in the VGH-TPE cohort. In samples positive for AKT1, internal tandem duplication (V33T and V23T), large PCR products corresponding to the internal tandem duplications were detected (arrow), in addition to the expected wild-type products. In samples V01T and V40T, which are negative for AKT1 internal tandem duplication, only wild-type products were detected. In sample V39T, an additional PCR product, slightly larger than the wild-type product, was also detected (arrowhead), which corresponds to the complex AKT1 mutation of p.Q79_W80delinsR and p.T81delinsMPGA in this case. b Sanger sequencing from a representative sample (V23T) positive for AKT1 internal tandem duplication showed the duplicated sequence. N normal lung, WT wild type

Full size image

In addition to AKT1-internal tandem duplication, we also detected other forms of AKT1 mutations in the sclerosing pneumocytoma tumor samples (Table 2), including p.E17K (six cases) (Fig. 4), p.E17K/p.E40K (one case), p.E17K/p.I19F (one case), p.E17K/p.W80G (one case), p.E49K/p.N53H (one case), p.Q79K/p.W80G (six cases), p.Q79K/p.W80R (two cases), pW80R/p.L78_Q79insKK (one case), and p.Q79_W80delinsR/p.T81delinsMPGA (one case). Most importantly, all these AKT1 mutations were mutually exclusive with AKT1-internal tandem duplications. Taking all forms of AKT1 mutations together (including point mutations, indels, and internal tandem duplications), all sclerosing pneumocytoma tumor samples in the VGH-TPE cohort contained one of the AKT1 mutations.

Fig. 4
figure 4

a Screenshot for a representative sample (V01T) showing AKT1 p.E17K (c.49G>A) mutation, using Integrative Genomics Viewer. The reference sequence shown in Integrative Genomics Viewer is the antisense strand of AKT1, and the corresponding nucleotide change for p.E17K mutation is C>T mutation (arrow). b Sanger sequencing from the same sample confirmed the AKT1 p.E17K (c.49G>A) mutation (arrow)

Full size image

With regard to β-catenin genes, we detected p.S45P point mutation in one single case (V09T). However, this case also harbored the AKT1 p.E17K mutation. The variant allele frequency of AKT1 p.E17K mutation was 36%. In contrast, the variant allele frequency of β-catenin p.S45P mutation was only 9%. The low variant allele frequency of β-catenin p.S45P mutation suggested that this mutation was present in a subclone of tumor cells, and was likely to be a secondary event during tumor progression.

No AKT1 or β-catenin mutations were detected in normal samples of the VGH-TPE cohort. The pie charts of AKT1 and β-catenin mutations in the PRJNA297066 cohort and VGH-TPE cohort are shown in Fig. 5.

Fig. 5
figure 5

Pie charts of AKT1 and β-catenin mutations in the PRJNA297066 and VGH-TPE cohorts

Full size image

Correlation between clinicopathological characteristics and AKT1 mutations in the VGH-TPE sclerosing pneumocytoma cohort

We analyzed the correlation between clinicopathological characteristics and AKT1 mutations in the VGH-TPE cohort, and the results are summarized in Table 3. There were no significant differences in clinical characteristics or pathological features between patients with AKT1-internal tandem duplication or other types of AKT1 mutations (Fig. 6). We noticed that all the male patients in the VGH-TPE cohort belonged to the group with other types of AKT1 mutations. However, this was probably due to chance because we observed that two male patients in the PRJNA297066 cohort belonged to the group with AKT1-internal tandem duplication (SH53T and SH02T).

Table 3 Characteristics of patients with AKT1-internal tandem duplications or other forms of AKT1 mutations in the VGH-TPE cohort
Full size table
Fig. 6
figure 6

Representative histological features of cases with AKT1 point mutations or short indels (a, c, e, g), and cases with AKT1 internal tandem duplication (b, d, f, h). a A case (V01T) with AKT1 p.E17K mutation showing papillary pattern. b A case (V26T) with AKT1-internal tandem duplication showing papillary pattern. c A case (V07T) with AKT1 p.E17K mutation showing hemorrhagic pattern. d A case (V19T) with AKT1-internal tandem duplication showing hemorrhagic pattern. e A case (V12T) with AKT1 p.Q79K; p.W80R mutation showing solid pattern. f A case (V35T) with AKT1-internal tandem duplication showing solid pattern. g A case (V32T) with AKT1 p.W80R; p.L78_Q79insKK mutation showing sclerotic pattern. h A case (V18T) with AKT1-internal tandem duplication showing sclerotic pattern (original magnification:×200)

Full size image

Discussion

In this study, we demonstrate that AKT1 mutation is present in nearly all the sclerosing pneumocytomas, with internal tandem duplication and point mutations being the most common forms. AKT1 belongs to the AKT kinase family, the members of which serve as critical downstream effectors of phosphatidylinositol-3 kinase (PI3K) signaling. Under basal conditions, AKT1 localizes to the cytosol, and maintains an inactive closed “PH-in” conformation through the interaction between its Pleckstrin homology and kinase domains [19]. Following activation of phosphatidylinositol-3 kinase by proliferative signals, the inactive AKT1 in the cytosol is recruited to the plasma membrane through binding of the Pleckstrin homology domain. This results in a conformational change that separates the Pleckstrin homology and kinase domains, shifting AKT1 to an open “PH-out” state. The PH-out state enables phosphorylation of Thr308 and Ser473 residues of AKT1, which leads to maximal activation of the kinase [20].

The PI3K–AKT pathway is one of the most commonly dysregulated pathways in cancer. Gain-of-function mutations in AKT1 have been identified in a broad range of tumor types, with AKT1 p.E17K being the most frequent hotspot [21]. The pathogenic role of AKT1 p.E17K mutation has been well studied. Structural studies have shown that AKT1 p.E17K mutation directly alters the electrostatic interactions of the lipid-binding pocket of the Pleckstrin homology domain and activates AKT1 through pathological localization to the plasma membrane. This mechanism stimulates downstream signaling and leads to oncogenic transformation [22].

In contrast, little is known regarding the role of AKT1 internal tandem duplications in oncogenesis. In the literature, there are very few reports of AKT1 internal tandem duplications in human cancers. To the best of our knowledge, only juvenile granulosa cell tumor and breast cancer have been reported to harbor AKT1 internal tandem duplications [23, 24]. The affected genomic region in juvenile granulosa cell tumor and breast cancer is exactly the same as observed in sclerosing pneumocytoma. In juvenile granulosa cell tumor, the AKT1 internal tandem duplication was detected in 10 out of 16 case (63%). In addition, there was also a high frequency of recurrent AKT1 point mutations (10 out of 16 cases, 63%) [23]. However, unlike in sclerosing pneumocytoma, the AKT1 internal tandem duplication and point mutations in juvenile granulosa cell tumor do not appear to be mutually exclusive. Another interesting finding is that the sites of AKT1 point mutations in sclerosing pneumocytoma are restricted to the Pleckstrin homology domain, whereas in juvenile granulosa cell tumor, many AKT1 point mutations occur outside the Pleckstrin homology domain. In breast cancer, the existence of AKT1 internal tandem duplication was identified very recently in a large-scale study by Chang et al. [24]. In this study, AKT1 point mutations (p.E17K) were detected in 80 out of 2561 cases (3%), whereas AKT1 internal tandem duplications were only detected in 2 out of 2559 cases (0.08%), which is very intriguing. In sclerosing pneumocytoma and juvenile granulosa cell tumor, the frequencies of AKT1 internal tandem duplications and point mutations were nearly equal. However, there appears to be a substantial difference between the frequency of AKT1 point mutations and AKT1 internal tandem duplications in breast cancer.

The pathogenic mechanisms of AKT1 internal tandem duplications are not well understood. It is noticeable that the insertion points of duplicated sequences in AKT1 internal tandem duplications clustered at amino acids 76–85 of AKT1 protein, which contains the region of Pleckstrin homology domain and kinase domain interaction interface (Fig. 7) [19]. It has been shown that disruption of Pleckstrin homology domain and kinase domain interactions by point mutations at the Pleckstrin homology domain and kinase domain interaction interface leads to oncogenic activation of AKT in human cancers [19]. It seems plausible that the AKT1 internal tandem duplications occurring at this region might also disturb Pleckstrin homology domain and kinase domain interactions, and lead to oncogenic activation of AKT. To verify this speculation, further structural studies are required. However, in studies on cell lines, it has been shown that HeLa cells and human mammary gland cells (MCF10A) transfected with AKT1 internal tandem duplication mutation constructs display high level of AKT1 phosphorylation and activation [23, 24]. Furthermore, cells transfected with AKT1 internal tandem duplication mutation constructs demonstrated sensitivity to the AKT kinase inhibitor, AZD5363, indicating the therapeutic implications of AKT1 internal tandem duplication [24, 25].

Fig. 7
figure 7

Ribbon representation of Pleckstrin homology domain (green) and kinase domain (blue) of AKT1 protein (Protein Data Bank ID code 3O96). The insertion points of duplicated sequences in AKT1 internal tandem duplication clustered at amino acids 76–85 (red), which involve the region of the interaction interface between the Pleckstrin homology domain and kinase domain

Full size image

In conclusion, our results show that AKT1 mutations, including internal tandem duplication, point mutations, and short indels, are present in nearly all sclerosing pneumocytomas. These mutations are mutually exclusive of each other. All these mutations occur in the Pleckstrin homology domain, a critical component in the activation of the AKT1 protein. Our results suggest that AKT1 mutation is the genetic hallmark of sclerosing pneumocytoma.